0618611045

T FOR
Applied Calculus
FOR SALEFOR
SALEFOR
Frank C. Wilson
Chandler-Gilbert Community College
FOR INSTRUCT
STRUCTOR
TOR
Scott Adamson
Chandler-Gilbert Community College
HOUGHTON MIFFLIN HARCOURT PUBLISHING COMPANY
Boston New York
Publisher: Richard Stratton Senior Sponsoring Editor: Molly Taylor Senior Marketing Manager: Jennifer Jones Development Editor: Peter Galuardi Senior Project Editor: Tamela Ambush Senior Content Manager: Maren Kunert Art and Design Manager: Jill Haber Cover Design Manager: Anne S. Katzeff Senior Photo Editor: Jennifer Meyer Dare Composition Buyer: Chuck Dutton New Title Project Manager: James Lonergan Editorial Associate: Andrew Lipsett Editorial Assistant: Joanna Carter-OConnell Cover photographs: Sam Stockwell/Getty Images; Bob Krist/CORBIS; Karl Petzke/FoodPix/Jupiter Images; J. Garcia/photocuisine/CORBIS Photo Credits: p. 1, Lou Dematteis/Reuther/CORBIS; p. 47, Richard Cummins/ CORBIS; p. 61, Frederick J. Brown/Getty Images; p. 115, Reuters/CORBIS; p. 176, Alex Fezer/CORBIS; p. 188, Tony Duffy/Getty Images; p. 256, Royalty Free/ CORBIS; p. 275, Taxi/Getty Images; p. 306, Ariel Skelley/CORBIS; p. 377, Stone/Getty Images; p. 392, Karen Kasmauski/CORBIS; p. 411, Reuters/CORBIS; p. 471, Peter Beck/CORBIS; p. 515, Royalty Free/CORBIS; p. 582, Reuters/ CORBIS; p. 636, Zen Sekizawa/Getty Images; p. 679, Arthur Morris/CORBIS; p. 686 (top left), Lawson Wood/CORBIS; p. 686 (top right), Jeffrey L. Rotman/ CORBIS; p. 686 (bottom right), Stuart Westmoreland/CORBIS; p. 691, Richard Hamilton Smith/CORBIS; p. 712, David Fleetham/CORBIS; p. 717, Associated Press/H. Rumph, Jr.; p. 779, 2008 Getty Images; p. 802, Paul Hawthorne/Getty Images Copyright 2009 by Houghton Mifflin Harcourt Publishing Company. All rights reserved. No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or by any information storage or retrieval system without the prior written permission of Houghton Mifflin Harcourt Publishing Company unless such copying is expressly permitted by federal copyright law. Address inquiries to College Permissions, Houghton Mifflin Harcourt Publishing Company, 222 Berkeley Street, Boston, MA 02116-3764. Printed in the U.S.A. Library of Congress Control Number: 2008925563 Instructors exam copy: ISBN 13: 978-0-547-00488-4 ISBN 10: 0-547-00488-5 For orders, use student text ISBNs: ISBN 13: 978-0-618-61104-1 ISBN 10: 0-618-61104-5 1 2 3 4 5 6 7 8 9 CRK 12 11 10 09 08
Contents
Preface vii Features x
1 Functions and Linear Models

1.1 Functions 2 1.2 Linear Functions 18 1.3 Linear Models 33
Chapter 1 Study Sheet 57 Chapter 1 Review Exercises 58 Make It Real Project 1 60
2 Nonlinear Models
2.1 2.2 2.3 2.4 2.5
61
Quadratic Function Models 62 Higher-Order Polynomial Function Models 93 Exponential Function Models 115 Logarithmic Function Models 135 Choosing a Mathematical Model 150 Chapter 2 Study Sheet 166 Chapter 2 Review Exercises 168 Make It Real Project 2 175 176
3 The Derivative
3.1 3.2 3.3 3.4 3.5 3.6
Average Rate of Change 177 Limits and Instantaneous Rates of Change 188 Limits and Continuity 205 The Derivative as a Slope: Graphical Methods 217 The Derivative as a Function: Algebraic Method 233 Interpreting the Derivative 242 Chapter 3 Study Sheet 249 Chapter 3 Review Exercises 251 Make It Real Project 3 255 256
4 Differentiation Techniques
4.1 4.2 4.3 4.4 4.5
Basic Derivative Rules 257 The Product and Quotient Rules 266 The Chain Rule 275 Exponential and Logarithmic Rules 285 Implicit Differentiation 293 iii
iv
Contents
5 Derivative Applications
5.1 5.2 5.3 5.4
306
Maxima and Minima 307 Applications of Maxima and Minima 327 Concavity and the Second Derivative 352 Related Rates 377 Chapter 5 Study Sheet 386 Chapter 5 Review Exercises 387 Make It Real Project 5 391 392
6 The Integral
6.1 6.2 6.3 6.4 6.5
Indefinite Integrals 393 Integration by Substitution 402 Using Sums to Approximate Area 411 The Definite Integral 433 The Fundamental Theorem of Calculus 451 Chapter 6 Study Sheet 463 Chapter 6 Review Exercises 466 Make It Real Project 6 470
7 Advanced Integration Techniques and Applications 471

7.1 Integration by Parts 472 7.2 Area Between Two Curves 7.3 Improper Integrals 498
481
8 Multivariable Functions and Partial Derivatives 515

8.1 8.2 8.3 8.4
Multivariable Functions 516 Partial Derivatives 528 Multivariable Maxima and Minima 545 Constrained Maxima and Minima and Applications 561 Chapter 8 Study Sheet 576 Chapter 8 Review Exercises 578 Make It Real Project 8 581
Contents
9 Trigonometric Functions
582 583
9.1 Trigonometric Functions, Equations, and Graphs 9.2 Derivatives of Trigonometric Functions 604 9.3 Integrals of Trigonometric Functions 618
10 Differential Equations
10.1 10.2 10.3 10.4 10.5
636
Slope Fields 637 Eulers Method 649 Separable Differential Equations and Applications 662 Differential Equations: Limited Growth and Logistic Models 673 First-Order Linear Differential Equations 691 Chapter 10 Study Sheet 707 Chapter 10 Review Exercises 709 Make It Real Project 10 715 717
11 Sequences and Series

11.1 11.2 11.3 11.4
Sequences 718 Series and Convergence 733 Taylor Polynomials 746 Taylor Series 763 Chapter 11 Study Sheet 774 Chapter 11 Review Exercises 776 Make It Real Project 11 778 779
12 Probability and Calculus

12.1 12.2 12.3 12.4
Continuous Probability Models 780 Uniform and Exponential Distributions 796 Expected Value, Variance, and Standard Deviation 804 Normal Distributions 820 Chapter 12 Study Sheet 834 Chapter 12 Review Exercises 836 Make It Real Project 12 840
Answers to Odd-Numbered Exercises Index I1
A1
This page intentionally left blank
Preface
To the Student
Have you ever asked, When am I ever going to use this? or Why should I care? after learning a new procedure or concept in math? Youre not alone many students have. This book seeks to answer those questions by teaching math concepts and skills in context and applying them to realistic situations you may encounter. Whether calculating how long it will take to pay off a car loan, or predicting what tuition will cost over the next few years, examples and exercises are based on interesting and engaging real-life data. Consequently, when you find a solution to a problem, you are learning something about the world around you. To make real-life data analysis even more meaningful, Make It Real projects are included. These projects help you become skilled in collecting and analyzing data from your own life. The skills learned through these projects will remain with you long after you have left this course. This book is written in a reader-friendly style. Although important terms and concepts are appropriately addressed, the focus is conceptual understanding, not mathematical jargon. If you are pursuing a degree in business, social science, or a similar field, this book is written specifically for you. We encourage you to read the book to deepen your understanding of the concepts studied in the classroom. The Just In Time algebra reviews and Algebra Flashback exercises are designed for those of you whose algebra skills are a bit rusty. These timely reviews will help refresh your memory and allow you to practice the algebra skills you will need to master the calculus concepts. You may find that your understanding of concepts and your ability to analyze data are enhanced through the use of technology. The graphing calculator (especially the TI83 Plus or TI84 Plus) is ideal for investigating graphs and estimating numerical answers. Spreadsheet programs (especially Microsofts Excel) are ideal for data analysis. Because learning how to use a calculator or Excel can be a challenge, Technology Tips are integrated throughout the book. These tips detail how to graph a function, solve an equation, find the maximum value of a function, and so on. Rather than giving a broad overview of a procedure, the tips take you through the actual keystrokes and mouse clicks while showing you screenshots so you can verify youre doing each step correctly. Our students help us become better teachers. We deeply value their input. Likewise, were interested in hearing from you. Let us know how this book works for you and feel free to share any feedback on how to improve this book. Contact us on our personal website: www.makeitreallearning.com. Enjoy!
To the Instructor
Thank you for your interest in this book. We believe you will find its approach refreshing and its content interesting to you and your students. It is written specifically for students pursuing degrees in business, social science, or a related field. As you know, many of these students do not enjoy mathematics and are taking this vii
viii
Preface
course only because it is a requirement. We hope to make a positive impact on these students attitudes toward mathematics. Several features are included in the book to make the course content more accessible to students, including:
An informal writing style emphasizes conceptual understanding without becoming bogged down in mathematical jargon. Examples and exercises throughout the text are based on interesting and engaging real-life data. Over 600 real-life applications, featuring over 70 businesses, products, and associations, help to make math real for your students. Make It Real projects allow students to collect and analyze data relevant to their personal lives. Detailed Technology Tips teach students how to use a graphing calculator and Excel as tools to analyze real-life data. Just In Time algebra reviews review important algebra content, refreshing students memories. Algebra Flashback exercises provide students with the opportunity to practice the algebra skills they will need in order to master the calculus concepts. Show You Know exercises focus on concepts rather than skills. These provide students a chance to reflect on the material and think about what is really going on in each section.
Candid feedback from our colleagues helps us become better teachers. Please feel free to contact us with any recommendations, comments, or other feedback you feel will enhance the effectiveness of this book. You can contact us on our personal website: www.makeitreallearning.com.
Disclaimer In this book, we have attempted to incorporate real-world data from the financial markets to the medical field. In each case, we have done our best to present the data accurately and interpret the data realistically. However, we do not claim to be experts in financial, medical, and other similar fields. Our interpretations of realworld data and associated conclusions may not adequately consider all relevant factors. Therefore, readers are encouraged to seek professional advice from experts in the appropriate fields before making decisions related to the topics addressed herein. Despite the usefulness of mathematical models as representations of real-world data sets, most mathematical models have a certain level of error. It is common for model results to differ from raw data set values. Consequently, conclusions drawn from a mathematical model may differ (sometimes dramatically) from conclusions drawn by looking at raw data sets. Readers are encouraged to interpret model results with this understanding. Acknowledgements This textbook would not have been possible without the contributions of many colleagues. We greatly appreciate all of the people who contributed time and talent to bring this book to fruition. The feedback from the following reviewers was invaluable and helped to shape the final form of the text: Brenda Alberico, College of DuPage; Lewis
Preface
ix
Blake, Duke University; Jack Bookman, Duke University; Jennifer Fowler, University of TennesseeKnoxville; R. Baker Kearfott, University of Southwestern Louisiana; Roger Lee, Salt Lake Community College; Philip E. Miller, Indiana University Southeast; Jean Peterson, University of WisconsinOshkosh; Brian Rodas, Santa Monica University; Annie Stith-Willis, Virginia Commonwealth University; Denise Szecsei, Stetson University; Nader Vakil, Western Illinois University; and Terry Walters, University of TennesseeChattanooga. Parts of this book were derived from another book in our series, Brief Applied Calculus. We would like to thank the reviews who helped shape that text. They are: Bill Ardis, Collin County Community College; James J. Ball, Indiana State University; Michael L. Berry, West Virginia Wesleyan College; Marcelle Bessman, Jacksonville University; Mike Bosch, Iowa Lakes Community College; Emily Bronstein, Prince Georges Community College; Dean S. Burbank, Gulf Coast Community College; Andra Buxkemper, Bunn College; Roxanne Byrne, University of ColoradoDenver; Scott A. Clary, Florida Institute of Technology; David Collingwood, University of Washington; Mark A. Crawford Jr., Western Michigan University; Khaled Dib, University of Minnesota Duluth; Lance D. Drager, Texas Tech University; Klara Grodzinsky, Georgia Institute of Technology; Lucy L. Hanks, Virginia Polytechnic Institute and State University; Jean B. Harper, State University of N.Y.College at Fredonia; Kevin M. Jenerette, Coastal Carolina University; Cynthia Kaus, Metropolitan State University; Michael LaValle, Rochester Community and Technical College; Roger D. Lee, Salt Lake Community College; Lia Liu, University of Illinois Chicago; Alan Mabry, University of Texas at El Paso; Quincy Magby, Arizona Western College; Mary M. Marco, Bucks County Community College; Nicholas Martin, Shepherd College; William C. McClure, Orange Coast College; James McGlothin, Lower Columbia College; Victoria Neagoe, Goldey Beacom College; David W. Nelson, Green River Community College; Ralph W. Oberste-Vorth, Marshall University; Armando I. Perez, Laredo Community College; Cyril Petras, Lord Fairfax Community College; Mihaela Poplicher, University of Cincinnati; John E. Porter, Murray State University; David W. Roach, Murray State University; R. A. Rock, Daniel Webster College; Arthur Rosenthal, Salem State College; Kimmo I. Rosenthal, Union College; Sharon Mayhew Saxton, Cascadia Community College; Edwin Shapiro, University of San Francisco; Denise Szecsei, Stetson University; Abolhassan S. Taghavy, Richard J. Daley College; Muhammad Usman, University of Cincinnati; Jorge R. Viramontes Olivas, University of Texas at El Paso; Beverly Vredevelt, Spokane Falls Community College; Michael L. Wright, Cossatot Community College. We thank Peter Galuardi and Molly Taylor, editors at Houghton Mifflin, for their support in the development of the text. Their frank recommendations were invaluable in enhancing the quality of the final product. On a personal note, we could not have written this text without the tireless support of our families. We express deep appreciation to our wives and children for their love, patience, and support as we took on this challenging endeavor.
Dedicated to: My parents, Blaine and Joyce Wilson, who believed in me when no one else did. FCW
My wife, Kim, who supported, encouraged, and prayed for me. SLA
Features
Real Applications! Real Data! Real Learning!
Chapter Opener
Chapter
The Derivative
t is impossible to determine how quickly a person is running from a single photograph, since speed is calculated as a change in distance over a change in time. Nevertheless, we may estimate a persons speed at a particular instant in time by determining the distance traveled over a small interval of time (e.g., one second). A runners speed may be classified as a rate of change in distance. A key component of calculus is the study of rates of change.
Each chapter opens with a quick introduction to a key concept presented within a real-life context, accompanied by a related photo. Also, a detailed list of objectives provides a clear picture of the concepts and skills that will be developed in the section.
3.1
Average Rate of Change

I
Calculate the average rate of change of a function over an interval
3.2
Limits and Instantaneous Rates of Change

I I
Estimate the instantaneous rate of change of a function at a point Use derivative notation and terminology to describe instantaneous rates of change Perform mathematical operations with limits Determine if a function is continuous
3.3
Limits and Continuity

I I
3.4
The Derivative as a Slope: Graphical Methods

I I
Find the equation of the tangent line of a curve at a given point Numerically approximate derivatives from a table of data Graphically interpret average and instantaneous rates of change Use derivative notation and terminology to describe instantaneous rates of change
3.5
The Derivative as a Function: Algebraic Method

I
Use the limit definition of the derivative to find the derivative of a function Use derivative notation and terminology to describe instantaneous rates of change Interpret the meaning of the derivative in the context of a word problem
3.1 Average Rate of Change
177
3.6 176
Interpreting the Derivative

I

I
Colleges and universities periodically raise their tuition rates in order to cover rising staffing and facilities costs. As a result, it is often difficult for students to know how much money they should save to cover future tuition costs. By calculating the average rate of change in the tuition price over a period of years, we can estimate projected increases in tuition costs. In this section, we will demonstrate how to calculate the average rate of change in the value of a function over a specified interval [a, b]. (The interval notation [a, b] is equivalent to a # x # b.)
Getting Started
Each section opens with Getting Started, real-life applications or mathematical scenarios, which shows the relevance of the section content to a students everyday life. Section objectives are also provided.
THE DIFFERENCE QUOTIENT: AN AVERAGE RATE OF CHANGE
The average rate of change of a function y 5 f ( x ) over an interval 3a, b4 is f (b) 2 f (a) b2a This expression is referred to as the difference quotient. For a linear function, the difference quotient gives the slope of the line.
In calculating the difference quotient, we answer the question, Over the interval 3a, b4 , on average, how much does a one-unit increase in the x value change the y value of the function?
EXAMPLE
Calculating an Average Rate of Change

The quarterly cost of tuition for full-time resident students at Green River Community College is shown in Table 3.1.
TABLE 3.1
Years (since 19941995) (t) 0 1 2 3 4 5 6 7
Quarterly Tuition Cost (dollars) 3 f (t) 4 432 450 467 486 505 528 547 581
Change in Tuition Cost from Prior Year (dollars)
18 17 19 19 23 19 34
Source: Green River Community College.
Features
xi
Technology Tips
The understanding of core concepts can be enhanced through the use of a graphing calculator. Technology Tips, incorporated throughout the text, guide students through new techniques on the calculator such as graphing a function, solving an equation, and finding the value of a function. All tips show actual calculator keystrokes, often with multiple, sequential screenshots, teaching students how to use graphing calculators as a tool to analyze real-life data. When the use of graphing technology is advised, a graphing calculator icon will appear .
A similar Technology Tip for Microsoft Excel is given at the end of the section.
11.1 Sequences
727
Using Technology to Graph a Sequence

The TI83 Plus calculator works nicely in creating a graph of a sequence.
Graphing a Sequence 1. Press MODE . Change the mode to Sequence mode by using the arrows to highlight SEQ and pressing ENTER .
2. Edit the parameters using the Y Editor. The parameter nMin refers to the minimum term number you choose to display. The parameter u(n) represents the formula. The u is accessed by pressing 2nd and the number 7 (note the blue u above the 7 button). The parameter n is accessed by pressing the X,T,U,n key. The parameter u(nMin) represents the first term of the sequence. 3. Edit the WINDOW parameters. The parameters nMin and nMax refer to the number terms in the sequence to be evaluated. PlotStart and PlotStep determine which term to plot first and the incremental value of n. The rest of the parameters are as usual, where x represents the term number and y represents the value in the sequence.
4. Press
GRAPH
(Continued)
Graphing a Sequence 1. In column A, write a list of whole numbers from 1 through the number of terms you wish to compute. In cell B2, write the initial term of the sequence. In this case, u( 1 ) 5 20.
Excel Technology Tips

Upon entering the working world, many students will use Excel or some other spreadsheet program as their primary tool for generating graphs, analyzing data, and generating models. With that in mind, Technology Tips focusing exclusively on Excel use are included. Many tips show actual keystrokes, often with multiple, sequential screenshots, teaching students how to use Excel as a tool to analyze real-life data.
2. In cell B3, type 0.305B2+20. This is equivalent to stating that u( n 1 1 ) 5 0.305u( n ) 1 20.
(Continued)
xii
Features
EXAMPLE
Analyzing Shuttle Bus Wait Times

Shuttle buses arrive at a hotel every 15 minutes to take visitors to Disneyland. Compute and interpret the expected value for this situation. Then calculate the variance and standard deviation.
SOLUTION Since the wait time is a uniform continuous random variable on the interval 30, 154 , we will use the uniform probability density function
Examples
Real-life applications, examples, and data help engage studentseven those who have never enjoyed mathematics. Cited sourcesespecially those of interest to business and social science studentshelp to make it real.
P( x ) 5 15 2 0 5 15 to compute the expected value. 1 E( x ) 5 ( 0 1 15 ) 2 5 15 2
5 7.5 minutes On average, a visitor would expect to wait 7.5 minutes for the shuttle bus. Next we compute the variance. Var( x ) 5 ( 15 2 0 ) 2 12
152 5 12 5 225 12
Units reminder
Accompanying many examples is a ruler icon reminding students to pay close attention to unit analysis.
According to the model, 5.395 years after the end of fiscal year 1998, the annual revenue is expected to reach $3 billion. To add meaning to the result, we convert 0.395 years into months. 0.395 years ? 12 months < 5 months 1 year
During the one-year period prior to the end of the fifth month of fiscal year 2004, we anticipate that $3 billion in revenues will be earned. (Note: Since t 5 5 is the end of fiscal year 2003, t 5 5.395 is in fiscal year 2004.)
3.6 Interpreting the Derivative
247
Exercises
Over 500 real-life applications featuring over 70 businesses, products, and associations help to make math real for your students! Using realworld data from real companies such as Starbucks and Wal-Mart, and interesting topics such as debit cards and student loans, the exercises bring a current and immediate motivation for learning mathematical concepts. A globe icon indicates that the exercise
million dollars, where x is the net number of CDs shipped in millions. (Net means after returns.) Interpret the meaning of V( 900 ) 5 12,261.9 and Vr( 900 ) 5 17.9. 14. Based on data from 1998 to 2001, the value of DVD videos shipped by recording industry manufacturers may be modeled by V( x ) 5 2 0.2173x2 1 25.84x 2 0.02345 million dollars, where x is the net number of DVDs shipped in millions. (Net means after returns.) (Source: Modeled from Recording Industry
DVDs
19.
(Source: Modeled from Recording Industry Association of America data.)
Based on data from 1991 to 2000, the number of homicides resulting from an alcohol-related brawl may be modeled by H( t ) 5 2 0.3551t 3 1 9.867t 2 2 104.7t 1 600.4
Deadly Alcohol-Related Brawls
(Source: Modeled from Crime in the United States 2000, Uniform Crime Report, FBI.)
deaths, where t is the number of years since 1990.
Interpret the meaning of H( 10 ) 5 185 and Hr( 10 ) 5 2 14 . 20.

Body Height The average height of a girl between the ages of 2 and 13 years may be modeled by H( a ) 5 2 0.0392a2 1 2.987a 1 29.69 inches, where a is the age of the girl. (Source:
Association of America data.)
Interpret the meaning of V( 7.5 ) 5 181.6 and V r( 7.5 ) 5 22.6 . 15. Based on data from 1904 to 1992, the number of women in the Summer Olympics may be modeled by W( t ) 5 0.0002344t 4 2 0.03440t 3 1 1.701t 2 2 19.69t 1 70.80
Summer Olympics
Modeled from www.babybag.com data.)
Interpret the meaning of H( 5 ) 5 43.6 and Hr( 5 ) 5 2.6 . 21.

Company Revenue Based on data from 1999 to 2001, the gross revenue from sales of Johnson & Johnson and its subsidiaries may be modeled by R( t ) 5 665t 2 1 1150t 1 27,357
(Source: Modeled from www.olympicwomen.co.uk data.)
women, where t is the number of years since 1900.
includes real-life data. A globe icon with an M indicates that the exercise includes a model based on real-life data.
Interpret the meaning of W( 92 ) 5 2662 and Wr( 92 ) 5 150. 16.

Summer Olympics
million dollars, where t is the number of years since the end of 1999. (Source: Modeled
from Johnson & Johnson 2001 Annual Report.)
Refer to the results of Exercise 15. Do you think W( 96 ) will be greater than 2662 and W r( 96 ) will exceed 150? Defend your conclusions. Based on data from 1790 to 2000, the population of the United States may be modeled by P( t ) 5 0.006702t 2 2 24.11t 1 21,696
U.S. Population
Interpret the meaning of R( 2 ) 5 32,317 and Rr( 2 ) 5 3810 . 22.

Company Costs Based on data from 1999 to 2001, the cost of goods sold for Johnson & Johnson and its subsidiaries may be modeled by C( t ) 5 103t 2 1 315t 1 8539
17.
(Source: Modeled from U.S. Bureau of the Census data.)
million people, where t is the calendar year.
since the end of 1999. (Source: Modeled from Johnson & Johnson 2001 Annual Report.)
million dollars, where t is the number of years Interpret the meaning of C( 2 ) 5 9581 and Cr( 2 ) 5 727.
Interpret the meaning of P( 2000 ) 5 284 and Pr( 2000 ) 5 2.70 . 18.
Homicide Rate
Based on data from 1990 to 2000, the homicide rate (deaths per 100,000 people) in the United States may be modeled by H( t ) 5 0.01129t 3 2 0.2002t 2 1 0.4826t 1 9.390 people, where t is the number of years since 1990. Interpret the meaning of H( 10 ) 5 5.5 and Hr( 10 ) 5 2 0.1 .
23.
Company Profit Based on data from 1999 to 2001, the operating profit of Frito-Lay North America may be modeled by P( t ) 5 2 47.5t 2 1 283.5t 1 1679
since the end of 1999. (Source: Modeled from PepsiCo 2001 Annual Report.)
million dollars, where t is the number of years
Interpret the meaning of P( 2 ) 5 2056 and Pr( 2 ) 5 93.5 .
Features
xiii
Show You Know Conceptual Exercises

Show You Know exercises are focused on a conceptual understanding of the objectives in each section. With a focus on critical thinking, these exercises ask students to think abstractly about a problem and explain their results.
41. Explain why it makes sense to use rectangles to approximate the area between the graph of a function and the horizontal axis. 42. Graphically speaking, what is the difference between a left-hand sum and a right-hand sum? 43. Why does increasing the number of rectangles used in a Riemann sum increase the accuracy of the area estimate? 44. A classmate claims that averaging the left- and right-hand sums always yields a better estimate for the area. Give an example to show that your classmate is incorrect. 45. If a graph gives the velocity of a car (in miles per hour) as a function of the number of hours the car has been moving, what will be the units of the area of the rectangles used to estimate the area between the graph and the horizontal axis? Explain.
6.4 The Definite Integral
433
47. The left-hand sum and right-hand sum of the function f ( x ) 5 2 x 2 1 4x on the interval 30, 44 are both equal to 10 when four rectangles are used. Does this mean that the area between the graph of f ( x ) 5 2 x 2 1 4x and the x-axis is equal to 10? Explain. 48. A positive function f is an increasing function. (That is, f ( a ) # f ( b ) whenever a # b.) Which sum will best approximate the area between the graph of f and the x -axis on the interval 3a, b4 : the left-hand sum, the right-hand sum, or the average of the left- and right-hand sums? Explain. 49. Calculate the left-hand sum for the function f ( x ) 5 Z 2 2x 1 2 Z on the interval 30, 24 using n 5 1, n 5 2, n 5 3, n 5 4, n 5 5, and n 5 6. For which value(s) of n does the left-hand sum best approximate the actual area between the graph of f and the x -axis on the interval 3a, b4 ? 50. For a nonnegative function f on the interval 3a, b4 , does reducing the width of the rectangles used in a Riemann sum ever worsen the Riemann sum estimate of the area between the graph of f and the x -axis on the interval 3a, b4 ? Explain. (Hint: Consider the function f ( x ) 5 Z 2 2x 1 2 Z using left-hand sums with n 5 2 and n 5 3.)
Challenge Exercises
The Challenge exercises are more difficult questions situated at the end of an exercise set. The exercises challenge students understanding of one or more of the sections topics.
46. Give an example of a positive function f on an interval 3a, b4 that has the property that the lefthand sum and right-hand sum approximation of the area between the graph of f and the x-axis are equal for all values of D x.
6.3 Using Sums to Approximate Area
431
Algebra Flashback Exercises

Algebra Flashback exercises give students an opportunity to refresh the algebra skills they will need to successfully master the objectives of the section.
6.3 Summary
In this section, you learned how to use left- and right-hand sums to approximate the area between the graph of a function and the horizontal axis. A solid understanding of these concepts will help you grasp the material presented in the next section.
6.3 Algebra Flashback

1. What is the equation for the area of a rectangle? 2. Each rectangle in a group of four rectangles has a width of 2 centimeters. The heights of the four rectangles are 1 centimeter, 4 centimeters, 9 centimeters, and 16 centimeters, respectively. What is the combined area of the group of rectangles? 3. Each rectangle in a group of four rectangles has a width of 0.5 centimeter. The heights of the four rectangles are 3 centimeters, 4 centimeters, 5 centimeters, and 6 centimeters, respectively. What is the combined area of the group of rectangles? 4. The interval 33,54 is to be divided into four equally sized subintervals. What is the width of each subinterval? 5. The interval 31,44 is to be divided into eight equally sized subintervals. What is the width of each subinterval? 6. What is the average of the numbers 210 and 140? 7. Convert cubic feet per second into cubic yards per minute. 8. Convert miles per hour into feet per second. (Hint: There are 5280 feet in a mile.)
6.3 Exercises
In Exercises 15, draw the rectangles used to calculate the left-hand sum estimate of the area between the graph of the function and the horizontal axis on the specified interval. (In each case, use four rectangles.) Then calculate the left-hand sum. 1. f ( x ) 5 2; 31, 54 2. g( x ) 5 2 2x 1 9; 30, 44 3. h( x ) 5 2x 1 x; 30, 24 3 4. f ( t ) 5 ; 31, 34 t 5. v( t ) 5 0.5t 1 20; 33, 74 In Exercises 615, use the left-hand sum to estimate the area between the graph of the function and the horizontal axis on the specified interval. For each exercise, calculate the sum with n 5 2, n 5 4, and n 5 10 rectangles. 6. s( t ) 5 2 3t 2 1 3t; 30, 14 7. s( t ) 5 t 3 2 3t 2 1 3t 2 1; 31, 24 8. g( x ) 5 2 x 2 1 4; 30, 24 9. h( x ) 5 2 1 1; 31, 54 x
10. f ( x ) 5 2 2x 1 20; 30, 104 11. s( x ) 5 ln( x ) ; 3e, e24 12. y 5 x2 2 2x; 32, 44 13. y 5 4t2 2 1; 32, 44 14. q 5 20 2 15. f ( t ) 5 1000 ; 3100, 2004 p
t12 ; 33, 114 t22
In Exercises 1635, use left- and right-hand sums (with n 5 4) to estimate the area between the graph of the function and the horizontal axis on the specified interval. In each exercise, calculate the left-hand sum, the right-hand sum, and the average of the two sums. The exact area, A, is given so that you can compare your estimates to the actual area. 16. f ( x ) 5 6x 1 1 on 32, 44 ; A 5 38 17. g( x ) 5 x 2 2 2x 1 2 on 33, 54 ; A 5 20 3
2
xiv
Features
Just In Time
Vertical and Horizontal Shifts
Just In Time Algebra Review

These reviews cover important algebra material at the point where students need it most.
The graph of any function may be moved on the coordinate grid through the use of vertical and horizontal shifts. Vertical shifts are achieved by changing the value of the output variable by a constant amount. Horizontal shifts are achieved by changing the value of the input variable by a constant amount. For the following examples, we use the function f ( x ) 5 2x. In each case, k and c are assumed to be nonnegative constants. Shift Vertical, upward Vertical, downward Horizontal, left Horizontal, right General Form y 5 f (x) 1 k y 5 f (x) 2 k y 5 f (x 1 c) y 5 f (x 2 c) Example g( x ) 5 2x 1 4 shifted up by 4 units g( x ) 5 2x 2 4 shifted down by 4 units g( x ) 5 2x 1 3 shifted left by 3 units g( x ) 5 2x 2 3 shifted right by 3 units
Chapter Study Sheet

A Chapter Study Sheet can be found at the end of each chapter before the Review Exercises. It acts as a summary and review of major concepts addressed in the chapter. Unlike the chapter summaries in many competing textbooks, the Chapter Study Sheet poses conceptual questions focused on the big ideas of the chapter. This helps students focus on the key concepts of the chapter in addition to critical skills as they prepare for chapter exams.
Chapter 10 Study Sheet
Chapter 10
Study Sheet
707
After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. How is a slope field created? What does each of the short line segments in a slope field represent? (10.1) dy 2. When creating a slope field for a particular differential equation dx, how should the axes be labeled? Why? (10.1) 3. How is a particular differential equation used to approximate solutions using Eulers Method? (10.2) 4. What role does step size play in the accuracy of the solution when using Eulers Method to approximate the solutions of a differential equation? (10.2) 5. How is solving a differential equation using separation of variables different from solving a differential equation using Eulers Method? (10.3) 6. Why is it helpful to separate the variables when solving particular differential equations? (10.3) 7. How does a limited growth differential equation model compare with a logistic growth differential equation model? (10.4) 8. What does the parameter M represent in the differential equations
dy dt
5 k( M 2 y ) and
dy dt
5 ky( M 2 y ) ? (10.4)
9. What types of differential equations can be solved using the integrating factor method? (10.5) 10. How is the integrating factor computed, and how is it helpful when solving certain differential equations? (10.5) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
Definitions
I I
Differential Equation: An equation that contains an unknown function and one or more of its derivatives First-Order Differential Equation: A differential equation that contains only first derivatives of the dependent variable Solution of a Differential Equation: A function in numerical, graphical, or symbolic form that satisfies a differential equation Slope Field: An array of slope marks in the xy plane representing solutions to a differential equation Step Size: The increment used for the independent variable when numerically approximating solutions to a differential equation using Eulers Method Eulers Method: A numerical process of computing function values over small intervals Separable Differential Equation: A first-order differential equation is said to be separable if it may be written as dx 5 g( y ) for some functions f ( x ) and g( y ) .
dy f (x)
Features
xv
Review Exercises
These exercises appear at the end of the chapter and help review the problems from each section. The Review Exercises provide an excellent way for students to study for an upcoming exam.
Chapter 10 Review Exercises
Section 10.1 In Exercises 16, match the given slope fields with the differential equations A through H. Note that there are six slope fields but eight differential equations, so two equations will not have matches.
A. C. E. G. 1. dy 5y dx dy y 5 x dx dy 5 x2 dx dy 1 5 y dx
y 8 6 4 2 6 0 0 y 2 4 6 8 x 4 2
Chapter 10
Review Exercises
709
4.
y 8 6 4 2 0 2 2 4 6 8 x
B. D. F. H.
dy 5 y2 dx dy 5 sin y dx dy 5 cos y dx dy 5 2y2 dx
5.
y 8 6 4 2 0 0 2 4 6 8 x
6.
y 8
2.
Make It Real Projects

P R O J E C T 1 2
What to do
1. Choose a situation that you are interested in researching for which probability density functions may be constructed. Consider the data sets given in this chapter for ideas and for examples of the types of data sets that will work best. 2. Once you have identified a data set, construct a probability density function. Depending on the situation, you may or may not be able to use one of the special probability density functions. 3. Use your probability density function to find probabilities that are interesting to you. Also find the expected value, variance, and standard deviation. 4. Repeat this process for an additional two or three data sets related to the topic you are researching. 5. Write a technical report describing the situation, the associated probability density function, and how it was constructed. Write an explanation of the expected value, variance, and standard deviation.
These end-of-chapter projects ask students to collect and analyze data from their own experiences and interests. By providing choices in the selection, students are better able to process these concepts and connect math to their own lives.
Where to find data

Data for this project can be found at the following sites, as well as others: www.census.gov/statab/www/ www.cancer.org www-nrd.nhtsa.dot.gov www.IQTest.com www.google.com www.nba.com www.mlb.com www.nfl.com
840
xvi
Features
Additional ResourcesGet the Most out of Your Textbook!

Supplements for the Instructor
Online Instructors Solutions Manual Found on the instructor website, this manual contains the complete, worked-out solutions for all the exercises in the text. HM Testing (Powered by Diploma) Testing the way you want it HM Testing offers all the tools needed to create, author, deliver, and customize multiple types of testsincluding authoring and editing algorithmic questions.
Supplements for the Student

Student Solutions Manual Written by Frank Wilson, Scott Adamson, and others, this manual offers step-bystep solutions for all odd-numbered text exercises. Excel Guide The guide provides an introduction to Excel, as well as step-by-step examples designed to teach students how to use Excel to solve selected types of problems found in this course.
Instructional DVDs
Hosted by Dana Mosely, these text-specific DVDs cover selected sections of the text and
provide explanations of key concepts, examples, exercises, and applications in a lecture-based format. HM MathSPACE encompasses the interactive online products and services integrated with Houghton Mifflin textbook programs. Available through text-specific student and instructor websites and via the online course management system, HM MathSPACE includes homework powered by WebAssign, a multimedia eBook, videos, and tutorials. WebAssign Developed by teachers for teachers, WebAssign allows instructors to focus on teaching rather than grading. Instructors can create assignments from a ready-to-use database of algorithmic questions based on end-of-section exercises or write and customize their own. With WebAssign, students can access homework, quizzes, and tests any time of day or night.
SMARTHINKING Live, online tutoring SMARTHINKING provides an easy-to-use and effective online, text-specific tutoring service. A dynamic Whiteboard and Graphing Calculator function enable students and e-structors to collaborate easily. Visit smarthinking.college.hmco.com for more information. Online Course Content for Blackboard, WebCT, and eCollege Deliver program or text-specific Houghton Mifflin content online using your institutions local course management system. Houghton Mifflin offers homework, tutorials, videos, and other resources formatted for Blackboard, WebCT, eCollege, and other course management systems. Add to an existing online course or create a new one by selecting from a wide range of powerful learning and instructional materials. For more information, visit college.hmco.com/pic/wilsonAC or contact your Houghton Mifflin sales representative.
Functions and Linear Models

M
athematical functions are a powerful tool used to model real-world phenomena. Whether simple or complex, functions give us a way to forecast expected results. Remarkably, anything that has a constant rate of change may be accurately modeled with a linear function. For example, the cost of filling your cars gas tank is a linear function of the number of gallons purchased.
1.1
Functions
Distinguish between functions and nonfunctions in tables, graphs, and words Use function notation Graph functions using technology Determine the domain of a function
1.2
Linear Functions
Calculate and interpret the meaning of the slope of a linear function Interpret the physical and graphical meaning of x- and y-intercepts Formulate the equation of a line given two points Recognize the slope-intercept, point-slope, and standard forms of a line
1.3
Linear Models
Use technology to model linear and near-linear data Use a linear equation to describe the relationship between directly proportional quantities Recognize and model naturally occurring linear, near-linear, and piecewise linear relationships
CHAPTER 1 Functions and Linear Models
1.1 Functions
Distinguish between functions and nonfunctions in tables, graphs, and words Use function notation Graph functions using technology Determine the domain of a function
Our society is a complex system of relationships among people, places, and things. Many of these relationships are interconnected. In mathematics, we often model the relationship between two or more interdependent quantities by using a function. In this section, we will show how to distinguish between functions and nonfunctions and will practice using function notation. We will also demonstrate how to use technology to draw a function graph, and discuss how to find the domain of a function.
DEFINITION: FUNCTION
A function is a rule that associates each input with exactly one output.
Often the rule is represented in a table of data with the inputs on the left-hand side and the outputs on the right-hand side. For example, the amount of money we pay to fill up our gas tank is a function of the number of gallons pumped (see Table 1.1).
TABLE 1.1
Gallons Pumped 10 15 20
Total Fuel Cost $15.99 $23.99 $31.98
In this case, the input to the function is gallons pumped and the output of the function is total fuel cost. Fuel cost is a function of the number of gallons pumped because each input has exactly one output. Similarly, the weekly wage of a service station employee is a function of the number of hours worked. The number of hours worked is the input to the function, and the weekly wage is the output of the function. Since weekly wage is a function of the number of hours worked, an employee who works 40 hours expects to be paid the same wage each time she works that amount of time.
EXAMPLE
Determining If a Table of Data Represents a Function

A car dealer tracks the number of blue cars in each of three shipments and records the data in a table (see Table 1.2). Is the number of blue cars a function of the number of cars in the shipment?
1.1 Functions
TABLE 1.2
Number of Cars in the Shipment 22

same input
Number of Blue Cars 6 7 5

C ( g)
24 24
different outputs
SOLUTION According to the definition of a function, each input must have exactly one output. The input value 24 has two different outputs: 5 and 7. Since the input 24 has more than one output, the number of blue cars is not a function of the number of cars in the shipment.
Function Notation
When we encounter functions in real life, they are often expressed in words. To make functions easier to work with, we typically use symbolic notation to represent the relationship between the input and the output. Lets return to the fuel cost table introduced previously (Table 1.3).
TABLE 1.3
Gallons Pumped 10 15 20
Total Fuel Cost $15.99 $23.99 $31.98
Observe that the fuel cost is equal to $1.599 times the number of gallons pumped. We represent this symbolically as C( g ) 5 1.599g , where C( g ) represents the total fuel cost when g gallons are pumped. [C( g ) is read C of g]. The letter C is used to represent the name of the rule, and the letter g in the parentheses indicates that the rule works with different values of g (see Figure 1.1).
g C
input value
function name
output value
FIGURE 1.1
We call the output variable of a function the dependent variable because the value of the output variable depends upon the value of the input variable. The input variable is called the independent variable. (One way to remember the meaning of the terms is to observe that both input and independent begin with in.) From the table, we see that C( 10 ) 5 15.99 C( 15 ) 5 23.99 C( 20 ) 5 31.98
For this function, the independent variable took on the values 10, 15, and 20, and the dependent variable assumed the values 15.99, 23.99, and 31.98.
EXAMPLE 2
Determining a Linear Model from a Verbal Description

An electronics store employee earns $8.50 per hour. Write an equation for the employees earnings as a function of the hours worked. Then calculate the amount of money the employee earns (in dollars) by working 30 hours.
SOLUTION Since the employee earns $8.50 for each hour worked, the employees total earnings are equal to $8.50 times the total number of hours worked. That is, E( h ) 5 8.50h
where E is the employees earnings (in dollars) and h is the number of hours worked. To calculate the amount of money earned by working 30 hours, we evaluate this function at h 5 30. E( 30 ) 5 8.5( 30 ) 5 255 The employee earns $255 for 30 hours of work.
SALEFOR IN-
Function notation is extremely versatile. Suppose we are given the function f ( x ) 5 x 2 2 2x 1 1 . We may evaluate the function using either numerical values or nonnumerical values. For example, f ( 2 ) 5 ( 2 ) 2 2 2( 2 ) 1 1 542411 51 f ( ^ ) 5 ( ^ ) 2 2 2( ^ ) 1 1
f ( a 1 2 ) 5 ( a 1 2 ) 2 2 2( a 1 2 ) 1 1 5 ( a2 1 4a 1 4 ) 2 2a 2 4 1 1 5 a2 1 2a 1 1 In each case, we replaced the value of x in the function f ( x ) 5 x 2 2 2x 1 1 with the quantity in the parentheses. Whether the independent variable value was 2, ^ , or a 1 2, the process was the same.
EXAMPLE 3
Evaluating a Function Using Function Notation

Evaluate the function s( t ) 5 t 3 1 4t at t 5 3, t 5 ^ , and t 5 a2.
SOLUTION
s( 3 ) 5 ( 3 ) 3 1 4( 3 ) 5 27 1 12 5 39
s( ^ ) 5 ( ^ ) 3 1 4( ^ )
s( a2 ) 5 ( a2 ) 3 1 4( a2 ) 5 a6 1 4a2
1.1 Functions
Graphs of Functions
Functions are represented visually by plotting points on a Cartesian coordinate system (see Figure 1.2). The horizontal axis shows the value of the independent variable (in this case, x), and the vertical axis shows the value of the dependent variable (in this case, y).
y 4 3 2 1 4 3 2 1 1 2 3 4 x
1 2 3 4
FIGURE 1.2
y 4 3 (3, 2) 2 1 4 3 2 1 1 2 3 (1, 4) 4 1 2 3 4 x (2, 4)
When using the coordinate system, y is typically used in place of the function notation f ( x ) . That is, y 5 f ( x ) . This is true even if the function has a name other than f. The point of intersection of the horizontal and vertical axes is referred to as the origin and is represented by the ordered pair ( 0, 0 ) . To graph an ordered pair ( a, b ) , we move from the origin 0 a 0 units horizontally and 0 b 0 units vertically and draw a point. If a . 0, we move to the right. If a , 0, we move to the left. Similarly, if b . 0, we move up, and if b , 0, we move down. For example, consider the table of values with its associated interpretation in Table 1.4 and the graph in Figure 1.3.
TABLE 1.4
x 23 21 1 2
y 2 24 22 4
Horizontal left 3 left 1 right 1 right 2
Vertical up 2 down 4 down 2 up 4
(1, 2)
FIGURE 1.3
When we are given the equation of a function, we can generate a table of values and then plot the corresponding points. Once we have drawn a sufficient number of points to be able to determine the basic shape of the graph, we typically connect the points with a smooth curve. For example, the function y 5 x 3 2 9x has the table of values and graph shown in Figure 1.4.
CHAPTER 1 Functions and Linear Models y 30 25 20 15 (2, 10) 10 (3, 0) (1, 8) 5 (0, 0) 4 3 2 1 5 10 15 20 25 30 1
x 24 23 22 21 0 1 2 3 4
y 2 28 0 10 8 0 28 2 10 0 28
(4, 28)
(3, 0) 2 3 (2, 10) 4 x
(1, 8)
( 4, 28)
FIGURE 1.4
EXAMPLE
Estimating Function Values from a Graph

Estimate f ( 2 3 ) and f ( 2 ) using the graph of f ( x ) 5 x 3 2 16x shown in Figure 1.5.
y 30 20 10 4 3 2 1 10 1 2 3 4 x
y 30 20 10 4 3 2 1 10 20 30 1 2 3 4 x
20 30
FIGURE 1.5
SOLUTION It appears from Figure 1.6 that f ( 2 3 ) < 20 and f ( 2 ) < 2 25. Calculating these values with the algebraic equation, we see that
f ( 2 3 ) 5 ( 2 3 ) 3 2 16( 2 3 ) 5 2 27 1 48 5 21 and
f ( 2 ) 5 ( 2 ) 3 2 16( 2 ) 5 8 2 32 5 2 24
FIGURE 1.6
One drawback of using a graph to determine the values of a function is that it is difficult to be precise. For this reason, algebraic methods are typically preferred when precision is important. Not all data sets represent functions. If any value of the independent variable is associated with more than one value of the dependent variable, the table of data and its associated graph will not represent a function. For example, consider y 5 6"2x (see Figure 1.7).
1.1 Functions y 4
x 0 0.5 0.5 2 2 4.5 4.5 8 8
y 0 21 1 22 2 23 3 24 4
3 2 1 1 1 2 3 4 1 2 3 4 5 6 7 8 x
FIGURE 1.7
Each positive value of x is associated with two different values of y. We can easily determine this from the graph by observing that a vertical line drawn through any positive value of x will cross the graph twice. This observation leads us to the Vertical Line Test.
VERTICAL LINE TEST
If every vertical line drawn on a graph intersects the graph in at most one place, then the graph is the graph of a function. Otherwise, the graph is not the graph of a function.
EXAMPLE
Determining If a Graph Represents a Function

The graph of y 5 x 2 is shown in Figure 1.8. Does the graph represent a function?
y 16 14 12 10 8 6 4 2 4 3 2 1 2 1 2 3 4 x
FIGURE 1.8
SOLUTION
y 16 14 12 10 8 6 4 2 4 3 2 1 2 1 2 3 4 x
FIGURE 1.9
Each of the vertical lines drawn crosses the graph in exactly one place, as shown in Figure 1.9. Therefore, the graph passes the Vertical Line Test and y is a function of x.
EXAMPLE
Determining If a Graph Represents a Function

The graph of y 5 6"x is shown in Figure 1.10. Does the graph represent a function?
y 3 2 1 1 1 2 3 1 2 3 4 5 6 7 8 9 x
FIGURE 1.10
SOLUTION Notice that many of the vertical lines in Figure 1.11 cross the graph in two places. The graph does not pass the Vertical Line Test, so y is not a function of x. Each positive value of x corresponds with two values of y.
1.1 Functions y 3 2 1 1 1 2 3 1 2 3 4 5 6 7 8 9 x
FIGURE 1.11
The graphs shown thus far have been continuous. That is, each graph could be drawn without lifting the pencil. However, some function graphs are discontinuous. A discontinuous graph has a break in the graph. The break in the graph is referred to as a discontinuity.
EXAMPLE
Determining If a Discontinuous Graph Represents a Function

The graph of h( x ) 5 e represents a function.
y 2 1 x
x if x # 1 is shown in Figure 1.12. Determine if the graph 2 if x . 1
1 1 2
FIGURE 1.12
SOLUTION The graph is discontinuous at x 5 1 . The value of the function at x 5 1 is 1. [The open dot at ( 1, 2 ) indicates that the point ( 1, 2 ) is not part of the graph.] Although the graph is discontinuous, each value of the independent variable is associated with exactly one value of the dependent variable. Therefore, the graph represents a function.
10
Graphing a Function 1. Bring up the graphing list by pressing the Y button.
2. Type in the function using the X,T,,n button for the variable and the ^ button to place an expression in an exponent. Make sure you use parentheses as needed.
3. Specify the size of the viewing window by pressing the WINDOW button and editing the parameters. The Xmin is the minimum x value, Xmax is the maximum x value, Ymin is the minimum y value, and Ymax is the maximum y value. The Xscl and Yscl are used to specify the spacing of the tick marks on the graph. 4. Draw the graph by pressing the button.
GRAPH
Domain and Range

We are often interested in the set of all possible values of the independent variable and the set of all possible values of the dependent variable.
DOMAIN AND RANGE
The set of all possible values of the independent variable of a function is called the domain. The set of all possible values of the dependent variable of a function is called the range. It is easy to remember the meaning of the terms if we observe that input, independent variable, and domain all contain the word in.
1.1 Functions
11
Consider a common kitchen blender. If we put an orange into a blender and turn the blender on, the orange is transformed into orange juice. We say that orange is in the domain of the blender function and orange juice is in the range of the blender function. On the other hand, if we put a rock into the blender and turn it on, the blender will self-destruct. We say that rock is not in the domain of the blender function, since blenders are unable to process rocks. Finding the domain of a function frequently involves solving an equation or an inequality. Recall that solving an equation means finding the value of the variable that makes the equation a true statement. The domain of most frequently used mathematical functions is the set of all real numbers. However, there are three common situations in which the domain of a function is restricted to a subset of the real numbers. They are: 1. A zero in the denominator 2. A negative value under a square root symbol (radical) 3. The context of a word problem Lets look at an example for each of the situations.
EXAMPLE
Determining the Domain of a Function

What is the domain of g( x ) 5
3x 2 1 ? 2x 1 6
SOLUTION We know that the value in the denominator must be nonzero, since division by zero is undefined. That is, 2x 1 6 2 0. To find the value of x that must be excluded from the domain, we must solve the following equation:
2x 1 6 5 0 2x 1 6 2 6 5 0 2 6 2x 5 2 6 26 2x 5 2 2 x 5 23
Subtract 6 from both sides
Divide both sides by 2
The domain of the function is all real numbers except 2 3. We may rewrite the function equation with the domain restriction as follows: g( x ) 5 3x 2 1 , 2x 1 6 x 2 23
EXAMPLE

What is the domain of y 5 "x 2 3?
SOLUTION We know that the value underneath the radical must be nonnegative, since the square root of a negative number is undefined in the real number system. Therefore, x 2 3 $ 0 . Solving for x, we get x23$0 x2313$013 Add 3 to both sides x$3
The domain of the function is all real numbers greater than or equal to 3.
12
EXAMPLE 10

The revenue R from the sale of x gallons of gasoline is given by the equation R( x ) 5 1.379x What is the domain of the function?
SOLUTION In the context of this problem, it doesnt make sense to talk about selling a negative number of gallons of gas, so x $ 0. The domain of the function is all nonnegative real numbers.
Graphing a Function 1. Enter a list of domain values for the function.
2. In the cell adjacent to the first domain value, type and then type in the equation of the function you want to graph. Use the cell reference with the domain value as the variable. The equation y 5x 4 is represented at the right.
3. Click the lower right-hand corner of the box in cell B2 and drag down to the cell in the B column that is adjacent to the last domain value. This pastes the formula in cell B2 to the remainder of the cells. The reference to A2 automatically updates to reference the adjacent cell.
(Continued)
1.1 Functions
13
4. Click and drag to select the entire domain and range.
5. Click the icon on the toolbar. This will open the Chart Wizard.
6. Select XY (Scatter) and the indicated chart subtype.
7. Press Finish to display the graph of the function that passes through the points. In general, the more points used to generate the graph, the more accurate the graph will be.
50 40 30 20 10 0 10 0 2 4 6 8 10 12
14
1.1 Summary
In this section, you learned to distinguish between functions and nonfunctions, and you practiced using function notation. You learned how to use technology to graph a function, and you discovered how to find the domain of a function. Mastering each of these techniques will help you understand the subsequent concepts covered in this chapter.

1. For a point ( a, b ) in a rectangular coordinate system, what will be the sign of a and of b if the point is in the second quadrant? 2. Why is division by zero undefined? 3. For the function c( t ) , which variable is the independent variable? 4. Describe how the following terms are related: output, range, input, dependent variable, domain, independent variable. 5. What does it mean to evaluate a function f ( x ) at x 5 4? 6. What does it mean to solve a function equation y 5 f (x) ? 7. Evaluate f ( x ) 5 4x 2 1 3x 2 1 at x 5 2 2. 8. Solve 4x 2 1 3x 2 1 5 6 for x.
1.1 Exercises
In Exercises 16, determine whether the output is a function of the input. 1. W( a ) 5 your weight in pounds when you were a years old. 2. S( n ) 5 your score on test number n in a finite math course. (Assume you could take the test only once.) 3. Time of Day 11:00 a.m. 1:00 p.m. 3:00 p.m. 4. Time of Day 10:00 a.m. 12:00 p.m. 2:00 p.m. Temperature (F) 68 73 75 Temperature (F) 66 74 74 5. Fish Caught 169 182 182 6. Fish Caught 252 276 301 Salmon in Catch 24 32 47 Salmon in Catch 74 74 92
1.1 Functions
15
In Exercises 713, calculate the value of the function at the designated input and interpret the result. 7. C( x ) 5 39.95x at x 5 4 , where C( x ) is the cost of buying x pairs of shoes. 8. C( x ) 5 29.95x 1 200 at x 5 300 , where C( x ) is the cost of making x pairs of shoes. 9. H( t ) 5 2 16t 2 1 120 at t 5 2, where H( t ) is the height of a cliff diver above the water t seconds after he jumped from a 120-foot cliff. 10. Find R( t ) at t 5 4, where R( t ) is the quarterly revenue for an international tortilla producer t quarters after December 1999, as shown below.
Revenue
12.
Stock Price Find P( t ) at t 5 3, where P( t ) is the stock price of an e-learning company at the end of the day and t is the number of days after November 4, 2001.
10.00 Stock price (in dollars) 9.50 9.00 8.50 8.00 7.50 7.00 0 1 2 3 4 5 Number of days after 11/4/01 6
Quarter 1 2 3 4 5 11.
Revenue (in millions of dollars) 460.0 452.0 474.0 466.0 483.8
Source: Digital Think Corporation.
13.
Stock Price Find P( t ) at t 5 4, where P( t ) is the stock price of a computer company at the end of the day and t is the number of days after November 4, 2001.
19.70 19.60 19.50 19.40 19.30 19.20 19.10 19.00 18.90 18.80 18.70 0 1 2 3 4 5 Number of days after 11/4/01 6
Source: Gruma, S.A. de C.V .

Earnings per Share Find E( t ) at t 5 4, where E( t ) is the quarterly earnings per share for an international tortilla producer t quarters after December 1999.
Quarter 1 2 3 4 5 6
Earnings per Share (in dollars) 0.13 2 0.11 0.15 0.06 0.03 0.06
Source: Apple Computer Corporation.
In Exercises 1419, determine whether the graphs represent functions by applying the Vertical Line Test. 14.
y 20 10 4 3 2 1 10 20 1 2 3 4 x
Stock price (in dollars)
15.
y 2 1
Source: Gruma, S.A. de C.V .
4 3 2 1 1
1 2 3 4 x
16 16.
CHAPTER 1 Functions and Linear Models y 3 2 1 4 3 2 1 0 1 2 3 4 x
In Exercises 2427, use the graph to estimate the value of the function at the indicated x value. Then calculate the exact value algebraically. 24. y 5 0.25( x 1 1 ) 2( x 2 3 ) ; x 5 4
y 20 15 10 5 4 3 2 15 10 15 20 1 2 3 4 x
17.
y 4 3 2 1 1 2 3 4 x
25. y 5 5 Z x Z 2 x 2 ; x 5 1
y 6 5 4 3 2 1 5 4 32 1 0 1 2 3 4 5x
18.
y 1.0 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 8 9x
26. y 5
Zx 1 2 Z ;x53 x2 1 1
y 2.0 1.5 1.0 0.5 4 3 2 1 0 1 2 3 4 x
19.
y 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0 1 2 3 4 5 6 7 8 9x
27. y 5 2
4( x 2 2 4 ) ; x 5 22 x 1 2x 2 1 3x 1 6
3
In Exercises 2023, graph the function on your graphing calculator using the specified viewing window. Note that a # x # b means x min 5 a and x max 5 b . Similarly, c # y # d means y min 5 c and y max 5 d . 20. y 5 x 2 2 5x; 2 3 # x # 7, 2 8 # y # 8 21. y 5 2 x 2 2 4; 2 3 # x # 3, 2 10 # y # 1 22. y 5 2 x 1 2; 2 3 # x # 3, 2 2 # y # 6 23. y 5 2 2x 2 1 1; 2 2 # x # 2, 2 3 # y # 2
y 3.0 2.5 2.0 1.5 1.0 0.5 4 3 2 1 0.5 1 2 3 4 x
In Exercises 2841, determine the domain of the function. 28. f ( x ) 5 5x 3 2 5x 30. C( r ) 5 r2 2 1 r11 29. f ( p ) 5 p2 2 2p 1 1 31. h( r ) 5 r2 2 1 r2 1 1
1.1 Functions
17
32. S( t ) 5
34. H( x ) 5 "x 2 5 36. f ( x ) 5 "x 1 5 x12
t23 4t 1 1
33. g( t ) 5
35. f ( a ) 5 a3 2 " a 1 1 37. f ( x ) 5 " 2x 1 6 x2 1 3
3t 2 3 4t 2 4
38. C( n ) 5 the cost of buying n apples 39. P( n ) 5 the profit earned from the sale of n bags of candy 40. W( n ) 5 the average birth weight of someone born in year n 41. H( n ) 5 the average height of someone born in year n
47. Your classmate claims that a students quarterly tuition cost is not a function of the number of credits the student takes. To support her claim, she points out that she and her friend are enrolled in the same classes, but she has to pay $2400 more in tuition than her friend because she is from outof-state. Do you agree with your classmates claim that a students quarterly tuition cost is not a function of the number of credits taken? Explain.
48. Graph the function f ( x ) 5
calculator. What do you think is the domain of f ? 49. Determine the domain of f ( x ) 5
x21 x2 2 1
x21 x2 2 1
on your
42. During a group activity, your professor asks your group to determine the value of t that makes the function g( t ) 5 3t 2 4 equal to 53. One of your group members sets up and solves the equation as follows: g( 53 ) 5 3t 2 4 53g 5 3t 2 4 53g 1 4 5 3t 53g 1 4 t5 3 Explain to your groupmate what his mistake was and how he might fix it. 43. Explain the relationship between the coordinate ( 3, 8 ) and the equation h( 3 ) 5 8. 44. Write in words the mathematical equation f ( t ) 5 5t 2 2 2t 1 7. 45. In the statement New car sales are a function of used car sales, what is the independent variable and what is the dependent variable of the function? 46. At many private colleges, a semesters tuition cost depends only on the number of credits taken by the student. What is a reasonable domain for the tuition cost function?
algebraically and compare your answer to the solution of Exercise 48. 50. Are all lines functions? Explain. 51. Does the table of data represent a function? Explain. x 0 1 1 3 4 y 5 6 6 0 2
52. For a given date and location, is air temperature a function of the time of day? Explain.
18
1.2 Linear Functions
Calculate and interpret the meaning of the slope of a linear function Interpret the physical and graphical meaning of x- and y-intercepts Formulate the equation of a line given two points Recognize the slopeintercept, point-slope, and standard forms of a line
The cost of filling our cars gas tank, the number of calories we consume by eating a few bags of fruit snacks, and the amount of sales tax we pay when we buy new clothes are examples of linear functions. In this section, we will show how to calculate the slope of a linear function. We will discuss how to interpret the physical and graphical meaning of slope, x-intercept, and y-intercept. We will also show three different ways to write a linear equation and demonstrate how to find the equation of a linear function from a data set.
DEFINITION: GRAPH OF A LINEAR FUNCTION
The line passing through any two points ( x1, y1 ) and ( x2, y2 ) with x1 2 x2 is referred to as the graph of a linear function.
Linear functions are characterized by a constant rate of change. That is, increasing a domain value by one unit will always change the corresponding range value by a constant amount. The converse is also true. Any table of data with a constant rate of change represents a linear function. The constant rate of change is referred to as the slope of the linear function.
DEFINITION: SLOPE
The slope of a linear function is the change in the output that occurs when the input is increased by one unit. The slope m may be calculated by dividing the difference of two outputs by the difference in the corresponding inputs. That is, m5 y2 2 y1 x2 2 x1
where ( x1, y1 ) and ( x2, y2 ) are data points of the linear function.
The Just in Time box on the following page details common errors that occur in calculating the slope.
19
Just In Time
Common Errors with Slope
Slope is, by definition, the change in the output over the change in the input. Remembering this will help you avoid these common errors. The points ( 2, 4 ) and ( 5, 9 ) will be used in each example. The correct slope is m5 y2 2 y1 x2 2 x1 924 5 522 5 5 3 Example of Error m5 522 924 3 5 5 924 225 5 5 23 5 52 3
Error m5 x2 2 x1 y2 2 y1
Why Wrong This is the change in the input over the change in the output. This is the reciprocal of the slope. The order of the terms in the denominator is reversed. The result is the opposite of the slope.
m5
y2 2 y1 x1 2 x2
m5
TABLE 1.5
Bags (n) 1 2 3
Calories (C) 100 200 300
Source: Fruit Smiles 9-oz. box label
The number of calories C in n bags of fruit snacks is shown in Table 1.5. Notice that the calorie count increases by 100 calories for each additional bag. Since the dependent variable C is changing at a constant rate (100 calories per bag), the data may be modeled by a linear function. In this case, the linear function is C 5 100n. To calculate the slope of the function, we may use any two data points. Using the data points ( x1, y1 ) 5 ( 1, 100 ) and ( x2, y2 ) 5 ( 3, 300 ) , the slope is m5 5 300 2 100 calories 3 2 1 bags 200 calories 2 bags
5 100 calories per bag A one-bag increase in the input results in a 100-calorie increase in the output. Will we get the same result if we use different points? Lets check using (1, 100) and (2, 200) (see Figure 1.13).
20
CHAPTER 1 Functions and Linear Models C 400 350 300 250 200 150 100 50 0 0 1
100 units 1 unit 2 3 4 n
FIGURE 1.13
m5 5
100 2 200 calories 1 2 2 bags 2 100 calories 21 bag
5 100 calories per bag The result is the same! With linear functions, we may use any two points to calculate the slope. Not all lines are functions. Vertical lines fail to pass the Vertical Line Test, so they are not functions. As shown in Example 1, vertical lines have an undefined slope.
EXAMPLE 1
Determining the Slope of a Line from Two Points on the Line

What is the slope of the line going through (2, 4) and (2, 8) (Figure 1.14)?
y 8 7 6 5 4 3 2 1 0 0 1 2 3 4
FIGURE 1.14
SOLUTION
m5 5
824 222 4 0
5 undefined Since division by zero is not defined, the line has an undefined slope. Vertical lines are the only lines that have an undefined slope.
21
EXAMPLE
Determining the Slope of a Line from Two Points on the Line

What is the slope of the line going through (2, 4) and (5, 4) in Figure 1.15?
y 5 4 3 2 1 0 0 1 2 3 4 5 x
FIGURE 1.15
SOLUTION
m5 5
424 522 0 3
50 Any line with a zero slope is a horizontal line. The absolute value of the slope is referred to as the magnitude of the slope. In a general sense, slope is a measure of steepness: The greater the magnitude of the slope, the greater the steepness of the line. The graph of a line with a negative slope falls as the independent variable increases. The graph of a line with a positive slope rises as the independent variable increases.
EXAMPLE 3
Determining the Sign and Magnitude of a Lines Slope from a Graph

Use the graph in Figure 1.16 to determine which lines have a negative slope and which lines have a positive slope. Then identify the line whose slope has the greatest magnitude.
y f
g x
FIGURE 1.16
22
SOLUTION Since lines h and j fall as the value of x increases, they have negative slopes. Since lines f and g rise as the value of x increases, they have positive slopes. The steepest line is the line whose slope has the greatest magnitude. Although both f and j are steep, line f is steeper. Therefore, f has the slope with the greatest magnitude.
Intercepts
In discussing linear functions, it is often useful to talk about where the line crosses the x and y axes. Knowing these intercepts helps us to determine the equation of the line.
DEFINITION: y -INTERCEPT
The y-intercept is the point on the graph where the function intersects the y axis. It is the point at which the value of the independent variable is 0. It is formally written as an ordered pair (0, b), but b itself is often called the y-intercept.
EXAMPLE
Finding the y-intercept of the Graph of a Linear Function

What is the y-intercept of the linear function y 5 3x 1 5 ?
SOLUTION At the y-intercept, the x coordinate is 0.
y 5 3( 0 ) 1 5 55
Substitute 0 for x
So the y-intercept of the function is (0, 5).
DEFINITION: x -INTERCEPT
The x-intercept is the point on the graph where the function intersects the x axis. It is the point at which the value of the dependent variable is 0. It is formally written as an ordered pair ( a, 0 ) , but a itself is often called the x-intercept.
EXAMPLE
Finding the x-intercept of the Graph of a Linear Function

What is the x-intercept of the linear function y 5 3x 1 5 ?
SOLUTION At the x-intercept, the y coordinate is 0.
y 5 3x 1 5 0 5 3x 1 5 2 3x 5 5 x52 5 3
5
Substitute 0 for y
So the x-intercept of the function is A 2 , 0 B . 3
23
EXAMPLE
Determining the Slope and Intercepts of the Graph of a Linear Function

Determine the slope, the y-intercept, and the x-intercept of the linear function from its graph (Figure 1.17).
y 8 6 4 2 1 1 2 3 4 x
FIGURE 1.17
SOLUTION Since the graph of the function crosses the y axis at ( 0, 6 ) , the y-intercept is ( 0, 6 ) . Since the graph crosses the x axis at ( 3, 0 ) , the x-intercept is ( 3, 0 ) . Since the line is falling as x increases, the slope will be negative. The slope of the line is
m5 5 5
y2 2 y1 x2 2 x1 620 023 6 23
5 22
Linear Equations
The graph of any line may be represented by a linear equation. Vertical and horizontal lines have the simplest linear equations.
EQUATIONS OF VERTICAL AND HORIZONTAL LINES

The equation of a vertical line passing through a point ( a, b ) is x 5 a . The equation of a horizontal line passing through a point ( a, b ) is y 5 b .
If we know the slope and y-intercept of a linear function, we are able to easily determine the slope-intercept form of the linear equation.
24
SLOPE-INTERCEPT FORM OF A LINE
A linear function with slope m and y-intercept ( 0, b ) has the equation y 5 mx 1 b
EXAMPLE
Determining the Slope-Intercept Form of a Line

What is the slope-intercept form of the linear function with slope 5 and y-intercept (0, 4)?
SOLUTION Since m 5 5 and b 5 4,
y 5 5x 1 4 is the slope-intercept form of the line.
EXAMPLE
Finding a Linear Model from a Verbal Description

For breakfast, we decide to eat an apple containing 5.7 grams of dietary fiber and a number of servings of Cheerios, each containing 3 grams of dietary fiber. (Source: Cheerios box label.) Write the equation for our dietary fiber intake as a function of number of servings of cereal.
SOLUTION We know that if we dont eat any cereal, we will consume 5.7 grams of dietary fiber (from the apple). So the y-intercept of the fiber function is (0, 5.7). Since each serving of Cheerios contains the same amount of fiber, we know that the function is linear. The slope of the function is 3 grams per serving. So the equation of the fiber function is
F( n ) 5 3n 1 5.7 where F( n ) is the amount of dietary fiber (in grams) and n is the number of servings of cereal. To check our work, we directly calculate the number of grams of dietary fiber in a breakfast containing two servings of cereal and one apple. Fiber 5 3 1 3 1 5.7 5 11.7 grams Using the fiber function formula, we get F( 2 ) 5 3( 2 ) 1 5.7 5 11.7 grams The results are the same, so we are confident that our formula is correct.
EXAMPLE
Determining a Linear Model from a Table of Data

The amount of sales tax paid on a clothing purchase in Seattle is a function of the sales price of the clothes, as shown in Table 1.6.
25
TABLE 1.6
Sales Price ( p ) $20.00 $30.00 $40.00
Tax ( T ) $1.72 $2.58 $3.44
Source: www.cityofseattle.net.
If the function is linear, write the equation for the sales tax as a function of the sales price.
SOLUTION We must first determine if the function is linear. Since each $10 increase in sales price increases the sales tax by a constant $0.86, we conclude that the function is linear. The slope of the function is
m5 5
2.58 2 1.72 dollars 30 2 20 dollars 0.86 dollar 10 dollars
5 0.086 tax dollar per sales price dollar (In other words, for each dollar increase in the sales price, the sales tax increases by 8.6.) Since a sale of $0 results in $0 sales tax, we know that the y-intercept is (0, 0). The equation of the function is T 5 0.086p To check our work, we substitute the point ( 40, 3.44 ) into the equation. 3.44 5 0.086( 40 ) 3.44 5 3.44 The statement is true, so we are confident that our equation is correct.
p 5 40 and T 5 3.44
EXAMPLE 10
Finding a Linear Model from the Graph of a Linear Function

The graph in Figure 1.18 shows the balance of a checking account as a function of the number of ATM withdrawals from the account.
y Account balance (in dollars) 200 150 100 50 0 0 1 2 3 4 5 6 7 8 9 10 x
FIGURE 1.18
26
(a) Identify the x-intercept and y-intercept and interpret their physical meanings. (b) Write the linear equation for the function.
SOLUTION
(a) The x-intercept is (10, 0). When there have been 10 withdrawals, the account balance is $0. The y-intercept is (0, 200). When there have been no withdrawals, the account balance is $200. (b) The slope of the function is m5 200 2 0 dollars 0 2 10 withdrawals
5 2 $20 per withdrawal So the slope-intercept form of the linear equation is B 5 2 20w 1 200 where B is the account balance and w is the number of withdrawals.
Finding the Equation of a Line

As demonstrated in Example 10, to find the slope-intercept form of a line from two points, we need to calculate the slope and the y-intercept of the function. To do this, we proceed as follows: 1. Calculate the slope. 2. Write the function in slope-intercept form, substituting the slope for m. 3. Select one of the points and substitute the output value for y and the input value for x. 4. Solve for b. 5. Write the function in slope-intercept form, substituting the y-intercept for b.
EXAMPLE 11
Finding the Slope-Intercept Form of a Line from Two Points

Find the equation of the line passing through the points (3, 5) and (7, 1).
SOLUTION
m5 5
521 327 4 24
5 21 So the slope is 2 1. Substituting the slope and the point (3, 5), we get y 5 21 ? x 1 b 5 5 2 1( 3 ) 1 b b58 The y-intercept is (0, 8). The slope-intercept form of the line is y 5 2 1 ? x 1 8 or y 5 2 x 1 8 .
27
Other Forms of Linear Equations

There are two additional forms of linear equations that are commonly used: standard form and point-slope form.
STANDARD FORM OF A LINE
A linear equation may be written as ax 1 by 5 c where a, b, and c are real numbers. If a 5 0, the graph of the equation is a horizontal line. If b 5 0, the graph of the equation is a vertical line. In the equation, a and b cannot both be zero.
The standard form of a linear equation is extremely useful when working with systems of equations or solving linear programming problems. We will discuss the standard form of a line further when we introduce these topics. The point-slope form of a line is especially useful when we know a lines slope and the coordinates of a point on the line.
POINT-SLOPE FORM OF A LINE
A linear function written as y 2 y1 5 m( x 2 x1 ) has slope m and passes through the point ( x1, y1 ) .
EXAMPLE 12
Finding a Linear Model from a Verbal Description

Based on data from 1980 to 1999, per capita consumption of milk as a beverage has been decreasing by approximately 0.219 gallon per year. In 1997, the per capita consumption of milk as a beverage was 24.0 gallons. (Source: Statistical Abstract of the United States, 2001, Table 202, p. 129.) Find an equation for the per capita milk consumption as a function of years since 1980.
SOLUTION The slope of the line is m 5 2 0.219. The point ( 17, 24.0 ) lies on the line, since 1997 is 17 years after 1980. The point-slope form of the line is
y 2 24.0 5 2 0.219( t 2 17 ) If preferred, the equation may be rewritten in slope-intercept form, y 5 2 0.219( t 2 17 ) 1 24.0 y 5 2 0.219t 1 3.723 1 24.0 y 5 2 0.219t 1 27.723 or in standard form, 0.219t 1 y 5 27.723
28
Graphing Linear Functions

To graph a linear function, we first generate a table of values by substituting different values of x into the equation and calculating the corresponding values of y. For example, if we are given the linear equation y 5 4x 2 8 , we may choose to evaluate the function at x 5 2 1 , x 5 0 , x 5 1 , x 5 2 , and x 5 3 , as shown in Table 1.7.
TABLE 1.7
y 6 4 2 1 2 4 6 8 10 12 14 1 2 3 x
x 21 0 1 2 3
y 2 12 28 24 0 4
FIGURE 1.19
We then plot the points and connect them with a straight line, as shown in Figure 1.19. (Although we plotted multiple points, only two points are necessary to determine the line.)
EXAMPLE 13
Graphing a Linear Function

Graph the function y 5 2x 2 3 .
SOLUTION From the equation, we see that the y-intercept is ( 0, 2 3 ) . We need to find only one more point. Evaluating the function at x 5 2 yields y 5 1 , so ( 2, 1 ) is a point on the line. We plot each point and connect the points with a straight line, as shown in Figure 1.20.
y 3 2 1 1 1 2 3 4 5 6 7 1 2 3 x
FIGURE 1.20
Recall that the standard form of a line is ax 1 by 5 c. Graphing a line from standard form is remarkably easy. The y-intercept of the graph occurs when the x value is equal to zero. Similarly, the x-intercept of the graph occurs when the
29
y value is equal to zero. Using these facts, we can quickly determine the x- and y-intercepts of the function. To find the x-intercept, we set y 5 0 . ax 1 by 5 c ax 1 b( 0 ) 5 c ax 5 c x5 c a
Notice that the x coordinate of the x-intercept is the constant term divided by the coefficient on the x term. To find the y-intercept, we set x 5 0 . ax 1 by 5 c a( 0 ) 1 by 5 c by 5 c y5 c b
Notice that the y coordinate of the y-intercept is the constant term divided by the coefficient on the y term. Using this procedure to find intercepts will allow us to graph linear equations in standard form quickly by hand.
EXAMPLE 14
Graphing a Linear Function

Graph the linear function 2x 1 y 5 4.
SOLUTION The x-intercept is found by dividing the constant term by the coefficient on the x term.
y 6 5 4 3 2 1 1 1 1 2 x
x5
4 2
52 The point ( 2, 0 ) is the x-intercept. The y-intercept is found by dividing the constant term by the coefficient on the y term. y5 4 1
54 The point ( 0, 4 ) is the y-intercept. We graph the x- and y-intercepts and then draw the line through the intercepts, as shown in Figure 1.21.
FIGURE 1.21
1.2 Summary
In this section, you learned how to calculate the slope of a linear function and how to interpret the physical and graphical meaning of slope, x-intercept, and yintercept. You also discovered how to calculate the equation of a linear function from a set of data and learned three different ways to write a linear equation.
30

1. What is the relationship between the formula for the slope of a line, m 5
y2 2 y1 , x2 2 x1
and the formula
5. How can you easily tell if the line between ( a, b ) and ( c, d ) is a horizontal line? 6. Why do we say that a vertical line has an undefined slope? 7. Why are the terms undefined slope and zero slope preferable to the term no slope? 8. The average rate of change for a line is constant regardless of which points on the line are used to calculate the rate of change. Determine the average rate of change for each of the following lines: y 5 mx 1 b, Ax 1 By 5 C, and y 2 y1 5 m( x 2 x1 ) .
avg rate of change 5
for calculating the average rate of change in a function f over an interval 3a, b4 ,
f (b) 2 f (a) ? b2a
2. Convert the linear equation Ax 1 By 5 C to a linear equation in slope-intercept form. 3. Convert the linear equation y 5 mx 1 b to a linear equation in standard form. 4. How can you easily tell if the line between ( a, b ) and ( c, d ) is a vertical line?
1.2 Exercises
In Exercises 16, calculate the slope of the linear function passing through the points. 1. (2, 5) and (4, 3) 3. (1.2, 3.4) and (2.7, 3.1) 4. (7, 11) and (9, 2) 6. (4, 3) and (4, 7) In Exercises 712, find the x-intercept and y-intercept of the linear function. 7. y 5 5x 1 10 9. y 5 2x 1 11 11. 3x 2 y 5 4 8. y 5 2 3x 1 9 1 10. y 5 x 2 2 2 12. 4x 2 2y 5 5 5. (2, 2) and (5, 2) 2. ( 2 3, 4 ) and ( 0, 2 2 ) 17. ( 2 2, 2 ) and (5, 2) 18. ( 2 3.1, 4.5 ) and ( 2.1, 2 3.4 ) In Exercises 1925, graph the line. 19. y 5 4x 2 2 21. y 2 4 5 0.5( x 2 2 ) 23. 2x 2 3y 5 5 2 4 25. y 5 2 x 1 3 3 20. x 2 y 5 3 22. y 5 2 5x 1 10 24. y 2 9 5 2 3( x 1 2 )
In Exercises 2630, use the graph to determine the equation of the line. 26.
y 6 5 4 3 2 1 2 1 1 2 3 1 2 3 4 x
In Exercises 1318, write the equation of the linear function passing through the points in slope-intercept form, point-slope form, and standard form. 13. (2, 5) and (4, 3) 14. ( 2 3, 4 ) and ( 0, 2 2 ) 15. (1.2, 3.4) and (2.7, 3.1) 16. (7, 11) and (9, 2)
31
27.
y 6 5 4 3 2 1
28.
(3, 5)
y 5 4 3 2 1 1 2 3 4 x
33.
Take-Home Pay
Months (since Sept. 2001) 0 1
Take-Home Pay (dollars) 3,167.30 4,350.31
4 3 2 11
1 2 3 4 x
4 3 2 11 2
Source: Employee pay stubs.
29.
y 3 2 1 1 2 3 4 x
30.
y 3 2 1 (0, 1.25) 1 2 3 4 5x
34.
Museum Admission Cost
1 1 2 3
5 4 321
People in Group 3 4 5 6
Total Cost of Admission for a Group Visiting the Experience Music Project $59.85 $79.80 $99.75 $119.70
In Exercises 3135, determine if the table of data represents a linear function. If so, calculate the slope and interpret its real-life significance. 31.
Average Personal Income
Source: www.emplive.com.
Year 1989 1999
U.S. Average Personal Income (in terms of year 2000 dollars) 18,593 28,525
35.
Solid Waste Disposal
Clean Wood (pounds) 500 700 900 1,000
Cost to Dispose of Clean Wood at Enumclaw Transfer Station $18.75 $26.25 $33.75 $37.50
Source: www.census.gov.
32.
Internet Access
Year 1997 1998 1999 2000

Number of People with Internet Access at Home or at Work (in thousands) 46,305 62,273 83,677 112,949
Source: www.dnr.metrokc.gov.
32
In Exercises 3638, find the linear equation that models the data and answer the specified questions. 36. The Recommended Daily Allowance of dietary fiber is 25 grams for a person on a 2,000-calorie-per-day diet. A
Nutrition
3 4
40. A particular line has the property that it has one y-intercept but no x-intercepts. What is the slope of the line? 41. Given that a workers current annual income is $29,450 and that the income is increasing at a constant rate of 3% per year, is the workers income a linear function of the years worked? Explain. 42. Explain how you would quickly determine the x- and y-intercepts of 2x 1 5y 5 30 using only mental math. 43. A local newspaper article states that Taxes, which are currently $4000, are slated to increase by $200 per year (5% per year) for the next 10 years. Does the articles statement make sense? Explain why or why not.
-cup serving of Post Fruity Pebbles contains
0.2 gram of dietary fiber. (Source: Package labeling.) A 1-cup serving of 2 percent milk fortified with vitamin A doesnt contain any
7
dietary fiber. A large banana A 8 to 8 8 inches long B contains 3.3 grams of dietary fiber. (Source: www.nutri-facts.com.) How many servings of Fruity Pebbles with milk would you have to eat along with a large banana to consume 8 grams of dietary fiber? (Round up to the nearest number of servings.) 37. The Recommended Daily Allowance of dietary fiber is 25 grams for a person on a 2,000-calorie-per-day diet. A
Nutrition
3 4
-cup serving of General Mills Wheaties contains
44. What is the equation of the line that passes through ( 4, 7 ) and ( 4, 2 1 ) ? 45. Do the two equations represent the same line? Explain. 3x 1 4y 5 12; y 1 3 5 2 0.75( x 2 8 ) 46. Explain why the equation of a vertical line cannot be written in slope-intercept or point-slope form. 47. The U.S. Census Bureau projects that the population of Alaska will grow at an approximate rate of 9,321 people per year between 1995 and 2025. It estimates that 791,000 people will be living in Alaska in 2025. (Source: www.census.gov.) What is the projected population of Alaska in 2035?
Alaska Population
2.1 grams of dietary fiber. (Source: Package labeling.) A 1-cup serving of 2 percent milk fortified with vitamin A doesnt contain any dietary fiber. A large A 8 to
7 88
banana contains 3.3 grams of dietary fiber. (Source: www.nutri-facts.com.) How many servings of Wheaties with milk would you have to eat along with a large banana to consume 8 grams of dietary fiber? (Round up to the nearest number of servings.) 38. The Recommended Daily Allowance of dietary fiber is 30 grams for a person on a 2,500-calorie-per-day diet. A
Nutrition
inches long B
large apple
7
8 8 inches long B contains 3.3 grams of dietary
5.7 grams of dietary fiber. A large banana A 8 to fiber. (Source: www.nutri-facts.com.) How many apples would you have to eat along with a banana to consume 30 grams of dietary fiber?
1 34
inches in diameter B contains
48. Are all lines functions? Explain. 49. Two lines are said to be parallel if they do not intersect. If f ( x ) 5 mx 1 b and g( x ) 5 nx 1 c are parallel lines, what conclusions can you draw about b, c, m, and n? 50. Two lines are perpendicular if the slope of one line is the negative reciprocal of the slope of the other line. Find the equation of the line perpendicular to y 5 4 that passes through the point ( 3, 2 2 ) . 51.
World Population Based on data from 1980 to 2001, the population of the world may be modeled by P 5 81.3t 1 4460 million people, where t is the number of years
39. Although the equation of any vertical line may easily be written in standard form, it may not be written in slope-intercept or point-slope form. Explain.
1.3 Linear Models
33
since 1980. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 1324, p. 829.) According to the model, how quickly will the world population be increasing in 2010? 52. The equation of a nonvertical line is given by ax 1 by 5 c. What are the slope, x-intercept, and y-intercept of the line? 53. Is the function shown in the table linear? Explain. x 2 5 8 10 14 y 11 17 23 29 35
55. In order for the function shown in the table to be linear, what must be the value of a? x 2 5 8 a 14 y 6 12 18 26 30
56. What is the point of intersection of the lines x 5 a and y 5 b ?
54. In order for the function shown in the table to be linear, what must be the value of b? x 2 5 8 10 14 y 7 13 19 b 31
1.3 Linear Models
Use technology to model linear and near-linear data Use a linear equation to describe the relationship between directly proportional quantities Recognize and model naturally occurring linear, near-linear, and piecewise linear relationships
The amount of money deducted from a paycheck for Social Security, the annual income of a sales representative, and the profit from a club car wash may be modeled by linear functions. In this section, we will look at a wide variety of information encountered in everyday life and learn how to determine if the data are linear, near-linear, or piecewise linear. We will also demonstrate how to use technology to determine the linear equation that best fits a data set. Additionally, we will show how to find linear equations to model real-life phenomena.
34
Direct Proportions
Two quantities are said to be directly proportional if the ratio of the output to the input is a constant. Quantities that are directly proportional to each other may be modeled by a linear equation of the form y 5 kx The constant k 5
y x
and is called the constant of proportionality.
EXAMPLE
Determining If a Data Set Is Directly Proportional

For the data in Table 1.8, determine whether the employees weekly income is directly proportional to the number of hours worked. If so, find a linear function that models the data.
TABLE 1.8
Hours Worked (h) 35.0 38.5 42.3

SOLUTION Calculate the ratio
I h
Income ( I ) $210.00 $231.00 $253.80
for each ordered pair.
$210 5 $6> hr 35 hours $231 5 $6> hr 38.5 hours $253.80 5 $6> hr 42.3 hours Since h 5 6 for each ordered pair, the weekly income is directly proportional to the hours worked with a constant of proportionality k 5 6. The linear function is I 5 6h
I
EXAMPLE

Determine whether the cost of filling a gas tank as given in Table 1.9 is directly proportional to the number of gallons put into the tank. If so, find a linear function that models the data.
TABLE 1.9
Number of Gallons (g) 3.50 6.28 17.34
Cost (C) $5.25 $9.41 $25.99
1.3 Linear Models C g
35
$5.25 5 1.500 dollars per gallon 3.50 gallons $9.41 < 1.498 dollars per gallon 6.28 gallons $25.99 < 1.499 dollars per gallon 17.34 gallons The ratios are close to constant, but they are different. But wait! Although gas prices are given to the tenth of a cent (e.g., $1.599), the cost of buying gasoline is rounded to the nearest cent. Could the difference in ratios be due to round-off error? Lets see. It looks as if k < 1.499. The associated linear model is C 5 1.499g How good is the model?
TABLE 1.10
Number of Gallons 3.50 6.28 17.34
Cost $5.25 $9.41 $25.99
Model Estimate $5.2465 $9.4137 $25.993
As we see in Table 1.10, if we round to the nearest cent, the model fits the data perfectly. From our personal experience purchasing gasoline, we know that the cost of filling the gas tank is directly proportional to the number of gallons we put in. Our personal experience validates the accuracy of the model.
EXAMPLE

A teachers salary may fluctuate over the course of the year. (Teaching contracts typically cover a nine-month period.) A teacher recorded his paycheck amount and the OASI (Social Security) deduction over a three-month period in Table 1.11.
TABLE 1.11
Paycheck Amount (p) $1,474.77 $2,386.17 $2,795.74
OASI Deduction (S) $91.44 $145.58 $170.98
Is the OASI deduction directly proportional to the paycheck amount? If so, find a linear function that models the data.
36
CHAPTER 1 Functions and Linear Models S p
91.44 < 0.06200 OASI dollar per paycheck dollar 1474.77 145.58 < 0.06101 OASI dollar per paycheck dollar 2386.17 170.98 < 0.06116 OASI dollar per paycheck dollar 2795.74 The ratios are not constant, so, technically, we may conclude that the data are not directly proportional. (The tax law requires that 6.2 percent of the gross pay amount reduced by the cost of medical premiums be paid into OASI. In the first check, no amount of money was paid toward the teachers medical plan. For the second two checks, the gross pay amount was reduced by the cost of the medical premiums before calculating the 6.2 percent OASI amount.)
Although the data in Example 3 are not linear, the fact that the ratios are nearly constant indicates that the data are near-linear. It looks as if k < 0.0615. A linear model for the data is S 5 0.0615p How good is the model? Lets check. S 5 0.0615( 1474.77 ) 5 90.70 S 5 0.0615( 2386.17 ) 5 146.75 S 5 0.0615( 2795.74 ) 5 171.94 Table 1.12 summarizes the results.
TABLE 1.12
OASI Deduction (S) $91.44 $145.58 $170.98
Model Estimate (M) $90.70 $146.75 $171.94
Error (S 2 M) 0.74 2 1.17 2 0.96
The error represents the difference between the actual data and the model estimate. The closer the magnitude of the error is to 0, the better the model fits the data. Is there a linear equation that better fits the data? Using technology, we can find the linear equation that best fits the data.
Linear Regression and the Line of Best Fit

Any number of lines may be used to model a near-linear set of data. We want to find the line of best fit, the line that best fits the data. To do this, we minimize the sum of the squares of the errors. Figure 1.22 shows three data points and a line used to model the data.
1.3 Linear Models y

E3 E1 E2
37
FIGURE 1.22
The vertical lines show the difference between the data output and the model estimate. The difference between the actual value and the estimated value of the individual data points (the error) is designated by E1, E2, and E3. To find the line of best fit, we square each of the errors, add the results together, and then minimize the sum using advanced calculus techniques. That is, we minimize E 5 ( E1 ) 2 1 ( E2 ) 2 1 ( E3 ) 2 . This process is called linear regression. Fortunately, the TI-83 Plus calculator has a linear regression feature that does the necessary calculations automatically.
Linear Regression 1. Press , scroll to and press ENTER twice. This will ensure that the correlation coefficient r and the coefficient of determination r 2 will appear.
2nd 0
then
DiagnosticOn,
2. Bring up the Statistics menu by pressing the STAT button.
3. Bring up the List Editor by selecting 1:Edit... and pressing ENTER .
(Continued)
38
4. Clear the lists. If there are data in the list, use the arrows to move the cursor to the list heading L1. Press the CLEAR button and press ENTER . This clears all of the list data. Repeat for each list with data. (Warning: Be sure to use CLEAR instead of DELETE . DELETE removes the entire column.) 5. Enter the numeric values of the inputs in list L1, pressing ENTER after each entry.
6. Enter the numeric values of the outputs in list L2, pressing ENTER after each entry.
7. Return to the Statistics menu by pressing the STAT button.
8. Bring up the Calculate menu by using the arrows to select CALC.
9. Calculate the linear equation of the model by selecting 4:LinReg(axb) and pressing ENTER twice. The line of best fit is y 5 0.06008x 1 2.6928 and has correlation coefficient r 5 0.9999.
1.3 Linear Models
39
The correlation coefficient is a number r between 2 1 and 1 that measures the degree to which two variables are linearly related. If the line of best fit has a positive slope, r . 0. If the line of best fit has a negative slope, r , 0. If there is a perfect linear relationship between the two variables, Z r Z 5 1. The more closely related the variables, the closer the correlation coefficient will be to 61. If the correlation coefficient is 0, the variables are not linearly related. The coefficient of determination is equal to r 2. For linear functions, the coefficient of determination is the square of the correlation coefficient. (The formulas for calculating the correlation coefficient and coefficient of determination will not be shown here but may be found in many college-level statistics textbooks.) In Example 3, the correlation coefficient was r 5 0.9999474638. This number is extremely close to 1, so the line of best fit fits the data well. The linear function that best fits the data from Example 3 is S 5 0.06008p 1 2.6928. We can readily see from Table 1.13 that the model we found using linear regression fits the data better than the first model.
TABLE 1.13
OASI Deduction (S) $91.44 $145.58 $170.98
First Model S 5 0.0615p $90.70 $146.75 $171.94
Linear Regression Model S 5 0.06008p 1 2.6928 $91.30 $146.05 $170.66
Sometimes it is useful to graph the model. One method of doing this is detailed in the following Technology Tip.
Graphing a Regression Equation 1. Press the Y button to bring up the Graphing List, immediately after calculating the regression equation.
2. Bring up the Variables menu by pressing the VARS key.
(Continued)
40
3. Bring up the Variables: Statistics menu by selecting 5:Statistics... and pressing ENTER .
4. Bring up the Equation menu by moving the cursor to EQ.
5. Select the regression equation by choosing 1:RegEQ and pressing ENTER . This pastes the regression equation into the Graphing List.
6. Graph the function by pressing the GRAPH button. You may need to adjust the viewing window to see the graph.
The graph of a set of discrete data is called a scatter plot. We can draw a scatter plot by hand, using graph paper, or we can use a calculator to draw the scatter plot.
1.3 Linear Models
41
Drawing a Scatter Plot 1. Bring up the Statistics Plot menu by pressing the 2nd button, then the Y button.
2. Open Plot1 by pressing
ENTER
3. Turn on Plot1 by moving the cursor to On and pressing ENTER . Confirm that the other menu entries are as shown.
4. Graph the scatter plot by pressing ZOOM and scrolling to 9:ZoomStat. Press ENTER . This will graph the entire scatter plot along with any functions in the Graphing List. The ZoomStat feature automatically adjusts the viewing window so that all of the data points are visible.
The graph shows the data points along with the linear model. We see that the linear model fits the data well.
42
EXAMPLE
Using the Line of Best Fit to Forecast Sales

The annual sales of McDonalds Corporation are shown in Table 1.14.
TABLE 1.14
Years Since 1990 (t) 1 2 3 4 5 6 7 8 9 10 11
Franchised Sales (S) (millions of dollars) 12,959 14,474 15,756 17,146 19,123 19,969 20,863 22,330 23,830 24,463 24,838
Source: www.mcdonalds.com.
(a) Draw a scatter plot of the data. Will a linear function fit the data well? (b) Find the equation of the linear model that best fits the data. (c) Graph the linear model simultaneously with the scatter plot. How well does the model predict the data? (d) What does the slope of the function tell us about the companys sales?
SOLUTION
(a) Figure 1.23 shows a scatter plot of the data.

26,000 Sales (in millions of dollars) 24,000 22,000 20,000 18,000 16,000 14,000 12,000 0 1 2 3 4 5 6 7 8 Years (since 1990) 9 10 11 12
FIGURE 1.23
The sales revenue increases every year, but the amount by which it increases varies from year to year. Since the rate of increase is not constant, a linear function wont fit the data perfectly. However, the data do appear to be nearlinear, so a linear model may fit the data well.
1.3 Linear Models
43
(b) Using linear regression, we determine that the linear model for sales is S 5 1233.5t 1 12,213 million dollars, where t is the number of years since the end of 1990. (c) 28,000
Sales (in millions of dollars) 26,000 24,000 22,000 20,000 18,000 16,000 14,000 12,000 0 1 2 3 4 5 6 7 8 Years (since 1990) 9 10 11 12 y = 1233.5x + 12,213
FIGURE 1.24
As shown in Figure 1.24 and Table 1.15, the linear model fits the data fairly well.
TABLE 1.15
Years Since 1990 (t) 1 2 3 4 5 6 7 8 9 10 11
Franchised Sales (S) (millions of dollars) 12,959 14,474 15,756 17,146 19,123 19,969 20,863 22,330 23,830 24,463 24,838
Estimated Sales (E) (millions of dollars) 13,447 14,680 15,914 17,147 18,381 19,614 20,848 22,081 23,315 24,548 25,782
Error (S 2 E ) (millions of dollars) 488 206 158 1 743 355 16 249 516 85 944
% Error S2E a b S 3.76% 1.42% 1.00% 0.01% 3.88% 1.78% 0.07% 1.12% 2.16% 0.35% 3.80%
Although the actual sales and the estimated sales from the model differ by as much as 944 million dollars, the model projections are within 3.88 percent of the actual values. (d) The slope of the sales function is 1233.5 million dollars per year. According to the model, we expect that, on average, the sales will increase by about 1233.5 million dollars per year. We may use this information to help us determine whether to buy stock in the company. We should be aware that past performance is no guarantee of future results. Before making a stock purchase, we would want to look at other indicators, such as earnings per share, recent news, analyst ratings, and so on.
44
Finding Linear Functions to Model Real-Life Phenomena

Not all of the data we encounter in the real world are laid out nicely in an inputoutput table. Often we have to ferret out the information needed to formulate a model. The next two examples will illustrate real-life situations in which using mathematics can give us financial leverage.
EXAMPLE
Using Linear Models to Make Business Decisions

Many companies offer their salespeople a base salary plus a commission based on the employees sales. In November 2001, a Nebraska firm posted an ad on an electronic bulletin board advertising a sales representative position paying a $36,000 base salary plus commission. (Source: www.dice.com.) Based on her past sales experience, a saleswoman estimates that she can generate $600,000 in sales annually. To maintain her standard of living, she needs to earn a total of $63,000 annually. (a) What commission rate (as a percentage of sales) must the saleswoman earn to maintain her standard of living? (b) Write the equation for her annual income as a function of her annual sales dollar volume. (c) If she wants to increase her annual income to $75,000, by how much will she have to increase her annual sales? (d) If she wants to increase her annual income to $75,000 without increasing her sales volume, what commission rate must she receive? (e) How could the answers to the previous questions help her during salary negotiations?
SOLUTION
(a) We have I 5 ms 1 b where I is the annual income (in thousands of dollars) and s is the annual sales (in thousands of dollars). If the saleswoman doesnt sell anything (assuming she doesnt get fired), she will still earn $36,000, so (0, 36) is the y-intercept. Thus I 5 ms 1 36 When s 5 600, I 5 63, so 63 5 m ? 600 1 36 27 5 600m m 5 0.045 She must earn a 4.5 percent commission to maintain her standard of living. (b) The equation for her annual income is I 5 0.045s 1 36 thousand dollars, where s is the sales volume (in thousands of dollars). (c) Setting I 5 75 and solving for s, we get 75 5 0.045s 1 36 39 5 0.045s s 5 866.667
1.3 Linear Models
45
She must increase her annual sales from $600,000 to $866,667 to increase her annual income to $75,000. (d) Letting m be the variable and setting I equal to 75, we get 75 5 m ? 600 1 36 39 5 600m m 5 0.065 So to increase her income to $75,000 without increasing sales, she must earn a 6.5 percent commission. (e) We know that her minimum acceptable commission rate is 4.5 percent. To increase her salary without having to increase her workload, she will have to convince her employer to increase her commission rate to 6.5 percent. If her employer is unwilling to raise the commission rate, she will have to bring in an additional $266,667 in sales to earn $75,000. This information can give her leverage in salary negotiations.
EXAMPLE
Using Linear Models to Make Business Decisions

In 2001, BeautiControl Cosmetics sold a new consultant a demonstration package for $119.82. (Source: BeautiControl Cosmetics consultant.) The consultants director told her that the average revenue from each in-home party is $250. The consultants cost for the cosmetics is 50 percent of the sales price. (a) Find the linear model for the consultants cost as a function of the dollar volume of cosmetics sold. (b) Find the linear model for the consultants revenue as a function of the dollar volume of cosmetics sold. (c) Find the linear model for the consultants profit as a function of the dollar volume of cosmetics sold. (d) How many dollars of cosmetics must the consultant sell before she begins to make a profit? (e) How many in-home parties must the consultant hold in order to begin making a profit?
SOLUTION
(a) Let x be the sales volume in dollars. Then the cost function C is C( x ) 5 0.5x 1 119.82 The cost of the demonstration package, $119.82, is the fixed cost. The consultants variable cost, 0.5x dollars, is dependent upon her sales dollar volume. (b) The revenue function R is R( x ) 5 x The consultants revenue is equal to her sales dollar volume. (c) The profit function is the difference between the revenue and cost functions. P( x ) 5 R( x ) 2 C( x ) 5 x 2 ( 0.5x 1 119.82 ) 5 0.5x 2 119.82
46
(d) We need to find the break-even point (the sales dollar volume that makes the profit equal to zero). Setting the profit equal to zero, we get 0 5 0.5x 2 119.82 119.82 5 0.5x x 5 239.64 So, to break even, the consultant must sell $239.64 worth of cosmetics. (e) Dividing the sales dollar volume needed to break even by the amount of revenue per party, we get 239.64 dollars 250.00
dollars party
5 0.95856 parties
It doesnt make sense to talk about fractions of a party, so we round up. The consultant should begin to make a profit at her first party.
Piecewise Linear Models

A function of the form f (x) 5 e ax 1 b x # c dx 1 g x . c
is called a piecewise linear function. The rule used to calculate f ( x ) depends upon the value of x. In this case, if x is less than or equal to c, the rule is f ( x ) 5 ax 1 b . If x is greater than c, the rule is f ( x ) 5 dx 1 g . A common piecewise linear function is the absolute value function f ( x ) 5 Z x Z . This function makes negative numbers positive and leaves positive numbers unchanged. In practical terms, it erases the negative sign of any negative number. The absolute value function is formally defined as a piecewise linear function. f (x) 5 Zx Z 5 e x x$0 2x x , 0
What is Z 2 5 Z ? Since 2 5 , 0, we apply the second rule of the piecewise function. Z25 Z 5 2(25) 55 Piecewise linear functions are very common in a variety of real-life situations, such as the ones given in Examples 7 and 8.
1.3 Linear Models
47
EXAMPLE
Using a Piecewise Linear Model to Forecast Monthly Payments

An orthodontist and the parents of a girl scheduled to receive braces agreed upon the following payment plan: a $550 down payment when treatment begins and $100 per month until the treatment is paid in full. The total cost for the treatment is $1250, and the treatment is expected to take 18 months to complete. (Source:
Authors personal payment plan.)
(a) How many months after treatment begins will the treatment be paid in full? (b) Write the equation for the balance due as a function of the number of months since treatment began.
SOLUTION
(a) The patients parents pay $550 when treatment begins, so $700 remains for the balance to be paid in full. Since $100 is paid each month, the treatment will be paid in full 7 months later. (b) B( t ) 5 e 2 100t 1 700 0 0#t#7 7 , t # 18
where B(t) is the balance of the account t months after treatment begins. Since payments are made monthly, the function is defined for whole-number values of t. We initially use a linear function for B( t ) , since the account balance starts at $700 and is reduced by a constant $100 per month. However, after 7 months, the account balance has reached $0. If we continued to use the linear function, the balance would end up being a negative number, which doesnt make sense in the context of the problem. Therefore, no further payments are made after the seventh month, and the account balance remains at $0 from the end of the seventh month through the eighteenth month.
EXAMPLE
Using a Piecewise Linear Model to Make Spending Decisions

Experience Music Project, an interactive music museum in Seattle, charges adult visitors a $19.95 admission fee. Groups of 15 or more may enter for $14.50 per person. (Source: www.emplive.com.) (a) Draw a scatter plot for group cost as a function of group size. (b) Write the total cost of admission as a function of the number of people in a group. Confirm that the function and the scatter plot are in agreement. (c) Calculate the cost of admitting a group of 14 people and the cost of admitting a group of 15 people. (d) Determine the largest group whose admission fee would still remain below the cost of a 14-person group.
48
SOLUTION
(a)
350.00 300.00 250.00 Dollars 200.00 150.00 100.00 50.00 0.00 0 2 4 6 8 10 12 Group size (people) 14 16 18 20
FIGURE 1.25
(b) C( x ) 5 e 19.95x x , 15 14.5x x $ 15
where x is the number of people in the group and C is the total cost in dollars. (c) C( 14 ) 5 19.95( 14 ) 5 $279.30 is the cost for a group of 14 people. C( 15 ) 5 14.5( 15 ) 5 $217.50 is the cost for a group of 15 people. (d) From the scatter plot, it looks as if the maximum number of people whose admission fee would still remain below the 14-person group price is 19 people. We can confirm the result algebraically. 279.30 5 14.5x x < 19.26 Since it doesnt make sense to talk about 0.26 of a person, we round down to 19. C( 19 ) 5 14.5( 19 ) 5 275.50 The cost for a 19-person group is $275.50. This is slightly less than the $279.30 cost for a 14-person group.
Determining Which Modeling Method to Use

Although linear regression is a powerful tool for creating linear models, many linear models can be constructed algebraically, as was shown in Examples 5 through 8. It is good practice to first attempt to find the linear model algebraically. A common pitfall of beginning students is to use the calculator to find the linear model when it is actually easier to find the model algebraically. Whenever the data set contains exactly two points, we should find the linear model algebraically. In addition, there are other key phrases that alert us to the fact that a linear model may be constructed algebraically. Table 1.16 details how to interpret the mathematical meanings of some commonly occurring phrases. By applying the techniques demonstrated in the table, you will increase your comfort level in working with real-world problems.
1.3 Linear Models
49
TABLE 1.16
Phrase constant rate of change increasing at a rate of 20 people per year tickets cost $37 per person sales increase by 100 units for every $1 decrease in price slope
Mathematical Meaning people . year
The slope of the linear function is 20 The slope is 37 The slope is m5 100 units 2 1 dollar dollars . person
5 2 100 units per dollar . initial price of $1.25 There are 350 chairs today. The number of chairs is increasing by 10 chairs per month. The price is $2.25 and is decreasing at a constant rate of $0.02 per day. The y-intercept is 1.25. The y-intercept is 350 and the slope is 10. The y-intercept is 2.25 and the slope is 20.02.
Drawing a Scatter Plot 1. Enter a table of data.
2. Click the left mouse button and drag to select the data to plot. Then click the icon on the toolbar. This will open the Chart Wizard.
(Continued)
50
3. Select XY(Scatter). Then click Next twice.
4. If desired, type labels for the axes and a title for the graph. Then click Next.
5. Specify whether you want the scatter plot to appear on its own sheet or in the current sheet. Then click Finish.
6. Verify that the scatter plot agrees with the table data.
12 10 8 6 4 2 0 0 1 2 3 4 5 6 7 8
1.3 Linear Models
51
Linear Regression 1. Draw a scatter plot of a data set.

50 40 30 20 10 0 10 0 2 4 6 8 10 12
2. Right-click on any point of the scatter plot to open a drop-down menu. From the drop-down menu, select Add Trendline to open the Add Trendline dialog box. Then select Linear.
3. Choose the Options tab and select Display equation on chart and Display R-squared value on chart. Click OK.
4. The graph and the equation of the line of best fit are displayed. The coefficient of determination r 2 is also displayed.
50 40 30 20 10 0 10 0 2 4
y = 5x 4 R2 = 1
10
12
52
1.3 Summary
In this section, you looked at a wide variety of information encountered in everyday life, and you learned how to model linear, near-linear, and piecewise linear data sets. You also learned how to use technology to determine the line of best fit when modeling real-life phenomena.

1. What mistake did the student make in calculating the slope of the line between ( 2, 4 ) and ( 5, 6 ) ? m5 624 225 2 5 23 2 52 3 4. If two quantities are directly proportional, what will be the y-intercept of the line that relates the two quantities? 5. If a persons weekly wage (in dollars) is a linear function of the number of hours worked, what are the units of the slope of the linear function that relates weekly wage to hours? 6. What is a piecewise linear function? 7. Evaluate f ( x ) 5 e 5x 2 9 10x 1 1 x,4 at x 5 4. x$4
2. What mistake did the student make in calculating the slope of the line between ( 6, c ) and ( 0, 6 ) ? m5 c26 620 c2 6 5 6 5c
8. The linear function y 5 mx 1 b may be used to represent the sentence Videotape sales (in millions of tapes) are a linear function of the price per videotape (in dollars). Explain the meanings of y, m, x, and b in the context of videotape sales.
3. What mistake did the student make in calculating the slope of the line between ( 2 2, 5 ) and ( 2 2, 14 ) ? m5 14 2 5 22 2 (22) 9 5 0 50
1.3 Exercises
In Exercises 15, do the following: (a) Draw the scatter plot. (b) Find the equation of the line of best fit. (c) Interpret the meaning of the slope and the y-intercept of the model. (d) Explain why you do or do not believe that the model would be a useful tool for businesses and/or consumers.
1.3 Linear Models
53
1.
Share Price
4. Harbor Capital Appreciation Fund Share Price on Last Day of Month $48.16 $41.50 $35.58
Per Capita Personal Income
Per Capita Personal IncomeUtah Years Since 1993 (t) 0 1 2 3 4 5 6 Personal Income (P) (dollars) 16,830 17,638 18,508 19,514 20,613 21,594 22,305 23,436
Month in 2000 10 11 12
Source: www.quicken.com.
2.
Share Price
Month in 2001 3 6 9
CREF Stock Retirement Annuity Share Price on Last Day of Month $164.5760 $173.1588 $146.4102 5.
Source: Bureau of Economic Analysis (www.bea.gov).

Per Capita Income Ranking
Source: TIAA-CREF statement.
3.
Years Since 1995

University Enrollment
North Carolina Per Capita Income Ranking (out of 50 states) 42 34 32 28 28
0 1 2 3 4
Washington State Public University Enrollment Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10 Students (S) 81,401 81,882 83,052 84,713 85,523 86,080 87,309 89,365 90,189 91,543 92,821
Source: http://www.census.gov/statab.
In Exercises 616, formulate the linear or piecewise linear model for the real-life scenario. If a linear model represents a directly proportional relationship between the independent and dependent variables, so state. 6.
Personal Income In November 2001, Hall Kinion advertised an Account Executive position in Savannah, Georgia, for a salesperson with experience selling Internet services. The job advertised a $35K base salary and $24K to $48K in commissions in the first year. (Source: www.dice.com.) (a) Assuming a 20 percent commission rate, how many dollars in sales would the salesperson have to generate to earn $64K annually?
Source: Washington State Higher Education Coordinating Board.
54
(b) If you were in the sales field, why would this type of information be valuable to you? 7.
Solid Waste Disposal King County charges $82.50 per ton to dispose of garbage at area transfer stations. A minimum disposal fee of $13.72 per entry is charged for vehicles disposing of trash. In addition to the disposal fee, a moderate risk waste (MRW) fee of $2.61 per ton is assessed. A minimum MRW fee of $1 is charged on all transactions. Tax is charged on the combined price of the disposal and moderate risk waste fees. The tax rate is 3.6 percent. The weight of the trash is rounded to the nearest 20 pounds before the fees are calculated.
10.
Solid Waste Disposal The Enumclaw Transfer/Recycling Station accepts clean wood (stumps, branches, etc.) in addition to household garbage. The station charges $75 per ton with a minimum fee of $12.75 per entry for vehicles disposing of clean wood. (Source:
www.dnr.metrokc.gov.)
(Source: www.dnr.metrokc.gov.)
(a) How many pounds of clean wood could be dropped off at the station without exceeding the $12.75 minimum fee? (b) Write the equation for the disposal cost as a function of the number of pounds of clean wood. (c) If you ran a landscaping business, how could you use the results of parts (a) and (b)? 11.
Used Car Value In 2001, the average retail price of a 2000 Toyota Land Cruiser 4-Wheel Drive was $43,650. The average retail price of a 1995 Toyota Land Cruiser 4-Wheel Drive was $21,125. (Source:
(a) Write the equation for disposal cost as a function of the number of pounds of refuse. (b) How much will it cost to drop off 230 pounds of trash? (c) How much will it cost to drop off 513 pounds of trash? (d) If you ran a construction company, how could you use the results of this exercise? 8.
Cellular Phone Plan In 2002, Qwest offered a 150-minute cellular plan including free long distance and free roaming to its Seattle-area customers for $29.99 per month. Additional time cost $0.35 per minute or portion of a minute. (Source: Qwest customer monthly
www.nadaguides.com.)
statement.)
(a) Find a linear model for the value of a Toyota Land Cruiser in 2001 as a function of its production year. (b) Use your linear model to predict the value of a 1992 Toyota Land Cruiser in 2001. (c) A 1992 Toyota Land Cruiser had an average retail price of $14,325 in 2001. How good was your linear model at predicting the value of the vehicle? 12.
Nutrition The Recommended Daily
(a) Write the cellular phone cost equation as a function of the number of minutes used. (b) How could you use the cost equation if you were enrolled in the plan? 9.
Cellular Phone Plan Sprint PCS offered a service plan to its Seattle-area customers for $29.99 per month that included 200 anytime minutes. The plan also included free long distance. Additional minutes or portions of a minute were $0.40 each. (Source:
Allowance (RDA) of fat for a person on a 2,000-calorie-per-day diet is less than 65 grams. A McDonalds Big Mac sandwich contains 34 grams of fat. A large order of French fries contains 26 grams of fat. (Source:
www.mcdonalds.com.)
www1.sprintpcs.com.)
(a) Determine the cost equation for the Sprint PCS plan as a function of the number of anytime minutes used. (b) Is the Sprint plan or the Qwest plan (from Exercise 8) a better deal for a consumer who uses 300 anytime minutes monthly? Explain.
(a) Write the equation for fat grams consumed as a function of large orders of French fries. (b) Write the equation for fat grams consumed as a function of Big Macs. (c) How many large orders of French fries can you eat without exceeding the RDA of fat? (d) How many Big Macs can you eat without exceeding the RDA of fat? (e) How many combination meals (Big Mac and French fries) can you eat without exceeding the RDA for fat?
1.3 Linear Models
55
13.
Allowance (RDA) of fat for a person on a 2,500-calorie-per-day diet is less than 80 grams. A McSalad Shaker Chef Salad contains 8 grams of fat. A package (44.4 ml) of ranch salad dressing contains 18 grams of fat. (Source:
www.mcdonalds.com.)
exactly one segment per airplane. That is, if the plane stops, the passenger changes planes.) 16.
Airline Ticket Cost The total cost of an airline ticket includes the published fare, a federal flight segment tax, a federal security fee, and an airport Passenger Facility Charge (PFC). The flight segment tax is $3 on each segment of the itinerary. (A segment is a takeoff and landing.) The PFC varies from airport to airport; it ranges from $3 to $4.50, and it is imposed by an airport on enplaning passengers. When a passenger is departing from or connecting at any airport, that airports PFC will apply in addition to the fare. Likewise, a $2.50 security fee is imposed on all enplaning passengers. A Southwest Airlines refundable round-trip airline ticket from Seattle to Phoenix is advertised at $444. (Source: www.southwest.com.) Write the maximum total cost of a round-trip airline ticket from Seattle to Phoenix as a function of the number of segments.
(a) Write the equation for fat grams as a function of Chef Salads. (b) How many salads can you eat and remain below the RDA of fat? (c) If you use one package of ranch salad dressing, how many salads can you eat and remain below the RDA of fat? (d) The RDA of fat for a person on a 2000calorie-per-day diet is 65 grams. If a person on this diet uses one package of ranch salad dressing, how many salads can the person eat and remain below the RDA of fat? 14. The Picture People photography studio in Tacoma, Washington, charges $40 for one of its promotional photo packages. The total purchase price for the promotional package, including tax, is $43.20. (Source: Picture People sales receipt.) (a) Write the equation for Tacoma sales tax as a function of the pretax purchase price. (b) A family has $60 cash to pay for pictures. What is the maximum pretax purchase price that the family can afford?
Sales Tax Airline Ticket Cost The total cost of an airline ticket includes the published fare, a federal flight segment tax, a federal security fee, and an airport Passenger Facility Charge (PFC). The flight segment tax is $3 on each segment of the itinerary. (A segment is a takeoff and landing.) The PFC varies from airport to airport; it ranges from $3 to $4.50, and it is imposed by an airport on enplaning passengers. When a passenger is departing from or connecting at any airport, that airports PFC will apply in addition to the fare. Likewise, a $2.50 security fee is imposed on all enplaning passengers. A Southwest Airlines refundable airline ticket from Seattle to Phoenix is advertised at $222. (Source: www.southwest.com.) Write the maximum total cost of a Southwest airline ticket from Seattle to Phoenix as a function of the number of segments. (Assume that a passenger travels
15.
17. A companys expenses are typically categorized as fixed costs or variable costs. Fixed costs are costs that the company must pay regardless of production levels, whereas variable costs (i.e. materials, labor) depend on the number of items produced. Under what conditions will a companys expenses be directly proportional to the number of items produced? 18. A linear regression model for a data set gives a correlation coefficient of r 5 1. Explain what this means. 19. Based on 10 years of sales data, a company creates a linear regression model to forecast future sales. The correlation coefficient for the model is r 5 0.9998, so the companys sales team is confident that the models projection of $450,000 in sales for the coming year is accurate. However, over the next year, the company brings in only $240,000 in sales instead of the $450,000 forecast by the model. Explain how this could have occurred, given that the correlation coefficient for the model was so close to 1. 20. What are some key verbal phrases that indicate that a linear model may be appropriate for a certain situation?
56
21. When is it appropriate to use a piecewise linear function to model a situation?
27. A linear model has a correlation coefficient of r 5 2 1. Does the fact that the linear model passes through the point ( 0, 0 ) imply that the output of the original data is directly proportional to the input of the original data? Explain. 28. Use your calculator to find the line of best fit for the data in the table. x 0 1 2 3 y 1 1 1 1
22. The standard form of a linear function in three dimensions is given by ax 1 by 1 cz 5 d , where x, y, and z are variables and a, b, c, and d are constants. A business determines that the revenue generated by selling x cups of coffee, y bagels, and z muffins is given by R 5 ax 1 by 1 cz What is the practical meaning of the constants a, b, and c? 23. A business determines that the cost of producing x cups of coffee, y bagels, and z muffins is given by C 5 ax 1 by 1 cz 1 d What is the practical meaning of the constants a, b, c, and d? 24. The equation of the line of best fit for the data in the table is y 5 0 . x 0 1 2 3 4 5 6 y 21 1 0 0 1 21 0
Why do you think the correlation coefficient is undefined? Does this mean that the model doesnt fit the data? Justify your answer. 29. Find the piecewise linear function that best fits the data in the table. x 0 1 2 3 4 5 6 y 2 4 6 7 8 9 10
30. Find the piecewise linear function that best fits the data in the table. x 2 2.0 1.0 2.5 4.0 8.0 10.0 y 14 8 5 2 4 5
This model passes through three of the seven points shown in the table. Is this model a good fit for the data? Justify your answer. 25. Are production costs directly proportional to the number of items produced for most businesses? Explain. 26. Does the fact that a linear model passes through the point ( 0, 0 ) imply that the dependent variable of the original data is directly proportional to the independent variable of the original data? Explain.
Chapter 1
Study Sheet
57

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. How do you determine if a data set, a graph, an equation, or a verbal expression represents a function? Your explanation should apply to each of the function representations listed. (1.1) 2. What are the relationships among the terms independent variable, dependent variable, input, output, domain, and range? (1.1) 3. How do you determine if a data set, a graph, an equation, or a verbal expression represents a linear function? (1.2) 4. What does it mean to say that a function has a constant rate of change? (1.2) 5. How are functions that have a constant rate of change related to each other? (1.2) 6. What are the benefits of each of the forms of a linear function? (1.2) 7. What is an efficient way to graph a linear function whose equation is given in standard form? In slope-intercept form? In point-slope form? (1.2) 8. How do you determine the equation of a linear function from two points? (1.3) 9. How do you determine the equation of the line of best fit for a data set? (1.3) 10. What does it mean for a line to be the line of best fit for a data set? (1.3) You should also be familiar with the following definitions, procedures and tests, and formulas that were emphasized in this chapter.
Definitions

Function: A rule that associates each input with exactly one output. Domain: The set of all possible values of the independent variable of a function. Range: The set of all possible values of the dependent variable of a function. Graph of a Linear Function: The graph of a straight line. Slope of a Line: The change in output that occurs when the input is increased by one unit. y-intercept: A point on the graph where the function intersects the y-axis. x-intercept: A point on the graph where the function intersects the x-axis. Piecewise Linear Function: A function that is defined by two or more linear functions over specific domain intervals.
Procedures and Tests
Vertical Line Test: If every vertical line drawn on a graph intersects the graph in at most one place, then the graph is the graph of a function. Direct Proportionality Test: If values ( x, y ) , then y is directly proportional to x.
y x
5 k for some constant k and for all
58
Formulas
Slope: m 5
y2 2 y1 x2 2 x1
Equation of a Vertical Line: x 5 a Equation of a Horizontal Line: y 5 b Slope-Intercept Form of a Line: y 5 mx 1 b Standard Form of a Line: Ax 1 By 5 C Point-Slope Form of a Line: y 2 y1 5 m( x 2 x1 )

Section 1.1 In Exercises 13, calculate the value of the function at the designated input and interpret the result.
1. C( x ) 5 49.95x at x 5 2 , where C is the cost (in dollars) of buying x pairs of shoes. 2. C 5 9.95x 1 1200 at x 5 400 , where C is the cost of making x pairs of shoes. 3. H 5 2 16t 2 1 100 at t 5 2, where H is the height of a cliff diver above the water t seconds after he jumps from a 100-foot cliff. In Exercises 47, determine the domain of the function. 4. f ( x ) 5 4x 3 2 10x 5. f ( p ) 5 p 2 9p 1 15
2
at the end of the day and t is the number of days after December 16, 2001.
3.40 Stock price (in dollars) 3.20 3.00 2.80 2.60 2.40 2.20 2.00 0 1 2 3 4 5 Number of days after 12/16/01 6
Source: Click2Learn Corporation.
9.
Stock Price Find P at t 5 2, where P is
a computer companys stock price at the end of the day, t days after December 16, 2001.
22.00 21.80 21.60 21.40 21.20 21.00 20.80 20.60 20.40 20.20 20.00 0 1 2 3 4 5 Number of days after 12/16/01 6
6. C 5 7. C 5
r2 2 1 2r 1 1 r2 1 1 r2 2 1
Stock price (in dollars)
In Exercises 89, interpret the meaning of the indicated point on the graph. 8.
Stock Price
Find P at t 5 4, where P is the stock price of an e-learning company
Source: Apple Computer Corporation.
Chapter 1
Review Exercises
59
In Exercises 1011, determine whether the graphs represent functions by applying the Vertical Line Test. 10.
y 20 15 10 5 4 3 2 15 10 15 20 1 2 3 4 x
A McDonalds Big N Tasty sandwich contains 32 grams of fat. A super size order of French fries contains 29 grams of fat. (Source:
www.mcdonalds.com.)
11.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
(a) Write the equation for fat grams consumed as a function of super size orders of French fries eaten. (b) Write the equation for fat grams consumed as a function of Big N Tasty sandwiches eaten. (c) How many combination meals (Big N Tasty sandwich and super size order of French fries) can you eat without exceeding the RDA of fat? 22.
Solid Waste Disposal The Enumclaw Transfer/Recycling Station accepts clean wood (stumps, branches, etc.) in addition to household garbage. The station charges $75 per ton with a minimum fee of $12.75 per entry for vehicles disposing of clean wood. (Source:
www.dnr.metrokc.gov.)
Section 1.2 In Exercises 1215, calculate the slope of the linear function passing through the points.
12. (2, 9) and (4, 3) 13. ( 2 3, 2 4 ) and ( 0, 2 2 ) 14. (1.3, 3.5) and (3.5, 1.3) 15. (7, 11) and (8, 0) In Exercises 1618, find the x-intercept and the y-intercept of the linear function. 16. y 5 2 5x 1 10 18. y 5 2x 2 12 In Exercises 1920, write the equation of the linear function passing through the points in slope-intercept form and in standard form. 19. (2, 5) and (4, 3) 20. ( 2 3, 4 ) and ( 0, 2 2 ) 17. y 5 2 3x 1 18 23.
(a) Write a piecewise linear function to model the cost of disposing of x pounds of clean wood. (Assume that the entire amount of wood is delivered in a single load.) (b) How much does it cost to dispose of 300 pounds of clean wood? 700 pounds?
Used Car Value In 2001, the average retail price of a 1998 Mercedes-Benz Roadster two-door SL500 was $51,400. The average retail price of a 2000 Mercedes-Benz Roadster two-door SL500 was $66,025. (Source:
www.nadaguides.com.)
(a) Find a linear model for the value of a Mercedes-Benz Roadster two-door SL500 in 2001 as a function of its production year. (b) Use your linear model to predict the 2001 value of a 1999 Mercedes-Benz Roadster two-door SL500. (c) A 1999 Mercedes-Benz Roadster two-door SL500 had an average retail price of $58,500 in 2001. How good was your linear model at predicting the value of the vehicle?
Section 1.3 In Exercises 2123, find a model for the data and answer the given questions.
21.
Allowance (RDA) of fat for a person on a 2,000-calorie-per-day diet is less than 65 grams.
P R O J E C T
What to do
1. Find a set of at least six data points from an area of personal interest. 2. Draw a scatter plot of the data and explain why you do or do not believe that a linear model would fit the data well. 3. Find the equation of the line of best fit for the data. 4. Interpret the physical meaning of the slope and y-intercept of the model. 5. Use the model to predict the value of the function at an unknown point and explain why you do or do not think the prediction is accurate. 6. Explain how a consumer and/or businessperson could benefit from the model. 7. Present your findings to the class and defend your conclusions.
Where to look for data

Box Office Guru
www.boxofficeguru.com Look at historical data on movie revenues.
Nutri-Facts
www.nutri-facts.com Compare the nutritional content of common foods based on serving size.
Quantitative Environmental Learning Project

www.seattlecentral.org/qelp Look at environmental information in easy-to-access charts and tables.
U.S. Census Bureau

www.census.gov Look at data on U.S. residents ranging from Internet usage to family size.
Local Gas Station or Supermarket

Track an items price daily for a week.
School Registrar
Ask for historical tuition data.
Utility Bills
Look at electricity, water, or gas usage.
Employee Pay Statements

Look at take-home pay or taxes.
Body Weight Scale

Track your body weight for a week.
60
Chapter
Nonlinear Models
M
athematical functions are commonly used to model real-world data. Although every model has its limitations, models are often used to forecast expected results. Selecting which mathematical model to use is relatively easy once you become familiar with the basic types of mathematical functions. Remarkably, these basic functions may be used to effectively model many real-world data sets. For example, based on data from 1993 to 2002, the sales revenue from Starbucks stores may be effectively modeled by a quadratic function.
2.1 Quadratic Function Models

Use nonlinear functions to model reallife phenomena Interpret the meaning of mathematical models in their real-world context Model real-life data with quadratic functions
2.2 Higher-Order Polynomial Function Models

Use nonlinear functions to model reallife phenomena Interpret the meaning of mathematical models in their real-world context Model real-life data with higher-order polynomial functions Use nonlinear functions to model reallife phenomena Interpret the meaning of mathematical models in their real-world context Graph exponential functions Find the equation of an exponential function from a table Model real-life data with exponential functions Use nonlinear functions to model reallife phenomena Interpret the meaning of mathematical models in their real-world context Graph logarithmic functions Model real-life data with logarithmic functions Apply the rules of logarithms to simplify logarithmic expressions Solve logarithmic equations Use critical thinking skills in selecting a mathematical model
2.3 Exponential Function Models

2.4 Logarithmic Function Models

2.5 Choosing a Mathematical Model
61
62
CHAPTER 2 Nonlinear Models
Use nonlinear functions to model real-life phenomena Interpret the meaning of mathematical models in their real-world context Model real-life data with quadratic functions
Many college-bound high school students enroll in Advanced Placement Program (AP) courses to earn college credit while still in high school. The College Board offers 35 courses in 19 disciplines ranging from Studio Art: 3D Design to Calculus AB. (Source: The College Board.) The number of students taking the Calculus AB test has increased dramatically since 1969, as shown in Figure 2.1.
160,000 140,000 120,000 AP Calculus exams 100,000 80,000 60,000 40,000 20,000
FIGURE 2.1
How many students will take the test in 2005? Nobody knows; however, using a mathematical model, we can predict what may happen. In this section, we will discuss the identifying features of quadratic functions. We will then demonstrate how to use algebraic methods and quadratic regression to model data sets whose graphs open upward or downward over their domain.
A polynomial is a function that is the sum of terms of the form ax n, where a is a real number and n is a nonnegative integer. For example, each of the following functions is a polynomial. f ( x ) 5 6x4 2 3x3 1 5x2 2 2x 1 9 g( x ) 5 4x2 1 2x h( x ) 5 2 1.3x 1 9.7 j( x ) 5 x2 2 2x 1 6 May a constant term like 9 be written in the form axn? Yes! Since x 0 5 1 for nonzero x, 9x0 5 9. Consequently, a constant function s( x ) 5 9 is also a polynomial. The degree of a polynomial is the value of its largest exponent. For example, the degree of the polynomial f ( x ) 5 6x 4 2 3x 3 1 5x 2 2 2x 1 9 is 4,
19 68 19 70 19 72 19 74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02
Year
63
since 4 is the largest exponent. Since the equation of any line may be written as y 5 ax 1 1 b, lines are polynomials of degree 1. Weve worked extensively with first-degree polynomials (lines) in the preceding chapter. In this section, we are interested in polynomials of degree 2. Polynomials of degree 2 are called quadratic functions.
QUADRATIC FUNCTION
A polynomial function of the form f ( x ) 5 ax2 1 bx 1 c with a 2 0 is called a quadratic function. The graph of a quadratic function is a parabola.
y 10 9 8 7 6 5 4 3 2 1 1 0
y = x2 2x + 3
FIGURE 2.2
y 10 9 8 7 6 5 4 3 2 1 1 0 1
When a parabola opens upward, we say that it is concave up. When the parabola opens downward, we say that it is concave down. The steepness of the sides of the parabola and its concavity are controlled by the value of a, the coefficient on the x 2 term in its equation. If a . 0, the graph is concave up. If a , 0, the graph is concave down. As the magnitude of a increases, the steepness of the graph increases. (For a . 0, the magnitude of a is a. For a , 0, the magnitude of a is 2 a.) Consider the graphs in Figures 2.2 and 2.3. These graphs have the same values for b and c ( b 5 2 2 and c 5 3 ) but differing values of a. In Figure 2.2, since a is the coefficient on the x 2 term, a 5 1. The magnitude of a is 1. Since a . 0, the graph is concave up. In Figure 2.3, since a is the coefficient on the x 2 term, a 5 2. The magnitude of a is 2. Increasing the magnitude of a increased the graphs steepness. Consider the graphs in Figures 2.4 and 2.5. These graphs have the same values for b and c ( b 5 4 and c 5 0 ) but differing values of a.
y 4 3 2 1 1 1 2 3 4 1 2 3 4 x
y = 2 x 2 2x + 3
y = x 2 + 4x
FIGURE 2.4
FIGURE 2.3
In Figure 2.4, since a is the coefficient on the x 2 term, a 5 2 1. The magnitude of a is 1. Since a , 0, the graph is concave down.
64
CHAPTER 2 Nonlinear Models y 4 3 2 1 1 1 2 3 4 1 2 3 4 x y = 2 x 2 + 4x
FIGURE 2.5
y decreasing
smallest y-value
y increasing
vertex x
FIGURE 2.6a Concave Up Parabola

y largest y-value y increasing vertex y decreasing
In Figure 2.5, since a is the coefficient on the x2 term, a 5 2 2. The magnitude of a is 2. Increasing the magnitude of a from 1 to 2 increased the graphs steepness. The coefficient b in the equation affects the horizontal and vertical placement of the parabola. The constant term c in the equation indicates that the point ( 0, c ) is the y-intercept of the graph. Recall that the graph of a function is said to be increasing if the value of y gets bigger as the value of x increases. Similarly, the graph of a function is said to be decreasing if the value of y gets smaller as the value of x increases. The vertex of a parabola is the point on the graph of a quadratic function where the curve changes from decreasing to increasing (or vice versa). The minimum y value of a concave up parabola occurs at the vertex (see Figure 2.6a). Similarly, the maximum y value of a concave down parabola occurs at the vertex (see Figure 2.6b). For parabolas with x-intercepts, the x coordinate of the vertex always lies halfway between the x-intercepts. Recall that as a result of the quadratic formula, we know that the x coordinate of the vertex is x 5 vertex is obtained by evaluating the quadratic function at this x value. For example, for the quadratic function y 5 x2 2 2x 1 3 , we know that a 5 1 and b 5 2 2. The x coordinate of the vertex is x5 5 2b 2a 2(22) 2( 1 ) 2 2
2b . 2a
The y coordinate of the
51
FIGURE 2.6b Concave Down Parabola
Evaluating the function at x 5 1 yields y 5 x 2 2 2x 1 3 5 ( 1 ) 2 2 2( 1 ) 1 3 512213 52 Therefore, the vertex of the function is ( 1, 2 ) .
65
Parabolas are symmetrical. That is, if we draw a vertical line through the vertex of the parabola, the portion of the graph on the left of the line is the mirror image of the portion of the graph on the right of the line. The line is referred to as the axis of symmetry (see Figure 2.7).
y axis of symmetry
vertex x
FIGURE 2.7
Since the axis of symmetry is a vertical line passing through the vertex, the equation of the axis of symmetry is x 5
2b . 2a
EXAMPLE
Describing the Graph of a Quadratic Function from Its Equation

Determine the concavity, the y-intercept, and the vertex of the quadratic function y 5 3x 2 1 6x 2 1 .
SOLUTION We have a 5 3, b 5 6, and c 5 2 1. Since a . 0, the parabola is concave up. Since c 5 2 1, the y-intercept is ( 0, 2 1 ) . The x coordinate of the vertex is given by
x5 5 5
2b 2a 26 2( 3 ) 26 6
5 21 The y coordinate of the vertex is obtained by evaluating the function at x 5 2 1 . y 5 3x 2 1 6x 2 1 5 3( 2 1 ) 2 1 6( 2 1 ) 2 1 532621 5 24 The vertex of the parabola is ( 2 1, 2 4 ) .
66
EXAMPLE
Determining the Equation of a Parabola from Its Graph

Determine the equation of the parabola shown in Figure 2.8.
y 20 18 16 14 12 10 8 6 4 2 0 0 1 2 3 4 5 6
FIGURE 2.8
SOLUTION The y-intercept is ( 0, 20 ) , so c 5 20. The vertex is ( 3, 2 ) . Since
the x coordinate of the vertex is
2b , 2a
we know that
2b 53 2a 2 b 5 6a b 5 2 6a Since f ( x ) 5 ax 2 1 bx 1 c , we have f ( x ) 5 ax 2 1 ( 2 6a ) x 1 ( 20 ) 5 ax 2 6ax 1 20

2
Since b 5 2 6a
The vertex is ( 3, 2 ) , so f ( 3 ) 5 2 . Therefore, f ( x ) 5 ax 2 2 6ax 1 20 2 5 a( 3 ) 2 2 6a( 3 ) 1 20 2 5 9a 2 18a 1 20 2 18 5 2 9a a52 Since b 5 2 6a, b 5 2 12. The equation of the parabola is f ( x ) 5 2x 2 2 12x 1 20. We can check our work by substituting in a different point from the graph. The parabola passes through the point ( 4, 4 ) . Consequently, this point should satisfy the equation f ( x ) 5 2x 2 2 12x 1 20 . f ( x ) 5 2x2 2 12x 1 20 4 5 2( 4 ) 2 2 12( 4 ) 1 20 4 5 2( 16 ) 2 48 1 20 4 5 32 2 48 1 20 454
? ? ?
Substitute in (3, 2)
Substitute in (4, 4)
67
Recognizing the relationship between the graph of a parabola and its corresponding quadratic function provides a quick way to evaluate the accuracy of a quadratic model. Quadratic models may be determined algebraically or by using quadratic regression. We will demonstrate both methods in the next several examples. Lets return to the AP Calculus exam data introduced at the beginning of the section. At first glance, the Calculus AB exam data dont look at all like a parabola; however, we do observe that the data appear to be increasing at an ever-increasing rate. (It took six years [1980 to 1986] for the number of exams to increase from 20,000 to 40,000, but it took only three years [1986 to 1989] for the number of exams to increase from 40,000 to 60,000.) Plotting the quadratic equation f ( t ) 5 157.8486t 2 2 622,378t 1 613,500,479 together with the data set, we observe that, in fact, a quadratic equation fits the data very well (Figure 2.9). (Source: The College Board.) This equation is found by performing quadratic regression on the data, a process that we will detail in the forthcoming Technology Tip.
160,000 140,000 120,000 AP Calculus exams 100,000 80,000 60,000 40,000 20,000
f (t ) = 157.8486t 2 622,378t + 613,500,479
FIGURE 2.9
Observe that the coefficients of the variables in the quadratic equation are very large. We can come up with an equation with smaller coefficients by aligning the data. Well let t 5 0 in 1969, t = 1 in 1970, and so on. Doing quadratic regression on the aligned data yields the equation f ( t ) 5 157.85t 2 2 770.64t 1 10,268 with the coefficient of determination r 2 5 0 . 9979. Recall that the coefficient of determination is a measure of how well the model fits the data. The closer r 2 is to 1, the better the model fits the data. Although both of the models fit the data, the second model will make computations easier because of the smaller coefficients. Using the model, we predict how many tests will be administered in 2005. In 2005, t 5 36. f ( 36 ) 5 157.85( 36 ) 2 2 770.64( 36 ) 1 10,268 5 187,099 We estimate that 187,099 AP Calculus AB exams will be administered in 2005. A quadratic function model for a data set may be generated by using quadratic regression on a graphing calculator or within Microsoft Excel. However, the fact that technology may be used to create a quadratic model does not guarantee that the model will be a good fit for the data.
19 68 19 70 19 72 19 74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02
Year
68
Quadratic Regression 1. Enter the data using the Statistics Menu List Editor. (Refer to Section 1.3 if youve forgotten how to do this.)
2. Bring up the Statistics Menu Calculate feature by pressing STAT and using the blue arrows to move to the CALC menu. Then select item 5:QuadReg and press ENTER .
3. If you want to automatically paste the regression equation into the Y5 editor, press the key sequence VARS Y-Vars; 1:Function; 1:Y1 and press ENTER . Otherwise press ENTER .
Although the quadratic model fits the AP Calculus AB exam data very well from 1969 to 2002, we must be cautious in using it to predict future behavior. (Predicting the output value for an input value outside of the interval of the input data is called extrapolation.) For this data set, we may feel reasonably comfortable with an estimate two or three years beyond the last data point; however, we would doubt the accuracy of the model 100 years beyond 2002. For example, in 2102 (t = 133), the estimated number of exams is 2,702,587. This figure is more than 17 times greater than the maximum number of exams that have ever been administered! We also need to look at the population of students who could possibly take the exam. Between 1970 and 1999, the number of twelfth graders (those who typically take the AP Calculus exam) fluctuated from a high of 3,026,000 in 1977 to a low of 2,381,000 in 1990. (Source: Statistical Abstract of the United States, 2001, Table 232, p. 149.) If the number of twelfth graders continues to fluctuate between these two values, at some point our model estimate for the exams would exceed the number of people who could conceivably take the exam. This illustrates the necessity to verify that a model makes sense in its real-world context. It is often possible to model a data set with more than one mathematical model. When selecting a model, we should consider the following: 1. The graphical fit of the model to the data 2. The correlation coefficient (r) or the coefficient of determination (r 2) 3. The known behavior of the thing being modeled
69
Recall that the closer the correlation coefficient is to 1 or 21, the better the model fits the data. Similarly, the closer the coefficient of determination is to 1, the better the model fits the data.
EXAMPLE
Using Quadratic Regression to Forecast Prescription Drug Sales

Retail prescription drug sales in the United States increased from 1995 to 2000 as shown in Table 2.1.
TABLE 2.1
Years Since 1995 (t) 0 2 3 4 5
Retail Sales (billions of dollars) [S(t)] 68.6 89.1 103.0 121.7 140.7
Source: Statistical Abstract of the United States, 2001, Table 127, p. 94.
Model the data using a quadratic function. Then use the model to predict retail prescription drug sales in 1996 and 2001.
SOLUTION We observe from the scatter plot that the data appear concave up everywhere (Figure 2.10).
150 140 130 120 110 100 90 80 70 60 0 1 2 3 Years (since 1995) 4 5
Retail prescription drug sales (in billions of dollars)
S (t ) = 1.411t 2 + 7.441t + 68.55
FIGURE 2.10
A quadratic function may fit the data well. We use quadratic regression to find the quadratic model that best fits the data. Based on data from 1995 to 2000, retail prescription drug sales in the United States may be modeled by S( t ) 5 1.411t 2 1 7.441t 1 68.55 billion dollars where t is the number of years since 1995.
70
The coefficient of determination (r2 5 0.9997) is extremely close to 1. The graph also appears to touch each data point. The model appears to fit the data well. In 1996, t 5 1, and in 2001, t 5 6. Evaluating the function at each t value, we get S( 1 ) 5 1.411( 1 ) 2 1 7.441( 1 ) 1 68.55 5 77.402 S( 6 ) 5 1.411( 6 ) 2 1 7.441( 6 ) 1 68.55 5 163.992 Since the original data were accurate to only one decimal place, we will round our solutions to one decimal place as well. We estimate that prescription drug sales were $77.4 billion in 1996 and $164.0 billion in 2001.
You may ask, Is there a way to find a quadratic model without using quadratic regression? There is. The model may not be the model of best fit, but it may still model the data effectively. In Example 4, we repeat Example 3 using an algebraic method to find a quadratic model.
EXAMPLE
Using Algebraic Methods to Model Prescription Drug Sales

Retail prescription drug sales in the United States increased from 1995 to 2000 as shown in Table 2.2.
TABLE 2.2
Years Since 1995 (t) 0 2 3 4 5
Retail Sales (billions of dollars) [S(t)] 68.6 89.1 103.0 121.7 140.7
Model the data using a quadratic function. Then use the model to predict retail prescription drug sales in 1996 and 2001.
SOLUTION Given any three data points, we can find a quadratic function that passes through the points, provided that the points define a nonlinear function. We will pick the points ( 0, 68.6 ) , ( 3, 103.0 ) , and ( 5, 140.7 ) from the
71
table. A quadratic function is of the form S( t ) 5 at 2 1 bt 1 c. Each of the points must satisfy this equation. 68.6 5 a( 0 ) 2 1 b( 0 ) 1 c c 5 68.6 103.0 5 a( 3 ) 2 1 b( 3 ) 1 c 103.0 5 9a 1 3b 1 c 140.7 5 a( 5 ) 2 1 b( 5 ) 1 c 140.7 5 25a 1 5b 1 c Since we know c 5 68.6, we can simplify the last two equations. 103.0 5 9a 1 3b 1 68.6 34.4 5 9a 1 3b 140.7 5 25a 1 5b 1 68.6 72.1 5 25a 1 5b We can now find the values of a and b by solving the system of equations. 9a 1 3b 5 34.4 25a 1 5b 5 72.1 c c c 9 25 9 2 30 3 5 3 0 34.4 d 72.1 34.4 d 2 44.3
5R1 2 3R2 Substitute t 5 5, S( t ) 5 140.7 Substitute t 5 3, S( t ) 5 103.0 Substitute t 5 0, S( t ) 5 68.6
10R1 1 3R2 0 30 211.1 d 2 30 0 2 44.3 2 30a 5 2 44.3 30b 5 211.1
a 5 1.477 A quadratic model for the data is
b 5 7.037
S( t ) 5 1.477t 2 1 7.037t 1 68.6 Graphing this model with the data shows that it fits the data relatively well (Figure 2.11).
S 145 135 125 115 105 95 85 75 65 0 0 1 2 3 4 5
S (t ) = 1.477t 2 + 7.037t + 68.6
FIGURE 2.11
72
In 1996, t 5 1, and in 2001, t 5 6. Evaluating the function at each t value, we get S( 1 ) 5 1.477( 1 ) 2 1 7.037( 1 ) 1 68.6 < 77.1 billion S( 6 ) 5 1.477( 6 ) 2 1 7.037( 6 ) 1 68.6 < 164.0 billion
These estimates are close to the estimates from Example 3. In Example 3, we estimated that prescription drug sales were $77.4 billion in 1996 and $164.0 billion in 2001.
EXAMPLE
Using Quadratic Regression to Forecast Nursing Home Care Costs

Because of their need for constant medical care, many elderly Americans are placed in nursing homes by their families. The amount of money spent on nursing home care has risen substantially since 1960, as shown in Table 2.3.
TABLE 2.3
Years Since 1960 (t) 0 10 20 30 40
Nursing Home Care (billions of dollars) [N(t)] 1 4 18 53 96
The U.S. Centers for Medicare and Medicaid Services predict that by 2010 the cost of nursing home care will reach $183 billion. Model the data with a quadratic function and calculate the cost of nursing home care in 2010. Compare your result to the U.S. Centers for Medicare & Medicaid Services estimate.
SOLUTION Using quadratic regression, we determine that the quadratic model
of best fit is N( t ) 5 0.07214t 2 2 0.4957t 1 1.029 billion dollars where t is the number of years since 1960. Since the coefficient of determination (r 2 5 0.9987) is extremely close to 1, we anticipate that the model fits the data well. Graphing the data and the model together yields Figure 2.12.
73
100 Nursing home expenditures (in billions of dollars) 80 N (t ) = 0.07214 t 2 0.49571t + 1.02857 60 40 20 0
10
15
20 25 30 Years (since 1960)
35
40
45
FIGURE 2.12
The graph passes near each data point. The model appears to fit the data fairly well. Zooming in, however, we see that the graph decreases from 1 billion to 0 billion between 1960 and 1963 (Figure 2.13). (We rounded to whole numbers, since the original data were in whole numbers.)
4.0 3.6 3.2 2.8 2.4 2.0 1.6 1.2 0.8 0.4 0
Nursing home expenditures (in billions of dollars)
N (t ) = 0.07214 t 2 0.49571t + 1.02857
4 5 6 Years (since 1960)
10
FIGURE 2.13
In reality, nursing home expenditures have increased every year since 1960. Nevertheless, despite this limitation in our model, we will use it to predict nursing home expenditures in 2010. In 2010, t 5 50. N( 50 ) 5 0.721( 50 ) 2 2 0.4957( 50 ) 1 1.0286 5 156.60 We estimate that in 2010, $157 billion will be spent on nursing home care. Our model estimate is substantially less than the $183 billion that the U.S. Centers for Medicare & Medicaid Services estimate. Their estimate probably anticipated the health care needs of the aging population of baby boomers, while ours did not. Models projecting health care costs are useful for legislators, insurance companies, and consumers. By considering future costs, people can prepare for the future and avert financial crises.
74
EXAMPLE
Using Algebraic Methods to Find a Quadratic Model for Net Sales

Based on the data in Table 2.4, find a quadratic model for the net sales of the Kellogg Company algebraically.
TABLE 2.4
Years Since 1999 (t) 0 1 2
Kellogg Company Net Sales (millions of dollars) R(t) 6,984.2 6,954.7 8,853.3
Source: Kellogg Company 2001 Annual Report, pp. 7, 27. SOLUTION Since we are given the y-intercept, ( 0, 6984.2 ) , we know that c 5 6984.2. We have
R( t ) 5 at 2 1 bt 1 c 5 at 2 1 bt 1 6984.2 R( 1 ) 5 a( 1 ) 2 1 b( 1 ) 1 6984.2 6954.7 5 a 1 b 1 6984.2 2 29.5 5 a 1 b R( 2 ) 5 a( 2 ) 2 1 b( 2 ) 1 6984.2 8853.3 5 4a 1 2b 1 6984.2 1869.1 5 4a 1 2b We must solve the system of equations a 1 b 5 2 29.5 4a 1 2b 5 1869.1 We will solve the system using the substitution method. Solving the first equation for a yields a 5 2 b 2 29.5. Substituting this result into the second equation 4a 1 2b 5 1869.1 yields 4( 2 b 2 29.5 ) 1 2b 5 1869.1 2 4b 2 118 1 2b 5 1869.1 2 2b 5 1987.1 b 5 2 993.55 Since a 5 2 b 2 29.5, a 5 2 ( 2 993.55 ) 2 29.5 5 964.05 The quadratic function that models the revenue of the Kellogg Company is R( t ) 5 964.05t 2 2 993.55t 1 6984.2, where t is the number of years since the end of 1999 and R( t ) is the revenue from sales in millions of dollars.
Since a 5 2 b 2 29.5
75
In each of the preceding examples, the quadratic model fit the data well. This is not always the case, as demonstrated in Example 7.
EXAMPLE
Determining When a Quadratic Model Should Not Be Used

The per capita consumption of ready-to-eat and ready-to-cook breakfast cereal is shown in Table 2.5.
TABLE 2.5
Years Since 1980 (t) 0 1 2 3 4 5 6 7 8 9
Cereal Consumption (pounds) [C(t)] 12 12 11.9 12.2 12.5 12.8 13.1 13.3 14.2 14.9
Years Since 1980 (t) 10 11 12 13 14 15 16 17 18 19
Cereal Consumption (pounds) [C(t)] 15.4 16.1 16.6 17.3 17.4 17.1 16.6 16.3 15.6 15.5
Explain why you do or do not believe that a quadratic function will model the data set well.
SOLUTION We first draw the scatter plot of the data set (Figure 2.14).
C 17 16 15 14 13 12 11 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 t
FIGURE 2.14
76
(Note: We have adjusted the viewing window so that we may better analyze the data. Recall that a parabola is either concave up everywhere or concave down everywhere. This scatter plot appears to be concave up between t 5 1 and t 5 10, concave down between t 5 10 and t 5 17, and concave up between t 5 17 and t 5 19. The fact that the scatter plot changes concavity causes us to doubt that a quadratic model will fit the data well. In short, the scatter plot doesnt look like a parabola or a portion of a parabola. The quadratic model that best fits the data is C( t ) 5 2 0.0179t 2 1 0.637t 1 10.8 and is shown in Figure 2.15.
C 17 16 15 14 13 12 11 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 t C (t ) = 0.0179 t 2 + 0.637 t + 10.8
FIGURE 2.15
The coefficient of determination ( r 2 5 0.853 ) is not as close to 1 as in our previous examples. Additionally, we can see visually that the model does not fit the data well.
Once a model has been created, we can often use it to make educated guesses about what may happen in the future. Forecasting the future is a key function for many businesses.
EXAMPLE
Using Quadratic Regression to Forecast Fuel Consumption

As vans, pickups, and SUVs have increased in popularity, the total fuel consumption of these types of vehicles has also increased (Table 2.6).
77
TABLE 2.6 Motor Fuel Consumption of Vans, Pickups, and SUVs
Years Since 1980 (t) 0 5 10 15 19
Fuel Consumption (billions of gallons) (F ) 23.8 27.4 35.6 45.6 52.8
Model the fuel consumption with a quadratic function and forecast the year in which fuel consumption will reach 81.0 billion gallons.
SOLUTION Using quadratic regression, we determine the model to be
F( t ) 5 0.0407t2 1 0.809t 1 23.3 billion gallons where t is the number of years since the end of 1980. We want to know at what value of t does F( t ) 5 81.0 . This problem may be solved algebraically or graphically using technology. We will solve the problem twice (once with each method) and allow you to use the method of your choice in the exercises.
ALGEBRAIC SOLUTION
81.0 5 0.0407t 2 1 0.809t 1 23.3 0 5 0.0407t 2 1 0.809t 2 57.7 This is a quadratic function with a 5 0.0407, b 5 0.809, and c 5 2 57.7 . Recall that the solution to a quadratic equation of the form at2 1 bt 1 c 5 0 is given by the Quadratic Formula, t5 2 b 6 "b2 2 4ac 2a
Substituting our values of a, b, and c into the Quadratic Formula yields t5 5 5 2 0.809 6 "( 2 0.809 ) 2 2 4( 0.0407 )( 2 57.7 ) 2( 0.0407 )
2 0.809 6 "10.0 0.0814 2 0.809 6 3.17 0.0814
78
In the context of the problem, we know that t must be nonnegative. Consequently, we will calculate only the nonnegative solution. t5 5 2 0.809 1 3.17 0.0814 2.36 0.0814
5 29.0 We anticipate that at the end of 2009 (29 years after the end of 1980), the fuel consumption will have reached 81.0 billion gallons.
GRAPHICAL SOLUTION
We graph the function F( t ) 5 0.0407t 2 1 0.809t 1 23.3 and the horizontal line y 5 81.0 simultaneously (Figure 2.16). F( t ) 5 81.0 at the point at which these two functions intersect.
F 90 80 70 60 50 40 30 20 10 0 0 y = 81.0 (29, 81.0)
F (t ) = 0.0407t 2 + 0.809 t + 23.3 4 8 12 16 20 24 28 32 t
FIGURE 2.16
Using the intersect feature on the graphing calculator, we determine that the point of intersection is ( 29, 81.0 ) . The interpretation of the solution is the same as that given in the discussion of the algebraic method.
79
Quadratic Regression 1. Draw a scatter plot of a data set.

60 50 40 30 20 10 0 0 2 4 6 8 10 12
2. Right-click on any point of the scatter plot to open a drop-down menu. From the drop-down menu, select Add Trendline to open the Add Trendline dialog box. Then select Polynomial and Order 2.
(Continued)
80
4. The graph and equation of the quadratic function that best fits the data are displayed. The coefficient of determination r2 is also displayed.
60 50 40 30 20 10 0 10 0
y = 1.0045x2 6.0875x + 9.1071 R2 = 0.9996
10
12
2.1 Summary
In this section, you learned how to use quadratic regression to model a data set. You also discovered the importance of analyzing a mathematical model before using it to calculate unknown values.

1. To calculate 2 1 times 2 1 on her calculator, a student types in 2 12 or 2 1 ` 2 and receives back the answer 2 1. But 2 1 times 2 1 is equal to 1, not 2 1. Why did the calculator return the incorrect answer? 2. What do we call the graph of a quadratic function? 3. For the quadratic equation y 5 3x 2 2 6x 1 4, what are the values of a, b, and c? 4. The discriminant of a quadratic equation is the quantity b2 2 4ac. What is the value of the discriminant for the equation y 5 2 x 2 2 3x 1 18? 5. When the discriminant of a quadratic function is negative, what do we know about the graph of the function? 6. If the vertex of a quadratic function lies on the x-axis, what do we know about the discriminant of the function? 7. What is the Quadratic Formula and what does it find? 8. What do you get when you simplify the expression ( 2x 1 3 )( 2 4x 1 5 ) ?
81
2.1 Exercises
In Exercises 15, determine the concavity, y-intercept, and vertex of the quadratic equation. 1. y 5 x 2 2 2x 1 1 3. g ( x ) 5 3x 1 3x
2
9.
y 5 0 5 10 15 20 1 2 3 4 5 6 7 x
2. f ( x ) 5 2 2x 2 1 4
4. h ( t ) 5 2 1.2t 2 1 2.4t 1 4.5 5. f ( t ) 5 2.8t 2 1.4t 1 2.1

2
In Exercises 610, determine the equation of the parabola from the graph. 6.
y 14 12 10 8 6 4 2 0
10.
y 60 40 20 0 20 40 60
1 2 3 4 5 6 7 x
0 1 2 3 4 5 6 7 x
7.
y 3 2 1 0 0 1 y 9 8 7 6 5 4 3 2 1 1 0 1 2 x 2 3 x
In Exercises 1115, do the following: (a) Find a quadratic model for the data algebraically. (b) Graph the model together with the scatter plot. (c) Explain why you do or do not believe that the function is a good model. 11.
Basketball Game Attendance
NCAA Basketball Game Attendance
Years Since 1985 (t) 0 5 10 14
Womens Games (thousands) (W ) 2,072 2,777 4,962 8,698
8.
82 12.
Ice Cream Cost
14.
Health Care Spending
Prepackaged Ice Cream Cost
Years Since 1990 (t) 0 3 4 5 6 7 8 9 10
Ice Cream Cost (dollars per 1/2 gallon) (C ) 2.54 2.59 2.62 2.68 2.94 3.02 3.30 3.40 3.66 15.
Health Care Spending by Insurance Companies
Years Since 1960 (t) 0 10 20 30 40
Spending (billions of dollars) (I ) 6 16 68 233 438
VHS Tape and CD-R Sales
VHS Tape Versus CD-R Disk Sales, 20002002
13.
DVD Player Shipments
DVD Player Shipments in North America
Millions of Blank VHS Tapes Sold Worldwide (v) 1,409 1,322 1,195
Millions of CD-R Disks Sold Worldwide (C ) 2,730 3,730 4,225
Years Since 1997 (t) 1 2 3 4
Shipments (thousands) (D) 1,080 4,100 9,700 16,700
Source: International Recording Media Association.
Source: DVD Entertainment Group.
In Exercises 1650, do the following: (a) Draw a scatter plot for the data. (b) Explain why a quadratic function might or might not fit the data. If it appears that a quadratic function may fit the data: (c) Use quadratic regression to find the quadratic function that best fits the data. (d) Graph the model together with the scatter plot. (e) Explain why you do or do not believe that the function is a good model. (f) Answer any additional questions that may be given.
83
16.
College Freshmen
18.
Male College Freshmen
Years Since 1970 (t) 0 10 15 20 25 27 28 29 30
Percentage of Freshmen That Are Male (F ) 52.1 48.8 48.9 46.9 45.6 45.5 45.5 45.3 45.2
Personal Income In the following table, personal income per capita is listed using constant (1996) dollars. That is, the actual income figures are adjusted for inflation.
Personal Income per Capita, Mississippi, ranked #50 in 2000
Years Since 1980 (t) 0 10 20
Income (dollars) (I ) 12,817 15,373 19,554
What is the vertex of the quadratic model, and what does it mean in the context of the problem? 19.
According to the model, in what year were there an equal number of male and female freshmen? 17.
Personal Income Because of inflation, the buying power of a dollar decreases over time. As a result, your personal income may increase while your buying power decreases. In the following table, personal income per capita is listed using constant (1996) dollars. That is, the actual income figures are adjusted for inflation.
Game Sales
U.S. Video Game Industry Sales
Years Since 1997 (t) 0 1 2 3 4
Sales (billions of dollars) (S ) 5.1 6.2 6.9 6.6 9.4
Personal Income per Capita, Connecticut, Ranked #1 in 2000
Years Since 1980 (t) 0 10 20
Income (dollars) (I ) 22,530 31,223 37,854
Source: www.npdfunworld.com.
84 20.
Milk Consumption
22.
Home Square Footage
Per Capita Milk Beverage Consumption
Percentage of New One-Family Houses Under 1,200 Square Feet
Years Since 1980 (t) 0 5 10 15 17 18 19
Consumption (gallons) (M ) 27.6 26.7 25.7 24.3 24.0 23.7 23.6
Years Since 1970 (t) 0 10 20 25 29
Percentage of New Houses (percent) (H ) 36 21 11 10 7
21.
What is the concavity of the graph of the model, and what does it mean in the context of the problem? 23.
Home Square Footage
Conventional Mortgages
Percentage of New Privately Owned One-Family Houses Financed with a Conventional Mortgage
New One-Family Homes: Average Number of Square Feet
Years Since 1970 (t) 0 10 20 25 29
Conventional Mortgages (percent) (M ) 47 55 62 74 79
Years Since 1970 (t) 0 10 20 25 29
Houses (square feet) (H ) 1,500 1,740 2,080 2,095 2,225
According to the model, in what year were 50 percent of the new privately owned homes financed with a conventional mortgage?
85
24.
Air Conditioning in Homes
26.
Home Sales Price
New One-Family Homes with Central Air Conditioning
Median Sales Price of a New One-Family House in the Southern United States
Years Since 1970 (t) 0 10 20 25 29
Years Since 1980 (t) 0 5 10 15 20
Price (thousands of dollars) (P) 59.6 75.0 99.0 124.5 148.0
25.
Homes with Garages
27.
Home Sales Price
New One-Family Homes with a Garage
Years Since 1970 (t) 0 10 20 25 29
Median Sales Price of a New One-Family House in the Northeastern United States
Years Since 1980 (t) 0 5 10 15 20
In what year were 90 percent of new onefamily homes expected to have a garage?
86 28.
Home Sales Price
30.
Fuel Consumption
Median Sales Price of a New One-Family House in the Western United States
Motor Fuel Consumption of Vans, Trucks, and SUVs
Years Since 1980 (t) 0 5 10 15 20
Years Since 1980 (t) 0 5 10 15 19
Fuel Consumption (billions of gallons) (F ) 23.8 27.4 35.6 45.6 52.8
29.
Vehicle Leasing
31.
Coffee Sales
Percentage of Households Leasing Vehicles
Starbucks Corporation Sales
Years Since 1989 (t) 0 3 6 9
Percentage of Households (percent) (P) 2.5 2.9 4.5 6.4
Years Since 09/93 (t) 0 1 2 3 4 5 6 7 8 9
Income from Sales (millions of dollars) (S ) 163.5 284.9 465.2 696.5 966.9 1,308.7 1,680.1 2,169.2 2,649.0 3,288.9
In what year did the percentage of households leasing vehicles reach 5 percent?
Source: moneycentral.msn.com.
According to the model, when did Starbucks Corporation sales reach $5000 million?
87
32.
Organic Cropland
34.
Prison Rate
Certified Organic Cropland
Federal and State Prison Rate
Years Since 1992 (t) 0 1 2 3 4
Acres (thousands) (A) 403 465 557 639 850
Years Since 1980 (t) 0 2 4 6 8 10 12 14 16 18
Rate (prisoners per 100,000 people) (R) 139 171 188 217 247 297 332 389 427 461
33.
Poultry Pricing
Average Retail Price of Fresh Whole Chicken
Years Since 1985 (t) 0 5 6 7 8 9 10 11 12 13 14 15
Price (dollars per pound) (P) 0.78 0.86 0.86 0.88 0.91 0.90 0.94 1.00 1.00 1.06 1.05 1.08 35.
College Enrollment
Private College Enrollment
Years Since 1980 (t) 0 2 4 6 8 10 12 14 16 18
Students (in thousands) (S) 2,640 2,730 2,765 2,790 2,894 2,974 3,103 3,145 3,247 3,373
88 36.
Military Personnel
38.
Personal Income
Active-Duty Military Personnel
Per Capita Personal Income: Colorado
Years Since 1990 (t) 1 2 3 4 5 6 7 8 9
Personnel (thousands) (P) 1,263 1,214 1,171 1,131 1,085 1,056 1,045 1,004 1,003
Years Since 1993 (t) 0 1 2 3 4 5 6 7
Personal Income (dollars) (P) 22,196 23,055 24,289 25,514 27,067 28,764 30,206 32,434
39.
Personal Income
37.
Personal Income
Per Capita Personal Income: Louisiana
Per Capita Personal Income: California
Years Since 1993 (t) 0 1 2 3 4 5 6 7
Years Since 1993 (t) 0 1 2 3 4 5 6 7
89
40.
Personal Income
42.
Pharmaceutical Income
Per Capita Personal Income: Minnesota
Johnson & Johnson Net Income
Years Since 1993 (t) 0 1 2 3 4 5 6 7
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10 11
Net Income (millions of dollars) (I ) 1,195 1,441 1,572 1,786 1,998 2,418 2,958 3,385 3,798 4,348 4,998 5,899
41.
Pharmaceutical Company Profit
Johnson & Johnson Pharmaceutical Operating Profit
Years Since 1997 (t) 0 1 2 3 4
Operating Profit (millions of dollars) (P) 2,332 3,114 3,735 4,394 4,928
Source: Johnson & Johnson Annual Report, 2001, p. 18.
In what year was Johnson & Johnson net income projected to reach $7000 million?
Source: Johnson & Johnson Annual Report, 2001, p. 8.
In what year were Johnson & Johnson profits expected to reach $6000 million?
90 43.
Company Payroll
Payroll of Private-Employer Firms*
According to the model, in what year is the number of public-school teachers expected to reach 3 million?
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8
Payroll (billions of dollars) (F ) 2,104 2,145 2,272 2,363 2,488 2,666 2,849 3,048 3,309
45.
Private-School Teachers
Projected Number of Private-School Teachers
Years Since 2000 (t) 0 1 2 3 4 5 6 7 8 9 10
Private-School Teachers (thousands) (G) 402 403 404 405 407 408 409 410 411 411 412
*Firms are an aggregation of all establishments owned by a parent company. Source: Statistical Abstract of the United States, 2001, Table 726, p. 486.
44.
Public-School Teachers
Projected Number of Public-School Teachers
Years Since 2000 (t) 0 1 2 3 4 5 6 7 8 9 10
Public-School Teachers (thousands) (G) 2,850 2,865 2,877 2,891 2,905 2,914 2,919 2,927 2,932 2,937 2,940
91
46.
Apparel Production Wages
48.
McDonalds Restaurant Sales
Apparel Production Workers Wages
McDonalds Franchised Sales
Years Since 1980 (t) 0 10 15 16 17 18 19 20
Wage of Mens and Boys Furnishings Production Workers (dollars per hour) (M ) 4.23 6.06 7.19 7.40 7.72 7.97 8.27 8.54
Years Since 1990 (t) 1 2 3 4 5 6 7 8 9 10 11
Franchised Sales (millions of dollars) (S ) 12,959 14,474 15,756 17,146 19,123 19,969 20,863 22,330 23,830 24,463 24,838
47.
Apparel Production Wages
49.
Advertising Expenditures
Apparel Production Workers Wages
Advertising Expenditures: Billboards
Years Since 1980 (t) 0 10 15 16 17 18 19 20
Wage of Womens and Misses Outerwear Production Workers (dollars per hour) (W ) 4.61 6.26 7.27 7.49 7.84 8.15 8.41 8.40
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Advertising Expenditures (millions of dollars) (A) 1,084 1,077 1,030 1,090 1,167 1,263 1,339 1,455 1,576 1,725 1,870
92
In what year were billboard advertising expenditures projected to reach $2 billion? 50.
52. A quadratic equation is written in factored form as y 5 2 2( x 2 6 )( x 2 10 ) . Explain how you can use mental math to quickly determine the x coordinate of the vertex from the equation. 53. Explain the difference between the term concave up and the term increasing. 54. Recall that the average rate of change of a function f over an interval 3d, d 1 h4 is given by
f (d 1 h) 2 f (d) . (d 1 h) 2 d
Advertising Expenditures: Direct Mail
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Show that the average rate of
change of a quadratic function is always a linear function. 55. A classmate claims that the graph of y 5 2 2x2 1 2 is concave down and increasing on the interval ( 2 ` , 1 ) , and concave up and decreasing on the interval ( 1, ` ) . Do you agree? Explain why or why not.
56. Find two different quadratic functions that pass through the points ( 0, 0 ) and ( 4, 0 ) . 57. A classmate claims that any group of three points defines a unique quadratic function. Is your classmate correct? Explain. 58. Is there a single quadratic function that passes through each of the following points: ( 1, 3 ) , ( 2, 5 ) , ( 4, 9 ) ? Explain. 59. What is the graphical significance of the point with x coordinate x 5 graph of y 5 ax2 1 bx 1 c ?
2 b 1 "b2 2 4ac 2a
on the
In what year were direct-mail advertising expenditures expected to reach $50 billion?
60. If ax2 1 bx 1 c . 3 for all values of x, what do we know about a and the graph of y 5 ax2 1 bx 1 c ?
51. A classmate claims that the x coordinate of the vertex of a parabola lies halfway between its x-intercepts. Use the Quadratic Formula, x5 claim.
2 b 6 !b2 2 4ac , 2a
to support or refute this
93
Use nonlinear functions to model real-life phenomena Interpret the meaning of mathematical models in their real-world context Model real-life data with higher-order polynomial functions
Since 1975, the average wage of sailors working on the East Coast of the United States has trailed that of sailors working on the West Coast (Table 2.7). At one point, West Coast sailors made nearly 1 times the wage of East Coast sailors.
1 2
TABLE 2.7 Typical Basic Monthly Wage for Able-Bodied Seamen in Addition to Room and Board
Years Since 1980 (t) 0 5 10 13 14 15 16 17 18 19 20
East Coast Wage (dollars) (E ) 967 1,419 1,505 1,721 1,790 1,918 2,014 2,094 2,178 2,265 2,453
West Coast Wage (dollars) (W ) 1,414 2,029 2,218 2,438 2,536 2,637 2,769 2,879 2,994 3,114 3,114
Salaries of both groups of sailors have increased substantially since 1980, although pay increases have been somewhat irregular. Will East Coast sailors ever make as much as West Coast sailors? In this section, we will use cubic and quartic polynomials to model data. To keep computations simple, we use these models only if were dissatisfied with the fit of simpler models.
Cubic Functions
A constant function is a polynomial of degree 0. A nonconstant linear function is a polynomial of degree 1. A quadratic function is a polynomial of degree 2. A cubic function is a polynomial of degree 3 and has the form f ( x ) 5 ax 3 1 bx 2 1 cx 1 d. The graph of a cubic function has exactly one inflection point (a point where the graph of the function changes concavity). The
94
term concavity refers to the curvature of the graph. Recall that when the graph curves upward, we say that it is concave up, and when the graph curves downward, we say that it is concave down. The following rhyme is helpful in remembering the meaning of the terms. Concave up is like a cup. Concave down is like a frown. The graphs in Figure 2.17 are concave up.
y y y
FIGURE 2.17
The graphs in Figure 2.18 are concave down.

y y y
x x
FIGURE 2.18
Cubic function graphs consist of two pieces: a concave up piece and a concave down piece. If a . 0, the graph of the cubic function is first concave down and then concave up. The graph of the cubic function will take on one of the basic shapes shown in Figure 2.19.
y y
FIGURE 2.19
95
The size of the bends in the graph will vary depending upon the values of b and c. When a is positive, the graph of a cubic function will be increasing (rising from left to right) for sufficiently large values of x. If a , 0, the graph of the cubic function is first concave up and then concave down. The graph of the cubic function will take on one of the shapes shown in Figure 2.20.
y y
FIGURE 2.20
Again, the size of the bends in the graph will vary depending upon the values of b and c. When a is negative, the graph of a cubic function will be decreasing (falling from left to right) for sufficiently large values of x. Returning to the sailors wages, we can draw some conclusions by looking at a scatter plot of the data. Since a portion of the scatter plot appears linear, we first attempt to model the data with a linear function (Figure 2.21).
2500 Sailors wages on the East Coast, E (in dollars per month) 2250 2000 1750 1500 1250 1000 750 0 2 4 6 8 10 12 14 Years, t (since 1980) 16 18 20 22
FIGURE 2.21
The linear model does not fit the data as well as we would like. We also note that from 1980 to 1990, the graph appears to be concave down, and from 1990 to 2000, the graph appears to be slightly concave up. We expect that a cubic model with a . 0 may fit the data well. Using the cubic regression feature on our calculator, we determine that the cubic function that best fits the data is E( t ) 5 0.2792t 3 2 7.037t 2 1 102.6t 1 983.6 dollars per month
96
where t is the number of years since 1980. The model graph is shown in Figure 2.22.
2500 Sailors wages on the East Coast, E (in dollars per month) 2250 2000 1750 1500 1250 1000 750 0 2 4 6 8 10 12 14 Years, t (since 1980) 16 18 20 22 E (t ) = 0.2792t 3 7.037t 2 + 102.6t + 983.6
FIGURE 2.22
The model fits the data fairly well, especially after 1993. The coefficient of determination of r 2 5 0.9905 further increases our confidence in the model. Lets now look at the scatter plot of the data from the West Coast (Figure 2.23).
3500 Sailors wages on the West Coast, W (in dollars per month) 3250 3000 2750 2500 2250 2000 1750 1500 1250 0 2 4 6 8 10 12 14 Years, t (since 1980) 16 18 20 22
FIGURE 2.23
Again, the graph appears to be concave down from 1980 to 1990. From 1990 to 1999, the graph appears to be concave up. From 1999 to 2000, the wage did not change. Using the cubic regression feature on our calculator, we determine that the cubic equation that best fits the data is W( t ) 5 0.2546t 3 2 7.277t 2 1 132.2t 1 1437 dollars per month where t is the number of years since 1980. The model graph is shown in Figure 2.24.
97
3500 Sailors wages on the West Coast, W (in dollars per month) 3250 3000 2750 2500 2250 2000 1750 1500 1250 0 2
W (t ) = 0.2546t 3 7.277t 2 + 132.2 t + 1437
8 10 12 14 Years, t (since 1980)
16
18
20
22
FIGURE 2.24
Will the two models ever be equal? In 2017, the graphs intersect with a monthly wage of nearly $9100 (Figure 2.25)! Since 2017 is 17 years beyond the last point in our data set, we are somewhat skeptical of the result. Nevertheless, we can conclude that the wage gap isnt likely to be eliminated in the next couple of years.
10,000 Sailors wages on the East and West Coasts, S (in dollars per month) 8000 W (t ) 6000 4000 E (t ) 2000 0 0 5 10 15 20 25 Years, t (since 1980) 30 35 40
FIGURE 2.25
The following Technology Tip details how to use your TI-83 Plus calculator to do cubic and quartic regression.
98
Cubic and Quartic Regression 1. Enter the data using the Statistics Menu List Editor. (Refer to Section 1.3 if youve forgotten how to do this.)
2. Bring up the Statistics Menu Calculate feature by pressing STAT and using the blue arrows to move to the CALC menu. Then select item 6:CubicReg or item 7:QuadReg and press ENTER .
3. If you want to automatically paste the regression equation into Y1 in the Y= editor, press the key sequence VARS Y-VARS; 1:Function; 1:Y1 and press ENTER . Otherwise, press ENTER .
EXAMPLE
Choosing a Polynomial Function to Model a Companys Revenue

Electronic Arts, Inc., is one of the premier producers of interactive electronic games playable on game platforms such as the Sony PlayStation 2, Microsoft Xbox, Nintendo GameCube, and Sony PlayStation. Despite the sluggish economy in the early 2000s, Electronic Arts showed an overall strong growth in revenue (Table 2.8).
TABLE 2.8
Years Since End of Fiscal Year 1998 (t) 0 1 2 3 4 5
Net Revenues (millions of dollars) [R(t)] 908.852 1,221.863 1,420.011 1,322.273 1,724.675 2,482.244
Source: Electronic Arts, Inc., Annual Reports, March 31, 2002 and 2004.
99
Find a polynomial model for the annual revenue of Electronic Arts, Inc., and forecast in what year the revenue will reach $3 billion.
SOLUTION From the scatter plot, it appears that a cubic function will best fit the data (Figure 2.26). Using cubic regression, we determine that the cubic model that best fits the data is
R( t ) 5 43.87t 3 2 274.8t 2 1 594.7t 1 900.0 million dollars where t is the number of years since the end of fiscal year 1998.
R 2500 2000 1500 1000 500 0 0 1 2 3 4 t R (t ) = 43.87t 3 274.8t 2 + 594.7t + 900.0
FIGURE 2.26
Since $3 billion is the same as $3000 million, we must determine when R( t ) 5 3000. To do this, we simultaneously graph R( t ) 5 43.87t 3 2 274.8t 2 1 594.7t 1 900.0 and y 5 3000 . Then, using the intersection feature on our graphing calculator, we determine when the two functions will be equal (Figure 2.27).
R 3000 2500 2000 1500 1000 500 0 0 1 2 3 4 5 t y = 3000 R (t ) = 43.87t 3 274.8t 2 + 594.7t + 900.0 (5.395, 3000)
FIGURE 2.27
According to the model, 5.395 years after the end of fiscal year 1998, the annual revenue is expected to reach $3 billion. To add meaning to the result, we convert 0.395 years into months. 0.395 years ? 12 months < 5 months 1 year
During the one-year period prior to the end of the fifth month of fiscal year 2004, we anticipate that $3 billion in revenues will be earned. (Note: Since t 5 5 is the end of fiscal year 2003, t 5 5.395 is in fiscal year 2004.)
100
Quartic Functions
A fourth-degree polynomial is a function of the form f ( x ) 5 ax 4 1 bx 3 1 cx 2 1 dx 1 e and is called a quartic function. A quartic function may have zero or two inflection points. A quartic graph with a . 0 will have one of the basic shapes shown in Figure 2.28.
y y y
Inflection points
Inflection points
x x x
Concave up everywhere No inflection points FIGURE 2.28
Concave up, then down, then up Two inflection points
Concave up, then down, then up Two inflection points
Notice that each graph opens upward. The size of the bends in the graph will vary depending upon the values of b, c, and d. When a is positive, the graph of a quartic function will be increasing (rising from left to right) for sufficiently large values of x. A quartic graph with a , 0 will have one of the basic shapes shown in Figure 2.29.
y x y x x Inflection points Inflection points y
Concave down everywhere No inflection points FIGURE 2.29
Concave down, then up, then down Two inflection points
Concave down, then up, then down Two inflection points
Notice that each graph opens downward. Again, the size of the bends in the graph will vary depending upon the values of b, c, and d. When a is negative, the graph of a quartic function will be decreasing (falling from left to right) for sufficiently large values of x.
101
EXAMPLE
Choosing a Nonlinear Function to Model Interest Rates

The average fixed interest rate for a conventional mortgage on a new home fluctuated between 1985 and 1995 as shown in Table 2.9.
TABLE 2.9 Average Fixed Interest Rates on New Home Conventional Mortgages
Years Since 1985 (t) 0 2 4 6 8 10
Percentage Rate (percent) (R) 11.90 9.50 10.20 9.32 7.27 7.95
Model the data with a nonlinear function. Do you think the model will be a good indicator of future interest rates? Explain.
SOLUTION
12.0 11.5 11.0 10.5 10.0 9.5 9.0 8.5 8.0 7.5 7.0 6.5 0 2 4 6 8 Years (since 1985) 10 12
FIGURE 2.30
New home fixed interest rates (in percent)
102
As shown in Figure 2.30, the graph appears to be concave up between 1985 and 1988, concave down from 1988 to 1992, and concave up from 1992 to 1995. A quartic model may work well (Figure 2.31).
12.0 11.5 11.0 10.5 10.0 9.5 9.0 8.5 8.0 7.5 7.0 R (t ) = 0.0112t 4 0.2242 t 3 + 1.403t 2 3.180 t + 11.89 6.5 0 2 4 6 8 10 12 Years (since 1985)
FIGURE 2.31
Using the quartic regression feature on our calculator, we determine that fixed interest rates for mortgages on new homes between 1985 and 1995 may be modeled by the equation R( t ) 5 0.0112t 4 2 0.2242t 3 1 1.403t 2 2 3.1796t 1 11.89 percent where t is the number of years since the end of 1985. The model shows that interest rates will rise rapidly every year after 1995. Lets calculate the predicted rate at the end of 1997 ( t 5 12 ) . R( 12 ) 5 0.0112( 12 ) 4 2 0.2242( 12 ) 3 1 1.403( 12 ) 2 2 3.1796( 12 ) 1 11.89 5 20.59 According to the model, interest rates reached 20.59 percent in 1997. This figure seems a bit unrealistic. (Between 1970 and 2000, interest rates remained below 15 percent.) The model may be a better indicator for intermediate years, such as 1992 or 1994.
EXAMPLE
Choosing a Function to Model the Number of Mutual Funds

The number of mutual funds available to investors has increased dramatically since 1970, as shown in Table 2.10.
New home fixed interest rates (in percent)
103
TABLE 2.10 Funds
Number of Mutual
Years Since 1970 (t) 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Mutual Funds (M ) 361 410 431 452 505 564 857 1,241 1,835 2,708 3,081 3,826 5,330 6,254 7,314 8,171
Find a function that models the data and estimate the number of mutual funds in 2005.
SOLUTION We could model the data with a quadratic function (Figure 2.32); however, even though the correlation coefficient is close to 1, the quadratic model doesnt fit the data very well between 1970 and 1980.
12,000 Number of mutual funds 10,000 8000 6000 4000 2000 0 0 5 10
y = 12.65x 2 117.4 x + 545.3
15 20 Years (since 1970)
25
30
35
FIGURE 2.32
104
A quartic model will give a better fit (Figure 2.33).

12,000 Number of mutual funds 10,000 8000 6000 4000 2000 0 0 5 10 15 20 Years (since 1970) 25 30 35
M (t ) = 0.02524t 4 + 1.519 t 3 16.18t 2 + 62.46t + 351.2
FIGURE 2.33
The number of mutual funds is modeled by the quartic equation M( t ) 5 2 0.02524t 4 1 1.519t 3 2 16.18t 2 1 62.46t 1 351.2 mutual funds where t is the number of years since 1970. For the year 2005, t 5 35. To estimate the number of mutual funds in 2005, we evaluate M( t ) at t 5 35. M( 35 ) 5 2 0.02524( 35 ) 4 1 1.519( 35 ) 3 2 16.18( 35 ) 2 1 62.46( 35 ) 1 351.2 5 9968 According to the model, in 2005 (t 5 35), there will be 9968 mutual funds. Given the historically rapid growth in the number of mutual funds, this estimate seems reasonable.
Cubic and Quartic Regression 1. Draw a scatter plot of a data set.

1200 1000 800 600 400 200 5 200 0 0 5 10 15
(Continued)
105
2. Right-click on any point of the scatter plot to open a drop-down menu. From the drop-down menu, select Add Trendline to open the Add Trendline dialog box. Then select Polynomial and Order 3 (for cubic) or 4 (for quartic).
(Continued)
106
4. The graph and equation of the cubic (or quartic) function that best fits the data are displayed. The coefficient of determination r2 is also displayed.
1200 1000 800 600 400 200 5 200 0 0
y = 0.9874x3 + 0.1012x2 3.4458x 0.2125 R2 = 0.9997
10
15
2.2 Summary
In this section, you learned how to use cubic and quartic polynomials to model data. You also discovered the importance of being cautious in using model predictions.

1. What is a polynomial? 2. What is an inflection point? 3. How do cubic functions differ from quadratic functions? 4. How do quartic functions differ from cubic functions? 5. What do quadratic and quartic functions have in common? 6. What is the degree of a polynomial? 7. A cubic polynomial has at most how many bends in its graph? 8. What effect does the leading coefficient of a polynomial have on its graph?
107
2.2 Exercises
In Exercises 120, find the quadratic, cubic, or quartic model that best fits the data. If multiple models appear to fit the data well, use the simplest model to answer the questions. 1.
Public Ninth- to Twelfth-Grade Enrollment
2.
Student-to-Teacher Ratio at Private Elementary and Secondary Schools
Years Since 1960 (t) 0 5 10 11 12 13 14 15 20 25 26 27 28 29 30 31 32 33 34 35 36 37 38
Students per Teacher (R) 30.7 28.3 23.0 22.6 21.6 21.2 20.4 19.6 17.7 16.2 15.7 15.5 15.2 14.2 14.7 14.7 14.6 14.8 14.9 14.9 14.9 15.1 15.2
Years Since 1984 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Students (in thousands) (S ) 12,304 12,388 12,333 12,076 11,687 11,390 11,338 11,541 11,735 11,961 12,213 12,500 12,847 13,054 13,193 13,369
According to your model, how many students were enrolled in high school (ninth to twelfth grades) in 2000?
According to your model, what will be the student-to-teacher ratio in 2006? Does this seem reasonable? Explain.
108 3.
Public-School Teachers
4.
Principals Salaries
Public Elementary- and Secondary-School Teachers
High School Principals Salaries
Years Since 1980 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Teachers (thousands) (R) 2,211 2,192 2,158 2,134 2,144 2,175 2,215 2,249 2,282 2,324 2,362 2,409 2,429 2,466 2,512 2,568 2,607 2,673 2,745 2,813 2,886
Years Since 1980 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Salary (dollars) (S ) 29,207 32,231 34,776 37,602 39,334 42,094 44,986 47,896 50,512 52,987 55,722 59,106 61,768 63,054 64,993 66,596 69,277 72,410 74,380 76,768 79,839
According to the model, in what year is the number of teachers expected to reach 3 million?
According to the model, when will the average high school principals salary reach $100,000?
109
5.
School Internet Access
6.
Banks
Percentage of Public-School Classrooms with Internet Access
Number of Different Banks (Not Bank Branches)
Years Since 1994 (t) 0 1 2 3 4 5 6
Percentage of Classrooms (percent) (C ) 3 8 14 27 51 64 77
Years Since 1984 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Banks (B) 17,900 18,033 17,876 17,325 16,562 15,829 15,192 14,517 13,891 13,261 12,641 12,002 11,478 10,923 10,463 10,221 9,908
According to the model, what percentage of classrooms had Internet access in 2001? In 2002? Do the model estimates seem reasonable? Explain.
According to the model, when will the number of banks drop below 9000? What do you think is causing the number of different banking companies to decrease?
110 7.
Debit Cards
Number of Debit Cards
According to your model, when will the NASDAQ market have an average daily volume of 2 billion shares? Visit www.marketdata.nasdaq.com to see if your model estimate is accurate. 9.
Stock Trading
Years Since 1990 (t) 0 5 6 7 8 9
Cards (millions) (C ) 164 201 205 211 217 228
New York Stock Exchange Shares Traded
Years Since 1980 (t) 0 2 4 6 8 10 12 14 16 18 20
Shares (millions) (S) 11,562 16,669 23,309 36,009 41,118 39,946 51,826 74,003 105,477 171,188 265,499
A consulting firm has projected that the number of debit cards will reach 270 million by 2005. Does this projection agree with your model? 8.
NASDAQ Volume
NASDAQ Average Daily Volume
Years Since 1980 (t) 0 2 4 6 8 10 12 14 16 18 20
Shares (millions) (S) 27 33 60 114 123 132 191 295 544 802 1,757
According to the model, how many shares will be traded in 2005? Does this seem reasonable? Explain.
111
10.
Listed Stocks
Companies with Stocks Listed on the NYSE
In what year do you expect the average wage to exceed $40,000? 12.
Computer Wholesalers Wages
Years Since 1980 (t) 0 2 4 6 8 10 12 14 16 18 20
Companies (C ) 1,570 1,526 1,543 1,575 1,681 1,774 2,088 2,570 2,907 3,114 2,862
Average Annual Wage per Worker for Computer and Equipment Wholesalers
Years Since 1992 (t) 1 2 3 4 5 6
Wages (dollars) (W ) 52,500 52,900 52,900 54,300 56,700 62,200
Calculate the average wage in 2000. Does this seem reasonable? Explain. 13.
Software Developers Wages
According to the model, in which years is the number of companies listed on the New York Stock Exchange expected to exceed 2700 companies? 11.
Computer Retailers Wages
Average Wage per Worker at Prepackaged Software Development Firms
Average Annual Wage per Worker for Computer and Equipment Retailers
Years Since 1992 (t) 0 1 2 3 4 5 6
Wages (dollars) (W ) 57,000 54,500 57,000 63,700 70,100 79,200 94,100
Years Since 1992 (t) 1 2 3 4 5 6
Wages (dollars) (W) 32,200 30,500 32,100 33,800 35,000 37,300
In what year will the wage reach $120,000?
112 14.
Software Retailers Wages
16.
Radio Broadcasters Wages
Average Wage per Worker at Prepackaged Software Retailers
Average Annual Wage per Worker in Radio Broadcasting
Years Since 1992 (t) 0 1 2 3 4 5 6
Wages (dollars) (W ) 32,200 30,500 32,100 33,800 35,000 37,300 40,400
Years Since 1992 (t) 0 1 2 3 4 5 6
Annual Wage (dollars) (W ) 23,500 24,300 26,000 27,200 29,300 31,300 34,200
What do you expect the average workers wage was in 1999? 15.
Software Wholesalers Wages
According to the model, what was the average wage of a radio broadcasting employee in 2006? Does this estimate seem reasonable? Explain. 17.
Television Broadcasters Wages
Average Wage per Worker at Prepackaged Software Wholesalers
Years Since 1992 (t) 0 1 2 3 4 5 6
Wages (dollars) (W) 52,500 52,900 52,900 54,300 56,700 62,200 69,700
Average Annual Wage per Worker in Television Broadcasting
Years Since 1992 (t) 0 1 2 3 4 5 6
Annual Wage (dollars) (W) 41,400 42,200 43,700 47,200 51,100 51,000 54,600
Compare the wholesalers average wage model with the retailers average wage model from Exercise 14. According to the models, will wages at retailers ever catch up with wages at wholesalers?
According to the model, when will the average annual wage for television broadcasting employees exceed $60,000? Does this estimate seem reasonable? Explain.
113
18.
Internet Usage
(Continued) 22 23 24 25 26 27 28 29 60.2 61.4 62.4 63.4 65.3 66.5 67.2 67.5
Projected Internet Usage, Hours per Person per Year (Based on 19951999 Data)
Years Since 1995 (t) 0 1 2 3 4 5 6 7 8 9
Usage per Person (hours per year) (H ) 5 10 34 61 99 135 162 187 208 228 20.
According to the model, what percentage of TV homes had cable in 2002? Does this seem reasonable? Explain.
Homes with VCRs
Percentage of TV Homes with a VCR
Years Since 1980 (t) 0 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Homes with a VCR (percent) (C ) 1.1 20.8 36.0 48.7 58.0 64.6 68.6 71.9 75.0 77.1 79.0 81.0 82.2 84.2 84.6 84.6
What do you project the Internet usage will be in 2006? Do you believe your projection is a good estimate? Explain. 19.
Homes with Cable
Percentage of TV Homes with Cable
Years Since 1970 (t) 0 5 10 15 16 17 18 19 20 21
TV Homes with Cable (percent) (C ) 6.7 12.6 19.9 42.8 45.6 47.7 49.4 52.8 56.4 58.9 (Continued)
114
Using the model from Exercise 19, determine in what year the percentage of homes with VCRs and the percentage of homes with cable was the same. (Hint: The meaning of t in the two exercises is different.)
24. If a cubic and a quartic model both appear to fit the same data set equally well, which model should you choose? Explain. 25. How many inflection points can the graph of a quartic function have? Draw the graphs of one or more quartic functions to support your conclusion.
21. The slope of a graph at a single point may be referred to as its instantaneous rate of change at that point. If the instantaneous rate of change of a graph becomes larger as x increases, what can you say about the concavity of the graph? 22. The graph of a particular cubic equation is concave up and decreasing on the interval ( 2 ` , 2 2 ) , concave up and increasing on the interval ( 2 2, 0 ) , concave down and increasing on the interval (0, 2), and concave down and decreasing on the interval ( 2, ` ) . Sketch the graph of a cubic function that meets these criteria. Then determine the value of x at which the graph of the function is increasing most rapidly. 23. The scatter plot of a data set is shown below. What type of function do you believe will best fit the data? Explain.
90 80 70 60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 9 10
26. A bend in a graph occurs when the graph changes from increasing to decreasing or from decreasing to increasing. A quadratic function has exactly one bend. A cubic function has zero or two bends. A quartic function has one or three bends. In general, how many bends may the graph of a polynomial of degree n have? 27. Can the number of x-intercepts of a graph of a polynomial function ever exceed the degree of the polynomial? Explain. 28. A function of the form f ( x ) 5 x n is called a power function. If n is a positive even integer, what is the relationship between f ( x ) and f (2 x )? 29. Find the equation of the cubic function that passes through the points ( 2 1, 0 ), ( 0, 0 ), ( 1, 0 ), and ( 3, 24 ). 30. Find the equation of the quartic function that passes through the points ( 2 1, 0 ), ( 0, 0 ), ( 1, 0 ), ( 2, 0 ), and ( 3, 24 ) .
115
Use nonlinear functions to model real-life phenomena Interpret the meaning of mathematical models in their real-world context Graph exponential functions Find the equation of an exponential function from a table Model real-life data with exponential functions
Michael Jordan is arguably the most renowned athlete of all time. With millions of dollars in endorsements in addition to his multimillion-dollar salary, he is one of the wealthiest athletes in history. Since his entry into the NBA in 1984, the average salary for NBA players has increased exponentially. In this section, we will demonstrate how exponential functions can be used to model rapidly increasing data sets such as the average salary for NBA players. We will develop exponential models from tables of data and verbal descriptions. In addition, we will show what an exponential function graph looks like.
TABLE 2.11 Average NBA Athletes Salary
NBA player average salary (in thousands of dollars)
Years Since 1980 (t) 0 5 10 15 16 17 18
Annual Salary (thousands of dollars) (S ) 170 325 750 1,900 2,000 2,200 2,600
The average annual salary of an NBA player increased from $170 thousand in 1980 to $2.6 million in 1998, as shown in Table 2.11. If we plot the NBA salary data together with the graph of the function S( t ) 5 161.4( 1.169 ) t, we see that the graph of S fits the data fairly well (Figure 2.34).
3000 2500 2000 1500 1000 500 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Years (since 1980) S (t ) = 161.4 (1.169) t
FIGURE 2.34
116
Notice that the independent variable, t, appears as an exponent. The function S is called an exponential function.
EXPONENTIAL FUNCTION
If a and b are real numbers with a 2 0, b . 0, and b 2 1, then the function y 5 abx is called an exponential function. The value b is called the base of the exponential function.
Why must b be positive? Consider the function y 5 ( 2 1 ) x . For integer values, y oscillates between 1 and 1 (Table 2.12). However, the function is undefined for numerous noninteger values of x (Table 2.13).
TABLE 2.12 TABLE 2.13
x 22 21 0 1 2
y 1 21 1 21 1
x 0.5 0.3 0.1 1.4 1.5
y Undefined Undefined Undefined 1 Undefined
On the other hand, if b . 0, then the value of bx is defined for all integer and noninteger values of x. Why dont we allow b to be 1? If b 5 1, then y 5 abx y 5 a( 1 ) x y5a
Since (1)x 1 for all x
The graph of the function y 5 a is a horizontal line and does not exhibit the same graphical behavior as all other functions of the form y 5 abx with b . 0. By eliminating the case of b 5 1, we are able to talk about a family of like functions. Exponential functions are used frequently to model growth and decay situations, such as growth in population, depreciation of a vehicle, or growth in a retirement account. When the growth or decay of a quantity is modeled by an exponential function y 5 abx , the independent variable is frequently time. The beginning value of the quantity at time x 5 0 is referred to as the initial value of the function and, as demonstrated below, is equal to a. y 5 ab0 5 a( 1 ) 5a Graphically speaking, the constant a in y 5 abx corresponds with the y-intercept of the exponential graph. In most real-life applications, a will be positive. The initial value of the NBA player salary function S( t ) 5 161.4( 1.169 ) t is 161.4. That
Since b0 1 for all nonzero real numbers b
117
is, the model estimates that the average NBA players salary was about $161 thousand in the year t 5 0 (1980). The base b of y 5 abx is often referred to as the growth factor of the function. Increasing the value of x by one unit increases y by a factor of b. The growth factor of the NBA player salary function S( t ) 5 161.4( 1.169 ) t is 1.169, since b 5 1.169. That is, the average NBA players salary is increasing by a factor of 1.169 annually. Next years salary is forecast to be 1.169 times this years salary. Recall that the x-intercept of a function occurs when y 5 0. For what values of x does abx 5 0? Lets consider the exponential function y 5 2x. Well generate a table of values and plot a few points to get an idea about what is happening graphically (Figure 2.35). x 4 3 2 1 0 1 2 3 4 y 5 2x
1 16 1 8 1 4 1 2 y 14 12 10 8 6 4 2 4 3 2 1 0 1 2 3 4 x
1 2 4 8 16
FIGURE 2.35
Observe that as x increases, y also increases. Also, observe that y is positive for all values of x. That is, there are no x-intercepts. This remains true even if we pick a negative number with a larger magnitude, say x 5 2 33. 2233 5 2233 5 1 233 1 8,589,934,592
We call the line y 5 0 a horizontal asymptote of the function y 5 2x. For sufficiently small values of x, the graph of y 5 2x approaches the graph of the line y 5 0. All exponential functions of the form y 5 abx have a horizontal asymptote at y 5 0.
Exponential Function Graphs

Exponential function graphs will take on one of the four basic shapes specified in Table 2.14. Each graph has a horizontal asymptote at y 5 0 and a y-intercept at ( 0, a ) .
118
TABLE 2.14 Exponential Function Graphs: y abx
Value of a
Value of b
Concavity of Graph
Increasing/ Decreasing
Sample Graph
y
a.0
b.1
Concave up
Increasing
x y
a.0
0,b,1
Concave up
Decreasing
x y x
a,0
b.1
Concave down
Decreasing
y x
a,0
0,b,1
Concave down
Increasing
EXAMPLE
Comparing Exponential Graphs

Compare and contrast the graphs of f ( x ) 5 3( 2 ) x and g( x ) 5 4( 0.5 ) x. (You may sketch the graphs by hand by plotting several points, or you may use technology to graph the functions.)
SOLUTION Both f and g are exponential functions. In the function equation of f, a 5 3 and b 5 2. The graph of f is concave up, since a . 0, and is increasing, since b . 1. The graph of f has a y-intercept at (0, 3). In the function equation of g, a 5 4 and b 5 0.5. The graph of g is concave up, since a . 0, and is decreasing, since b , 1. The graph of g has a y-intercept
119
at (0, 4). The graphs of both functions have a horizontal asymptote at y 5 0 (Figure 2.36).
y 25 20 15 10 g (x ) = 4(0.5) x 3 2 1 5 f (x ) = 3(2) x x
FIGURE 2.36
Properties of Exponents
As seen earlier in this section, it is often necessary to apply the properties of exponents when working with exponential functions. Since these properties are typically covered in depth in an algebra course, we will only summarize them here.
PROPERTIES OF EXPONENTS
If b, m, and n are real numbers with b . 0, the following properties hold. Property 1. b2n 5 1 bn Example 1. 223 5 1 23
2. bm ? bn 5 bm 1 n 3. bm 5 bm 2 n bn
2. 32 ? 34 5 36 3. 56 5 52 54
4. bmn 5 ( bm ) n 5 ( bn ) m
4. 62 ? 3 5 ( 62 ) 3 5 ( 63 ) 2
TABLE 2.15
Finding an Exponential Function from a Table

We can easily determine if a table of values models an exponential function by calculating the ratio of consecutive outputs for evenly spaced inputs. The ratio will be constant if the table of values models an exponential function. Consider Table 2.15.
x 0 2 4 6
y 6 24 96 384
120
The domain values are equally spaced (two units apart). Calculating the ratios of the consecutive range values, we get 24 54 6 96 54 24 384 54 96
In each case, the ratio was 4. Therefore, the table of values models an exponential function. But what is the equation of the function? We can find the equation of the function algebraically. We know that an exponential function must be of the form y 5 abx. Substituting the point (0, 6) into the equation and solving, we get y 5 abx 6 5 ab0 65a Since a 5 6, y 5 6b . Substituting the point (2, 24) into the equation y 5 6bx, we get
x
Substitute x 0 and y 6
y 5 6bx 24 5 6b2 45b b52

2
Substitute x 2 and y 24 b 2 2 2 since the base of an exponential function must be positive
Therefore, the exponential function that models the table data is y 5 6( 2 ) x .
EXAMPLE
Finding an Exponential Equation from a Table

Find the equation of the exponential function modeled by Table 2.16.
TABLE 2.16
x 1 2 3 4
y 21 63 189 567
SOLUTION We know that an exponential function must be of the form y 5 abx .
Substituting the point (2, 63) into the equation, we get y 5 abx 63 5 ab2 y 5 abx 21 5 ab1
Substitute x 1 and y 21 Substitute x 2 and y 63
Substituting the point (1, 21) into the equation y 5 abx, we get
We can eliminate the a variable by dividing the first equation by the second equation. (Note: What we are really doing is dividing both sides of the first equation by the same nonzero quantity, expressed in two different forms.) 63 ab2 5 1 21 ab 35 a 221 ?b a
Since bm 5 bm 2 n bn
351?b b53
121
We may then substitute the value of b into either equation to find a. 21 5 ab1 21 5 a ? 31 21 5a 3 a57 The exponential function that models the table data is y 5 7( 3 ) x. When using the method of dividing one equation by the other, computations will tend to be easier if we divide the equation with the largest exponent by the equation with the smallest exponent. The technique shown in Example 2 may be generalized for all exponential functions. Using the result from the generalized solution will allow us to determine the exponential equation more quickly. Consider an exponential function y 5 abx whose graph goes through the points ( x1, y1 ) and ( x2, y2 ) . We have y 5 abx y1 5 abx and y 5 abx y2 5 abx
2 1
Substitute b 5 3
Substitute x 5 x1 and y 5 y1
Substitute x 5 x2 and y 5 y2
Dividing the second equation by the first equation yields a quick way to calculate b. y2 ab x2 5 y1 ab x1 y2 a 5 b x22 x1 y1 a y2 5 b x22 x1 y1
1 1 y a 2 b x22 x1 5 ( b x22 x1 ) x22 x1 y1 1 y 2 2 x1 a 2 b x22 x1 5 b x x2 2 x1 y1 1 y b 5 a 2 b x22 x1 y1
122
Using the points (1, 21) and (3, 189) from Example 2, we get
1 y b 5 a 2 b x22 x1 y1
b5 a
1 189 32 b 1 21
Substitute x1 1, x2 3, y1 21, and y2 189
5 91>2 53
5 "9
Recall that for b $ 0, b1>2 5 "b
This method is especially useful when checking to see if a data table with unequally spaced inputs is an exponential function. If it is an exponential function, the value of b will be constant regardless of which two points we substitute into the formula.
Using Exponential Regression to Model Data

Data sets with near-constant ratios of change may be modeled using the exponential regression feature on our graphing calculator. It is often helpful to do a scatter plot of the data first to see if the graph looks like an exponential function.
EXAMPLE
Using Exponential Regression to Model the Population of Akron, Ohio

The population of Akron, Ohio, is shown in Table 2.17. Model the population with an exponential function and forecast the population of Akron in 2010.
SOLUTION We first draw a scatter plot so that we can visually predict whether an exponential function will fit the data well (Figure 2.37). (Note that because we have zoomed in on the data, the x axis is given by y 5 210 instead of y 5 0 .)
280 270 Population, y (in thousands) 260 250 240 230 220 210 0 10 20 30 Years, x (since 1970) 40
TABLE 2.17
Years Since 1970 (x) 0 10 20 30
Population (thousands) (y) 275 237 223 217
FIGURE 2.37
Since the scatter plot is decreasing and concave up, an exponential function may fit the data well. Using techniques given in the Technology Tip following
123
this example, we can calculate an exponential model for the data set (Figure 2.38).
280 270 Population, y (in thousands) 260 250 240 230 220 210 0 10 20 30 Years, x (since 1970) 40 y = 266.1(0.9923) x
FIGURE 2.38 TABLE 2.18
Year (x) 0 10 20 30
Population 2 210 (thousands) ( y) 65 27 13 7
The model y 5 266.1( 0.9923 ) x is a terrible fit for the data. What happened? Recall that an exponential function of the form y 5 abx has a horizontal asymptote at y 5 0 . From the scatter plot of the data, it appears that the population is approaching a constant value of y 5 210 instead of y 5 0 . (This value is not precise. We are guessing what the horizontal asymptote would be based on the scatter plot of the data set.) If we subtract 210 from each of the y values, we get a new data set with a horizontal asymptote of y 5 0 . This is referred to as an aligned data set (Table 2.18). Using exponential regression to find a model for the aligned data set yields y 5 60.81( 0.9285 ) x , a function that fits the aligned data well (Figure 2.39).
70 Population minus 210, y (in thousands) 60 50 40 30 20 10 0 0 10 20 30 Years, x (since 1970) 40 y = 60.81(0.9285) x
FIGURE 2.39
Subtracting 210 from each of the y values of the original data set had the graphical implication of moving the data down 210 units. To return the aligned data model to the position of the original data, we need to move the model up 210 units. To do this, we add back the 210. P( x ) 5 60.81( 0.9285 ) x 1 210
Add 210 to shift the data graph upward
The model created from the aligned data fits those data much better than the unaligned data model (Figure 2.40).
124
280 270 260 Population, y (in thousands) 250 240 230 220 210 200 0
P(x) = 60.81(0.9285) x + 210 y = 266.1(0.9923) x
10
20 30 Years, x (since 1970)
40
FIGURE 2.40
To estimate the population of Akron, Ohio, in 2010, we evaluate P( x ) 5 60.81( 0.9285 ) x 1 210 at x 5 40 . P( 40 ) 5 60.81( 0.9285 ) 40 1 210 < 213 We forecast that the population of Akron, Ohio, in 2010 will be 213,000.
Just In Time
Vertical and Horizontal Shifts
The graph of any function may be moved on the coordinate grid through the use of vertical and horizontal shifts. Vertical shifts are achieved by changing the value of the output variable by a constant amount. Horizontal shifts are achieved by changing the value of the input variable by a constant amount. For the following examples, we use the function f ( x ) 5 2x. In each case, k and c are assumed to be nonnegative constants. Shift Vertical, upward Vertical, downward Horizontal, left Horizontal, right General Form y 5 f (x) 1 k y 5 f (x) 2 k y 5 f (x 1 c) y 5 f (x 2 c) Example g( x ) 5 2x 1 4 shifted up by 4 units g( x ) 5 2x 2 4 shifted down by 4 units g( x ) 5 2x 1 3 shifted left by 3 units g( x ) 5 2x 2 3 shifted right by 3 units
125
The methods of aligning data and of performing exponential regression are detailed in the following Technology Tips.
Aligning a Data Set 1. Enter the data using the Statistics Menu List Editor. (Refer to Section 1.3 if youve forgotten how to do this.)
2. Move the cursor to the top of L3. We want the entries in L3 to equal the entries in L2 minus the amount of the vertical shift (in this case 210). To do this, we must enter the equation L3=L2-210. 3. Press 2nd then 2 to place L2 on the equation line at the bottom of the viewing window. Then press and 2 1 0 to subtract 210.
4. Press ENTER to display the list of aligned values in L3.
126
Exponential Regression 1. Enter the data using the Statistics Menu List Editor. (Refer to Section 1.3 if youve forgotten how to do this.)
2. Bring up the Statistics Menu Calculate feature by pressing STAT and using the blue arrows to move to the CALC menu. Then select item :ExpReg and press ENTER .
3. If the data to be evaluated are in L1 and L2, press ENTER . Otherwise, go to step 4. (If you want to automatically paste the regression equation into the Y editor, press the key sequence VARS Y-VARS; 1:Function; 1:Y1 before pressing ENTER .) 4. If the data to be evaluated are in L1 and L3, press the key sequence 2nd , 2nd 3 to place the 1 entries L1 and L3 on the home screen.
5. Press ENTER to display the exponential model. (If you want to automatically paste the regression equation into the Y editor, press the key sequence VARS Y-VARS; 1:Function; 1:Y1 before pressing ENTER .)
127
Finding an Exponential Function from a Verbal Description

Much of the information we encounter in the media is given to us in terms of percentages. Consider these typical news headlines. Gasoline Prices Flare Up 9 Percent and Home Values Increase by 6 Percent. Anything that increases or decreases at a constant percentage rate may be modeled with an exponential function. The value of the growth factor b is 1 1 r, where r is the decimal form of the percentage (i.e., 5 percent = 0.05). Note that if r , 0, then b , 1. A negative rate of growth is referred to as depreciation, while a positive rate of growth is referred to as appreciation.
ANNUAL GROWTH RATE
Let y 5 abt model the amount of a quantity at time t years. The annual growth rate r of the quantity y is given by r 5 b 2 1. Similarly, the annual growth factor is b 5 r 1 1.
For example, the author ordered an appraisal of his home in August 2004. In her report, the appraiser asserted that home values in the area were increasing at a rate of 5 percent per year. In other words, r 5 0.05. The corresponding growth factor is b 5 1.05. To forecast the value of his home in August 2005, the author multiplied the appraised value by the growth factor.
EXAMPLE
Using an Exponential Model to Forecast a Cars Value

New cars typically lose 50 percent of their value in the first three years. Many cars lose more than 25 percent of their value in the first year alone. (Source: Runzheimer International.) Depreciating approximately 13.4 percent annually, the Saturn Sedan 2 holds its value better than most vehicles. Estimate the value of a $12,810 Saturn Sedan 2 three years after it is purchased.
SOLUTION Since the car is depreciating at a constant percentage rate, an exponential function may be used to model the data. The initial value of the car is $12,810, so a 5 12,810. The car is losing value at 13.4 percent per year, so r 5 2 0.134. Since b 5 1 1 r, the exponential function is
V( t ) 5 12,810( b ) t
5 12,810 31 1 ( 2 0.134 ) 4 t 5 12,810( 0.866 ) dollars

t
Since b 5 1 1 ( 2 0.134 )
where t is the age of the car in years. We want to know the value of the car after three years, so t 5 3. V( 3 ) 5 12,810( 0.866 ) 3 5 12,810( 0.649 ) 5 8320 We estimate that the car will be valued at $8320 when it is three years old.
128
EXAMPLE
Using an Exponential Function to Model Inflation

Inflation (rising prices) causes money to lose its buying power. In the United States, inflation hovers around 3 percent annually. (a) If a candy bar costs $0.59 today, what will it cost 10 years from now? (b) When will a candy bar cost $1.00?
SOLUTION Since were assuming a constant percentage increase in the price of the candy bar, an exponential model should be used. The initial value is $0.59, so a 5 0.59. The price of the candy bar is increasing by 3 percent annually, so r 5 0.03. The cost of the candy bar is given by
C( t ) 5 0.59( 1 1 0.03 ) t dollars 5 0.59( 1.03 ) t dollars where t is the number of years from today. (a) Evaluating the function at t 5 10, we get C( 10 ) 5 0.59( 1.03 ) 10 5 0.59( 1.344 ) 5 0.79 We estimate that the candy bar will cost $0.79 ten years from now. (b) We want to know when C( t ) 5 1. 1 5 0.59( 1.03 ) t 1 5 1.03t 0.59 1.695 5 1.03t At this point, we are stuck. We havent yet developed the mathematical machinery (logarithms) to get the variable out of the exponent. We can, however, attack the problem graphically by breaking the equation 1.695 5 1.03t into two separate functions: y1 5 1.695 and y2 5 1.03t . By graphing these functions simultaneously, we can determine the point of intersection of the two functions using our graphing calculator. Using the graphing calculator, we generate the graph in Figure 2.41 and determine the point of intersection.
FIGURE 2.41
The functions intersect when t 5 17.85. In about 18 years, we estimate that the cost of a candy bar will be $1.00.
129
We also could have used the guess-and-check method to approximate the result by evaluating C( t ) 5 0.59( 1.03 ) t at various values of t. Since C( 17 ) 5 0.59( 1.03 ) 17 5 0.9752 and C( 18 ) 5 0.59( 1.03 ) 18 5 1.004
we determine that the price reaches $1.00 between the seventeenth and eighteenth years.
Aligning a Data Set 1. Enter a data set in columns A and B.
2. We want the entries in column C to equal the entries in column B minus the amount of the vertical shift (in this case, 210). To accomplish this, enter the expression y 2 210 in cell C2 using the cell reference B2 to represent y. 3. Click the lower right-hand corner of the box in cell C2 and drag down to the cell in the C column adjacent to the last domain value. This pastes the formula in cell C2 to the remainder of the cells. The reference to B2 automatically updates to reference the adjacent cell. The aligned data now appear in column C.
130
Exponential Regression 1. Draw a scatter plot of a data set.

70 60 50 40 30 20 10 0 0 5 10 15 20 25 30 35
2. Right-click on any point of the scatter plot to open a drop-down menu. From the drop-down menu, select Add Trendline to open the Add Trendline dialog box. Then select Exponential.
(Continued)
131
4. The graph and equation of the exponential function that best fits the data are displayed. The coefficient of determination r2 is also displayed.
70 60 50 40 30 20 10 0 0 5 10 15 20 25 30 35 y = 60.808e 0.0742x R2 = 0.9939
2.3 Summary
In this section, you learned how exponential functions can be used to model rapidly increasing (or decreasing) data sets. You graphed exponential functions and developed exponential models from tables of data and verbal descriptions.

1. How do linear and exponential functions differ from each other? 2. In order for y 5 abx to be an exponential function, what restrictions must be placed on b? 3. What is a horizontal asymptote? 4. How can you tell by looking at the equation that y 5 247( 0.971 ) x represents a decreasing exponential function? 5. How do exponential functions differ from polynomial functions? 6. Do the graphs of exponential functions have vertical asymptotes? 7. Is y 5 3x 22 an exponential function? Explain. 8. What effect does the value of a have on the graph of y 5 ab x ?
2.3 Exercises
In Exercises 110, do the following: (a) Determine if the graph of the function is increasing or decreasing. (b) Determine if the graph of the function is concave up or concave down. (c) Identify the coordinate of the y-intercept. (d) Graph the function to verify your conclusions. 1. y 5 4( 0.25 ) x 3. y 5 0.5( 2 ) x 2. y 5 2 2( 0.5 ) x 4. y 5 6( 0.1 ) x
132
5. y 5 0.4( 5 ) x 7. y 5 2 1.2( 2.3 ) 9. y 5 3( 0.9 )

x x
6. y 5 2 0.1( 0.2 ) x 8. y 5 5( 0.4 )

x x
18.
x 2 5 9 11
y 4 32 512 2,048 y 256 8 2 0.25 x 0 2 5 7 y 10,000 900 24.3 2.187
10. y 5 2 5( 3 )
In Exercises 1120, use algebraic methods to find the equation of the exponential function that fits the data in the table. 11. x 0 1 2 3 13. x 1 2 3 4 15. x 2 4 6 8 16. x 2 4 6 8 17. x 1 2 4 5 y 1 4 64 256 100 1 0.01 0.0001 y 2 6 18 54 y 10 20 40 80 y 16 4 1 0.25 y 14. 12. x 0 1 2 3 x 1 2 3 4 y 3 12 48 192 y 10 50 250 1,250
19.
x 0 5 7 10
20.
In Exercises 2125, use exponential regression to model the data in the table. Use the model to predict the value of the function when t 5 25, and interpret the real-world meaning of the result. The Consumer Price Index is used to measure the increase in prices over time. In each of the following tables, the index is assumed to have the value 100 in the year 1984. 21.
Dental Prices
Price of Dental Services
Years Since 1980 (t) 0 5 10 15 20
Price Index (I ) 78.9 114.2 155.8 206.8 258.5
133
22.
Wine Prices
25.
Entertainment Prices
Price of Wine Consumed at Home
Years Since 1980 (t) 0 5 10 15 20
Price of Admission to Entertainment Venues
Price Index (I ) 89.5 100.2 114.4 133.6 151.6
Years Since 1980 (t) 0 5 10 15 20
Price Index (I ) 83.8 112.8 151.2 181.5 230.5
23.
Television Prices
Price of a Television Set
In Exercises 2630, find the exponential function that fits the verbal description. 26. Salaries Instructors salaries are $45,000 per year and are expected to increase by 3.5 percent annually. What will instructors salaries be five years from now? 27. Savings Account A savings account balance is currently $235 and is earning 2.32 percent per year. When will the balance reach $250? 28. The cost of a 27-inch flat-screen television in 2002 was $599.99. (Source: www.bestbuy.com.) Television prices are expected to decrease by 16 percent per year. How much is the flat-screen television expected to cost in 2007?
Television Price
Years Since 1980 (t) 0 5 10 15 20
Price Index (I ) 104.6 88.7 74.6 68.1 49.9
24.
Alcohol Prices
29.
Price of Distilled Spirits Consumed at Home
Years Since 1980 (t) 0 5 10 15 20
Price Index (I ) 89.8 105.3 125.7 145.7 162.3
Reserved seating at a Dave Matthews Band concert at the Gorge Amphitheater in George, Washington, cost $59.90 in September 2004. (Source: www.ticketmaster.com.) If admission fees for concerts are expected to increase by 28 percent per year, what is the expected price of a Dave Matthews Band ticket in September 2006?
Concert Admission
30. Tuition Tuition is currently $2024 per year and is increasing by 12 percent annually. In how many years from now will tuition reach $3567?
134
In Exercises 3135, find the solution by solving the equation graphically. 31. ( 1.2 ) x 5 5 33. ( 1.05 ) x 5 2 35. ( 0.3 ) x 5 0.09 32. ( 0.9 ) x 5 0.5 34. ( 1.12 ) x 5 2
42.
Retirement Investments Repeat Exercise 41, except this time invest twice as much in the Social Choice account as in the Retirement Annuity. What factors should the investor consider before changing the amount invested in each of the accounts? Retirement Investments Based on data from January 1, 2002, through December 31, 2002, the annual rate of return on the CREF Social Choice account was 22.98 percent, substantially lower than the 10-year average return of 8.67 percent. Over the same period, the CREF Stock account earned 20.73 percent, substantially lower than its 10-year average return of 7.69 percent. (Source: TIAA-
43.
36. What is the relationship between a growth factor and a percentage growth rate? 37. The graph of the exponential function y 5 abx does not have any x-intercepts. Explain. 38. Explain why it is sometimes important to align a data set before performing exponential regression. Then explain how the equation of a vertically aligned exponential model differs from that of a standard exponential model. 39. The phrase Its growing exponentially is commonly used in the media to describe anything that is growing rapidly. In order for this statement to be mathematically accurate, what properties must the data set have? 40. What effect does the growth factor b have on the shape of the graph of y 5 abx? 44.
CREF.)
An investor invested $1000 in each of the two accounts at the start of 2002. (a) What was the value of the investment in each account at the end of 2002? (b) Assuming that in each year following 2002, the investments earned an annual return equal to the 10-year average, predict how long it will take for the combined value of the accounts to reach $2000.
Retirement Investments Based on data from January 1, 2002, through December 31, 2002, the annual rate of return of the CREF Growth account was 30.06 percent, substantially lower than the five-year average return of 5.48 percent. Over the same period, the CREF Inflation-linked Bond account earned 16.32 percent, substantially higher than its fiveyear average return of 8.33 percent. (Source:
41.
Retirement Investments An investor has $3000 to invest in two different investment accounts: CREF Social Choice and TIAA Retirement Annuity. Based on data from January 1, 1993, through December 31, 2002, the average annual return on the CREF Social Choice account was 8.67 percent, and the average annual return on the TIAA Retirement Annuity was 6.93 percent. (Source: TIAA-CREF.) The investor plans to invest twice as much in the Retirement Annuity as in the Social Choice account. Assuming that the future rates of return will be the same as the past rates of return, do the following: (a) Find the equation of the function for the predicted value of the $3000 investment t years from now. (b) What is the predicted value of the investment 20 years from now?
TIAA-CREF.)
An investor invested $1000 in each of the two accounts at the start of 2002. (a) What was the value of the investment in each account at the end of 2002? (b) Assuming that in each year following 2002, the investments earned an annual return equal to the five-year average, predict how long it will take for the combined $2000 investment in the accounts to double. 45. How long will it take for a $1000 investment earning 10 percent annually to equal the value of a $2000 investment earning 6 percent annually?
Investments
135
Use nonlinear functions to model real-life phenomena Interpret the meaning of mathematical models in their real-world context Graph logarithmic functions Model real-life data with logarithmic functions Apply the rules of logarithms to simplify logarithmic expressions Solve logarithmic equations
In Section 2.3, we showed that the NBA players average salary could be modeled by S( t ) 5 161.4( 1.169 ) t thousand dollars where t is the number of years since 1980. We could use the model to estimate the average salary in any year t. But what if we wanted to algebraically determine the year in which the salary would be $1,000,000? In this section, we will demonstrate how logarithmic functions can be used to model data sets that are increasing at a decreasing rate. We will show how to graph logarithmic functions and will illustrate graphically the inverse relationship between logarithmic and exponential functions. We will explain how the rules of logarithms can be used to simplify logarithmic expressions and solve logarithmic equations. We will also show how to develop logarithmic models from tables of data, such as NBA players salaries, by using logarithmic regression.
We want to algebraically determine the year in which the NBA players average salary reached $1 million. We can estimate the year by looking at the data in Table 2.19.
TABLE 2.19 NBA Players Average Salary
Salary (thousands of dollars) (s) 170 325 750 1,900 2,000 2,200 2,600
Years Since 1980 (t) 0 5 10 15 16 17 18
Notice that salary is the independent variable and years is the dependent variable. The average salary was $750,000 in 1990 and $1,900,000 in 1995, so we know that $1,000,000 was reached sometime between 1990 and 1995. Returning to the exponential model for the NBA players average salary, we have s( t ) 5 161.4( 1.169 ) t thousand dollars
136
where t is the number of years since 1980. Recall that $1,000,000 is equal to $1000 thousand. We must determine when s( t ) 5 1000 . In other words, we must solve the equation 1000 5 161.4( 1.169 ) t 1000 161.4( 1.169 ) 5 161.4 161.4 6.196 5 1.169t If we want to solve this equation algebraically, at this point we are stuck. We must come up with some way to get the t out of the exponent. Observe that t is the exponent that we place on 1.169 in order to get 6.196. We need to create a symbol to represent the phrase is the exponent we place on. The chosen symbol, as bizarre as it may seem, is log. t is t= the exponent we place on log
1.169
t
Since the units of the salary are thousands of dollars
Compressing these terms into a single equation yields t 5 log1.169 6.196 Using the calculator LOG button and some log rules that will be introduced later, we determine that t < 11.7 years. That is, between 1991 ( t 5 11 ) and 1992 ( t 5 12 ) , the NBA players average salary reached $1 million.
Logarithmic Functions
The notation log is short for logarithm. It is helpful to think of log as representing the phrase the exponent we place on. The equation y 5 logb( x ), which is read y equals log base b of x, means y is the exponent we place on b in order to get x. For example, y 5 log2( 8 ) means y is the exponent we place on 2 in order to get 8. That is, 2y 5 8. Since 23 5 8, we conclude that y 5 3.
EXAMPLE 1
Calculating a Logarithm
Find the value of y given y 5 log3( 9 ) .
SOLUTION y 5 log3( 9 ) means y is the exponent we place on 3 in order to get 9. That is, 3y 5 9. Since 32 5 9, we conclude that y 5 2.
LOGARITHMIC FUNCTION
Let b and x be real numbers with b . 0 and b 2 1 and x . 0 . The function y 5 logb( x ) is called a logarithmic function. The value b is called the base of the logarithmic function. We read the expression logb( x ) as log base b of x. The use of parentheses around the x in y 5 logb( x ) is optional. That is, y 5 logb( x ) is equivalent to y 5 logb x. However, we will consistently use paren-
1.169
in order to get 6.196 6.196
137
theses throughout this text to remind ourselves that the expression logb is meaningless by itself. Although any positive number not equal to 1 may be used as the base of a logarithmic function, the most commonly used bases are 10 and e. The irrational number e is approximately equal to 2.71828. The number e, like , has many wonderful uses. Although we will see some of the power of e in this chapter, we will witness its full strength when we move into calculus. The number e is generated by calculating A 1 1 infinitely big as n goes to infinity. Surprisingly, A 1 1 n B < 2.718281828 for
1
n
1 n
for infinitely large n. You might think that A 1 1 n B will get

1
n
large values of n (see Table 2.20).
TABLE 2.20
n 1 10 100 1,000 10,000 100,000
1 n a1 1 b n 2.000 2.594 2.705 2.717 2.718 2.718
The value of e, 2.718281828. . . , may be accessed on the TI-83 Plus calculator by pressing the key sequence 2nd and . The expression log10( x ) is often written log( x ) and is called the common log. We will assume that the base is 10 when no base is written. The expression loge( x ) is typically written ln( x ) and is called the natural log.
Graphs of Logarithmic Functions

y 3 2 1 0 1 2 3 1 2 3 g (x) = log 3(x) h (x) = log 4(x) 4 5 f (x) = log 2(x)
Like polynomial and exponential function graphs, logarithmic function graphs share many common characteristics. Consider the logarithmic functions f ( x ) 5 log2( x ), g( x ) 5 log3( x ), and h( x ) 5 log4( x ) (Figure 2.42). Each graph is concave down and increasing. Each graph passes through the point ( 1, 0 ) and appears to have a vertical asymptote at x 5 0 (the y axis). In fact, for all positive values of b not equal to 1, the graph of y 5 logb( x ) has a vertical asymptote at x 5 0 and an x-intercept at (1, 0). Logarithmic functions are undefined for x , 0. The shape of the graph is determined by b. The base b controls how rapidly the graph increases (or decreases). For each of the graphs in Figure 2.42, b . 1. What if b , 1? Consider the logarithmic functions f ( x ) 5 log0.3( x ), g( x ) 5 log0.5( x ), and h( x ) 5 log0.7( x ) (Figure 2.43).
FIGURE 2.42
138
CHAPTER 2 Nonlinear Models y 2 1 0 1 2 3 4 g (x) = log 0.5(x) h (x) = log 0.7(x) 1 2 3 4 5 f (x) = log 0.3(x) x
FIGURE 2.43
Each graph is concave up and decreasing. However, these graphs also have a vertical asymptote at x 5 0 and pass through the point ( 1, 0 ) . Our graphical observations are summarized in Table 2.21.
TABLE 2.21 Logarithmic Function Graphs: y 5 logb( x )
Value of b
Concavity
Increasing/Decreasing
y
Graph
b.1
Concave down
Increasing
0,b,1
Concave up
Decreasing
x
EXAMPLE
Determining the Shape of a Logarithmic Function

Determine the concavity and increasing/decreasing behavior of y 5 log2( x ) and y 5 log0.5( x ). Then graph both functions to verify your results.
SOLUTION Since 2 . 1, y 5 log2( x ) will be concave down and increasing. In order to graph the function by hand, we must generate a table of values for the function. You may find it helpful to write y 5 log2( x ) as x 5 2y , then select the values of y and calculate the values of x (Table 2.22). Plotting the points and connecting the dots yields the graph shown in Figure 2.44.
139
TABLE 2.22
x52 1 2 4 8 16
y 0 1 2 3 4
4 3 2 1 0 1 2 3 4 2 4 6 8 10 12 14 16 x f (x) = log 2(x)
FIGURE 2.44
Since 0.5 , 1, y 5 log0.5( x ) will be concave up and decreasing. You may find it helpful to write y 5 log0.5( x ) as x 5 ( 0.5 ) y , then select the values of y and calculate the values of x (Table 2.23). Plotting the points and connecting the dots yields the graph shown in Figure 2.45.
TABLE 2.23
y 4
x 5 ( 0.5 ) y 16 8 4 2 1
y 24 23 22 21 0
3 2 1 0 1 2 3 4 2 4 6 8 10 12 14 16 f (x) = log 0.5(x) x
FIGURE 2.45
EXAMPLE
Comparing the Graphs of Logarithmic Functions

Graph f ( x ) 5 ln( x ) and g( x ) 5 log( x ) simultaneously. Then determine where f ( x ) , g( x ) and where g( x ) . f ( x ) .
SOLUTION
y f (x) = ln (x) 1 g (x) = log (x) x 3 4
0 1 1 2
FIGURE 2.46
140
(To graph the function on the TI-83 Plus, use the LOG and LN keys as appropriate.) Based on the graphs in Figure 2.46, we conclude that for x , 1, ln( x ) , log( x ) . For x . 1 , ln( x ) . log( x ) .
Relationship Between Logarithmic and Exponential Functions

Exponential and logarithmic functions are intimately related. Any logarithmic function may be rewritten as an exponential function (and vice versa).
LOGARITHMIC AND EXPONENTIAL FUNCTION RELATIONSHIP
For b . 0 with b 2 1 and x . 0 , the following statements are equivalent: 1. y 5 logb( x ) 2. by 5 x
Consider tables of values for the functions f ( x ) 5 10x and g( x ) 5 log( x ) (Table 2.24).
TABLE 2.24
x 22 21 0 1 2
f ( x ) 5 10x 0.01 0.1 1 10 100
x 0.01 0.1 1 10 100
g( x ) 5 log( x ) 22 21 0 1 2
y 4 3 2 1 1 1 1 2
g (x) = e x y=x
f (x) = ln(x)
FIGURE 2.47
Notice that the input of the exponential function is the output of the logarithmic function. Similarly, the output of the exponential function is the input of the logarithmic function. A point ( x1, y1 ) is on the graph of f if and only if the point ( y1, x1 ) is on the graph of g. Functions with this property are called inverse functions. Exponential functions and logarithmic functions are inverse functions of each other. The inverse relationship of logarithmic and exponential functions may also be seen by graphing y 5 ln( x ) and y 5 e x together with the graph of y 5 x (Figure 2.47). The graph of y 5 ln( x ) is the reflection of y 5 e x about the line y 5 x. That is, if we folded our paper along the line y 5 x, the graphs of the two functions would lie directly on top of each other. When we solve a logarithmic function y 5 logb( x ) for x, we get the exponential function x 5 b y. This relationship between logarithmic and exponential functions will be critical to solving the logarithmic and exponential equations.
Rules of Logarithms
There are several rules that are used to manipulate logarithms. Each rule is important; however, we will use Rule 7 most extensively. It is so essential that we comically call it the mother of all log rules.
141
RULES OF LOGARITHMS
If x, y, and n are real numbers with x . 0 and y . 0 , then the following rules apply. Common Logarithm 1. log( 1 ) 5 0 2. log( 10 ) 5 1 3. log( 10 ) 5 x
x
Natural Logarithm 1. ln( 1 ) 5 0 2. ln( e ) 5 1 3. ln( ex ) 5 x 4. eln(x) 5 x 5. ln( xy ) 5 ln( x ) 1 ln( y ) x 6. ln a b 5 ln( x ) 2 ln( y ) y 7. ln( xp ) 5 p ln( x )
4. 10log(x) 5 x 5. log( xy ) 5 log( x ) 1 log( y ) x 6. log a b 5 log( x ) 2 log( y ) y 7. log( xp ) 5 p log( x )
In fact, Rules 1, 5, 6, and 7 apply for any base, not just 10 and e. Rules similar to Rules 2 through 4 may easily be formulated for any other base. In our applications, we will be using base 10 and base e predominantly. This is because the following change-of-base formula allows us to convert any logarithm into a logarithm with one of these two bases.
CHANGE-OF-BASE FORMULA
logb( a ) 5
log( a ) ln( a ) 5 log( b ) ln( b )
The change-of-base formula is easily derived. Let y 5 logb( a ) by 5 a log( b ) 5 log( a )

y
Rewrite in exponential form Take the log of both sides Apply log Rule 7 Divide both sides by log(b)
y log( b ) 5 log( a ) y5 log( a ) log( b )
But y 5 logb( a ) so logb( a ) 5 log( a ) log( b )
The ability to apply the log rules to manipulate expressions and equations is one of the critical skills required to be successful in solving exponential and logarithmic equations. To assist you in mastering those skills, we will do several skill-and-drill examples.
142
EXAMPLE
Using Log Rules to Simplify a Logarithmic Expression

Use the log rules to rewrite the logarithmic expression as a single term. log( 3 ) 2 log( 27 ) 1 log( 12 )
SOLUTION
log( 3 ) 2 log( 27 ) 1 log( 12 ) 5 log a 5 log a 5 log a
3 b 1 log( 12 ) 27 3 ? 12 b 27 36 b 27
Rule 6 Rule 5
4 5 log a b 3
EXAMPLE
Writing a Logarithmic Expression as a Single Logarithm

Rewrite the logarithmic expression as a single logarithm. log( x ) 1 log( x 3 ) 2 4 log( x 2 )
SOLUTION
log( x ) 1 log( x 3 ) 2 4 log( x 2 ) 5 log( x ) 1 3 log( x ) 2 2 ? 4 log( x ) 5 4 log( x ) 2 8 log( x ) 5 2 4 log( x ) The solution may also be written as 2 log( x 4 ) , log( x 24 ) , or log a 1 b x4
Rule 7 Combine like terms Combine like terms
When checking a solution with the answers in the back of the book, it is important to recognize that the correct solution may be written in a variety of forms.
EXAMPLE
Writing a Logarithmic Expression as a Single Logarithm

Rewrite the expression 2 log A 3x B 2 log( x 3 ) as a single logarithm.
2
SOLUTION
2 log a
2 b 2 log( x 3 ) 3x 2 2 5 log a b 2 log( x 3 ) 3x 5 log c (2) 2 d 1 log( x 3 ) 21 ( 3x ) 2
Rule 7 Rule 7, rules of exponents
143
5 log a 5 log a 5 log a 5 log a
4 b 1 log( x 23 ) 9x 2 4 ? x 23 b 9x 2 4 1 ? 3b 2 9x x 4 b 9x 5
Rules of exponents Rule 5 Rules of exponents Rules of exponents and fractions
EXAMPLE
Solving a Logarithmic Equation

Solve the equation log3( x ) 5 4 for x.
SOLUTION
log3( x ) 5 4 34 5 x x 5 81
Rewrite in exponential form
EXAMPLE
Solving an Exponential Equation

Solve the equation 3x 5 12 for x.
SOLUTION
1
3x 5
1 12 1 b 12
Take the natural log of both sides Rule 7 and Rule 6 Rule 1
ln( 3x ) 5 ln a
x ln( 3 ) 5 ln( 1 ) 2 ln( 12 ) x ln( 3 ) 5 0 2 ln( 12 ) x ln( 3 ) 5 2 ln( 12 ) x52 ln( 12 ) ln( 3 )
ln(12 )
Divide by ln(3)
It is important to note that 2 ln( 3 ) 2 2 ln A 3 B .

12
x < 2 2.262
Logarithmic Models
Functions that are increasing at a decreasing rate may often be modeled by logarithmic functions. Using the logarithmic regression feature, LnReg, on our calculator returns a function of the form y 5 a 1 b ln( x ). The a shifts the graph of y 5 b ln( x ) vertically by Z a Z units. The resultant graph passes through the point ( 1, a ) instead of ( 1, 0 ) .
144
EXAMPLE
Using Logarithmic Regression to Forecast the Population of Hawaii

Find the logarithmic model that best fits the data in Table 2.25. Then evaluate the function at p 5 1700 and interpret your result.
SOLUTION Using the logarithmic regression feature on our calculator (as shown in the following Technology Tip), we get
TABLE 2.25 Projected Population of Hawaii
People (thousands) ( p) 1,187 1,257 1,342 1,553 1,812
Years Since 1995 (T ) 0 5 10 20 30
y 5 2 496.0 1 70.17 ln( x ) or, in terms of our variables, T( p ) 5 2 496.0 1 70.17 ln( p ) Evaluating the function when p 5 1700 , we get T( 1700 ) 5 2 496.0 1 70.17 ln( 1700 ) 5 2 496.0 1 70.17( 7.438 ) 5 2 496.0 1 522.0 5 26.0 The population of Hawaii is projected to reach 1,700,000 in 2021 (26 years after 1995).
Logarithmic Regression 1. Enter the data using the Statistics Menu List Editor. (Refer to Section 1.3 if youve forgotten how to do this.)
2. Bring up the Statistics Menu Calculate feature by pressing STAT and using the blue arrows to move to the CALC menu. Then select item 9:LnReg and press ENTER .
3. If you want to automatically paste the regression equation into the Y editor, press the key sequence VARS Y-VARS; 1:Function; 1:Y1 and press ENTER . Otherwise, press ENTER .
145
Logarithmic Regression 1. Draw a scatter plot of a data set.

35 30 25 20 15 10 5 0 1000 1500 2000
2. Right-click on any point of the scatter plot to open a drop-down menu. From the drop-down menu, select Add Trendline to open the Add Trendline dialog box. Then select Logarithmic.
(Continued)
146
4. The graph and equation of the logarithmic function that best fits the data are displayed. The coefficient of determination r 2 is also displayed.
35 30 25 20 15 10 5 0 1000
y = 70.174Ln(x) 496.01 R2 = 0.9977
1500
2000
2.4 Summary
In this section, you learned how to graph logarithmic functions. You discovered the inverse relationship between logarithmic and exponential functions and practiced using rules of logarithms to simplify logarithmic expressions and solve logarithmic equations. You also developed logarithmic models from data tables by using logarithmic regression.

1. What is a logarithm? 2. Write ln( 8 ) 2 ln( 4 ) in an alternative form. 3. Write log2 ( 12 ) in an alternative form. 4. What exponential equation is equivalent to y 5 log5 ( 30 ) ? 5. Without using a calculator, determine between which two integer values log5 ( 30 ) lies. 6. What type of asymptote is a logarithmic function guaranteed to have? 7. Simplify the expression ln ( 2 ) without using a calculator. 8. Is ln ( 2 ) equivalent to ln A 2 B ? Explain.
ln ( 8 ) 8 ln ( 8 )
2.4 Exercises
In Exercises 15, solve the logarithmic equation for y without using a calculator. 1. y 5 log5( 25 ) 2. y 5 log3( 81 ) 3. y 5 log2( 64 ) 4. y 5 log2( 421 ) 1 5. y 5 log3 a b 9 In Exercises 610, determine the concavity and increasing/decreasing behavior of the graph of the function. Then graph the function. (Hint: You will have to use the change-of-base formula if you graph the function on your calculator. That is, y 5 logb( x ) 1 y 5 ln( b ) .) 6. y 5 log2( x ) 8. y 5 log0.2( x ) 10. y 5 log0.8x 7. y 5 log4( x ) 9. y 5 log0.7x
ln( x )
147
In Exercises 1120, use the inverse relationship between logarithmic and exponential functions to solve the equations for x. Simplify your answers. 11. 2x 5 64 13. 2x 5 1 2 12. 5x 5 125 14. 3x 5 1 9
x 34. 2 2 ln a b 1 ln( 2 ) 2 35. 3 ln( 3x 2 ) 2 ln( x 6 ) 36. 2 ln( 3x ) 2 ln( 3x 2 ) 1 2 ln( 3 ) 37. 2 ln( 4x 2 ) 3 1 2 ln( x 6 ) 2 6 ln( x ) 38. 2 ln( 3x ) 2 2 3 ln( x 22 ) 2 5 ln( x ) 39. 4 ln( 3x ) 1 ln( 81x 2 ) 2 ln( 9x ) 40. ln( 3x ) 2 ln( 9x 2 ) 1 ln( 3 ) 2 ln( x ) In Exercises 4145, use logarithmic regression to model the data in the table. Use the model to predict the value of the function when i 5 125, and interpret the real-world meaning of the result. The Consumer Price Index is used to measure the increase in prices over time. In each of the following tables, the index is assumed to have the value 100 in the year 1984. 41.
Dental Services
15. 4x 5 64 17. log4( x ) 5 2 2 19. log3( x ) 5 5
16. log5( x ) 5 3 18. log2( x ) 5 6 1 20. log4( x ) 5 2
In Exercises 2140, use the rules of logarithms to rewrite each expression as a single logarithmic expression. 21. log( 2 ) 1 log( 8 ) 22. log( 2 ) 2 log( 8 ) 23. log( 2x ) 1 log( x 3 ) 24. log( 5x ) 1 log( 20 ) 25. log( j ) 1 log( a ) 1 log( m ) 26. 3 log( x 2 ) 2 4 log( x 2 ) 1 log( 2x 3 ) 27. 7 log( x 21 ) 2 4 log( 3x 2 ) 2 log( 4x ) 28. 2 log( 3x ) 2 1 log( x ) 212 1 29. 4 log a b 1 3 log( x ) x 2 30. 3 log a b 2 log( x 3 ) 5x 1 31. ln( x ) 2 ln a b x 32. 2 ln( 2x ) 2 1 ln a 1 b x2
Price of Dental Services
Price Index (i) 78.9 114.2 155.8 206.8 258.5
Years Since 1980 (t) 0 5 10 15 20
33. ln( 1 ) 2 log( 1 ) 1 ln( x 2 )
148 42.
Price of Wine
45.
Entertainment Admission Price
Price of Wine Consumed at Home
Price of Admission to Entertainment Venues
Price Index (i) 89.5 100.2 114.4 133.6 151.6
Years Since 1980 (t) 0 5 10 15 20
Price Index (i) 83.8 112.8 151.2 181.5 230.5
Years Since 1980 (t) 0 5 10 15 20
43.
Price of a TV
Price of a Television Set
Price Index (i) 104.6 88.7 74.6 68.1 49.9
Years Since 1980 (t) 0 5 10 15 20
46. Write a logarithmic equation that is equivalent to bx 5 c. 47. A classmate claims that logarithm is just a fancy word for exponent. Do you agree? Explain your reasoning. 48. Explain how the property of exponential functions bx ? by 5 bx 1 y is related to the logarithmic property logb ( xy ) 5 logb x 1 logb y. 49. Explain the type of behavior a scatter plot must exhibit to be effectively modeled by a logarithmic function of the form y 5 a 1 b ln( x ) , b . 0. 50. Describe how logarithmic growth compares to exponential growth.
44.
Price of Alcohol
Price of Distilled Spirits Consumed at Home
Price Index (i) 89.8 105.3 125.7 145.7 162.3
Years Since 1980 (t) 0 5 10 15 20
149
51.
Given the following table, perform each of the tasks identified as (a) through (c).
Homes with VCRs
Percentage of TV Homes with a VCR
Years Since 1984 (t) 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Homes with a VCR (percent) (V ) 20.8 36.0 48.7 58.0 64.6 68.6 71.9 75.0 77.1 79.0 81.0 82.2 84.2 84.6
(a) Use logarithmic regression to find the logarithmic function that best fits the data. (b) Solve the equation for the input variable, t. (c) Determine the year in which 90 percent of TV homes were expected to have VCRs. Do you think this prediction will be accurate? Explain. 52. The domain of the function y 5 logb( x ) is the set of all positive real numbers. Explain why the negative real numbers and zero are not in the domain of the function. 53. If b . c . 1, for what values of x is logb( x ) . logc( x ) ? 54. Does the equation ex 5 ln( x ) have a solution? Explain. 55. In order for the equation logb( x ) 5 2 logc( x ) to be true for all positive values of x, what must be the relationship between b and c?
150
Use critical thinking skills in selecting a mathematical model
Many data sets may be effectively fitted with more than one mathematical model. Determining which model to use requires critical thinking. In this section, we will discuss strategies and techniques for choosing a mathematical model. However, before delving into model selection strategies, we will briefly introduce one other type of mathematical model: the logistic function. More detailed coverage of logistic functions is given in Chapter 8.
Logistic Functions
A logistic function graph is an s-shaped curve that is bounded above and below by horizontal asymptotes (Figures 2.48 and 2.49). It may be either an increasing function or a decreasing function.
y y
Increasing
x Decreasing
FIGURE 2.48
FIGURE 2.49
The equation of a logistic function is of the form f (x) 5 c 1k 1 1 ae2bx
where a, b, c, and k are all constants. Logistic functions exhibit slow growth, then rapid growth, followed by slow growth or slow decay, then rapid decay, followed by slow decay.
Selecting a Mathematical Model

Graphing calculators and computer spreadsheets are extremely useful tools for mathematical modeling. When you are given a table of data to model, use the following model selection strategies.
151
Model Selection Strategies to Use When Given a Table of Data

1. Draw a scatter plot. 2. Determine whether the scatter plot exhibits the behavior of the graph of
one or more of the standard mathematical functions: linear, quadratic, cubic, quartic, exponential, logarithmic, or logistic. 3. Find a mathematical model for each function type selected in Step 2. 4. Use all available information to anticipate the expected behavior of the thing being modeled outside of the data set. Eliminate models that dont exhibit the expected behavior. (Sometimes it is convenient to switch the order of Steps 3 and 4.) 5. Choose the simplest model from among the models that meet your criteria. In accomplishing Step 2, it is helpful to recognize key graphical features exhibited by the data set. Often we can eliminate one or more model types by noting the concavity of the scatter plot. Table 2.26 identifies key graphical features to look for.
TABLE 2.26
Function Type Linear Quadratic Cubic Quartic Exponential Logarithmic Logistic

2
Model Equation y 5 mx 1 b y 5 ax 1 bx 1 c y 5 ax 3 1 bx 2 1 cx 1 d y 5 ax 4 1 bx 3 1 cx 2 1 dx 1 f y 5 abx with a . 0 y 5 a 1 b ln( x ) with b . 0 c y5 1k 1 1 ae2bx
Key Graphical Features Straight line Concave up everywhere or concave down everywhere Changes concavity exactly once; no horizontal asymptotes Changes concavity zero or two times; no horizontal asymptotes Concave up; horizontal asymptote at y 5 0 Concave down; vertical asymptote at x 5 0 Changes concavity exactly once; horizontal asymptotes at y 5 k and y 5 c 1 k
EXAMPLE
Choosing a Mathematical Model to Forecast a Room Rate

As shown in Table 2.27, the average room rate for a hotel/motel increased between 1990 and 1999. Find a mathematical model for the data and forecast the average hotel/motel room rate for 2004.
152
TABLE 2.27 Average Hotel/Motel Room Rate
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9
Room Rate (dollars) (R) 57.96 58.08 58.91 60.53 62.86 66.65 70.93 75.31 78.62 81.33
SOLUTION We first draw a scatter plot of the data on our graphing calculator (see Figure 2.50).
FIGURE 2.50
The graph is initially concave up, but it changes concavity near t 5 6. Since the graph changes concavity exactly once, a cubic or a logistic model (Figure 2.51) may fit the data.
153
R as a function of t
R as a function of t
R( t ) 5 2 0.04815t 3 1 0.8868t 2 2 1.495t 1 58.32

FIGURE 2.51
R( t ) 5
27.80 1 57 ( 1 1 47.23e20.6425t )
Both the cubic model and the logistic model appear to fit the data extremely well. Which model will be best for forecasting R in 2004 ( t 5 14 ) ? To answer this question, we extend the viewing rectangle for each graph to include the interval 32 1, 154 (see Figure 2.52). R as a function of t R as a function of t
R( t ) 5 2 0.04815t 3 1 0.8868t 2 2 1.495t 1 58.32

FIGURE 2.52
R( t ) 5
27.80 1 57 ( 1 1 47.23e20.6425t )
The cubic model forecasts a decrease in the room rate by 2004 ( t 5 14 ) . According to the cubic model, the rate will drop to $79.07. Since room rates have increased every year since 1990, it is unlikely that the room rate will drop at any point in the future. The logistic model forecasts a leveling off of the room rate. According to the logistic model, the 2004 room rate will be $84.64. This forecast is likely to be more accurate than the cubic model projection.
It is important to note that there is a certain degree of uncertainty when picking a model to use from a group of models that seem to fit the data. It is possible that two different people may select different models as the best model. For this reason, it is important to always explain the reasoning behind the selection of a particular model.
154
EXAMPLE
Choosing a Mathematical Model to Forecast a Movie Ticket Price

As shown in Table 2.28, the average price of a movie ticket increased between 1975 and 1999. Find a mathematical model for the data and forecast the average price of a movie ticket in 2004.
TABLE 2.28
Years Since 1975 (t) 0 5 10 15 20 24
Movie Ticket Price (dollars) (P) 2.05 2.69 3.55 4.23 4.35 5.08
Source: Statistical Abstract of the United States, 2001, Table 1244, p. 761. SOLUTION We first draw the scatter plot (Figure 2.53).
FIGURE 2.53
The first four data points appear to be somewhat linear, so our initial impression is that a linear model may fit the data well. However, the fifth data point is not aligned with the first four, so we know that the data set isnt perfectly linear. However, a linear model may still fit the data fairly well.
155
Alternatively, we may look at the scatter plot and conclude that the function is concave down on 30, 154 and concave up on 315, 244 . Since the scatter plot appears to change concavity once and does not have any horizontal asymptotes, a cubic model may work well. Figure 2.54 shows the two models. P as a function of t P as a function of t
P( t ) 5 0.1224t 1 2.148
FIGURE 2.54
P( t ) 5 0.00007426t 3 2 0.004278t 2 1 0.1846t 1 2.001
In 2004, t 5 29. Evaluating each function at t 5 29 yields the following: P( 29 ) 5 0.1224( 29 ) 1 2.148 < $5.70 P( 29 ) 5 0.00007426( 29 ) 3 2 0.004278( 29 ) 2 1 0.1846( 29 ) 1 2.001 < $5.57 It is difficult to know which of these estimates of the average price of a movie ticket is more accurate. The models seem to fit the data equally well. Furthermore, we have no other information that leads us to believe that one model would be better than the other. Therefore, we pick the simpler model: P( t ) 5 0.1224t 1 2.148 .
Certain verbal descriptions often hint at the particular mathematical model to use. By watching for key phrases, we can narrow the model selection process. Table 2.29 presents some examples of typical phrases, their interpretation, and possible models.
156
TABLE 2.29 Finding a Mathematical Model from a Verbal Description
Phrase Salaries are projected to increase by $800 per year for the next several years. The companys revenue has decreased by 5 percent annually for the past six years.
Interpretation The graph of the salary function will have a constant slope, m 5 800. Since the annual decrease is a percentage of the previous years revenue, the dollar amount by which revenue decreases annually is decreasing. The revenue graph is decreasing and concave up. The graph of the sales function may have a horizontal asymptote at or near y 5 0 and a horizontal asymptote slightly above the maximum projected sales amount.
Possible Model Linear S( t ) 5 800t 1 b Exponential R( t ) 5 a( 1 2 0.05 ) t 5 a( 0.95 ) t
Product sales were initially slow when the product was introduced, but sales increased rapidly as the popularity of the product increased. Sales are continuing to increase, but not as quickly as before. Enrollments have been dropping for years. Each year we lose more students than we did the year before.
Logistic
The graph of the enrollment function is decreasing, since enrollments are dropping. It will also be concave down, since the rate at which students are dropping continues to increase in magnitude. The graph of the profit function increases rapidly, then levels off, then increases rapidly again.
Quadratic
Company profits increased rapidly in the early 1990s but leveled off in the late 1990s. In the early 2000s, profits again increased rapidly.
Cubic
EXAMPLE
Choosing a Mathematical Model to Forecast Coca-Cola Production

In its 2001 Annual Report, the Coca-Cola Company reported the following: Our worldwide unit case volume increased 4 percent in 2001, on top of a 4 percent increase in 2000. The increase in unit case volume reflects consistent performance across certain key operations despite difficult global economic conditions. Our business system sold 17.8 billion unit cases in 2001.
(Source: Coca-Cola Company 2001 Annual Report, p. 46.)
Find a mathematical model for the unit case volume of the Coca-Cola Company.
157
SOLUTION Since the unit case volume is increasing at a constant percentage rate (4 percent), an exponential model will fit the data well. Furthermore, the initial number of unit cases sold was 17.8 billion.
V( t ) 5 abt where t is the number of years since the end of 2001 5 17.8( 1 1 0.04 ) t billion unit cases 5 17.8( 1 . 04 ) t billion unit cases The mathematical model for the unit case volume is V( t ) 5 17.8( 1 . 04 ) t .
Sometimes a data set may not be effectively modeled by any of the aforementioned functions. In these cases, we look to see if we can model the data with a piecewise function. That is, we use one function to model a portion of the data and a different function to model another portion of the data. Even with this approach, there are some data sets (e.g., daily stock prices) that can rarely be modeled by one of the standard functions.
EXAMPLE
Finding a Piecewise Model for AIDS Deaths in the United States

From 1981 to 1995, the number of adult and adolescent AIDS deaths in the United States increased dramatically. However, from 1995 to 2001, the annual death rate plummeted, as shown in Table 2.30.
TABLE 2.30 Adult and Adolescent AIDS Deaths in the United States
Years Since 1981 (t) 0 1 2 3 4 5 6 7 8 9 10
Number of Deaths During Year (D) 122 453 1,481 3,474 6,877 12,016 16,194 20,922 27,680 31,436 36,708 (Continued)
158
(Continued) Years Since 1981 (t) 11 12 13 14 15 16 17 18 19 20 Number of Deaths During Year (D) 41,424 45,187 50,071 50,876 37,646 21,630 18,028 16,648 14,433 8,963
Source: Centers for Disease Control and Prevention, HIV/AIDS Surveillance Report, December 2001, p. 30.
Find the mathematical model that best models the data and forecast the number of adult and adolescent AIDS deaths in 2004.
SOLUTION We first draw a scatter plot of the data (Figure 2.55).
FIGURE 2.55
The data set appears to exhibit logistic behavior up until 1995. After 1995, the graph appears to exhibit cubic behavior. We will use a piecewise function to model the data set. We determine each of the model pieces by using logistic and cubic regression. (Although the data point associated with t 5 14 was used in finding both pieces of the model, we must assign the domain value t 5 14 to one piece or the other. We choose to assign t 5 14 to the logistic piece of the model.) 53,955 1 1 38.834e 2 0.45127t P( t ) 5 2 381.06t 3 1 20,770t 2 2 379,469t 1 2,339,211 0 # t # 14 t . 14
159
FIGURE 2.56
Our piecewise model (Figure 2.56) appears to fit the data set very well. We are asked to forecast the number of AIDS deaths in 2004 ( t 5 23 ) . Since 23 . 14, we will use the second function in the piecewise model. P( 23 ) 5 2 381.06( 23 ) 3 1 20,770( 23 ) 2 2 379,469( 23 ) 1 2,339,211 5 2 37,603 adult and adolescent AIDS deaths It is impossible to have a negative number of deaths! Despite the fact that the model fit the data well, using the model to forecast the 2004 mortality rate yielded an unreasonable result. Returning to the data set, we estimate that the number of AIDS deaths in the years beyond 2001 will be somewhere between 0 and 8963 (the 2001 figure).
If a data set cannot be effectively modeled by one of the standard mathematical functions or a piecewise function, we may conclude that we dont know how to effectively model the data. In such cases, we may estimate a future result by identifying a range of seemingly reasonable values.
EXAMPLE
Analyzing Data Not Easily Modeled with a Common Function

The number of firearms detected during airport passenger screening is shown in Table 2.31.
TABLE 2.31
Years Since 1980 (t) 0 5 10 11 12 13
Firearms Detected (F ) 1,914 2,913 2,549 1,644 2,608 2,798
Years Since 1980 (t) 14 15 16 17 18 19
Firearms Detected (F ) 2,994 2,390 2,155 2,067 1,515 1,552
160
Estimate the number of firearms detected by airport screeners in 2002.

SOLUTION
FIGURE 2.57
The scatter plot (Figure 2.57) does not resemble any of the standard mathematical functions. It also does not appear that a piecewise model will fit the data well. The number of firearms detected in a given year appears to be somewhat random, ranging from about 1500 firearms to roughly 3000 firearms. After September 11, 2001, airline screening became much more thorough. The increased security may be having a deterrent effect. According to the Bureau of Transportation Statistics (www.bts.gov), there were 1071 firearms detected in 2001. As of March 2004, airline screening data for 2002 had not yet been published on the bureaus web site. We estimate that the number of firearms detected in 2002 will be in the 10002000 range. Our estimate is based in part on the additional data we discovered through research.
2.5 Summary
In this section, you learned strategies and techniques for selecting mathematical models. You also discovered that oftentimes more than one function may be used to model the same data set.

1. What types of functions are typically used to model data sets? 2. What two types of functions may be used to model an increasing data set that changes from concave up to concave down exactly once and has no other changes in concavity? 25 3. Solve the equation 21 5 for x. 1 1 17e20.25x 4. A data set has a near-constant percentage change for equally spaced inputs. What type of function should be used to model the data set? 5. A data set has a near-constant rate of change for equally spaced inputs. What type of function should be used to model the data set? 6. Describe the appearance of a data set that is modeled well by a quadratic function. 7. How can you solve the equation 3.95 5 2.17t 3 2 9.55t 1 6.84? 8. Can the equation 3x 5 log4( x ) be solved algebraically? Explain
161
2.5 Exercises
In Exercises 110, find the equation of the mathematical model (if possible) that you believe will most accurately forecast the indicated result. Justify your conclusions. 1.
Gaming Hardware Sales
3.
Community College Education Costs
Maricopa Community College District Tuition and Fees
Electronic Gaming Hardware Factory Sales
Years Since 199798 (t) 0 1 2 3 4 5 6
Cost per Credit (dollars) (C) 37 38 40 41 43 46 51
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9
Sales (millions of dollars) (S) 975 1,275 1,575 1,650 1,575 1,500 1,600 1,650 1,980 2,250 4.
Source: www.dist.maricopa.edu.
Forecast the cost per credit for students attending college in the district in 200506.
Breakfast Cereal Consumption
Per Capita Consumption of Breakfast Cereals
Forecast gaming hardware sales for 2002. 2.

Number of Farms
Years Since 1980 (t) 0 1 2 3 4 5 6 7 8 9 10 11
Consumption (pounds) (C ) 12 12 11.9 12.2 12.5 12.8 13.1 13.3 14.2 14.9 15.4 16.1 (Continued)
Years Since 1978 (t) 0 4 9 14 19
Farms (thousands) (f) 1,015 987 965 946 932
Forecast the number of farms in 2000.
162
(Continued) 12 13 14 15 16 17 18 19 16.6 17.3 17.4 17.1 16.6 16.3 15.6 15.5
6.
Community College Education Costs
Average Annual Undergraduate Tuition and Fees at Washington State Community Colleges
Years Since 198485 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Tuition and Fees (dollars) (F ) 581 699 699 759 780 822 867 945 999 1,125 1,296 1,350 1,401 1,458 1,515 1,584 1,641 1,743
Forecast the per capita consumption of breakfast cereal in 2001. 5.

State University Enrollment
Washington State Public University Enrollment
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Students (S) 81,401 81,882 83,052 84,713 85,523 86,080 87,309 89,365 90,189 91,543 92,821
Source: Washington State Higher Education Coordinating Board, Higher Education Statistics, September 2001.
Forecast the annual tuition and fees at a Washington state community college in 2005.
Source: Washington State Higher Education Coordinating Board, Higher Education Statistics, September 2001.
Forecast the Washington state public university enrollment in 2010.
163
7.
AIDS Incidence in Children
9.
Aviation
Estimated Pediatric AIDS Incidence (United States only)
Air Carrier Accidents
Years Since 1992 (t) 0 1 2 3 4 5 6 7 8 9
Number of Cases (C) 954 927 821 687 515 329 235 179 120 101
Year (t) 1992 1993 1994 1995 1996 1997 1998 1999 2000
Accidents (A) 18 23 23 36 37 49 50 52 54
Source: Centers for Disease Control and Prevention, HIV/AIDS Surveillance Report, December 2001, p. 36.
Forecast the number of air carrier accidents in 2002. 10.

Federal Funds for Elections
Forecast the number of pediatric AIDS cases in the United States in 2005. 8.
Per Capita Personal Income
Federal Funds for Presidential Election Campaigns
Per Capita Personal IncomeFlorida
Years Since 1993 (t) 0 1 2 3 4 5 6 7
Years Since 1980 (t) 0 4 8 12 16
Federal Funds (millions of dollars) (F ) 62.7 80.3 92.2 110.4 152.6
Forecast the amount of federal funds spent in the 2004 presidential election. In Exercises 1120, find mathematical models for each of the verbal descriptions. 11. Candy Bar Prices Candy bars currently cost $0.60 each. The price of a candy bar is expected to increase by 3 percent per year in the future.
Forecast the per capita personal income in Florida in 2004.
164 12.
On March 12, 2004, a builder priced a new home in Queen Creek, Arizona, at $198.9K. The builders sales representative told the author that the price for that home style would increase on March 16, March 30, and April 13 to $205.9K, $209.9K, and $212.9K, respectively. (Source: Fulton Homes.) Find a mathematical model for the price of the new home style.
Housing Prices
13. Calculator Prices A calculator currently costs $87. The price of the calculator is expected to increase by $3 per year. Find a mathematical model for the price of the calculator. 14. Mortality Rates There are presently 95 members of a high school graduating class who are still living. The number of surviving class members is decreasing at a rate of 4 percent per year. Find a mathematical model for the number of surviving class members. 15. Product Sales Growth We are introducing a new product next year. We anticipate that sales will initially be slow but will increase rapidly once people become aware of our product. We anticipate that our monthly sales will start to level off in 18 months at about $200,000. We predict that sales for the first two months will be $12,000 and $19,000, respectively. Develop a mathematical model to forecast monthly product sales. 16. Club Membership We are concerned about the decreasing number of members of our business club. Two years ago, we had 200 members. Last year we had 165 members, and this year we have 110 members. If something doesnt change, we expect to lose even more members next year than we lost this year. Develop a mathematical model for the club membership. 17.
Population Growth The town of Queen Creek, Arizona, was founded in 1989. In 1990, there were 2667 people living in the town. The town grew rapidly in the 1990s, in large part because of new home construction in the area. There were 4316 people living in the town in 2000 and 4940 people in 2001. The
Arizona Department of Commerce estimated the 2002 population of Queen Creek at 5555 people. If a logistic model is used to model the population of Queen Creek, what is the maximum projected population of the city? (Hint: First align the data.) The Arizona Department of Commerce estimated the 2003 population of Queen Creek at 7480. In light of this additional information, is there a different type of model that would have better predicted the 2003 population of Queen Creek? Explain. 18. On March 12, 2004, a builder priced a new home in Queen Creek, Arizona, at $143.9K. The builders sales representative told the author that the price for that home style would increase on March 16, March 30, and April 13 to $148.9K, $151.9K, and $153.9K, respectively. (Source: Fulton Homes.) Find a mathematical model for the price of the new home style.
Housing Prices
19.
Federal income tax rates are dependent upon the amount of taxable income received. In 2003, federal income taxes were calculated as follows. For single filers, the first $7000 earned was taxed at 10 percent. The next $21,400 earned was taxed at 15 percent. The next $40,400 earned was taxed at 25 percent. For example, the tax of a single woman who earned $25,000 would be calculated as follows: 10% tax on the first $7,000 $7,000 3 0.10 5 $700 Amount to be taxed at a higher rate $25,000 2 $7000 5 $18,000 15% tax on the next $18,000 $18,000 3 0.15 5 $2,700
Federal Tax Rates
The persons total tax is $700 1 $2,700 5 $3,400 Find a mathematical model for income tax as a function of taxable income for single filers. 20.
Federal Tax Rates In 2003, federal income taxes for married individuals filing jointly were calculated as follows. The first $14,000 was taxed at 10 percent. The next $42,800 was taxed at 15 percent. The next $57,850 was taxed at 25 percent.
165
The tax of a married couple who earned $65,000 would be calculated as follows: 10% tax on the first $14,000 $14,000 3 0.10 5 $1,400 Amount to be taxed at a higher rate $65,000 2 $14,000 5 $51,000 15% tax on the next $42,800 $42,800 3 0.15 5 $6420 Amount to be taxed at a higher rate $51,000 2 $42,800 5 $8200 25% tax on the next $8200 $8200 3 0.25 5 $2050 The couples total tax is $1400 1 $6420 1 $2050 5 $9870 Find a mathematical model for income tax as a function of taxable income for couples filing jointly.
an unknown data value and claims that the correlation coefficient of 1 implies that the forecasted value will be 100 percent accurate. Do you agree? Explain.
26. The graph of a mathematical model passes through all of the points of a data set. A student claims that the model is a perfect forecaster of future results. How would you respond? 27. A scatter plot is concave up and increasing on 30, 54 , concave down and increasing on 35, 84 , and decreasing at a constant rate on 38, 154 . Describe two different mathematical models that may fit the data set. 28. Describe how a business owner can benefit from mathematical modeling, despite the imprecision of a models results. 29. Daily fluctuations in the stock market make the share price of a stock very difficult to model. What approach would you take if you wanted to model the long-term performance of a particular stock? 30. You are asked by your boss to model the data shown in the following scatter plot. How would you respond?
21. Describe the limitations of using mathematical models to forecast unknown results. 22. What criteria should you use to determine which mathematical model best fits a particular situation? 23. Explain how the correlation coefficient r and the coefficient of determination r 2 are used when analyzing a mathematical model. 24. What strategies should be used to model a data set when a scatter plot of the data is not effectively modeled by one of the common mathematical functions? 25. You determine that the line of best fit for a particular data set has a correlation coefficient of r 5 1. Your classmate uses the model to forecast
166

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What does it mean for a graph to be concave up? concave down? Be sure to include the idea of rate of change in your explanation. (2.1) 2. Which of the following types of functions (if any) are polynomials: quadratic, cubic, quartic, exponential, logarithmic, logistic? Explain your reasoning. (2.1, 2.2) 3. What is the relationship between the degree of a polynomial and the shape of its graph? Define any special terms used in your explanation. (2.2) 4. Explain the difference between the phrases increasing by a constant amount each year and increasing by a constant percentage each year. (2.3) 5. Use graphs, tables, and/or symbols to describe how logarithmic and exponential functions are related. (2.3, 2.4) 6. Conceptually, what does y 5 logb( x ) mean? (2.4) 7. Exponential and logistic functions both have horizontal asymptotes. Explain why they have horizontal asymptotes by using equations of each function type. (2.3, 2.5) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
Definitions

Degree of a Polynomial: The value of a polynomials largest exponent. Parabola: The graph of a quadratic function. Concavity: The curvature of a graph. A graph curved upward over an interval is concave up on that interval. A graph curved downward over an interval is concave down on that interval. Inflection Point: A point on a graph at which the function changes concavity. Initial Value: The value a in the exponential function y 5 abx. Growth Factor: The base b in the exponential function y 5 abx. Horizontal Asymptote: The horizontal line that the graph of a function approaches as x S ` or as x S 2 ` . (All exponential functions have a horizontal asymptote.) Annual Growth Rate: The percentage by which a quantity changes each year. Common Logarithm: A logarithm with base 10. It is usually written as log(x). Natural Logarithm: A logarithm with base e. It is usually written as ln(x).
167
Procedures, Properties, and Tests
Properties of Exponents 1 b2n 5 bn bm ? bn 5 bm 1 n bm 5 bm 2 n bn bmn 5 ( bm ) n 5 ( bn ) m Relationship Between Exponential and Logarithmic Functions y 5 bx is equivalent to log y 5 x. b Rules of Logarithms log ( 1 ) 5 0 b log ( b ) 5 1 b log ( bx ) 5 x b blog (x) 5 x log ( xy ) 5 log ( x ) 1 log ( y ) b b b x log a b 5 log ( x ) 2 log ( y ) b b b y p log ( x ) 5 p log ( x ) b b
b
Strategies for Selecting a Mathematical Model Draw a scatter plot. Determine if the shape of the plot resembles a common function. Find a model for each of the common function types identified as candidates. Eliminate models that dont fit the data set well or that dont exhibit the known behavior of the data being modeled. Choose the simplest model from among those that meet the criteria.
Formulas

Quadratic Function: y 5 ax2 1 bx 1 c , a 2 0 Cubic Function: y 5 ax3 1 bx2 1 cx 1 d , a 2 0 Quartic Function: y 5 ax4 1 bx3 1 cx2 1 dx 1 k , a 2 0 Exponential Function: y 5 abx, b . 0 and b 2 1 Logarithmic Function: y 5 logb( x ), b . 0, b 2 1, and x . 0 Logistic Function: y 5 c 1k 1 1 ae2bx
Quadratic Formula: x 5
2 b 6 "b2 2 4ac 2a
x-Coordinate of the Vertex of a Parabola: x 5 2
b 2a
Growth Factor of a Function with Points ( x1, y1) and ( x2, y2): 1 y2 x 2 x b5 a b y1
2 1
Growth Rate of y = ab x: r 5 b 2 1 Change-of-Base Formula: logb( a ) 5 log( a ) ln( a ) 5 log( b ) ln( b )
168

Section 2.1 In Exercises 113, find the model that best fits the data. Use the model to answer the given questions.
1.
Advertising Expenditures: Magazines
2.
Advertising Expenditures: Cable TV
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
According to the model, how much money was spent on magazine advertising in 2002?
According to the model, how much money was spent on cable television advertising in 2002? 3. Using the models from Exercises 1 and 2, determine in what year cable television advertising expenditures are expected to exceed magazine advertising expenditures.
169
4.
Advertising Expenditures: Radio
According to the model, when will Yellow Pages advertising exceed $15 billion?
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Section 2.2
6.
Federal Credit Unions
Years Since 1975 (t) 0 5 10 15 20 25
Credit Unions (C) 12,737 12,440 10,125 8,511 7,329 6,336
According to the model, when will radio advertising exceed $25 billion? 5.
Advertising Expenditures: Yellow Pages
According to the model, how many federal credit unions were there in 1999? Does this seem reasonable? Explain. 7.
Advertising Expenditures: Newspapers
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
According to the model, how much money was spent on newspaper advertising in 2001?
170 8.
Full-Service Restaurant Sales
Years Since 1980 (t) 0 2 4 6 8 10 12 14 16 18 20
Sales (millions of dollars) (S ) 39,307 46,443 54,815 61,474 69,356 77,811 83,561 91,457 100,830 117,774 134,461
According to the model, when will broadcast television advertising expenditures exceed $50 billion? 10.
Manufacturing Full-Time Employees: Leather and Leather Products
Years Since 1995 (t) 0 1 2 3 4
Employees (thousands) (N ) 106 95 89 84 76
According to the model, in what year will the number of full-time employees in the leather and leather products manufacturing industry drop below 65,000?
According to the model, when will fullservice restaurant sales exceed $150 billion? 9.
Advertising Expenditures: Broadcast TV
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
171
11.
Wages
12.
Wages
Average Hourly Earnings in Manufacturing Industries: Michigan
Average Hourly Earnings in Manufacturing Industries: Florida
Years Since 1980 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Average Earnings (dollars per hour) (E) 9.52 10.53 11.18 11.62 12.18 12.64 12.80 12.97 13.31 13.51 13.86 14.52 14.81 15.36 16.13 16.31 16.67 17.18 17.61 18.38 19.20
Years Since 1980 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Average Earnings (dollars per hour) (E ) 5.98 6.53 7.02 7.33 7.62 7.86 8.02 8.16 8.39 8.67 8.98 9.30 9.59 9.76 9.97 10.18 10.55 10.95 11.43 11.83 12.28
According to the model, what will the average hourly wage in Michigan manufacturing industries be in 2003? 13.
According to the model, what will the average hourly wage in Florida manufacturing industries be in 2003? Based on the wage data in Exercises 1112, do you think it would be better to start up a manufacturing business in Florida or in Michigan? Justify your answer and explain what other issues might affect your decision.
172
Section 2.3 In Exercises 1415, determine if the graph of the function is increasing or decreasing and if it is concave up or concave down. Identify the coordinates of the y-intercept. Then graph the function to verify your conclusions.
14. y 5 2( 0.75 )
x
In Exercises 2021, find the exponential function that fits the verbal description. Calculate the value of the function five years from now and interpret its realworld meaning. 20. My car is depreciating at a rate of 17 percent per year. It is currently valued at $6000. 21. My monthly household expenses are increasing by 2.3 percent annually. It currently costs $4500 per month to maintain my household. In Exercises 2223, find the solution by graphically solving the equation. 22. 1.9( 2.6 ) x 5 10 23. 9.7( 0.4 ) x 5 2
15. y 5 2 0.3( 2.8 )
In Exercises 1617, find the equation of the exponential function that fits the data in the table algebraically. 16. x 0 1 2 3 y 3 15 75 375 17. x 2 4 6 8 y 36 144 576 2,304
Section 2.4 In Exercises 2425, solve the logarithmic equation for y without using a calculator.
24. y 5 log6( 36 ) 25. y 5 log5( 0.2 )
In Exercises 1819, use exponential regression to model the data in the table. Use the model to predict the value of the function when t 5 10, and interpret the real-world meaning of the result. 18.
Number of Subway Restaurants
In Exercises 2627, determine the concavity and increasing/decreasing behavior of the graph of the function. Then graph the function to verify your results. (Hint: You will have to use the change-of-base formula if you graph the function on your calculator. That is, y 5 logb( x ) 1 y 5 ( ) ln b ) 26. y 5 log5( x ) 27. y 5 log0.4( x )
ln( x )
Years Since 1996 (t) 0 1 2 3 4
Restaurants (N ) 12,516 13,066 13,600 14,162 14,662
In Exercises 2831, use the inverse relationship between logarithmic and exponential functions to solve the equations for x. Simplify your answers. 1 x 28. a b 5 64 2 30. log3( x ) 5 4 29. 5x 5 625 31. log5( x ) 5 2 2
Source: www.subway.com.
19.
Number of McDonalds Restaurants
Years Since 1997 (t) 0 1 2 3 4
In Exercises 3235, use the rules of logarithms to rewrite each expression as a single logarithm. 32. log( 5 ) 1 log( 3 ) 34. 3 log( 2x ) 2 log( 2x )
2 3
Restaurants (N ) 22,928 24,513 26,309 28,707 30,093
33. 2 log( 4x ) 2 log( 8 )
35. 2 log( 3x ) 2 1 log( 3x2 )
173
In Exercises 3637, use logarithmic regression to model the data in the table. Use the model to predict the value of the function when t 5 10, and interpret the real-world meaning of the result. 36.
McDonalds Systemwide Sales
Section 2.5 In Exercises 3840, find the equation of the mathematical model (if possible) that you believe will most accurately forecast the future behavior of the thing being modeled. Justify your conclusions.
38.
Projected Teacher Salaries
Years Since 1997 (t) 1 2 3 4 5
Sales (millions of dollars) (S) 33,638 35,979 38,491 40,181 40,630
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Public-School Teacher Average Annual Salary (thousands of dollars) (S) 31.4 33.1 34.1 35.0 35.7 36.7 37.7 38.5 39.5 40.6 41.7
37.
Non-Alcohol-Related Auto Accident Fatalities
Years Since 1989 (t) 1 3 4 5 6 7 8 9 10
Fatalities (percentage) (F ) 50.5 54.5 56.5 59.3 58.8 59.1 61.5 61.4 61.7
Does the model estimate at t 5 10 agree with the raw data value?
174 39.
Cassette Tape Market Share
Cassette Tape Sales
Years Since 1993 (t) 0 1 2 3 4 5 6 7 8 9
Percent of Music Market (percentage points) (P) 38.0 32.1 25.1 19.3 18.2 14.8 8.0 4.9 3.4 2.4
Source: Recording Industry Association of America.
40.
Music Market Size
Music Market Size
Years Since 1997 (t) 0 1 2 3 4 5
Dollar Volume (millions) (P) $12,236.80 $13,723.50 $14,584.50 $14,323.00 $13,740.89 $12,614.21
Source: Recording Industry Association of America.
P R O J E C T
What to do
1. Visit the Bureau of Labor Statistics web site (www.bls.gov/cpi) and access the most recent Consumer Price Index news report.
2. Select an expenditure category. 3. Record the annual percentage rate of change in that category. (This is called the compound annual rate in the report.) 4. Find the price of a product that belongs to the category you selected at a local retailer. 5. Using the Consumer Price Index information, estimate the price of the item five years from now.
Consumer Price Index explained

The Consumer Price Index (CPI) is a measure of the average change in prices over time of goods and services purchased by households. The CPI for All Urban Consumers (CPI-U) and the Chained CPI for All Urban Consumers (C-CPIU), which cover approximately 87 percent of the total population and include in addition to wage earners and clerical worker households, groups such as professional, managerial, and technical workers, the self-employed, shortterm workers, the unemployed, and retirees and others not in the labor force. (Source: Bureau of Labor Statistics News, July 2002, USDL-020-480.)
175
Chapter
The Derivative
t is impossible to determine how quickly a person is running from a single photograph, since speed is calculated as a change in distance over a change in time. Nevertheless, we may estimate a persons speed at a particular instant in time by determining the distance traveled over a small interval of time (e.g., one second). A runners speed may be classified as a rate of change in distance. A key component of calculus is the study of rates of change.
3.1
Average Rate of Change
3.2
Limits and Instantaneous Rates of Change
Estimate the instantaneous rate of change of a function at a point Use derivative notation and terminology to describe instantaneous rates of change Perform mathematical operations with limits Determine if a function is continuous
3.3
Limits and Continuity
3.4
The Derivative as a Slope: Graphical Methods
3.5
The Derivative as a Function: Algebraic Method
Use the limit definition of the derivative to find the derivative of a function Use derivative notation and terminology to describe instantaneous rates of change Interpret the meaning of the derivative in the context of a word problem
3.6 176
177
Colleges and universities periodically raise their tuition rates in order to cover rising staffing and facilities costs. As a result, it is often difficult for students to know how much money they should save to cover future tuition costs. By calculating the average rate of change in the tuition price over a period of years, we can estimate projected increases in tuition costs. In this section, we will demonstrate how to calculate the average rate of change in the value of a function over a specified interval [a, b]. (The interval notation [a, b] is equivalent to a # x # b.)
THE DIFFERENCE QUOTIENT: AN AVERAGE RATE OF CHANGE
The average rate of change of a function y 5 f ( x ) over an interval 3a, b4 is f (b) 2 f (a) b2a
This expression is referred to as the difference quotient. For a linear function, the difference quotient gives the slope of the line.
In calculating the difference quotient, we answer the question, Over the interval 3a, b4 , on average, how much does a one-unit increase in the x value change the y value of the function?
EXAMPLE

The quarterly cost of tuition for full-time resident students at Green River Community College is shown in Table 3.1.
TABLE 3.1
Years (since 19941995) (t) 0 1 2 3 4 5 6 7
Quarterly Tuition Cost (dollars) 3 f (t) 4 432 450 467 486 505 528 547 581
Change in Tuition Cost from Prior Year (dollars)
18 17 19 19 23 19 34
178
CHAPTER 3 The Derivative
What is the average rate of change in the quarterly tuition cost from the 19941995 academic year to the 20012002 academic year? Rounded to the nearest dollar, what do you estimate the 20022003 quarterly tuition cost will be?
SOLUTION One way to calculate the average rate of change is to average the annual rates of change over the entire period. To calculate the average, we sum each of the seven rate of change data values and divide by the number of data values.
Average rate of change 5 5
( 18 1 17 1 19 1 19 1 23 1 19 1 34 ) dollars 7 years 149 dollars per year 7
5 21.29 per year It is significant to note that the same result may be obtained more rapidly by applying the difference quotient formula. For the period 19941995 to 20012002, the interval 3a, b4 5 30, 74 . The average rate of change of the tuition is f (b) 2 f (a) f (7) 2 f (0) 5 b2a 720 5 5
581 2 432 dollars 7 years 149 dollars 7 years
5 21.29 dollars per year From 19941995 to 20012002, the tuition increased by an average of approximately $21 per year. We estimate that the 20022003 quarterly tuition will be $21 more than the 20012002 tuition, or $602.
When determining the meaning of an average rate of change in a real-life problem, it is essential to find the units of measurement of the result. Fortunately, the units are easily determined. The units of the rate of change are the units of the output divided by the units of the input. In Example 1, the units of the output were dollars and the units of the input were years. Consequently, the units of the average rate of change were dollars divided by years, or dollars per year.
EXAMPLE

A companys stock price often fluctuates dramatically from day to day. If an investor is preoccupied with daily price changes, she may sell too early or buy too late to get a good return on her investment. One way an investor can gauge the change in a companys stock price is to calculate the average rate of change in the price of the stock over a period of time.
179
Table 3.2 shows the closing stock price of Apple Computer Corporation over a nine-month period.
TABLE 3.2
Months (since August 2001) (t) August 2001 ( t 5 0 ) September ( t 5 1 ) October ( t 5 2 ) November ( t 5 3 ) December ( t 5 4 ) January 2002 ( t 5 5 ) February ( t 5 6 ) March ( t 5 7 ) April ( t 5 8 )
Source: www.quicken.com.
Stock Price at the End of the First Trading Day of the Month (dollars) 3 f (t) 4 19.06 18.25 15.54 18.59 21.05 23.30 24.41 24.29 24.46
What was the average rate of change in the stock price between (a) August 2001 and October 2001? (b) October 2001 and April 2002? (c) August 2001 and April 2002?
SOLUTION
(a) In the table, August is month 0 and October is month 2. The average rate of change of the stock over the interval 30, 24 is f (b) 2 f (a) f (2) 2 f (0) 5 b2a 220 5
15.54 2 19.06 dollars 2 months
5 2 1.76 dollars per month Between August and October 2001, the share price of Apple Computer stock decreased by an average of $1.76 per month. (b) In the table, October is month 2 and April is month 8. The average rate of change of the stock over the interval 32, 84 is f (b) 2 f (a) f (8) 2 f (2) 5 b2a 822 5
24.46 2 15.54 dollars 6 months
5 1.49 dollars per month
180
Between October 2001 and April 2002, the share price of Apple Computer stock increased by an average of $1.49 per month. (c) In the table, August is month 0 and April is month 8. The average rate of change of the stock over the interval 30, 84 is f (8) 2 f (0) 24.46 2 19.06 dollars 5 820 8 months 5 0.68 dollar per month Between October 2001 and April 2002, the share price of Apple Computer stock increased by an average of $0.68 per month.
Some investors choose to buy or sell a stock once it hits a certain limit price. In fact, many investment companies allow investors to activate a limit order for a period of 30 days. If the stock price hits the limit price within the 30-day period, the order is executed. If the stock price fails to hit the limit price during the 30-day period, the order expires with no trading action. Observing the overall trend in a stocks price helps the investor to pick a limit price and predict whether the order will be executed.
EXAMPLE

The population of Washington state may be modeled by the function P( t ) 5 0.8981( 1.019 ) t , where P is the population in millions of people and t is the number of years since 1900. (Source: Modeled from data at www.ofm.wa.gov.) What is the average rate of change in the population from 1990 to 2000?
SOLUTION We observe that in the model, t is the number of years since 1900. So for the model, t 5 90 represents 1990 and t 5 100 represents 2000. The average rate of change in the population over the interval 390, 1004 is
5.90 2 4.89 million people P( 100 ) 2 P( 90 ) 5 100 2 90 10 years
5 0.101 million people per year Between 1990 and 2000, the population of Washington state increased by an average of 101,000 people per year.
Graphical Interpretation of the Difference Quotient

A line connecting any two points on a graph is referred to as a secant line. Graphically speaking, the difference quotient for a function y 5 f ( x ) is the slope of the secant line connecting ( a, f ( a )) and ( b, f ( b )) (Figure 3.1).
181
y f (b) Secant line
y = f (x) (b, f (b))
f (b) f (a) (a, f (a)) ba x
f (a)
b
f (b) 2 f (a) b2a
FIGURE 3.1 Secant line slope 5
EXAMPLE
Finding the Slope of a Secant Line

The graph of the function f ( x ) 5 3x 2 2 6x 1 5 is shown in Figure 3.2. Calculate the slope of the secant line of f that passes through ( 1, 2 ) and ( 2, 5 ) .
y 11 10 9 8 7 6 5 4 3 2 1 1 0
y = 3x 2 6x + 5
FIGURE 3.2
y 11 10 9 8 7 6 5 4 3 2 1 1 0
y = 3x 6x + 5
2
SOLUTION We first plot the points ( 1, 2 ) and ( 2, 5 ) and draw the line connecting them (Figure 3.3). This line is the secant line of the graph between x 5 1 and x 5 2 . The slope of the secant line is given by
Secant line (2, 5)
m5 5
f (b) 2 f (a) b2a f (2) 2 f (1) 221 522 1
(1, 2) 1 2 x
FIGURE 3.3
The slope of the secant line is 3. That is, on the interval 31, 24 , a one-unit increase in x results in a three-unit increase in y, on average.
53
182
EXAMPLE
Calculating the Slope of a Secant Line

Calculate the slope of the secant line of the function f shown in Figure 3.4.
y 15 10 5 2 (2, 19) 1 (1, 1) 1 2 3 4 x (4, 17)
10 15
f (x) = x 3 3x 2 + 1 Secant line
FIGURE 3.4
SOLUTION From the graph, we see that the line intersects the graph of f at ( 2 2, 2 19 ) , ( 1, 2 1 ) , and ( 4, 17 ) . The slope of the given secant line may be determined by using any pair of these points.
m5 5 5
f (4) 2 f (22) 4 2 (22) 17 2 ( 2 19 ) 6 36 6
For this function f, the average rate of change in f over the interval 32 2, 14 is the same as the average rate of change in f over the intervals 31, 44 and 32 2, 44 .
56
3.1 Summary
In this section, you learned how to calculate the average rate of change of a function over a specified interval. You also learned that the units of the average rate of change are the units of the output divided by the units of the input. You saw that, graphically speaking, the difference quotient is the slope of a secant line. In the next section, we will demonstrate how to use the difference quotient to estimate an instantaneous rate of change.
183

1. What formula is used to calculate the slope of a line? 2. What do the expressions y2 2 y1 and x2 2 x1 in the slope formula mean? 3. What is the difference between an average temperature and an average rate of change in temperature? 4. What is the slope of the line that passes through ( 2, 2 3 ) and ( 4, 10 ) ? 5. How do you determine the units of an average rate of change? 6. A horizontal line passes through ( x1, y1 ) and ( x2, y2 ) . What do you know about y1 and y2 and the slope of the line? 7. A vertical line passes through ( x1, y1 ) and ( x2, y2 ) . What do you know about x1 and x2 and the slope of the line? 8. What is the average rate of change of f ( x ) 5 x2 on the interval 31, 1 1 h4 ? (Hint: The correct solution is given in terms of h.)
3.1 Exercises
In Exercises 110, calculate the average rate of change of the function over the given interval. 1. f ( x ) 5 2x 2 5 over the interval 33, 54 2. g( x ) 5 x 2 2 5x over the interval 32 2, 44 12.
Internet Access People with Internet access between 1997 and 2000.
3. v( x ) 5 2 x 3 over the interval 32 1, 14 4 4. g( t ) 5 3t 2 over the interval 32, 44 t 5. v( m ) 5 m2 2 m over the interval 32 3, 44 6. h( s ) 5 ln( s ) over the interval 34, 74 ln( x ) 7. z 5 over the interval 31, 54 x
Year 1997 1998 1999 2000

Number of People with Internet Access at Home or Work (in thousands) 46,305 62,273 83,677 112,949
2t 2 5 8. f ( t ) 5 over the interval 32 2, 24 t15 10. w( t ) 5 2t22 1 5 over the interval 31, 44 In Exercises 1120, calculate the average rate of change in the designated quantity over the given interval(s). 11. Air Temperature Temperature between 11:00 a.m. and 3:00 p.m. Time of Day 11:00 a.m. 1:00 p.m. 3:00 p.m. Temperature ( F) 68 73 75 9. q( x ) 5 " x 1 2 over the interval 30, 64
13.
Dow Jones Industrial Average Dow Jones Industrial Average between 1996 and 1998 and between 1997 and 1999.
Year 1996 1997 1998 1999

Dow Jones Industrial Average Closing Value at the End of the Year (points) 6,448.30 7,908.30 9,181.40 11,497.10
184 14.
Nasdaq Composite Index closing value between 1996 and 1998 and between 1997 and 1999.
Nasdaq Composite Index
17.
Daily newspaper subscriptions as the number of cable TV subscribers increased from 50.5 million (in 1990) to 67.7 million (in 2000).
Newspaper Subscriptions
Year 1996 1997 1998 1999

Nasdaq Composite Index Closing Value at the End of the Year (points) 1,291.0 1,570.4 2,192.7 4,069.3 18.
Cable TV Subscribers (millions) 50.5 60.9 66.7 67.7
Daily Newspaper Circulation (millions) 62.3 58.2 56.0 55.8
Personal Income Disposable personal income between 1990 and 2000.
15.
A third-grade students reading score between the first and third quarter.
Reading Scores
Quarter 1 2 3
Source: Authors data.
Reading Score (words per minute) 69 107 129
Year 1990 1995 1999 2000
Disposable Personal Income (dollars per person) 17,176 20,358 23,708 24,889
16.
The percentage of U.S. housing units that are owner-occupied when the number of housing units increased from 94.2 million (in 1990) to 107.0 million (in 2001).
Owner-Occupied Homes
19.
Poverty Level Percent of people below poverty level when the unemployment rate decreased from 5.6 percent (in 1990) to 4.0 percent (in 2000).
Number of Housing Units (millions) 94.2 100.0 104.9 105.7 107.0

Percent of Housing Units That Are Owner-Occupied 63.9 64.7 66.8 67.4 67.8
Unemployment Rate (percent) 5.6 4.2 4.0
People Below Poverty Level (percent) 13.5 11.8 11.3
185
20.
Net farm income when land in farms decreased from 987 million acres (in 1990) to 943 million acres (in 2000).
Net Farm Income
28.
Land in Farms (millions of acres) 987 943
Farm Income (billions of dollars) 44.6 46.4 29.
y 3 2 1 0 1 2 3
y = f (x)
y 3 2 y = f (x)
In Exercises 2125, graph each function. Then use the difference quotient, slope of the secant line through the points ( 1, f ( 1 )) and ( 3, f ( 3 )) . 21. f ( x ) 5 2x 23. f ( x ) 5 5 25. f ( x ) 5 ( x 2 2 )
2
f (b) 2 f (a) , b2a
1 1 x
to calculate the
0 1 2 3 y = f (x)
22. f ( x ) 5 5 2 x 2 24. f ( x ) 5 3x 1 1
30.
In Exercises 2630, use the difference quotient, through the points ( 1, f ( 1 )) and ( 3, f ( 3 )) for the given graph of f. 26.
y 9 8 7 6 5 4 3 2 1 0 0 y 4 3 2 1 0 0 1 2 3 x y = f (x) f (b) 2 f (a) , b2a
to calculate the slope of the secant line
y 8 7 6 5 4 3 2 1 0 1 2
In Exercises 3140, calculate the average rate of change in the state populations (in people per month) between April 1, 2000, and July 1, 2001. 31.
Population
y = f (x)
Montana Date
1 2 3 x
Population (in thousands) 902.2 904.4
April 2000 July 2001
27.
186 32.
Population
36.
Population
North Dakota Date April 2000 July 2001 33. Population (in thousands) 642.2 634.4 Date
Louisiana Population (in thousands) 4,469.0 4,465.4
April 2000 July 2001 37.
Population
Population
Massachusetts Date April 2000 July 2001 34. Population (in thousands) 6,349.1 6,379.3
West Virginia Date April 2000 July 2001 38. Population (in thousands) 1,808.3 1,801.9
Population
Population
Utah Date April 2000 July 2001 35. Population (in thousands) 2,233.2 2,269.8 Date
Wyoming Population (in thousands) 493.8 494.4
April 2000 July 2001 39.
Population
Population
Missouri Date April 2000 July 2001 Population (in thousands) 5,595.2 5,629.7
Pennsylvania Date April 2000 July 2001 Population (in thousands) 12,281.1 12,287.2
187
40.
Population
45. What does a driver mean when he says, We averaged 50 miles per hour between Phoenix and Salt Lake City? Arizona Population (in thousands) 5,130.6 5,307.3
Date April 2000 July 2001
46. Let f ( x ) 5 x 2 2 4x . Compare and contrast the average rate of change of f ( x ) and of 2( f ( x )) on the interval 31, 44 . 47. Given f ( x ) 5 x 2 2 4 , determine the average rate of change of f over the interval 3a, a 1 h4 . 48. The slope of the secant line passing through ( 2, f ( 2 )) and ( 2.01, f ( 2.01 )) is 4. Determine the vertical distance between ( 2, f ( 2 )) and ( 2.01, f ( 2.01 )) .
41. The difference quotient, b 2 a , is widely used throughout mathematics. Explain what the numerator and denominator of the difference quotient represent and what the difference quotient itself represents. 42. Explain the relationship among the following four expressions: y2 2 y1 , x2 2 x1 Dy , Dx f (b) 2 f (a) , b2a f (a 1 h) 2 f (a) h
f (b) 2 f (a)
49. The slope of a secant line of the function f ( x ) 5 x 2 passes through the points ( 0, 0 ) and ( h, h2 ) . Determine the slope of the secant line when h 5 1, h 5 0.1, h 5 0.01, and h 5 0.001. As h gets smaller, what value does the slope of the secant line appear to be approaching? 50. A secant line of the function f ( x ) 5 x 1>3 passes through the points ( 0, 0 ) and ( h, h1>3 ) . Determine the slope of the secant line when h 5 1, h 5 0.1, h 5 0.01, and h 5 0.001. As h gets smaller, what value does the slope of the secant line appear to be approaching?
43. Explain how a secant line of a graph is related to the average rate of change of the associated function over an interval. 44. If the graph of a function is concave up and decreasing on the interval ( a 1 h, b 1 h ) for all positive values of h, what do you know about the average rate of change of the function over the interval as h approaches infinity?
188
3.2 Limits and Instantaneous Rates of Change
Estimate the instantaneous rate of change of a function at a point Use derivative notation and terminology to describe instantaneous rates of change
In the July 1988 Olympic trials, Florence Griffith Joyner shattered the world record for the 100-meter dash with a time of 10.49 seconds. How fast was she running (in meters per second) when the picture at the left was taken? From a single photo, it is impossible for us to determine her speed. However, if we knew how long it took her to reach various checkpoints during the race, we could approximate her speed at the finish line. In this section, we will demonstrate how to use the difference quotient to estimate and the derivative to calculate an instantaneous rate of change.
While an average rate of change is calculated over an interval 3a, b4 , an instantaneous rate of change is calculated at a single value a. For example, the average highway speed of a person over a 200-mile trip may have been 59 miles per hour. However, when he passed a state patrol car exactly 124 miles into the trip, he was speeding at 84 miles per hour. His average speed on the interval 30, 2004 was 59 miles per hour; however, his instantaneous speed at d 5 124 was 84 miles per hour.
EXAMPLE
Estimating an Instantaneous Rate of Change

Suppose that a runner in the 100-meter dash recorded the times shown in Table 3.3.
TABLE 3.3
Time (seconds) (t) 0 5.24 9.46 9.97 10.39 10.49
Total Distance Traveled (meters) 3D( t ) 4 0 50 90 95 99 100
Estimate her speed when she crossed the finish line.

SOLUTION Her average speed over various distances may be calculated using
D( b ) 2 D( a ) . b2a
the difference quotient,
189
Her average speed over the 100-meter distance was Average speed 5 100 2 0 meters 10.49 2 0 seconds
5 9.53 meters per second Her average speed over the last 50 meters was Average speed 5 5 100 2 50 meters 10.49 2 5.24 seconds 50 meters 5.25 seconds
5 9.52 meters per second Her average speed over the last 10 meters was Average speed 5 5 100 2 90 meters 10.49 2 9.46 seconds 10 meters 1.03 seconds
5 9.71 meters per second Her average speed over the last 5 meters was Average speed 5 5 100 2 95 meters 10.49 2 9.97 seconds 5 meters 0.52 second
5 9.62 meters per second And her average speed over the last meter was Average speed 5 5 100 2 99 meters 10.49 2 10.39 seconds 1 meter 0.10 second
5 10.0 meters per second Although each calculation yielded a different result, all of these difference quotients estimate the runners finish-line speed. Which of the estimates do you think is most accurate? The last calculation best estimates her finish-line speed because it measures the change in distance over the smallest interval of time: 0.1 second. (Reducing the time interval to an even smaller amount of time, say 0.01 second, would further improve the estimate.) We estimate that the runners speed when she crossed the finish line was 10.0 meters per second.
To estimate the instantaneous rate of change of a function y 5 f ( x ) at a point ( a, f ( a )) , we calculate the average rate of change of the function over a very small interval 3a, b4 . If we let the variable h represent the distance between
190
x 5 a and x 5 b , then b 5 a 1 h. Consequently, the difference quotient may be rewritten as f (b) 2 f (a) f (a 1 h) 2 f (a) 5 b2a (a 1 h) 2 a 5 f (a 1 h) 2 f (a) h
THE DIFFERENCE QUOTIENT AS AN ESTIMATE OF AN INSTANTANEOUS RATE OF CHANGE
The instantaneous rate of change of a function y 5 f ( x ) at a point ( a, f ( a )) may be estimated by calculating the difference quotient of f at a, f (a 1 h) 2 f (a) h using an h arbitrarily close to 0. (If h 5 0, the difference quotient is undefined.)
EXAMPLE
Estimating an Instantaneous Rate of Change

The average income of an elementary or secondary teacher may be modeled by S( t ) 5 0.00713t3 2 0.111t2 1 1.42t 1 31.6, 0 # t # 10 where S( t ) is the annual salary in thousands of dollars and t is the number of years since 1990. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 237, p. 151.) According to the model, how quickly was the average income of a teacher increasing in 1999?
SOLUTION In 1999, t 5 9. Using the difference quotient
S( 9 1 h ) 2 S( 9 ) h and selecting increasingly small values of h, we generate Table 3.4.

TABLE 3.4
h 1.000 0.100 0.010 0.001
S( 9 1 h ) 2 S( 9 ) h 1.2432 1.1628 1.1554 1.1547
We conclude that in 1999, the average annual salary of a teacher was increasing by about $1.155 thousand ($1155) per year. The following Technology Tip details how to generate a table of values for the difference quotient using the TI-83 Plus.
191
Calculating the Difference Quotient for Different Values of h 1. Enter the function S( t ) as Y1 by pressing the Y button and typing the equation.
2. Press the
button, scroll to and press ENTER . (Hint: It is quicker to scroll up than to scroll down.)
MATH
:Solver...,
3. If the Solver already contains an equation, you may see something similar to the graphic shown. Press the blue up arrow and then press CLEAR to delete the equation.
4. Notice that the equation is set equal to zero. We may rewrite the difference quotient D 5 05
Y1( x 1 h ) 2 Y1( x ) h
as
Y1( x 1 h ) 2 Y1( x ) h
2 D. Type in the
second equation using the VARS ; Y-VARS; 1:Function menu sequence to enter the function Y1. To enter the variable H, press the ALPHA ^ key sequence. To enter the variable D, press the ALPHA X1 key sequence. Press ENTER . 5. Enter the value of x where you want to evaluate the difference quotient and enter a small positive value of H. Move your cursor to the D variable and press the ALPHA ENTER key sequence to solve for the difference quotient D. Repeat for various values of H.
192
EXAMPLE
Using Technology to Estimate an Instantaneous Rate of Change

Based on data from 19902000, the cost of 1/2 gallon of prepackaged ice cream may be modeled by C( t ) 5 0.0142t2 2 0.0272t 1 2.53 dollars where t is the number of years since 1990. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 706, p. 468.) Use the Solver on your calculator to estimate the instantaneous rate of change of the cost in 1999.
SOLUTION We begin by entering C( t ) into the
Y
editor.
In 1999, t 5 9. Entering x 5 9 and H 5 0.001 and solving for D yields
Repeating this process for various values of H, we create Table 3.5.

TABLE 3.5
h 0.001 0.0001 0.00001 0.000001
C( 9 1 h ) 2 C( 9 ) h 0.22841420 0.22840142 0.22840014 0.22840000
In 1999, the cost of 1/2 gallon of prepackaged ice cream was increasing by about $0.23 per year. That is, from 1999 to 2000, the price was expected to increase by about 23 cents.
193
Limits
In Examples 1 through 3, we estimated the instantaneous rate of change at a point by calculating the average rate of change over a short interval by picking small values of h. The terms short and small are vague. Numerically, what does small mean? Mathematicians struggled with this dilemma for years before developing the concept of the limit. We will explore the limit concept graphically before giving a formal definition. Consider the graph of the function f ( x ) 5 2 x2 1 4 over the interval 32 3, 34 (Figure 3.5). We ask the question, As x gets close to 2, to what value does f ( x ) get close? Observe from the graph of f that as the value of x moves from 0 to 2, the value of f ( x ) moves from 4 to 0. We represent this behavior symbolically with the notation
xS22
y 4 3 2 1 3 2 1 1 2 3 4 5 1
y = f (x)
lim f ( x ) 5 0
FIGURE 3.5
which is read, the limit of f ( x ) as x approaches 2 from the left is 0. This is commonly referred to as a left-hand limit because we approach x 5 2 through values to the left of 2. Observe from the graph of f that as the value of x moves from 3 to 2, the value of f ( x ) moves from 25 to 0. We write
xS21
lim f ( x ) 5 0
xSa2
which is read, the limit of f ( x ) as x approaches 2 from the right is 0. This is commonly referred to as a right-hand limit because we approach x 5 2 through values to the right of 2. On a number line, 2 ` lies to the left of 1` . For the left-hand limit, lim f ( x ) , the minus sign is used to indicate that we are approaching x 5 a from the direction of 2 ` . For the right-hand limit, lim f ( x ) , the plus sign is used to indicate that we are approaching x 5 a from the direction of 1` . The left- and right-hand limit behavior can also be seen from a table of values for f ( x ) (Table 3.6).
xSa1
TABLE 3.6
x 0.00 T Left of x 5 2 T 1.00 1.90 1.99 2.00 c Right of x 5 2 c 2.01 2.10 3.00
f (x) 4.000 3.000 0.390 0.040 0.000 0.040 0.410 5.000 f ( x ) gets close to 0 as x nears 2 f ( x ) gets close to 0 as x nears 2
194
When the left- and right-hand limits of f ( x ) approach the same finite value, we say that the limit exists. In this case, the left- and right-hand limits of f ( x ) neared the same value ( y 5 0 ) as x approached 2. We say that the limit of f ( x ) as x approaches 2 is 0 and write lim f ( x ) 5 0
xS2
Sometimes the left- and right-hand limits of a function at a point are not equal. Consider the graph of the piecewise function f (x) 5 b
y 3 2 1 3 2 1 1 2 3 4 5 6 1 2
y = f (x) x
on the interval 32 3, 34 (see Figure 3.6). We ask the question, As x gets close to 1, to what value does f ( x ) get close? Observe from the graph that as the value of x moves from 0 to 1, the value of f ( x ) moves from 2 to 1. We write
xS12
2 x2 1 2 x11
x#1 x.1
lim f ( x ) 5 1
That is, the left-hand limit of f ( x ) as x approaches 1 is 1. Similarly, as the value of x moves from 2 to 1, the value of f ( x ) moves from 3 to 2. We write
xS11
lim f ( x ) 5 2
FIGURE 3.6
That is, the right-hand limit of f ( x ) as x approaches 1 is 2. Since for this function, the left- and right-hand limits are not equal, we say that the limit of f ( x ) as x approaches 1 does not exist or, simply, the limit does not exist. This can also be seen from a table of values for f ( x ) (Table 3.7).
TABLE 3.7
x T Left of x 5 1 T 0.00 0.90 0.99 1.00 c Right of x 5 1 c 1.01 1.10 2.00 3.00
f (x) 2.000 1.190 1.020 1.000 2.010 2.100 3.000 4.000 f ( x ) gets close to 2 as x nears 1 f ( x ) gets close to 1 as x nears 1
EXAMPLE
Finding Left-hand and Right-hand Limits

Given Table 3.8, determine lim f ( x ) and lim f ( x ) .
xS02 xS01
195
TABLE 3.8
x 1.00 0.10 0.01 0.00 0.01 0.10 1.00
f (x) 0.84147 0.99833 0.99998 Undefined 0.99998 0.99833 0.84147
SOLUTION We observe that as x moves from 21 to 20.01, the value of f ( x ) nears 1. That is,
xS02
lim f ( x ) 5 1
Similarly, as x moves from 1 to 0.01, the value of f ( x ) nears 1. That is,

xS01
lim f ( x ) 5 1
The limit of f ( x ) as x approaches 0 is 1, since the left- and right-hand limits of f ( x ) are both equal to 1. That is, lim f ( x ) 5 1
xS0
It is important to note that even though f ( x ) is undefined at x 5 0 , as x gets close to 0, f ( x ) gets close to 1.
As illustrated in Example 4, one of the most powerful features of limits is that the limit of a function may exist at a point where the function itself is undefined. This feature will be used often when we introduce the limit definition of the derivative later in this section.
THE LIMIT OF A FUNCTION
If f ( x ) is defined for all values of x near c, then

xSc
lim f ( x ) 5 L
means that as x approaches c, f ( x ) approaches L. We say that the limit exists if 1. L is a finite number and 2. Approaching c from the left or right yields the same value of L.
196
We are most interested in the limit of the difference quotient as h approaches zero. That is,
hS0
lim
f (a 1 h) 2 f (a) h
Recall that the difference quotient represents the average rate of change of f ( x ) over the interval 3a, a 1 h4 . When we place the limit on the difference quotient, we are symbolically asking, As the distance between the two x values (a and a 1 h) gets smaller, what happens to the average rate of change of f ( x ) on the interval 3a, a 1 h4 ? The limit of the difference quotient as h approaches zero (if the limit exists) is the instantaneous rate of change in f ( x ) at x a . Note that even though the difference quotient is undefined when h 5 0, the limit may still exist.
EXAMPLE
Calculating an Instantaneous Rate of Change

Let f ( x ) 5 x2 . What is the instantaneous rate of change of f ( x ) when x 5 3 ?
SOLUTION We can calculate the instantaneous rate of change by taking the limit of the difference quotient as h approaches zero. The instantaneous rate of change of f ( x ) at x 5 3 is given by
hS0
lim
f (3 1 h) 2 f (3) (3 1 h)2 2 (3)2 5 lim hS0 h h 5 lim 5 lim

hS0
Since f ( x ) 5 x2
( 9 1 6h 1 h2 ) 2 9 h 6h 1 h2 h h( 6 1 h ) h
Since h 5 1 for h 2 0
h
hS0
5 lim
hS0 hS0
5 lim ( 6 1 h ) for h 2 0
As h nears 0, what happens to the value of 6 1 h? Lets pick values of h near 0 (Table 3.9).
TABLE 3.9
h 0.100 0.010 0.001 0.000 0.001 0.010 0.100
6 1 h, h 2 0 5.900 5.990 5.999 Undefined 6.001 6.010 6.100
197
As seen from the table, even though the difference quotient is undefined when h 5 0, the value of the simplified difference quotient, 6 1 h, gets close to 6 as h approaches 0. In fact, by picking sufficiently small values of h, we can get as close to 6 as we would like. So the instantaneous rate of change of f ( x ) when x 5 3 is 6. Observe that we can attain the same result by plugging in h 5 0 after canceling out the h in the denominator of the difference quotient. That is,
hS0
lim( 6 1 h ) 5 6 1 0 56
Throughout the rest of this chapter, we will substitute in h 5 0 after eliminating the h in the denominator of the difference quotient. This process will simplify our computations while still giving the correct result. The limit of the difference quotient as h approaches zero is used widely throughout calculus and is called the derivative.
THE DERIVATIVE OF A FUNCTION AT A POINT
The derivative of a function y 5 f ( x ) at a point ( a, f ( a )) is f r( a ) 5 lim

hS0
f (a 1 h) 2 f (a) h
f r( a ) is read f prime of a and is the instantaneous rate of change of the function f at the point ( a, f ( a )) .
EXAMPLE
Calculating the Derivative of a Function at a Point

Given f ( x ) 5 3x 1 1, find f r( 2 ) .
SOLUTION
f r( 2 ) 5 lim
hS0
f (2 1 h) 2 f (2) h
5 lim
hS0
33( 2 1 h ) 1 14 2 33( 2 ) 1 14 h
Since f (2 h) [3(2 h) 1] and f (2) 3(2) 1
5 lim 5 lim 5 lim
hS0
( 6 1 3h 1 1 ) 2 ( 7 ) h ( 3h 1 7 ) 2 ( 7 ) h 3h h Since
h h
hS0
hS0 hS0
5 lim 3 53
51
198
In this case, the difference quotient turned out to be a constant value of 3, so taking the limit of the difference quotient as h approached 0 did not alter the value of the difference quotient. For linear functions, the slope of the line is the instantaneous rate of change of the function at any value of x. Consequently, the derivative of a linear function will always be a constant value that is equal to the slope of the line.
Recall that in Example 1, we estimated the runners finish-line speed by finding the average rate of change in distance over a small interval of time. In other words, the distance traveled divided by the amount of time used to travel the distance yielded the average speed. One of the drawbacks of using the term speed is that it does not indicate direction. For example, consider two race cars traveling at 75 mph on the same road (Figure 3.7).
1
velocity = 75 mph
2
velocity = 75 mph
75 mph
75 mph
1
velocity = 75 mph
2
velocity = 75 mph
75 mph
75 mph
FIGURE 3.7
If the cars are traveling in the same direction, we dont foresee any problems. However, if the cars are traveling in opposite directions, a head-on collision may be imminent. The direction of the cars is a key piece of information that is not captured by the term speed. Fortunately, the term velocity combines the concepts of speed and direction. A negative sign placed on the speed is used to distinguish a difference in direction. In Figure 3.7, a positive velocity indicates that the car is traveling from left to right. A negative velocity indicates that the car is traveling from right to left. When talking about the velocity of objects that are rising or falling, it is customary to use a positive velocity to indicate that the object is rising and a negative velocity to indicate that the object is falling. It is common to model the vertical position of a free-falling object on earth with the equation s( t ) 5 2 16t2 1 v0t 1 s0 feet, where v0 is the velocity of the object (in feet per second) and s0 is the vertical position of the object (in feet) at time t 5 0 seconds. (Although this model neglects air resistance, it does give us an idea of what is happening physically.) To determine the velocity of the object at time t seconds, we simply need to find sr( t ) .
199
EXAMPLE
Calculating the Velocity of a Falling Object

A water balloon is dropped from the top of a 20-foot roof. How fast is it falling 1 second into its flight? How fast is it falling when it hits the ground?
SOLUTION The initial velocity of the balloon, v0, is 0 feet per second because the balloon was dropped, not thrown. The initial position of the balloon, s0, is 20 feet, since the balloon is being dropped from 20 feet above the ground. Therefore, the balloons position may be modeled by
s( t ) 5 2 16t 2 1 v0t 1 s0 5 2 16t 2 1 0t 1 20 5 2 16t 2 1 20 To calculate the velocity of the balloon at time t 5 1 second, we must find sr( 1 ) . sr( 1 ) 5 lim 5 lim 5 lim 5 lim 5 lim 5 lim 5 lim
hS0
s( 1 1 h ) 2 s( 1 ) h
hS0
32 16( 1 1 h ) 2 1 204 2 32 16( 1 ) 2 1 204 h
Since s( 1 1 h ) 5 2 16( 1 1 h ) 2 1 20
hS0
32 16( 1 1 2h 1 h2 ) 1 204 2 32 16 1 204 h ( 2 16 2 32h 2 16h2 1 20 ) 2 ( 4 ) h ( 2 32h 2 16h2 1 4 ) 2 ( 4 ) h 2 32h 2 16h2 h h( 2 32 2 16h ) h
Factor h out of the numerator Since h 5 1 for h 2 0 Plug in h 5 0
h
hS0
hS0
hS0
hS0 hS0
5 lim( 2 32 2 16h ) 5 2 32 2 16( 0 ) 5 2 32 feet per second
So 1 second after the balloon is dropped, it is traveling toward the ground at a speed of 32 feet per second. (As previously noted, a positive velocity typically indicates that an object is moving upward, while a negative velocity signifies that an object is moving downward.) To determine how fast the balloon is falling when it hits the ground, we must first determine when it will hit the ground. Since s( t ) is the height of the balloon above the ground, we must determine when s( t ) 5 0. s( t ) 5 2 16t 2 1 20 0 5 2 16t 2 1 20 16t 2 5 20 t2 5 20 16
t 2 5 1.25 t < 1.1 seconds
200
We estimate that the balloon will hit the ground in 1.1 seconds. To determine the velocity of the balloon at that time, we calculate sr( 1.1 ) . sr( 1.1 ) 5 lim 5 lim 5 lim 5 lim 5 lim
hS0
s( 1.1 1 h ) 2 s( 1.1 ) h
hS0
32 16( 1.1 1 h ) 2 1 204 2 32 16( 1.1 ) 2 1 204 h
hS0
32 16( 1.21 1 2.2h 1 h2 ) 1 204 2 32 16( 1.21 ) 1 204 h ( 2 19.36 2 35.2h 2 16h2 1 20 ) 2 ( 2 19.36 1 20 ) h ( 2 35.2h 2 16h2 1 0.64 ) 2 ( 0.64 ) h 2 35.2h 2 16h2 h h( 2 35.2 2 16h ) h
Factor h out of the numerator Since h 5 1 for h 2 0 Plug in h 5 0
h
hS0
hS0
5 lim 5 lim
hS0
hS0 hS0
5 lim( 2 35.2 2 16h ) 5 2 35.2 2 16( 0 ) 5 2 35.2 feet per second
We estimate that the balloon will be falling at a rate of a little over 35 feet per second when it strikes the ground.
EXAMPLE
Finding and Interpreting the Meaning of the Derivative of a Function at a Point

The population of Washington state may be modeled by the function P( t ) 5 0.8981( 1.019 ) r million people where t is the number of years since 1900. (Source: Modeled from www.ofm.wa.gov data.) Find and interpret the meaning of Pr( 104 ) .
SOLUTION Since t is the number of years since 1900, t 5 104 is the year in 2004. Pr( 104 ) is the instantaneous rate of change in the population in 2004, given in millions of people per year.
Pr( 104 ) 5 lim
hS0
P( 104 1 h ) 2 P( 104 ) h
5 lim
hS0
5 lim 5 lim
hS0
0.8981 3( 1.019 ) 104 1 h 2 ( 1.019 ) 1044 h
30.8981( 1.019 ) 104 1 h4 2 30.8981( 1.019 ) 1044 h

Factor out 0 . 8981 Since ( 1.019 ) 104 1 h 5 ( 1.019 ) 104( 1.019 ) h
hS0
0.8981 3( 1.019 ) 104( 1.019 ) h 2 ( 1.019 ) 1044 h
201
5 lim
hS0
( 0.8981 )( 1.019 )104 3( 1.019 ) h 2 14 h 6.360 3( 1.019 ) h 2 14 h
Factor out ( 1 . 019 ) 104
5 lim
hS0
Unlike in Examples 5, 6, and 7, we are unable to eliminate the h in the denominator algebraically and calculate the exact value of Pr( 104 ) . Nevertheless, by picking a small value for h (say h 5 0.001), we can estimate the instantaneous rate of change. 6.360 3( 1.019 ) 0.001 2 14 5 0 . 1197 million people per year 0.001 5 119 . 7 thousand people per year < 120 thousand people per year According to the model, the population of Washington was increasing by approximately 120 thousand people per year in 2004.
Although we were unable to obtain the exact value of the derivative of the exponential function in Example 8, we will develop the theory in later sections that will allow us to calculate the exact value of the derivative of an exponential function.
Calculating the Difference Quotient for Different Values of h 1. Label six cells as indicated in the figure.
2. Type in appropriate values in Row 2, using formulas as appropriate. Cells B2, D2, and E2 are calculated using formulas based on cells A2, A2 and C2, and D2, respectively. 3. Cell F2 will contain the difference quotient formula h terms of the cell references, ( E2 2 B2 ) > C2.
f (x 1 h ) 2 f (x )
or, in
(Continued)
202
4. Change the value of h to see the difference quotient for that value of h.
3.2 Summary
In this section, you learned how to use the difference quotient to estimate an instantaneous rate of change. You also discovered that the derivative may be used to calculate the exact value of an instantaneous rate of change of a function at a point.

1. What formula is used to calculate an average rate of change? 2. How may the formula for the average rate of change be used to estimate an instantaneous rate of change? 3. What does it mean when we say, as h S 0, f ( h ) S 10? 4. Given that f ( x ) 5 2 3x2 1 7x 2 1 , what is f (2 1 h) ? 5. Given that f ( x ) 5 2 3x2 1 7x 2 1, what is f (x 1 h) ? 6. Simplify 7. Simplify
3 ( 3 1 h )2 1 2( 3 1 h ) 4 2 3 32 1 2( 3 ) 4 . h
3 5( 2 1 1 h )3 1 9( 2 1 1 h ) 4 2 3 5( 2 1 )3 1 9( 2 1 ) 4 . h 3 2( x 1 h )2 2 8( x 1 h ) 4 2 3 2x2 2 8x 4 . h
8. Simplify
3.2 Exercises
In Exercises 110, use the difference quotient
f (a 1 h) 2 f (a) h
9. P( r ) 5 500( 1 1 r ) 2; r 5 0.07 10. P( r ) 5 100( 1 1 r ) 2; r 5 0.12 In Exercises 1120, use the derivative to calculate the instantaneous rate of change of the function at the given input value. (In each exercise, you can eliminate the h algebraically.) Compare your answers to the solutions of Exercises 110. 11. f ( x ) 5 x2; x 5 2 12. g( x ) 5 2 x2 1 1; x 5 2 13. s( t ) 5 2 16t2 1 64; t 5 2 14. s( t ) 5 2 16t2 1 64t 1 32; t 5 0 15. w( t ) 5 4t 1 2; t 5 5 16. P( t ) 5 t2 2 2t 1 1; t 5 1
(with h 5 0.1, h 5 0.01, and
h 5 0.001) to estimate the instantaneous rate of change of the function at the given input value. 1. f ( x ) 5 x2; x 5 2 2. h( x ) 5 2 x2 1 1; x 5 0 3. s( t ) 5 2 16t2 1 64; t 5 2 4. s( t ) 5 2 16t2 1 64t 1 32; t 5 0 5. w( t ) 5 4t 1 2; t 5 5 6. P( t ) 5 t2 2 2t 1 1; t 5 1 7. P( t ) 5 5; t 5 25 8. g( x ) 5 5x 2 x2; x 5 2 1
203
17. P( t ) 5 5; t 5 25 18. g( x ) 5 5x 2 x ; x 5 2 1
2
33.
19. P( r ) 5 500( 1 1 r ) 2; r 5 0.07 20. P( r ) 5 100( 1 1 r ) 2; r 5 0.12 In Exercises 2130, use the difference quotient (with h 5 0.1, h 5 0.01, and h 5 0.001) to estimate the instantaneous rate of change of the function at the given input value. You may find it helpful to apply the Technology Tip demonstrated in the section. 21. f ( x ) 5 2x23; x 5 3 22. P( t ) 5 50( 1.02 ) t; t 5 10 23. P( t ) 5 230( 0.9 ) t; t 5 25 24. g( x ) 5 2 2( 3 ) x; x 5 2 1 25. P( r ) 5 500( 1 1 r ) 10; r 5 0.07 26. P( r ) 5 100( 1 1 r ) 5; r 5 0.12 27. y 5 ln( x ) ; x 5 2 28. y 5 x ln( x ) ; x 5 1 29. g( x ) 5 e3x; x 5 1 30. g( x ) 5 8 2 2x; x 5 5 In Exercises 3140, determine the instantaneous rate of change of the function at the indicated input value. (You may find it helpful to apply the Technology Tip demonstrated in the section.) Then explain the real-life meaning of the result. 31. The annual salary, S, of public elementary and secondary teachers t years after 1990 may be modeled by S( t ) 5 0.00713t3 2 0.111t2 1 1.42t 1 31.6 thousand dollars. (Source: Modeled from Statistical
Public-School Teacher Salaries
The amount of money spent in the United States on prescription drugs may be modeled by D( t ) 5 0.003667t 3 2 0.0986t 2 1 0.8976t 1 2.829
Prescription Drug Spending
billion dollars, where t is the number of years after 1960. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 119, p. 91.) Find and interpret the meaning of Dr( 45 ) . 34.
Video-Game Sales The annual sales, S, of the U.S. video-game industry from 1997 to 2001 may be modeled by S( t ) 5 0.1958t 4 2 1.275t 3 1 2.254t 2 2 0.0750t 1 5.100
billion dollars, where t is the number of years since 1997. (Source: Modeled from data at www.npdfunworld.com.) Find and interpret the meaning of Sr( 5 ) . 35.
Video-Game Sales In the first quarter of 2001, U.S. video-game industry sales totaled $1.6 billion. Sales increased by 20 percent in the first quarter of 2002. The number one video game in the first quarter of 2002 was Grand Theft Auto 3. (Source: www.npdfunworld.com.) If this pattern continues, video-game industry sales in the first quarter, S, may be modeled by S 5 1.6( 1.2 ) t
billion dollars, where t is the number of years since 2001. Find and interpret the meaning of Sr( 6 ) . 36. The amount of money spent annually in the United States on nursing home care from 1960 to 2000 may be modeled by P( t ) 5 0.07214t2 2 0.4957t 1 1.029
Nursing Home Care
Find and interpret the meaning of Sr( 10 ) . 32.
Abstract of the United States, 2001, Table 237, p. 151.)
The annual payroll, P, of the apparel and textile manufacturing industry between 1992 and 1997 may be modeled by P 5 2 1560x 1 24,850 billion dollars, where x is the number of paid employees (in thousands). (Source: Modeled from
Apparel and Textile Industry Pay
billion dollars, where t is the number of years after 1960. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 119, p. 91.) Find and interpret the meaning of Dr( 45 ) . 37. The cost of full-time resident tuition at Green River Community College from 1994 to 2001 may be modeled by E( t ) 5 0.1869t3 2 1.081t2 1 19.57t 1 431.6
College Tuition
Statistical Abstract of the United States, 2001, Table 722, p. 482.) Find and interpret the meaning of
Pr( 15 ) .
of Er( 7 ) .
(Source: Modeled from Green River Community College data.) Find and interpret the meaning
dollars, where t is the number of years since 1994.
204 38.
Demand for VHS Tapes From 2000 to 2002, the worldwide demand for blank VHS tapes fell. The demand for the tapes may be modeled by V( t ) 5 2 20t2 2 67t 1 1409
would a penny dropped from the observation deck be traveling when it hit the ground? 43. A small rubber ball is thrown into the air by a child at a velocity of 20 feet per second. The child releases the ball 4 feet above the ground. What is the velocity of the ball after 1 second?
Velocity of a Ball
million tapes, where t is the number of years since 2000. (Source: Modeled from International Recording Media Association data.) Find and interpret the meaning of V r( 2 ) . 39. From 2000 to 2002, the worldwide demand for CD-R disks increased substantially. The demand for the disks may be modeled by C( t ) 5 2 252.5t2 1 1252.5t 1 2730
Demand for CD-R Disks
44.
A diving ring is thrown from a diving board at a velocity of 2 10 feet per second toward the surface of a diving pool. If the surface of the pool is 40 feet below the point of release, what will be the velocity of the ring when it hits the surface of the water?
Velocity of a Diving Ring
million disks, where t is the number of years since 2000. (Source: Modeled from International Recording Media Association data.) Find and interpret the meaning of C r( 2 ) . 40.
VHS Tapes and CD-R Disks From 2000 to 2002, the worldwide demand for blank VHS tapes fell, while the demand for CD-R disks increased. The demand for CD-R disks may be modeled by C( v ) 5 2 0.03550v2 1 85.45v 2 47,200
45.
A juggler tosses a bowling pin into the air at a velocity of 15 feet per second. If she catches the pin at the same point she released it, what will be the velocity of the pin when she catches it?
Velocity of a Juggling Pin
46. Explain the difference between an average rate of change and an instantaneous rate of change. 47. An alternative form of the difference quotient is f (a 1 h) 2 f (a) . Is this form of the difference h quotient more useful than the
f (b) 2 f (a) b2a
million disks, where v is the number of blank VHS tapes sold (in millions). (Source: Modeled Find and interpret the meaning of C r( 500 ) . Do you believe that the sales of blank VHS tapes and CD-R disks are related? Justify your answer. Exercises 4145 deal with the velocity of a freefalling object on earth. As explained in the section, the vertical position of a free-falling object may be modeled by s( t ) 5 2 16t2 1 v0t 1 s0 feet, where v0 is the velocity of the object and s0 is the vertical position of the object at time t 5 0 seconds. 41. A can of soda is dropped from a diving board 40 feet above the bottom of an empty pool. How fast is the can traveling when it reaches the bottom of the pool?
Velocity of a Dropped Object
from International Recording Media Association data.)
form of
the difference quotient when discussing the concept of instantaneous rate of change at a point? Explain. 48. Does the speedometer on a car give an average rate of change or an instantaneous rate of change? Explain. 49. What does it mean to say that lim g( t ) exists?
tS2
50. What does it mean to say that lim g( t ) 5 14?

t S 21
51. How is the difference quotient related to the derivative?
42.
The Space Needle in Seattle, Washington, is a popular tourist attraction. The observation deck is approximately 520 feet above the ground. Dropping anything from that height is extremely dangerous to people on the ground. How fast
Velocity of a Dropped Object
52. The derivative of a continuous function f is undefined at the point 3a, f ( a ) 4 . Draw a possible graph of f, including the point 3a, f ( a ) 4 .
53. A function f has the property that f r( a ) 5 f r( b ) . Does a 5 b? Explain.
3.3 Limits and Continuity
205
54. The instantaneous rate of change of a function f is equal to 3 for all values of x. How many x-intercepts does the graph of f have? 55. The instantaneous rate of change of a function f is equal to 0 for all values of x. How many x-intercepts does the graph of f have? 56. The Mean Value Theorem states that if f is continuous and smooth (no breaks, no sharp
points) on an interval 3a, b4 , then there exists a value c with a , c , b such that f (b) 2 f (a) f r( c ) 5 b2a 2 Given f ( x ) 5 x on the interval 32, 44 , find a value c with 2 , c , 4 such that f (4) 2 f (2) f r( c ) 5 422
Perform mathematical operations with limits Determine if a function is continuous
Few concepts in calculus are more important than the limit concept. The limit concept provides the bridge between algebra and calculus. In this section, we will continue our discussion of limits and formally define the term continuity.
Numerous mathematical operations may be performed when working with limits. Often, effective application of the operations makes limit calculations easier.
LIMIT OPERATIONS
Given that c and k are real numbers and n is a positive integer, the following properties and operations are valid, provided that lim f ( x ) and lim g( x ) exist. 1. lim k 5 k
xSc xSc xSc
Constant Property
xSc
2. lim k f ( x ) 5 k( lim f ( x ))
xSc
Constant Multiple Property

xSc
3. lim ( f ( x ) 6 g( x )) 5 lim f ( x ) 6 lim g( x )

xSc xSc
Addition/Subtraction Property Multiplication Property Division Property Power Property Rational Exponent Property
4. lim ( f ( x ) ? g( x )) 5 ( lim f ( x )) ? ( lim g( x ))

xSc xSc xSc
5. lim
lim f ( x ) f (x) 5 xSc , if lim g( x ) 2 0 x S c g( x ) lim g( x ) xSc

xSc
n
6. lim 3 f ( x )4 n 5 S lim f ( x )T
xSc xSc xSc xSc
7. lim 3 f ( x ) 4 1>n 5 S lim f ( x )T 1>n (If n is even, lim f ( x ) must be positive.)

xSc
206
EXAMPLE
Calculating a Limit with an Indeterminate Form

Calculate lim
x S 20
x2 2 400 . x 2 20
SOLUTION We first try direct substitution.
x S 20
lim
x2 2 400 202 2 400 5 x 2 20 20 2 20 0 5 0
The form 0 is called an indeterminate form. It means that we dont know whether the limit exists. We will try factoring the numerator and breaking the function into its various parts.
x S 20
lim
x2 2 400 ( x 1 20 )( x 2 20 ) 5 lim x 2 20 x S 20 x 2 20 x 2 20 5 lim ( x 1 20 ) c lim a bd x S 20 x S 20 x 2 20 5 ( 20 1 20 ) S lim ( 1 )T , x 2 20

x S 20
Product Property
5 40( 1 ) 5 40 The limit exists and equals 40.
Constant Property
the form ` . These forms are referred to as indeterminate forms. An indeterminate form implies that we dont know the value of the limit or whether it exists, but we may be able to determine the value with further manipulation. In Example 1, we 0 ` saw the form 0. In Example 2, we will see the form ` .
The process of trying to calculate a limit sometimes produces the form 0 or

`
Limits at Infinity
Recall that lim f ( x ) 5 L means that as x approaches c (from the left or right), the value of f ( x ) approaches L. In the previous section, we looked at limits with a finite value for c. What if c is 2 ` or ` ? Does the concept of limit make sense? Well investigate this question by looking at lim x . This expression asks, As x approaches infinity, to what value does values for f ( x ) 5
1 x 1 x
xS` xSc
get close? We construct a table of
to help us get an intuitive feel for the answer to the question.
207
x 10 100 1000 10,000 100,000 1,000,000
f (x) 5 0.1 0.01 0.001 0.0001
1 x
0.00001 0.000001
Observe that as x gets larger and larger, f ( x ) 5 this result graphically. See Figure 3.8.
y 0.04 0.03 0.02 0.01 500 0.01 0.02 0.03 0.04 500 1 f (x ) = x
fact, the larger x is, the closer f ( x ) is to 0. Therefore, lim
1 x
gets smaller and smaller. In

xS`
1 x
5 0. We can verify
FIGURE 3.8
Note that although we can approach ` only from the left, we are still able to determine a constant value that the function is approaching. For our function, f ( x ) 5 , what is L in lim f ( x ) 5 L? That is, what conx stant value does the function f ( x ) 5 approach as x approaches 2 ` ? A quick x look at the graph shows us that although f ( x ) is negative for all values of x , 0, 1 f ( x ) approaches 0 as x approaches 2 ` . That is, lim 5 0.
x S 2`
x S 2`
In general, for any function of the form f ( x ) 5

1 and lim n x x S 2`
x 1 xn
with n . 0, lim
xS`
1 xn
50
5 0. This property will be extremely useful in later examples and
exercises.
208
LIMITS AT INFINITY
A function f ( x ) has a limit at 2` if there exists some constant value L to which f ( x ) gets infinitely close for negative x of sufficiently large absolute value. Symbolically, we write
xS 2`
lim f ( x ) 5 L
Similarly, a function f ( x ) has a limit at ` if there exists some constant value L to which f ( x ) gets infinitely close for sufficiently large x. Symbolically, we write
xS`
lim f ( x ) 5 L
EXAMPLE
Determining If a Function Has a Limit at 6`

Find lim
2 x 11 x S 2`
2 x
and lim
2x xS` x 1 1
2
symbolically, graphically, and using a table.
denominator x 2 1 1 S ` . Since the numerator is approaching zero and the de2x nominator is approaching infinity, lim x2 1 1 5 0.
x S 2`
SOLUTION Observe that as x approaches 2 ` , the numerator 2x S 0 and the
Observe that as x approaches ` , the numerator 2x S ` and the denominator x2 1 1 S ` . So lim

2x xS` x 1 1
2
will result in an indeterminate form, ` . The question
becomes, Is the numerator or the denominator going to infinity faster? Since an increasing exponential function ultimately grows faster than any polynomial function, the numerator will grow more rapidly than the denominator, and there2x fore lim x2 1 1 5 ` .
xS`
We verify our conclusions graphically in Figure 3.9.

y 4 3 2 1 8 6 4 2 1 x f (x) = 2x x2 + 1
FIGURE 3.9
From the graph, it appears that as x approaches 2 ` , the function approaches zero. However, the function does not appear to approach a constant value as x approaches ` .
209
We can also verify our conclusions using a table. We evaluate the function at negative values with large magnitude and at large positive values. 2x x2 1 1
x 2 100 2 50 2 25 25 50 100
f (x) 5
7.9 3 10235 < 0 3.6 3 10219 < 0 4.8 3 10211 < 0 53,601 4.5 3 1011 1.3 3 1026
We observe that negative values of x with large magnitude yield function values close to zero. On the other hand, large positive values of x yield increasingly large 2x 2x function values. So lim x2 1 1 5 0 and lim x2 1 1 5 ` .
x S 2` xS`
EXAMPLE
Using Limits at Infinity in a Real-World Context

Based on data from 1980 to 2003, the number of golf course facilities in the United States may be modeled by g( t ) 5 4633 1 12,000 facilities 1 1 59.97e20.2567t
where t is the number of years since the end of 1980. Calculate lim g( t ) and tS` interpret your result in a real-world context.
SOLUTION We construct a table of values using increasingly large values of t.
t 0 25 50 100 1,000 10,000
g( t ) 12,076 16,220 16,632 16,633 16,633 16,633
From the table, we see that lim g( t ) 5 16,633. According to the model, the number
tS`
of golf course facilities in the United States will approach but not exceed 16,633 as times goes on.
210
In general, for any function of the form f ( x ) 5 x n with n . 0, lim

1 lim x n xS`
xS 2`
1 xn
50
5 0. This property will be extremely useful in later examples and exand ercises, including Examples 4 and 5.
EXAMPLE
Finding the Limit at Infinity for a Rational Function

Find lim
x 2 1 2x 1 5 x3 1 1 xS`
SOLUTION Observe that dividing the numerator and denominator by x3 yields
x2 1 2x 1 5 x 1 2x 1 5 x3 lim 5 lim xS` x3 1 1 xS` x3 1 1 x3

2
Divide numerator and denominator by x3
1 1 1 1 2a 2b 1 5a 3b x x x 5 lim xS` 1 11 3 x 5 01010 110

Since lim
xS`
1 5 0 for n . 0 xn
50 So lim
xS`
x 1 2x 1 5 x3 1 1
2
5 0.
EXAMPLE
Finding the Limit at Infinity for a Rational Function

Find lim x2
xS`
x3 , 14
if it exists.
SOLUTION Observe that dividing the numerator and denominator by x2 yields
x3 x3 x2 lim 2 5 lim 2 xS` x 1 4 xS` x 1 4 x2 5 lim x 11 4 x2

xS`
xS`
lim x
110
xS`
5 lim x 5`
211
Thus the rational expression approaches infinity. We verify this result graphically in Figure 3.10.
y 20 f (x) = 15 10 5 5 5 x x3 x2 + 4
10
15
20
FIGURE 3.10
EXAMPLE
Using Limits at Infinity in a Real-World Context

Based on data from 1990 to 2004, blank cassette tape sales may be modeled by C( t ) 5 327.2 1 50 million dollars 1 1 0.01003e0.5277t
where t is the number of years since 1990. Calculate and interpret the meaning of lim C( t ) . Then explain whether or not you think the model prediction will be
tS`
realistic in the long term.

SOLUTION
tS`
lim C( t ) 5 lim a
tS`
327.2 1 50 b 1 1 0.01003e0.5277t
tS`
5 327.2 a lim 5 50
1 b 1 lim 50 1 1 0.01003e0.5277t tS`

Since denominator S `
5 327.2( 0 ) 1 50
According to the model, blank cassette tape sales will approach 50 million dollars annually as time goes on. While this model may be realistic in the short term, we think that cassette tapes will ultimately be replaced by alternative forms of media storage and, as a result, sales will eventually reach zero.
Continuity
Many function graphs have the property that they may be drawn without lifting the pencil from the paper. We have previously used this loose definition of continuity to determine whether a graph is continuous. With the power of limits, we can formally define continuity.
212
CONTINUITY AT A POINT
A function f ( x ) is said to be continuous at x 5 c if all of the following criteria are satisfied. 1. lim f ( x ) exists.
xSc
2. f ( c ) exists. 3. lim f ( x ) 5 f ( c )
xSc
Each of the discontinuous functions in Figure 3.11 fails one or more of the criteria. Note that the third criterion cannot be satisfied if either of the first two criteria is not satisfied.
y 5 2.0 4 3 2 1 1 1 2 3 4 5 x 0.5 0 0 0.5 1.0 1.5 2.0
xS2
1.5 1.0
FIGURE 3.11a lim f ( x ) does not exist.
FIGURE 3.11b f ( 2 ) does not exist.
0 0 1.75 2.0
xS2
2.25
FIGURE 3.11c lim f ( x ) 2 f ( 2 ) .
A function is said to be continuous on an open interval if it is continuous at every point in the interval. A function is continuous everywhere if it is continuous at all possible input values. Such functions are referred to as continuous functions. All polynomial, exponential, logarithmic, and logistic functions are continuous.
213
EXAMPLE
Determining If a Function Is Continuous

Is the function f ( x ) 5
x2 2 1 x21
continuous? Explain.
SOLUTION From a function graph drawn using a graphing calculator or computer program, it appears that the function is continuous. See Figure 3.12.
y 3 2 1 2 1 1 2 1 2 3 x f (x) = x 1
FIGURE 3.12
Yet, from the function equation, we can see that the function is undefined when x 5 1 because the denominator becomes zero. Observe that f (x) 5 5 x2 2 1 x21 ( x 2 1 )( x 1 1 ) x21
5 x 1 1, if x 2 1 Well first determine lim f ( x ) .

xS1 xS1
lim f ( x ) 5 lim ( x 1 1 )
xS1
5111 52 The first criterion of continuity has been met. What is f ( 1 ) ? Initially we may think that f ( 1 ) 5 2. However, f ( x ) 5 x 1 1 only if x 2 1. When x 5 1, the function is undefined due to the zero in the denominator of the original function. So f ( 1 ) is undefined and the second criterion of continuity is not satisfied. x2 2 1 The function f ( x ) 5 x 2 1 is not continuous because of the discontinuity that occurs at x 5 1.
EXAMPLE
Determining If a Function Is Continuous

Is the function f ( x ) 5 e e x21 1 1, x , 1 continuous? Explain. 2 x2 1 3, x $ 1
SOLUTION We saw in Example 7 that the graph drawn by a calculator or computer may lead us to mistakenly conclude that a discontinuous function is continuous.
214
So we will first solve this problem symbolically and then verify our conclusions graphically. Since exponential and polynomial functions are continuous, each piece of the piecewise function is continuous. The only place where the piecewise function might not be continuous is at x 5 1. We first calculate lim f ( x ) . We have
xS1 x S 12
lim f ( x ) 5 lim ( ex 2 1 1 1 )
x S 12
x S 11
lim f( x ) 5 lim ( 2 x2 1 3 )
x S 11
5 e1 2 1 1 1 5 e0 1 1 52
and
5 2 ( 12 ) 1 3 5 21 1 3 52
Since the left- and right-hand limits are equal, lim f ( x ) 5 2. The first criterion xS1 of continuity is satisfied. We next calculate f ( 1 ) . f ( 1 ) 5 2 ( 12 ) 1 3 5 21 1 3 52 The second criterion of continuity is satisfied. Because lim f ( x ) 5 f ( 1 ) , the function satisfies the third criterion of continuity. Thus the function is continuous at x 5 1. Since the only possible point at which a discontinuity might have occurred is x 5 1, the function is continuous everywhere. A graph of the function (Figure 3.13) verifies our conclusion.
y 3 2 f (x) = e x 1 + 1 1 2 1 1 2 1 f (x) = x 2 + 3 x
xS1
Since f ( x ) 5 2 x 2 1 3 for x $ 1
FIGURE 3.13
Much of the study of calculus is focused on the behavior of continuous functions. In future chapters, many of the shortcuts for calculating the instantaneous rate of change of a function will rely on the concept of continuity.
3.3 Summary
In this section, you leaned how to work with limits, including limits at 6` . You also discovered how to formally determine whether a function is continuous.
215

1. Simplify the expression algebraically:
x 2 1 5x x2 1 6x 1 5 x 2 1 5x x 1 6x 1 5
2
x 2 1 5x . x2 1 6x 1 5
5. As x approaches 7 from the left, what value does

x 2 1 9x 1 14 x2 2 49
2. As x approaches 1 from the right, what value does near? 3. As x approaches 5 from the right, what value does near? 4. Simplify the expression algebraically:
x 2 1 9x 1 14 . x2 2 49
near?
6. As x approaches 2 7 from the left, what value does

x 2 1 9x 1 14 x 2 2 49
near?
x11
7. As x goes to infinity, what value does 2 get x 11 close to? 8. As x goes to negative infinity, what value does
2x 2 1 4x x2 1 3
get close to?
3.3 Exercises
In Exercises 110, calculate the limit, if it exists. 1. lim 2. 3. 4. 5. 6.
xS1
2x 2 4 x22
3 21 xS 2` x 1 x 29 19. lim ( 4x 1 x4 ) 17. lim

xS`
18. lim
xS 2`
e0.25x ln Z x Z
e3x 1 1 lim x xS0 e 1 2 x2 2 4 lim x S 22 x 1 2 2x( 5x 2 3 ) lim xS2 3x ln ( x 2 8 ) lim xS9 x29 lim (( 2x 2 x2 )( 4x 2 5 ))
xS1
20. lim
xS`
9x21 1 4 3x22 2 4
In Exercises 2130, determine if the function is continuous. 21. g( t ) 5 22. f ( t ) 5 23. h( x ) 5 24. g( x ) 5 t12 t2 1 4 t23 t2 2 9 3x 2 9x2 x23 x2 1 5x 1 6 x2 2 9
2x 2 4 2 b xS5 x22 8. lim ( ln( x ) ? ( 22x )) 7. lim a

xS4
9. lim
10. lim ( 4x( 16 2 x2 )) 1>2

xS 24
xS0
A" x22 1 4 )( log( x 1 1 ) B T
25. g( t ) 5 ln( t 2 1 1 ) 26. y 5 ln Z x Z 27. y 5 e x2 2 1, x , 4 15ex24, x $ 4 |x|, x , 2 2 2, x . 2 2 6x 1 4, x , 0 22x, x$0
In Exercises 1120, determine the limit at 1` or 2 ` , as indicated in the exercise. x2 11. lim 22x 12. lim x xS` xS` 2 13. lim ( ex 1 200 )
xS 2`
28. f ( x ) 5 e 29. g( x ) 5 e 30. h( t ) 5 e
ln x 15. lim x S ` 3x 2 4
2x 21 xS 2` 1 1 x 3x2 2 4 16. lim 2 xS` 1 1 x 14. lim
ln( t 2 1 1 ) , x # 0 1 2 ln( t ) , x . 0
216
In Exercises 3135, use limits in answering the indicated questions. 31.

Apple Computer Sales Based on data
35.
from 2001 to 2005, the net international sales for Apple Computer Corporation may be modeled by 3217 I( d ) 5 1 2400 1 1 9446e20.001704d million dollars, where d is the dollar amount of net domestic sales (in millions). What is lim I( d ) dS` and what does it represent? 32. Based on data from 1997 to 2004, the average price of a DVD player for a given year may be modeled by P( d ) 5 1024 2 90.98 ln( d ) dollars, where d is the number of DVD players sold that year. What is lim P( d ) and what does it mean in a real-world dS` context?
DVD Player Price
Liver Disease and Cirrhosis Based on data from 1970 to 2003, the age-adjusted death rate due to chronic liver disease and cirrhosis may be modeled by
9.900 19 1 1 0.1181e0.1771t deaths per 100,000 people, where t is the number of years since 1970. (Source: Modeled from r( t ) 5
Statistical Abstract of the United States, 2006, Table 106)
According to the model, what will be the death rate in 2010? in 2020? in 2030? What do you forecast the death rate will be in the long run?
36. Describe what lim g( x ) 5 43 means graphically.

xS`
33.
37. Explain why lim
Based on data from 1995 to 2004, the number of drug prescriptions written annually in the United States may be modeled by 1194 P( t ) 5 1 2100 1 1 17.04e20.6872t
Drug Prescriptions
xS`
1 x
5 0.
t 2 2 2t t2 2 4
38. The function f ( t ) 5
is undefined at t 5 2 2
and t 5 2. Although lim f ( t ) exists, lim f ( t ) does t S 22 not exist. Explain. t S 2 39. Describe in practical terms what we mean when we say that a graph is continuous. 40. Give an example of a function that meets the first two criteria of continuity but fails the last criterion.
million prescriptions, where t is the number of years since 1995. According to this model, what is the maximum number of prescriptions that will ever be written? Use limits to support your conclusion. 34.
Insecticide-Treated Nets One of the principal strategies used in Africa to reduce the transmission of malaria to children is insecticide-treated nets. Based on World Health Organization data from 1999 to 2005, the number of nets distributed annually in the African region may be modeled by N( t ) 5 491,600( 2.081 ) t
41. A classmate claims that if a function g( x ) 5 h( x ) has the property that lim f ( x ) 5 ` and lim h( x ) 5 ` , then lim g( x ) 5 ` . Do you agree?
xSc xSc xSc
f (x)
Why or why not? 42. A classmate claims that if a function g( x ) 5 h( x ) has the property that lim f ( x ) 5 k, where k is a constant, and lim h( x ) 5 ` , then lim g( x ) 5 0.
xSc xSc xSc
f(x)
nets and the population of the African region may be modeled by A( t ) 5 738,083,274( 1.022 ) t 2 5 people, where t is the number of years since 1999.
(Source: Modeled from data at www.afro.who.int.)
Do you agree? Why or why not? 43. Give an example of a function with domain all real numbers that is discontinuous at x 5 3.
According to these models, what will happen to the average number of people per net distributed in the African region over the long run? Explain.
3.4 The Derivative as a Slope: Graphical Methods
217
Based on data from 1990 to 2000, the average price for a half gallon of prepackaged ice cream may be modeled by C( t ) 5 0.0142t 2 2 0.0272t 1 2.53 dollars where t is the number of years since 1990. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 706, p. 468.) To help project future revenue, a creamery wants to know what the average annual increase in the price of ice cream was from 1990 to 2000 and at what rate the price will be increasing at the end of 2000. We will answer these questions by calculating the average and instantaneous rates of change in the price of ice cream. In Sections 3.1 and 3.2, you learned how to calculate the average rate of change of a function over an interval and the instantaneous rate of change of a function at a point. In this section, we will revisit these concepts from a graphical standpoint. We will also demonstrate how to use tangent-line approximations to estimate the value of a function.
As shown in Section 3.1, the difference quotient formula,
the average rate of change in the value of the function between the points ( a, f ( a )) and ( a 1 h, f ( a 1 h )) . If we let a 5 c 2 h, then the difference quotient formula becomes in our exploration of the price of ice cream, since we will be approaching t 5 c through values to the left of t 5 c. Recall from Section 3.1 that, graphically speaking, the difference quotient is the slope of the secant line connecting the two points on the graph of a function. According to the ice cream price model, C( 0 ) 5 2.530 and C( 10 ) 5 3.678. Were interested in the slope of the secant line between ( 0, 2.530 ) and ( 10, 3.678 ) . The slope of this line represents the average rate of change in the price of a half gallon of ice cream between 1990 and 2000 (Figure 3.14).
f (c) 2 f (c 2 h) . We will use this modified form of the difference quotient h
f (a 1 h) 2 f (a) , h
gives
218
CHAPTER 3 The Derivative C (t ) 3.90 3.80 3.70 3.60 3.50 3.40 3.30 3.20 3.10 3.00 2.90 2.80 2.70 2.60 2.50 0
(10, 3.678)
Price (in dollars)
(0, 2.530)
C (t ) = 0.0142t 2 0.0272t + 2.53 1 2 3 4 5 6 7 8 Years (Since 1990) 9 10 11 t
FIGURE 3.14
The ice cream graph secant line between ( 0, 2.530 ) and ( 10, 3.678 ) has the slope m5 5 C( 10 ) 2 C( 0 ) dollars 10 years 3.678 2 2.530 dollars 10 years
5 0.115 dollar per year Between 1990 and 2000, the average price of a half gallon of ice cream increased by $0.115 (about 12 cents) per year. Does this mean that from 2000 to 2001, the price will increase by about 12 cents? No. Looking at the graph of the model, we notice that the price of ice cream is rising at an increasing rate as time progresses (the steeper the graph, the greater the magnitude of the rate of change). We can approximate the instantaneous rate of change at the end of 2000 (t 5 10) by calculating the slope of a secant line through ( 10, 3.678 ) and a nearby point, as shown in Figure 3.15. The ice cream graph secant line between ( 5, 2.749 ) and ( 10, 3.678 ) has the slope m5 5 C( 10 ) 2 C( 5 ) dollars 5 years 3.678 2 2.749 dollars 5 years
5 0.186 dollar per year
3.4 The Derivative as a Slope: Graphical Methods C (t ) 3.90 3.80 3.70 3.60 3.50 3.40 3.30 3.20 3.10 3.00 2.90 2.80 2.70 2.60 2.50 0 1 2 3
219
(10, 3.678)
Price (in dollars)
(5, C (5)) = (5, 2.749)
4 5 6 7 8 Years (Since 1990)
10 11
FIGURE 3.15
As shown in Figure 3.16, the ice cream graph secant line between ( 9, 3.435 ) and ( 10, 3.678 ) has the slope m5 5 C( 10 ) 2 C( 9 ) dollars 10 2 9 years 3.678 2 3.435 dollars 1 years
5 0.243 dollar per year

C (t ) 3.90 3.80 3.70 3.60 3.50 3.40 3.30 3.20 3.10 3.00 2.90 2.80 2.70 2.60 2.50 0 1 2 3
(10, 3.678) [9, C (9)] = (9, 3.435)
Price (in dollars)
4 5 6 7 8 Years (Since 1990)
10 11
FIGURE 3.16
By zooming in (Figure 3.17), we can see that between t 5 9 and t 5 10, the secant line and the graph of the function are extremely close together.
220
CHAPTER 3 The Derivative C (t ) 3.68 3.63 Price (in dollars) 3.58 3.53 3.48 3.43 9 Years (Since 1990) (9, 3.435) 10 t C (10, 3.678)
FIGURE 3.17
The ice cream graph secant line between (9.9, 3.652) and (10, 3.678) has the slope m5 5 C( 10 ) 2 C( 9.9 ) dollars 10 2 9.9 years 3.678 2 3.652 dollars 0.1 years
5 0.26 dollar per year Through (9.9, 3.652) and (10, 3.678), the slope of the secant line is $0.26 per year. Again zooming in, we see that the secant line and the graph of the function are nearly identical between t 5 9.9 and t 5 10 (Figure 3.18). In fact, we are unable to visually distinguish between the two.
C (t )
3.680 Price (in dollars)
(10, 3.678)
3.670
3.660
3.650
(9.9, 3.652) 10 t Years (Since 1990)
9.9
FIGURE 3.18
221
As t gets closer and closer to 10, the slope of the secant line will approach Cr( 10 ) . This occurs because Since h is the horizontal distance between the points, h becomes increasingly small as t nears 10. Using the methods covered in Section 3.2, we determine algebraically that Cr( 10 ) 5 0.2568. At the end of 2000, ice cream prices were increasing at a rate of roughly $0.26 per year. In other words, we anticipate that ice cream prices increased by approximately $0.26 between 2000 and 2001. As the preceding example demonstrates, to find the instantaneous rate of change of a function y 5 f ( x ) at a point P 5 ( a, f ( a )) , we can find the limit of the slope of the secant line through P and a nearby point Q as the point Q gets closer and closer to P. Graphically speaking, we select values Q1, Q2, Q3, . . . , with each consecutive value of Qi being a point on the curve that is closer to the point P than the one before it. The limit of the slope of the secant lines (imagine the points P and Q finally coinciding) is the line tangent to the curve at the point P 5 ( a, f ( a )) (Figure 3.19).
y f (x) C( 10 ) 2 C( 10 2 h ) h
< Cr( 10 ) for small values of h.
P Q4 Q3
Q2 Q1 Tangent line x
FIGURE 3.19
The tangent line to a graph f at a point ( a, f ( a )) is the line that passes through ( a, f ( a )) and has slope f r( a ) .
THE GRAPHICAL MEANING OF THE DERIVATIVE
The derivative of a function f at a point ( a, f ( a )) is the slope of the tangent line to the graph of f at that point. The slope of the tangent line at a point is also referred to as the slope of the curve at that point.
EXAMPLE
Finding the Equation of a Tangent Line

Find the equation of the tangent line to the graph of f ( x ) 5 x2 that passes through ( 2, 4 ) . Then graph the tangent line and the graph of f.
222
SOLUTION The slope of the tangent line is f r( 2 ) .
f r( 2 ) 5 lim
hS0
f (2 1 h) 2 f (2) h ( 2 1 h ) 2 2 22 h 4 1 4h 1 h2 2 4 h 4h 1 h2 h h( 4 1 h ) h
Since f ( x ) 5 x2
5 lim 5 lim 5 lim
hS0
hS0
hS0
5 lim
hS0
5 lim( 4 1 h )
hS0
5410 54 The slope of the tangent line is 4. Using the slope-intercept form of a line, we have y 5 4x 1 b . Substituting in the point ( 2, 4 ) , we get 4 5 4( 2 ) 1 b 4581b b 5 24 The equation of the tangent line is y 5 4x 2 4 . We generate a table of values for y 5 4x 2 4 and f ( x ) 5 x2 . Then we graph the results (Figure 3.20).
y 16 14 12 10 8 6 4 2 1 2 4 6 f (x) = x 2
x 21 0 1 2
x
y 28 24 0 4 8 12
f (x) 1 0 1 4 9 16
(2, 4)
3 4
y = 4x 4
FIGURE 3.20
EXAMPLE
Finding the Equation of a Tangent Line

Find the equation of the tangent line of the graph of f ( x ) 5 2 x2 1 2x 1 2 that passes through ( 1, 3 ) . Then plot the tangent line and the graph of f.
SOLUTION The slope of the tangent line is f r( 1 ) . Because of the complex nature of f, we will calculate f ( 1 1 h ) and f ( 1 ) individually, then substitute the simplified values into the derivative formula.
223
f ( 1 1 h ) 5 2 ( 1 1 h ) 2 1 2( 1 1 h ) 1 2 5 2 ( 1 1 2h 1 h2 ) 1 2 1 2h 1 2 5 2 1 2 2h 2 h 1 2h 1 4
2
f ( 1 ) 5 2 ( 1 ) 2 1 2( 1 ) 1 2 5 21 1 2 1 2 53
5 2h 1 3
2
f r( 1 ) 5 lim
hS0
f (1 1 h) 2 f (1) h ( 2 h2 1 3 ) 2 3 h 2 h2 h
5 lim
hS0
5 lim
hS0 hS0
5 lim( 2 h ) 50 The slope of the tangent line is 0. Using the slope-intercept form of a line, we have y 5 0x 1 b . Substituting in the point ( 1, 3 ) , we get 3 5 0( 1 ) 1 b b53 The equation of the tangent line is y 5 3 . We generate a table of values for y 5 3 and f ( x ) 5 2 x2 1 2x 1 2 . Then we graph both functions (Figure 3.21).
y (1, 3) 3 2 1 x y=3
x 21 0 1 2 3
y 3 3 3 3 3
f (x) 21 2 3 2 21
1 1
f (x) = x 2 + 2x + 2
FIGURE 3.21
y 3.01 (1, 3) 3.00
y=3
As shown in Example 2, a function y 5 f ( x ) will have a horizontal tangent line at ( a, f ( a )) if f r( a ) 5 0.
Tangent-Line Approximations
2.99 2.98 f (x) = x 2 + 2x + 2 0.9 1.0 1.1 x
FIGURE 3.22
In Examples 1 and 2, you may have noticed that the tangent line lies very near to the graph of f for values of x near a. In fact, if we zoom in to the region immediately surrounding ( a, f ( a )) , the graph of f and the tangent line to the graph of f at ( a, f ( a )) appear nearly identical. Zooming in on the graph in Example 2, we can see that for values of x near 1, the tangent-line y value is a good approximation of the actual function value (see Figure 3.22).
224
Since it is frequently easier to calculate the values of the tangent line than the values of the function, sometimes the tangent line is used to estimate the value of the function. For example, suppose we wanted to estimate f ( 0.9 ) given f ( x ) 5 2 x2 1 2x 1 2. Since x 5 0.9 is near x 5 1 , we may use the equation of the tangent line, y 5 3 , to estimate f ( 0.9 ) . That is, f ( 0.9 ) < 3. The actual value is f ( 0.9 ) 5 2 ( 0.9 ) 2 1 2( 0.9 ) 1 2 5 2 0.81 1 1.8 1 2 5 2.99 Our tangent-line estimate ( y 5 3 ) was remarkably close to the actual value of the function ( f ( 0.9 ) 5 2.99 ) and required no additional computations.
EXAMPLE 3
Using a Tangent-Line Approximation to Forecast DVD Player Demand

Based on data from 1998 to 2001, the number of DVD players shipped in North America may be modeled by P( t ) 5 995t2 1 271t 2 245 thousand players where t is the number of years since the end of 1997. (Model based on data from DVD Entertainment Group.) According to the model, 16,759 thousand DVD players were shipped in 2001. Determine how quickly DVD player shipments were increasing at the end of 2001. Then use a tangent line to estimate how many DVD players were shipped in 2002 and compare the estimate to the actual value predicted by the model.
SOLUTION Since t 5 4 corresponds with the year 2001, the instantaneous rate of change in DVD player shipments at the end of 2001 is given by
Pr( 4 ) 5 lim
hS0
P( 4 1 h ) 2 P( 4 ) thousand players h years
We already know that P( 4 ) 5 16,759 . Well calculate P( 4 1 h ) and then substitute the simplified value into the derivative formula. P( 4 1 h ) 5 995( 4 1 h ) 2 1 271( 4 1 h ) 2 245 5 995( 16 1 8h 1 h2 ) 1 1084 1 271h 2 245 5 15,920 1 7960h 1 995h2 1 271h 1 839 5 995h2 1 8231h 1 16,759 Pr( 4 ) 5 lim P( 4 1 h ) 2 P( 4 ) h ( 995h2 1 8231h 1 16,759 ) 2 16,759 h 995h2 1 8231h h h( 995h 1 8231 ) h
hS0
5 lim 5 lim
hS0
hS0
5 lim
hS0
225
5 lim ( 995h 1 8231 )

hS0
5 995( 0 ) 1 8231 5 8231 thousand players year
At the end of 2001, DVD player shipments were increasing at a rate of 8231 thousand (8.231 million) players per year. That is, between 2001 and 2002, the number of DVD players shipped annually was expected to increase by approximately 8.231 million. The slope-intercept form of the tangent line is y 5 8231t 1 b . Using the point ( 4, 16,759 ) , we determine 16,759 5 8231( 4 ) 1 b 16,759 5 32,924 1 b b 5 2 16,165 The tangent-line equation is y 5 8231t 2 16,165 . At t 5 5, we have y 5 8231( 5 ) 2 16,165 5 41,155 2 16,165 5 24,990 Using the tangent-line equation, we estimate that 24.990 million DVD players were shipped in 2002. According to the model, the actual number of DVD players shipped was slightly greater. P( 5 ) 5 995( 5 ) 2 1 271( 5 ) 2 245 5 25,985 The model indicates that 25.985 million DVD players were shipped in 2002.
Why was there a discrepancy between the two estimates in Example 3? Look at the graph of the model and the tangent line (Figure 3.23).
26,000 25,000 DVD Players (in thousands) 24,000 23,000 22,000 21,000 20,000 19,000 18,000 17,000 16,000 4 Years (since 1997) 5 y = 8231t 16,165 P (t ) = 995t 2 + 271t 245
FIGURE 3.23
226
Although both functions were equal at t 5 4, by the time t reached 5, the value of the model exceeded the tangent-line estimate by about 1,000 thousand DVD players. Although tangent-line estimates of a functions value are not exact, they are often good enough for their intended purpose.
Numerical Derivatives
TABLE 3.10
x 0 1 2 3 4
f (x) 0 1 4 9 16
Often we encounter real-life data in tables or charts. Is it possible to calculate a derivative from a table of data? Well investigate this question by looking at a table of data for f ( x ) 5 x2 (see Table 3.10). For this function, f r( 2 ) 5 4 . We can estimate f r( 2 ) by calculating the slope of the secant line through points whose x values are equidistant from x 5 2 . That is, f r( 2 ) < 5 5 f (3) 2 f (1) 321 921 2 8 2
54
y 16 14 12 10 8 6 4 2 1 0 1 (2, 4) (1, 1) 2 3 4 5 6 x Function Secant Tangent (3, 9) m=4 f ' (2) = 4 f (x) = x 2
In this case, our estimate was equal to the tangent-line slope. Lets look at the situation graphically (see Figure 3.24). When estimating a derivative numerically, we typically select the two closest data points that are horizontally equidistant from our point of interest. Doing so often yields a line that is parallel to the tangent line. Picking two points that are equidistant from the point of interest will tend to give the best estimate of the derivative and can be used as long as the point of interest is not an endpoint. (If the point of interest is an endpoint, we find the slope of the secant line between the endpoint and the next closest point.) If we assume that each output in a table of data represents f ( a ) for a corresponding input a, then we can symbolically represent the process of numerically estimating a derivative as follows.
FIGURE 3.24
NUMERICAL ESTIMATE OF THE DERIVATIVE
The derivative of a function f at a point ( a, f ( a )) may be approximated from a table by f r( a ) < < f (a 1 h) 2 f (a 2 h) (a 1 h) 2 (a 2 h) f (a 1 h) 2 f (a 2 h) 2h
where h is the horizontal distance between a and a 1 h.
If a is the largest domain value in the data set, then f r( a ) < < f (a) 2 f (a 2 h) (a) 2 (a 2 h) f (a) 2 f (a 2 h) h
227
If a is the smallest domain value in the data set, then f r( a ) < < f (a 1 h) 2 f (a) (a 1 h) 2 (a) f (a 1 h) 2 f (a) h
In both cases, h is the distance between a and the next closest domain value.
EXAMPLE
Estimating the Derivative from a Table of Data

Use Table 3.11 to estimate how quickly home sales prices in the southern United States were increasing at the end of 1990. [That is, estimate the slope of the tangent line at ( 10, 99.0 ) .]
TABLE 3.11 Median Sales Price of a New One-Family House in the Southern United States
Years Since 1980 (t) 0 5 10 15 20
Source: Statistical Abstract of the United States, 2001, Table 940, p. 598. SOLUTION Well estimate the slope of the tangent line at ( 10, 99.0 ) by calculating the slope of the secant line between ( 5, 75.0 ) and ( 15, 124.5 ) .
f r( 10 ) < <
f ( 15 ) 2 f ( 5 ) thousand dollars 15 2 5 years 124.5 2 75.0 10
< 4.95 thousand dollars per year We estimate that home prices in the southern United States were increasing at a rate of 5.0 thousand dollars per year at the end of 1990. (Since the original data were accurate to one decimal place, we rounded our result to one decimal place.)
228
EXAMPLE
Estimating the Derivative from a Table of Data

Given the data in Table 3.12, how quickly was the percentage of households leasing vehicles increasing in 1995 and 1998?
TABLE 3.12 Percentage of Households Leasing Vehicles
Years Since 1989 (t) 0 3 6 9
Percent (P) 2.5 2.9 4.5 6.4
Source: Statistical Abstract of the United States, 2001, Table 1086, p. 680. SOLUTION We want to find f r( 6 ) and f r( 9 ) .
f r( 6 ) < <
f ( 9 ) 2 f ( 3 ) percentage points 923 years 6.4 2 2.9 6
< 0.583 percentage point per year In 1995, the percentage of households with leased vehicles was increasing by about 0.6 percentage point per year. Since an endpoint occurs at t 5 9, we will estimate f r( 9 ) by calculating the slope of the secant line between the endpoint and the next closest point. f r( 9 ) < < f ( 9 ) 2 f ( 6 ) percentage points 926 years 6.4 2 4.5 3
< 0.633 percentage point per year In 1998, the percentage of households with leased vehicles was also increasing by about 0.6 percentage point per year.
3.4 Summary
In this section, you learned that the derivative at a point is the slope of the tangent line to the graph at that point. You discovered that the tangent line may be used to approximate the value of a function. You also learned how to estimate a derivative from a table of data.
229

1. How do you calculate the slope of a secant line? 2. What formula do you use to calculate the slope of a tangent line? 3. On what interval(s) does the graph appear to have a positive slope?
y 40 3 2 1 80 120 160 200 240 1 2 3 x
4. At what points does the graph shown in Exercise 3 have a zero slope? 5. What is the relationship between the instantaneous rate of change of a function at a point and the slope of the graph of that function at the same point? 6. In order to find the equation of a tangent line at a given point, what piece of information do you need? 7. What is the slope of a parabola at its vertex? 8. What is the relationship between the slope of a linear function and its instantaneous rate of change?
3.4 Exercises
In Exercises 110, determine the equation of the tangent line of the function at the given point. Then graph the tangent line and the function together. 1. f ( x ) 5 x 2 4x; ( 1, 2 3 )
2
motor fuel consumed annually by pickups, SUVs, and vans may be modeled by F( t ) 5 0.0407t 2 1 0.8091t 1 23.33 billion gallons, where t is the number of years since 1980. (Source: Modeled from Statistical
2. f ( x ) 5 2 x2 1 6; ( 2, 2 ) 3. g( x ) 5 x2 1 2x 1 1; ( 0, 1 ) 4. g( x ) 5 x 2 4; ( 3, 5 )
2
5. g( x ) 5 x2 2 4x 2 5; ( 4, 2 5 ) 6. h( x ) 5 x3; ( 2, 8 ) 7. h( x ) 5 x ; ( 0, 0 )
3
How quickly was fuel consumption increasing in 1999? What was the estimated fuel consumption in 2000? (Use a tangent-line approximation.) 12.
High School Principal Salaries Based
8. h( x ) 5 x3; ( 1, 1 ) 9. f ( x ) 5 ( x 2 3 ) 2; ( 3, 0 ) 10. f ( x ) 5 ( x 1 2 ) ; ( 2 1, 1 )
2
on data from 1980 to 2000, the average salary of a high school principal may be modeled by S( t ) 5 2499t 1 30,039
(Source: Modeled from Statistical Abstract of the United States, 2001, Table 238, p. 152.)
In Exercises 1120, answer the questions by calculating the slope of the tangent line and the tangent-line equation, as appropriate. 11.
Motor Fuel Consumption Based on data from 1980 to 1999, the amount of
How quickly were high school principals salaries increasing in 2000? What was the average high school principals salary in 2001? (Calculate the tangent-line approximation for the 2001 salary
230
and compare it to the model calculation for the 2001 salary.) 13.
New One-Family Home Size Based on
data from 1970 to 1999, the average number of square feet in new one-family homes may be modeled by H( t ) 5 2 0.1669t2 1 29.95t 1 1490 square feet, where t is the number of years after 1970. (Source: Modeled from Statistical Abstract of
the United States, 2001, Table 938, p. 597.)
According to the model, at what rate is the value of the cassette tapes shipped changing when the number of cassette tapes shipped is 250 million? What was the estimated shipment value when 251 million cassette tapes were shipped? (Use a tangent-line approximation.) 18.
Decreasing Popularity of the Cassette Tape Since the advent of the compact
At what rate was home size increasing in 1999? What was the estimated size of a new home in 2000? (Use a tangent-line approximation.) 14.
New One-Family Home Size How
could a real estate developer or home builder use the results of Exercise 13? 15.
Student-to-Teacher Ratio Based on
disk and mp3 player, cassette tape sales have dropped off dramatically. In fact, in 2004 there were 5.2 million cassettes shipped compared to 442.2 million in 1990. (Source: Statistical Abstract of the United States, 2006, Table 1131.) Use the model in Exercise 17 to calculate the rate of change in cassette tape shipment value in 1990 and in 2004. 19.
College Attendance Based on data from 2000 to 2002 and Census Bureau projections for 2003 to 2013, private college enrollment may be modeled by
data from 1960 to 1998, the student-toteacher ratio at private elementary and secondary schools may be modeled by R( t ) 5 0.01484t 2 1.000t 1 31.60
2
students per teacher, where t is the number of years since 1960. (Source: Modeled from Statistical
P( x ) 5 0.340x 2 457 thousand students where x is the number of students (in thousands) enrolled in public colleges. (Source: Modeled from
Statistical Abstract of the United States, 2006, Table 204.) Calculate the rate of change in private
According to the model, how quickly was the student-to-teacher ratio changing in 1998? What was the estimated student-to-teacher ratio for 1999? (Use a tangent-line approximation.) 16.
Student-to-Teacher Ratio Private
college enrollment when there are 12,752 thousand public college students enrolled. Use a tangent-line approximation to estimate the number of students enrolled in private colleges when there are 12,753 thousand public college students. 20.
Radio Broadcasting Wages Based on
schools succeed financially by convincing families that children taught at their school will obtain a better educational experience than that which is available at other public and private schools in the surrounding area. How could a private elementary school use the model and results of Exercise 15 in its marketing plan? 17.
Cassette Tape Shipment Value Based
data from 1992 to 1998, the average wage of a worker in the radio broadcasting industry may be modeled by W( t ) 5 164.3t 2 1 778.6t 1 23,500 dollars, where t is the number of years since 1992.
on data from 1990 to 2004, the value of music cassette tapes shipped may be modeled by V( s ) 5 2 0.00393s2 1 9.74s 2 63.0 million dollars where s represents the number of music cassette tapes shipped (in millions). (Source: Modeled from
Statistical Abstract of the United States, 2006, Table 1131.)
How quickly were wages of radio broadcasting employees increasing at the end of 1998? What was the average salary in 1999? (Use a tangent-line approximation.)
231
In Exercises 2130, estimate the specified derivative by using the data in the table. Then interpret the result. 21.
Television Broadcasting Wages
25.
Tuition at Green River Community College
Full-Time Resident Tuition
Average Annual Wage per Worker: Television Broadcasting
Years Since 1994 (t) 0 1 2 3 4 5 6 7
Quarterly Tuition (dollars) [E( t ) ] 432 450 467 486 505 528 547 581
Year Since 1992 (t) 0 1 2 3 4 5 6
Annual Wage (dollars) [W( t ) ] 41,400 42,200 43,700 47,200 51,100 51,000 54,600
Estimate Er( 6 ) and Er( 7 ) . What can you conclude about the rate of increase in tuition at Green River Community College? 26.
Bread Prices
Estimate W r( 5 ) . 22. 23. Using the table in Exercise 21, estimate W r( 6 ) .

Daily Newspapers
Whole Wheat Bread Cost
Years Since 1990 (t) 3 4 5 6 7 8 9 10
Cost (dollars per loaf) [C( t ) ] 1.12 1.12 1.15 1.30 1.30 1.32 1.36 1.36
Different Daily Newspapers
Years Since 1970 (t) 0 5 10 15 20 25 30
Newspapers [N( t ) ] 1,748 1,756 1,745 1,676 1,611 1,533 1,480
Estimate C r( 4 ) and C r( 5 ) .
Estimate N r( 25 ) . 24. Use the table in Exercise 23 to estimate N r( 30 ) .
232 27.
Baseball Game Attendance
30.
Movie Prices
Major League Baseball Attendance
Movie Ticket Prices
Years Since 1994 (t) 0 1 2 3 4 5
American League Games (millions of people) [A( t ) ] 24.2 25.4 29.7 31.3 31.9 31.8
Years Since 1975 (t) 0 5 10 15 20 24
Price (dollars) [P( t ) ] 2.05 2.69 3.55 4.23 4.35 5.08
Estimate Ar( 3 ) . 28.

Baseball Game Attendance
Estimate Pr( 24 ) .
Major League Baseball Attendance
Years Since 1994 (t) 0 1 2 3 4 5
National League Games (millions of people) [N( t ) ] 25.8 25.1 30.4 31.9 38.4 38.3
31. What does the difference quotient represent graphically?
f (a 1 h) 2 f (a) h
32. What is the difference between a secant line and a tangent line? Explain. 33. Explain why the derivative of a function f at a point A may often be estimated accurately by calculating the average rate of change between two points on the graph of f horizontally equidistant from A. 34. A classmate claims that a tangent line touches a graph in exactly one place, whereas a secant line touches the graph in exactly two places. Is your classmates statement correct? Explain why or why not. 35. Explain why a tangent line of a function at a point ( a, f ( a )) may be used to approximate the value of the function f at values of x near a.
Estimate Nr( 3 ) . 29.

Major League Baseball Using the tables from Exercises 27 and 28, estimate the overall rate of change in major league baseball game attendance in 1997. (Hint: First make a table with the combined attendance of both leagues.)
3.5 The Derivative as a Function: Algebraic Method
233
36. Given f ( x ) 5 x2 , find two pairs of points ( a, f ( a )) and ( b, f ( b )) such that f r( 2 ) 5

f (b) 2 f (a) . b2a
38. Given f ( x ) 5 x3, calculate f r( 2 1 ) and f r( 1 ) . What can you conclude about the graph of f at ( 2 1, 2 1 ) and ( 1, 1 ) ? 39. Given f r( 3 ) 5 4 and g( x ) 5 x2 2 f ( x ) , find g r( 3 ) . 40. Given p( x ) 5 f ( x ) 1 g ( x ) , what is the relationship between pr( 0 ) , f r( 0 ) , and g r( 0 ) ?
What do both soltuions have
in common? 37. Given f ( x ) 5 x2 2 2x , find two pairs of points ( a, f ( a )) and ( b, f ( b )) such that f r( 2 ) 5
f (b) 2 f (a) . b2a
What do both solutions have
in common?
Use the limit definition of the derivative to find the derivative of a function Use derivative notation and terminology to describe instantaneous rates of change
Based on data from 1990 to 2000, the cumulative number of homicides resulting from a romantic triangle may be modeled by R(t) 5 220.45t 2 1 471.8t 152.5 homicides between the start of 1991 and the end of year t, where t is the number of years since 1990. (Source: Modeled from Crime in the United States 2000, Uniform Crime Report, FBI.) How quickly was the cumulative number of homicides increasing at the end of 1997, 1998, and 1999? Although we could calculate the derivative at t 5 7, t 5 8, and t 5 9, we can save time by finding the derivative function and then substituting in the various values of t. In this section, we will introduce the derivative function and show how to find it algebraically. Your skill in finding the derivative at a point will prove to be especially useful in this section. Well begin with a couple of simple examples before returning to the romantic triangle problem.
EXAMPLE
Calculating the Derivative of a Function at Multiple Points

Determine the instantaneous rate of change of f ( x ) 5 3x2 at ( 1, 3 ) , ( 3, 27 ) , and ( 10, 300 ) .
SOLUTION Although we could calculate f r( 1 ) , f r( 3 ) , and f r( 10 ) individually, it will be more efficient to find the derivative function itself and then substitute in the different values of x.
234
We begin with the derivative formula; however, instead of substituting a specific value for a, we replace a with the variable x . f r( x ) 5 lim
hS0
5 lim
hS0
33( x 1 h ) 24 2 ( 3x2 ) h
f (x 1 h) 2 f (x) h
Since f ( x ) 5 3x2
5 lim 5 lim 5 lim
hS0
33( x2 1 2hx 1 h2 ) 4 2 ( 3x2 ) h ( 3x2 1 6hx 1 3h2 ) 2 ( 3x2 ) h 6hx 1 3h2 h h( 6x 1 3h ) h
hS0
hS0
5 lim
hS0 hS0
5 lim( 6x 1 3h ) 5 6x 1 3( 0 ) 5 6x The result f r( x ) 5 6x is the derivative function for f ( x ) 5 3x2 . It can be used to calculate the instantaneous rate of change of f at any point ( a, f ( a )) . f r( 1 ) 5 6( 1 ) 56 The instantaneous rate of change of f at ( 1, 3 ) is 6. f r( 3 ) 5 6( 3 ) 5 18 The instantaneous rate of change of f at ( 3, 27 ) is 18. f r( 10 ) 5 6( 10 ) 5 60 The instantaneous rate of change of f at ( 10, 300 ) is 60.
As demonstrated in Example 1, the techniques used to find the derivative function are virtually identical to the procedures used to find the derivative at a point. However, knowing the derivative function allows us to calculate the derivative at a number of different points more quickly than calculating the derivative at each point separately. The derivative function for a function f is typically called the derivative of f.
235
THE DERIVATIVE FUNCTION
The derivative of a function f is given by f r( x ) 5 lim if the limit exists.

hS0
f (x 1 h) 2 f (x) h
EXAMPLE
Finding the Derivative of a Function

Find the derivative of f ( x ) 5 4x3 2 2x 1 1 . Then calculate the slope of the tangent line at x 5 0, x 5 1, and x 5 2.
SOLUTION
f r( x ) 5 lim
hS0
5 lim 5 lim 5 lim 5 lim
hS0
34( x 1 h ) 3 2 2( x 1 h ) 1 14 2 ( 4x3 2 2x 1 1 ) h
f (x 1 h) 2 f (x) h
Since f (x) 5 4x3 2 2x 1 1
hS0
34( x3 1 3hx2 1 3h2x 1 h3 ) 2 2x 2 2h 1 14 2 ( 4x3 2 2x 1 1 ) h
hS0
( 4x3 1 12hx2 1 12h2x 1 4h3 2 2x 2 2h 1 1 ) 2 ( 4x3 2 2x 1 1 ) h ( 12hx2 1 12h2x 1 4h3 2 2h ) h h( 12x2 1 12hx 1 4h2 2 2 ) h
hS0
5 lim
hS0 hS0
5 lim( 12x2 1 12hx 1 4h2 2 2 ) 5 12x2 1 12( 0 ) x 1 4( 0 ) 2 2 2 5 12x 2 2

2
Replace h with 0
The derivative of f ( x ) 5 4x3 2 2x 1 1 is f r( x ) 5 12x2 2 2 . Graphically speaking, the derivative is the slope of the tangent line. Therefore, f r( 0 ) 5 12( 0 ) 2 2 2 5 22 means that at x 5 0 the slope of the tangent line is 2 2. Similarly, f r( 1 ) 5 12( 1 ) 2 2 2 5 10 means that at x 5 1 the slope of the tangent line is 10. Likewise, f r( 2 ) 5 12( 2 ) 2 2 2 5 46 means that at x 5 2 the slope of the tangent line is 46.
236
Just In Time
Pascals Triangle
Pascals Triangle makes the expansion of a binomial raised to a power relatively easy. The expanded form of ( x 1 h ) n is an x n 1 an 2 1 x n 2 1h 1 an 2 2 x n 2 2h2 1 c 1 a2 x 2hn 2 2 1 a1xhn 2 1 1 a0 hn, where the ai are constants from Pascals Triangle. n n n n n n n 50 51 52 53 54 55 56 1 1 1 1 1 1 1 6 5 15 4 10 20 3 6 10 15 2 3 4 5 6 1 1 1 1 1 1
The triangle has 1s along the outside. Each term within the table is found by adding the two terms immediately above it. Using the triangle, we see that ( x 1 h ) 3 5 1x3 1 3x2h 1 3xh2 1 1h3
5 5 4 3 2
and
2 3
( x 1 h ) 5 1x 1 5x h 1 10x h 1 10x h 1 5xh4 1 1h5
EXAMPLE
Calculating the Instantaneous Rate of Change of a Function at Multiple Points

Based on data from 1990 to 2000, the cumulative number of homicides resulting from a romantic triangle may be modeled by R( t ) 5 2 20.45t 2 1 471.8t 2 152.5 homicides between the start of 1991 and the end of year t, where t is the number of years since 1990. (Source: Modeled from Crime in the United States 2000, Uniform Crime
Report, FBI.)
How quickly was the cumulative number of homicides increasing at the end of 1997, 1998, and 1999?
SOLUTION We will begin by finding the derivative Rr( t ) .
Rr( t ) 5 lim
hS0
R( t 1 h ) 2 R( t ) h
Because of the complex nature of R( t ) , we will first calculate R( t 1 h ) and then substitute the result into the derivative formula. R( t 1 h ) 5 2 20.45( t 1 h ) 2 1 471.8( t 1 h ) 2 152.5 5 2 20.45( t2 1 2ht 1 h2 ) 1 471.8t 1 471.8h 2 152.5 5 2 20.45t2 2 40.90ht 2 20.45h2 1 471.8t 1 471.8h 2 152.5
237
We already know that R( t ) 5 2 20.45t 2 1 471.8t 2 152.5 . Substituting both of these quantities into the derivative formula yields Rr( t ) 5 lim
hS0
R (t 1 h) 2 R (t) h ( 2 20.45t 2 2 40.90ht 2 20.45h2 1 471.8t 1 471.8h 2 152.5 ) 2 ( 2 20.45t 2 1 471.8t 2 152.5 ) h 2 40.90ht 2 20.45h2 1 471.8h h h( 2 40.90t 2 20.45h 1 471.8 ) h
Every term in the numerator without an h canceled out Since h 5 1 for h Z 0
h
5 lim 5 lim
hS0
hS0
5 lim
hS0 hS0
5 lim( 2 40.90t 2 20.45h 1 471.8 ) 5 2 40.90t 2 20.45( 0 ) 1 471.8 5 2 40.90t 1 471.8
So Rr( t ) 5 2 40.90t 1 471.8 homicides per year. We can now compute the instantaneous rate of change in the cumulative number of homicides in 1997, 1998, and 1999. Rr( 7 ) 5 2 40.90( 7 ) 1 471.8 5 185.5 < 186 homicides per year Rr( 8 ) 5 2 40.90( 8 ) 1 471.8 5 144.6 < 145 homicides per year Rr( 9 ) 5 2 40.90( 9 ) 1 471.8 5 103.7 < 104 homicides per year The cumulative number of homicides resulting from a romantic triangle was increasing at a rate of 186 homicides per year in 1997, 145 homicides per year in 1998, and 104 homicides per year in 1999. According to the model, although the cumulative number of homicides continued to increase, the rate at which these homicides were increasing slowed between the end of 1997 and the end of 1999.
EXAMPLE
Finding the Derivative of a Function

Find the derivative of g ( t ) 5 2t 3 2 4t 1 3 .
SOLUTION We must find g r( t ) 5 lim
hS0
g (t 1 h) 2 g (t) . Well first find g( t h
1 h)
and then substitute the result into the derivative formula. g ( t 1 h ) 5 2( t 1 h ) 3 2 4( t 1 h ) 1 3 5 2( t 3 1 3t 2h 1 3th2 1 h3 ) 2 4t 2 4h 1 3 5 2t 3 1 6t 2h 1 6th2 1 2h3 2 4t 2 4h 1 3
238
gr( t ) 5 lim
hS0
g (t 1 h) 2 g (t) h ( 2t 3 1 6t 2h 1 6th2 1 2h3 2 4t 2 4h 1 3 ) 2 ( 2t 3 2 4t 1 3 ) h ( 6t 2h 1 6th2 1 2h3 2 4h ) h h( 6t 2 1 6th 1 2h2 2 4 ) h

Every term in the numerator without an h canceled out
5 lim
hS0
5 lim 5 lim
hS0
hS0 hS0
5 lim( 6t 2 1 6th 1 2h2 2 4 ) 5 6t 2 1 6t( 0 ) 1 2( 0 ) 2 2 4 5 6t 2 2 4 The derivative of g ( t ) 5 2t 3 2 4t 1 3 is gr( t ) 5 6t 2 2 4 .
Estimating Derivatives
For polynomial functions, all terms in the numerator of the derivative formula without an h will cancel out. This allows us always to eliminate the h in the denominator. However, with some other types of functions, the h in the denominator cannot be eliminated algebraically. In this case, we can estimate the derivative function by substituting in a small positive value (e.g., 0.001) for h. The closer the value of h is to zero, the more accurate the estimate of the derivative will be.
EXAMPLE
Calculating the Instantaneous Rate of Change of a Function at Multiple Points

The per capita consumption of bottled water in the United States may be modeled by W 5 2.593( 1.106 ) t gallons where t is the number of years since the end of 1980. (Source: Modeled from Statistical
Determine how quickly bottled water consumption was increasing at the end of 2002, 2004, and 2006.
SOLUTION We must evaluate W r( t ) at t 5 22, 24, and 26.
Wr( t ) 5 lim
hS0
W( t 1 h ) 2 W( t ) h
5 lim 5 lim 5 lim
hS0
32.593( 1.106 ) t 1 h4 2 32.593( 1.106 ) t 4 h
hS0
hS0
2.593( 1.106 ) t 3( 1.106 ) h 2 14 h

hS0
32.593( 1.106 ) t ( 1.106 ) h4 2 32.593( 1.106 ) t 4 h
Since ( 1.106 ) t 1 h 5 ( 1.106 ) t( 1.106 ) h Factor out 2.593( 1.106 ) t
5 2.593( 1.106 ) t ? lim
3( 1.106 ) h 2 14 h
239
We can move the expression 2.593( 1.106 ) t to the other side of the limit because it does not contain an h. Since
hS0
lim
3( 1.106 ) h 2 14 3( 1.106 ) 0.001 2 14 < h 0.001 < 0.1008
we have W r( t ) 5 2.593( 1.106 ) t ? lim

t
hS0
3( 1.106 ) h 2 14 h
< 2.593( 1.106 ) ? ( 0.1008 ) < 0.2614( 1.106 ) t We will now evaluate the derivative function at t 5 22, 24, and 26. W r( 22 ) < 0.2614( 1.106 ) 22 < 2.398 According to the model, bottled water consumption was increasing by 2.398 gallons per year at the end of 2002. W r( 24 ) < 0.2614( 1.106 ) 24 < 2.934 Bottled water consumption was increasing by 2.934 gallons per year at the end of 2004. W r( 26 ) < 0.2614( 1.106 ) 26 < 3.589 Bottled water consumption is expected to be increasing at a rate of 3.589 gallons per year at the end of 2006.
3.5 Summary
In this section, you learned how to find the derivative function algebraically. You discovered that it is often easier to expand and simplify f ( x 1 h ) before substituting it into the derivative formula. Additionally, you found that when the h in the denominator cannot be eliminated algebraically, you can estimate the derivative numerically.

1. Given g ( x ) 5 3x2 1 2x 1 1, calculate g ( x 1 h ) . 2. Given f ( x ) 5 5x 2 10, calculate f ( x 1 h ) . 3. Given g ( x ) 5 x3, calculate g ( x 1 h ) 2 g ( x ) . 4. Given f ( x ) 5 x2 2 6x, calculate f ( x 1 h ) . 5. Given g ( x ) 5 2 3x2 1 7x 1 2, calculate g (x 1 h) . 6. Given f ( x ) 5 x , calculate f ( x 1 h ) 2 f ( x ) . 7. Given g ( x ) 5 2x, calculate g ( x 1 h ) . 8. Given f ( x ) 5 3( 4x ) , calculate f ( x 1 h ) .
1
240
3.5 Exercises
In Exercises 110, find the derivative of the function. 1. f ( x ) 5 x2 2 4x 3. g ( x ) 5 x2 1 2x 1 1 5. g ( x ) 5 x2 2 4x 2 5 7. j( x ) 5 x3 1 2 9. f ( t ) 5 ( t 2 3 )
2
2. f ( x ) 5 2 x2 1 6 4. g ( x ) 5 x2 2 4 6. w( x ) 5 x3 8. j( x ) 5 x3 1 x 10. f ( t ) 5 ( t 1 2 )
2
thousand dollars, where t is the number of years since 1980. (Source: Modeled from Statistical
According to the model, was the median sales price increasing more quickly at the end of 1990 or the end of 2000? 28.
New Home Prices Based on data from 1980 to 2000, the median sales price of a new one-family home in the northeastern United States may be modeled by P( t ) 5 2 0.01242t 2 1 8.279t 1 68.27 thousand dollars, where t is the number of years since 1980. (Source: Modeled from Statistical
In Exercises 1120, find the slope of the tangent line of the function at x 5 1 , x 5 3 , and x 5 5 . 11. g ( x ) 5 2x2 1 x 2 1 13. f ( x ) 5 x2 2 2x 15. j( x ) 5 2 5 17. W( x ) 5 2 4x 1 9 12. g ( x ) 5 3x2 2 4x 2 5 14. f ( x ) 5 4x 2 5 16. j( x ) 5 2 2x 1 4 18. W( x ) 5 x3 2 1
According to the model, was the median sales price increasing more quickly at the end of 1990 or the end of 2000? 29. New Home Prices Using the models from Exercises 27 and 28, determine whether median new home prices were increasing faster in the South or the Northeast in 1999. 30.
Restaurant Sales Based on data from
19. S( x ) 5 3x2 2 2x 1 1 20. S( x ) 5 2 4x2 1 9 In Exercises 2125, estimate the derivative of the function. When you are unable to eliminate the h in the denominator of the derivative formula algebraically, use h 5 0.001. 21. P( x ) 5 3x 23. C( x ) 5 2 3 ? 4x 25. R( x ) 5 5.042 ? ( 0.98 ) x In Exercises 2635, use the derivative function to answer the questions. 26. Based on data from 1970 to 1999, the average size of a new one-family home may be modeled by H( t ) 5 2 0.1669t 2 1 29.95t 1 1490
New Home Size
22. R( x ) 5 2 ? 3x 24. P( x ) 5 2.03x
1980 to 2000, full-service restaurant sales may be modeled by S( t ) 5 12.77t2 2 276.9t 1 5222t 1 38,203 million dollars, where t is the number of years since 1980. (Source: Modeled from Statistical
Were restaurant sales growing faster at the end of 1984 or the end of 1994? Explain. 31.
Gaming Software Sales Based on data
square feet, where t is the number of years since 1970. (Source: Model based on Statistical Abstract of
from 1990 to 1999, electronic gaming software factory sales may be modeled by S( t ) 5 39.79t 2 2 48.41t 1 2378 million dollars, where t is the number of years since the end of 1990. (Source: Modeled from
Statistical Abstract of the United States, 2001, Table 1005, p. 634.)
Was the average size of a new home growing faster at the end of 1989 or the end of 1999? 27. Based on data from 1980 to 2000, the median sales price of a new one-family home in the southern United States may be modeled by P( t ) 5 0.0506t 2 1 3.515t 1 58.49
New Home Prices
Were software sales growing faster at the end of 1997 or the end of 1999? 32.
Certified Organic Cropland Based on data from 1992 to 1996, the amount of certified organic cropland may be modeled by A( t ) 5 20.57t 2 2 24.51t 1 410.3
241
thousand acres, where t is the number of years since the end of 1992. (Source: Modeled from
How much more rapidly were enrollments increasing in 1998 than they were increasing in 1995?
In addition to showing that the amount of organic cropland increased between 1992 and 1996, how could you use the model to convince organic farming investors that there is still room for growth in the organic farming market? 33.
Prescription Drug Spending Based on
data from 1990 to 2003, per capita prescription drug spending may be modeled by P( t ) 5 2.889t 2 2 2.613t 1 158.7 dollars
(Source: Statistical Abstract of the United States, 2006, Table 121.)
36. The derivative of a function at a point is the limit of the slopes of the secant lines of the function that contain the point. For the absolute value function, the limit of the slopes of the secant lines containing the point (0, 0) is given by
hS0
where t is the number of years since 1990.
lim
Z0 1 h Z 2 Z0 Z . h
What are the slopes of the secant lines for the absolute value function near x 5 0? 37. Explain why the derivative does not exist at a sharp point in the graph of a function. (Hint: You may find it helpful to refer to Exercise 1.) 38. In the limit definition of the derivative, f r( x ) 5 lim h represent and what does h S 0 mean? 39. You may have previously learned how to calculate derivatives using a collection of derivative shortcuts referred to as derivative rules. In light of these shortcuts, why is it still important to understand and be able to use the limit definition of the derivative? 40. Sometimes it is algebraically impossible to eliminate the h in the denominator of the limit definition of the derivative. In such a case, describe a strategy that could be used to come up with a fairly accurate estimate of the derivative.
hS0
In what year was prescription drug spending increasing twice as fast as it was increasing in 2000? 34.
Federal and State Prison Rate Based
on data from 1980 to 1998, the federal and state prison rate may be modeled by R( t ) 5 0.3750t 2 1 11.70t 1 138.7 prisoners per 100,000 people, where t is the number of years since the end of 1980. (Source:
Modeled from Statistical Abstract of the United States, 2001, Table 332, p. 200.)
f( x 1 h ) 2 f( x ) h
, what does the variable
State lawmakers are debating whether or not the state should invest more money in a crime prevention program. An opponent of the measure argues that between 1980 and 1998, the prison rate increased by an average of 18 prisoners per hundred thousand people per year. She contends that additional money should not be spent on crime prevention unless the crime rate is increasing by at least 25 prisoners per 100,000 people per year. How could a proponent of the crime prevention measure use the model to convince her that crime prevention legislation should be passed now? (Assume that the debate takes place in 2000.) 35.
Private College Enrollment Based on
41. Given f ( x ) 5 |x| , find f r( x ) . 42. Given f r( x ) 5 3x and g ( x ) 5 x2 1 3 1 f ( x ) , find gr( x ) . 43. Given f ( x ) 5 x2 2 2x , determine where f r( x ) 5 0 . 44. Given f ( x ) 5 x3 2 3x , determine where f r( x ) 5 0 . 45. Graph f ( x ) 5 x3 2 3x . What does the result of Exercise 44 tell you about the graph of f ?
data from 1980 to 1998, the number of students enrolled at private colleges may be modeled by S( t ) 5 0.9195t 2 1 23.25t 1 2652 thousand students, where t is the number of years since the end of 1980. (Source: Modeled from
242
Interpret the meaning of the derivative in the context of a word problem
Many of us feel inundated by the advertisements we are sent through the mail. Dont expect this to let up anytime soon: Spending on direct-mail advertising has risen every year since 1990. The amount of money spent on direct-mail advertising may be modeled by A( t ) 5 70.54t2 1 1488t 1 22,828 million dollars where t is the number of years since 1990. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 1272, p. 777.) According to the model, A( 9 ) 5 41,934 and Ar( 9 ) 5 2758 . But what does this mean? In this section, we will discuss how to interpret the meaning of a derivative in the context of a real-life problem.
Recall that the units of the derivative are the units of the output divided millions of dollars by the units of the input. In this case, the units of Ar are or year millions of dollars per year. Note that t 5 9 corresponds to the year 1999. We conclude that in 1999, 41,934 million dollars were spent on direct-mail advertising, and spending was increasing by 2758 million dollars per year. In other words, according to the model, 41,934 million dollars were spent on direct-mail advertising in 1999 and spending increased by about 2758 million dollars between 1999 and 2000. The term about, approximately, or roughly must be used when using the second interpretation of the derivative, since we are using a tangent-line approximation to estimate the increase over the next year. (For a graphical discussion of tangent-line approximations, refer to Section 3.4.)
INTERPRETING THE DERIVATIVE
Let f ( x ) be a function. The meaning of f r( a ) 5 c may be written in either of the following two ways: When x 5 a, the value of the function f is increasing (decreasing) by c units of output per unit of input. The value of the function f will increase (decrease) by about c units of output between a units of input and a 1 1 units of input.
243
EXAMPLE
Interpreting the Meaning of the Derivative

Based on data from 1969 to 2002, the number of students taking the AP Calculus AB exam may be modeled by E( t ) 5 157.8t2 2 770.6t 1 10,268 students where t is the number of years since 1969. (Source: Modeled from College Board
data.)
Interpret the meaning of E( 33 ) 5 156,682 and Er( 33 ) 5 9644 . Then estimate E( 34 ) .

SOLUTION Since t is the number of years since 1969, t 5 33 corresponds to 2002. E( 33 ) 5 156,682 means that 156,682 students took the exam in 2002.
The units of the derivative are
students . year
Er( 33 ) 5 9644 means that in 2002,
the number of students taking the exam was increasing by 9644 students per year. In other words, the number of students taking the exam will increase by about 9644 students between 2002 and 2003. We use a tangent-line approximation to estimate E( 34 ) . Since Er( 33 ) < Er( 33 ) < E( 34 ) 2 E( 33 ) 34 2 33 E( 34 ) 2 E( 33 ) 1
Er( 33 ) < E( 34 ) 2 E( 33 ) E( 34 ) < E( 33 ) 1 Er( 33 ) < 156,682 1 9644 < 166,326 We estimate that 166,326 students took the exam in 2003. Each year the College Board hires high school and college educators from around the globe to score the free-response portion of the Calculus AB exam. If an average grader can score 270 exams during the week-long grading,
9644 270
5 35.7 < 36 additional graders were needed to score the exams in 2003.
EXAMPLE

Based on data from 1940 to 2000, the average monthly Social Security benefit for men may be modeled by P( t ) 5 0.3486t 2 2 5.505t 1 35.13 dollars where t is the number of years since 1940. (Source: Modeled from Social Security Interpret the meaning of P( 60 ) 5 959.79 and Pr( 60 ) 5 36.33 . Then use a tangent-line approximation to estimate P( 61 ) .
SOLUTION Since t is the number of years since 1940, t 5 60 corresponds to 2000. P( 60 ) 5 959.79 means that (according to the model) the average Social Security benefit for men in 2000 was $959.79. Administration data.)
244
dollars . year
Pr( 60 ) 5 36.33 means that in 2000,
the average Social Security benefit for men was increasing by $36.33 per year. In other words, the average Social Security benefit for men was expected to increase by about $36.33 between 2000 and 2001. We estimate the benefit in 2001 by approximating P( 61 ) . P( 61 ) < P( 60 ) 1 Pr( 60 ) < 959.79 1 36.33 < 996.12 In 2001, we estimate that the average Social Security benefit for men was $996.12.
EXAMPLE
Using a Tangent-Line Approximation to Estimate a Function Value

The average weight of a boy between the ages of 2 and 13 years old may be modeled by W( a ) 5 0.215a2 1 2.993a 1 23.78 pounds where a is the age of the boy in years. (Source: Modeled from www.babybag.com Interpret the meaning of W( 10 ) 5 75.21 and W r( 10 ) 5 7.29. Then use a tangent-line approximation to estimate W( 11 ) .
SOLUTION Since a is the age of the boy, a 5 10 corresponds to a 10-year-old boy. W( 10 ) 5 75.21 means that the average weight of a 10-year-old boy is 75.21 pounds. data.)
pounds . year of his age
W r( 10 ) 5 7.29 means that the
average weight of a 10-year-old boy is increasing by 7.29 pounds per year of his age. In other words, the average weight of a boy will increase by about 7.29 pounds between his tenth and eleventh years. We use a tangent-line approximation to estimate W( 11 ) . W( 11 ) < W( 10 ) 1 W r( 10 ) < 75.21 1 7.29 < 82.5 We estimate that the average weight of an 11-year-old boy is 82.5 pounds.
EXAMPLE

Based on data from 1970 to 1999, the percentage of new one-family homes that are 1200 square feet or less may be modeled by H( t ) 5 0.02676t 2 2 1.753t 1 35.93 percentage points where t is the number of years since 1970. (Source: Modeled from Statistical Abstract
of the United States, 2001, Table 938, p. 597.)
245
Interpret the meaning of H( 14 ) 5 16.6 and Hr( 14 ) 5 2 1.00 .

SOLUTION Since t is the number of years since 1970, t 5 14 corresponds to 1984. H( 14 ) 5 16.6 means that in 1984, 16.6 percent of new homes were 1200 square feet or less in size.
percentage points . year
Hr( 14 ) 5 2 1.00 means
that in 1984, the percentage of new homes that are 1200 square feet or less in size was decreasing by 1 percentage point per year. In other words, the percentage of new homes that are 1200 square feet or less in size was expected to decrease by about 1 percentage point (from 16.6 percent to 15.6 percent) from 1984 to 1985.
3.6 Summary
In this section, you learned how to interpret the meaning of the derivative verbally in its real-world context. You also discovered two different ways of representing the concept.

1. The cost c of buying x 3.5-pound bags of Boston Baked Beans is given by c( x ) 5 5.95x 1 13.85 dollars. Interpret the real-world meaning of c( 10 ) 5 73.35. 2. The number of miles per gallon a car gets, m, is a function of the cars speed, s (miles per hour). Interpret the real-world meaning of m( 50 ) 5 23. 3. Degrees Celsius, c, are a function of the number of degrees Fahrenheit, f. Interpret the real-world meaning of c( 212 ) 5 100. 4. An employees weekly wage in dollars, w, is a function of the number of hours worked, h. What does w( 20 ) 5 256 mean? 5. The cost c of buying x gallons of gasoline is given by c( x ) 5 3.219x dollars. Interpret the real-world meaning of c( 20 ) 5 64.38. 6. The area of a square, a, is a function of the length of one of its sides, s. What does a( 4 ) 5 16 mean? 7. The volume of a cube, v, is a function of its surface area, a. What does v( 6 ) 5 1 mean? 8. Given s 5 f ( t ) , determine which variable is dependent and which variable is independent.
3.6 Exercises
In Exercises 130, interpret the real-life meaning of the indicated values. Answer additional questions as appropriate. 1. The weight of a 2-year-old to 13-year-old girl may be modeled by
Body Weight
Interpret the meaning of W ( 10 ) 5 76.64 and W r( 10 ) 5 8.24. Then estimate W ( 11 ) . 2. Compare the results of Example 3 and Exercise 1. Were boys or girls expected to gain more weight between their tenth and eleventh years? Explain.
Body Weight Per Capita Income Based on data from 1959 to 1989, the annual per capita
W( a ) 5 0.289a2 1 2.464a 1 23.10 pounds, where a is the age of the girl. (Source:
3.
246
income of residents of Washington state may be modeled by P( t ) 5 2 2.98t 2 1 327.4t 1 7881 dollars, where t is the number of years since 1959.
(Source: Modeled from www.census.gov.)
9.
Radio Broadcasting Wages Based on
data from 1992 to 1998, the average annual wage of an employee in the radio broadcasting industry may be modeled by W( t ) 5 164.3t 2 1 778.6t 1 23,500
Interpret the meaning of P( 25 ) 5 14,204 and Pr( 25 ) 5 178.4 . Then estimate P( 26 ) .
dollars, where t is the number of years since 1992. Interpret the meaning of W( 6 ) 5 34,086 and Wr( 6 ) 5 2750. Estimate W( 7 ) .
Restaurant Sales Based on data from
4.
Per Capita Income Using the model from Exercise 3, Pr( 20 ) 5 208.2 , Pr( 25 ) 5 178.4, Pr( 30 ) 5 148.6 , and Pr( 35 ) 5 118.8. Based on these results, what conclusions can you draw about per capita income in the state of Washington? Nasdaq Market Volume Based on data
10.
5.
1989 to 1998, restaurant sales may be modeled by S( r ) 5 0.9291r 2 160.0 billion dollars, where r is the number of restaurants (and other eating places) in thousands.
from 1980 to 2000, the average daily volume of the Nasdaq market may be modeled by V( t ) 5 0.06754t 4 2 1.936t 3 1 17.97t 2 2 41.27t 1 39.03 million shares, where t is the number of years since 1980. (Source: Model based on Statistical
Interpret the meaning of S( 400 ) 5 212 and Sr( 400 ) 5 0.9291. (Hint: Convert the units of the derivative into thousands of dollars per restaurant.) 11.
Billboard Advertising Based on data
Interpret the meaning of V( 18 ) 5 918 and V r( 18 ) 5 299. Then estimate V( 19 ) .

Nasdaq Market Using the model
6.
from Exercise 5, V r( 18 ) 5 299, V r( 19 ) 5 398, and V r( 20 ) 5 516. Based on these results, what conclusions can you draw about the average daily volume of the Nasdaq market?
New York Stock Exchange Volume
from 1990 to 2000, the amount of money spent on billboard advertising may be modeled by A( t ) 5 9.396t 2 2 11.58t 1 1063 million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical
7.
Interpret the meaning of A( 10 ) 5 1887 and Ar( 10 ) 5 176 .

Cassette Tapes
Based on data from 1980 to 2000, the annual volume of the New York Stock Exchange may be modeled by V( t ) 5 4.807t 4 2 98.56t 3 1 541.3t 2 1 2936t 1 10,612 million shares, where t is the number of years since 1980. (Source: Modeled from Statistical
12.
Based on data from 1992 to 2000, the net number of cassette tapes shipped by recording industry manufacturers may be modeled by C( t ) 5 0.6527t 3 2 8.348t 2 2 10.30t 1 367.5 million cassettes, where t is the number of years since 1992. (Source: Modeled from Recording
Industry Association of America data.)
Interpret the meaning of V( 18 ) 5 168,659 and V r( 18 ) 5 38,760. Then estimate V( 19 ) . model from Exercise 7, V r( 18 ) 5 38,760, V r( 19 ) 5 48,650, and V r( 20 ) 5 60,140. Based on these results, what conclusions can you draw about the annual volume of the New York Stock Exchange?
Interpret the meaning of C( 8 ) 5 85.0 and Cr( 8 ) 5 2 18.5.
8.
New York Stock Exchange Using the
13.
Compact Disks Based on data from 1992 to 2000, the value of CDs shipped by recording industry manufacturers may be modeled by V( x ) 5 0.008086x2 1 3.308x 1 2735
247
million dollars, where x is the net number of CDs shipped in millions. (Net means after returns.) Interpret the meaning of V( 900 ) 5 12,261.9 and Vr( 900 ) 5 17.9. 14. Based on data from 1998 to 2001, the value of DVD videos shipped by recording industry manufacturers may be modeled by V( x ) 5 2 0.2173x2 1 25.84x 2 0.02345
DVDs
19.
(Source: Modeled from Recording Industry Association of America data.)
Deadly Alcohol-Related Brawls Based on data from 1991 to 2000, the number of homicides resulting from an alcohol-related brawl may be modeled by H( t ) 5 2 0.3551t 3 1 9.867t 2 2 104.7t 1 600.4
deaths, where t is the number of years since 1990.
Interpret the meaning of H( 10 ) 5 185 and Hr( 10 ) 5 2 14 . 20.

Body Height The average height of a
million dollars, where x is the net number of DVDs shipped in millions. (Net means after returns.) (Source: Modeled from Recording Industry Interpret the meaning of V( 7.5 ) 5 181.6 and V r( 7.5 ) 5 22.6 . 15.
Summer Olympics Based on data from
Association of America data.)
girl between the ages of 2 and 13 years may be modeled by H( a ) 5 2 0.0392a2 1 2.987a 1 29.69 inches, where a is the age of the girl. (Source:
Interpret the meaning of H( 5 ) 5 43.6 and Hr( 5 ) 5 2.6 .
1904 to 1992, the number of women in the Summer Olympics may be modeled by W( t ) 5 0.0002344t 4 2 0.03440t 3 1 1.701t 2 2 19.69t 1 70.80
(Source: Modeled from www.olympicwomen.co.uk data.)
21.
women, where t is the number of years since 1900.
Company Revenue Based on data from 1999 to 2001, the gross revenue from sales of Johnson & Johnson and its subsidiaries may be modeled by R( t ) 5 665t 2 1 1150t 1 27,357
Interpret the meaning of W( 92 ) 5 2662 and Wr( 92 ) 5 150. 16.
of Exercise 15. Do you think W( 96 ) will be greater than 2662 and W r( 96 ) will exceed 150? Defend your conclusions.
U.S. Population Based on data from
Summer Olympics Refer to the results
Interpret the meaning of R( 2 ) 5 32,317 and Rr( 2 ) 5 3810 .
22.
17.
1790 to 2000, the population of the United States may be modeled by P( t ) 5 0.006702t 2 2 24.11t 1 21,696
(Source: Modeled from U.S. Bureau of the Census data.)
Company Costs Based on data from 1999 to 2001, the cost of goods sold for Johnson & Johnson and its subsidiaries may be modeled by C( t ) 5 103t 2 1 315t 1 8539
million people, where t is the calendar year.
Interpret the meaning of P( 2000 ) 5 284 and Pr( 2000 ) 5 2.70 . 18.
Interpret the meaning of C( 2 ) 5 9581 and Cr( 2 ) 5 727.

Company Profit Based on data from 1999 to 2001, the operating profit of Frito-Lay North America may be modeled by P( t ) 5 2 47.5t 2 1 283.5t 1 1679
Homicide Rate Based on data from 1990 to 2000, the homicide rate (deaths per 100,000 people) in the United States may be modeled by H( t ) 5 0.01129t 3 2 0.2002t 2 1 0.4826t 1 9.390
23.
people, where t is the number of years since 1990. Interpret the meaning of H( 10 ) 5 5.5 and Hr( 10 ) 5 2 0.1 .
from PepsiCo 2001 Annual Report.)
Interpret the meaning of P( 2 ) 5 2056 and Pr( 2 ) 5 93.5 .
248 24.
Based on data from 1991 to 2001, the average hourly earnings for a Ford Motor Company employee may be modeled by W( t ) 5 2 0.003931t 4 1 0.1005t 3 2 0.8295t 2 1 3.188t 1 16.48
Employee Earnings
P( t ) 5 2 19.652t 3 1 296.96t 2 1 502.89t 1 29,285 dollars, where t is the number of years since the end of 1993. (Source: Modeled from Bureau of Interpret the meaning of P( 5 ) 5 36,767 and Pr( 5 ) 5 1998.6 . 29.
Income Analysis If a high per capita personal income is considered to be one indicator of a strong economy, compare and contrast the income and projected income growth in California, Colorado, and Connecticut between 1998 and 1999, using the results of Exercises 2628.
Economic Analysis data.)
dollars, where t is the number of years since the end of 1990. (Source: Ford Motor Company 2001 Interpret the meaning of W( 11 ) 5 27.39 and Wr( 11 ) 5 0.49. 25.
River Flow Based on data from
Annual Report.)
November 22 to November 24, 2002, the flow of the Snoqualmie River near Snoqualmie Falls, Washington, may be modeled by f ( t ) 5 25t2 2 145t 1 490 cubic feet per second, where t is the number of 24-hour periods since noon on November 22, 2002. Interpret the meaning of f ( 2 ) 5 300 and f r( 2 ) 5 2 45 . How could a river rafting company use this type of information? 26.
Personal Income: California Based
30.
Based on data from 1997 to 2003, the average price of a VCR may be modeled by
Price of a VCR
(Source: Modeled from www.dreamflows.com data.)
123.0 1 60 1 1 0.2649e0.9733t dollars, where t is the number of years since the end of 1997. (Source: Modeled from Consumer P( t ) 5
Electronics Association data.)
on data from 1993 to 2000, the per capita personal income of California may be modeled by P( t ) 5 123.73t 2 2 445.58t 1 22,847 dollars, where t is the number of years since the end of 1993. (Source: Modeled from Bureau of Interpret the meaning of P( 5 ) 5 23,712 and Pr( 5 ) 5 791.7. 27.
Personal Income: Colorado Based
Interpret the meaning of Pr( 1 ) 5 2 29.00 , Pr( 3 ) 5 2 16.82 , and Pr( 5 ) 5 2 3.285 .
31. Is it preferable for the derivative of your household electricity billing rate function to be positive, zero, or negative? Explain. 32. A company earned $520,000 in 2008. At the end of 2008, company executives forecasted that the companys earnings were increasing at a rate of $95,000 per year. Describe how this situation may be represented mathematically using the concept of a function and its derivative. 33. A vehicles odometer and speedometer reveal important information to the driver of the vehicle. Using the concepts of function and derivative, explain the relationship between the odometer reading and the speedometer reading. 34. What is a tangent-line approximation? 35. Explain why f ( x 1 1 ) < f ( x ) 1 f r( x ) .
on data from 1993 to 2000, the per capita personal income of Colorado may be modeled by P( t ) 5 80.994t 2 1 890.16t 1 22,158 dollars, where t is the number of years since the end of 1993. (Source: Modeled from Bureau of Interpret the meaning of P( 5 ) 5 28,634 and Pr( 5 ) 5 1700.1 . 28.
Personal Income: Connecticut Based
on data from 1993 to 2000, the per capita personal income of Connecticut may be modeled by
Chapter 3
Study Sheet
249
Adult and adolescent AIDS deaths (in thousands)
60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Years (since the end of 1981)
36.
Market Analysis The DVD player was
introduced in 1997. Based on the results of Exercise 30, what effect did the introduction of the DVD player have on the price of a VCR? As t gets larger, what value do you think P r( t ) will approach? 37.
Company Sales Income Based on data
from 1993 to 2002, the annual sales income of Starbucks Corporation may be modeled by S( t ) 5 29.23t 2 1 79.33t 1 177.4 million dollars, where t is the number of years since the end of 1993. According to the model, in what year (between 1993 and 2002) was annual sales income increasing most rapidly? Justify your answer. 38.
AIDS Deaths in the United States
The highest number of AIDS deaths (50,876) occurred in 1995. (Source: Centers for Disease
Control and Prevention.)
Estimate from the graph the year in which f r was the greatest and the year in which f r was the most negative.
Based on data from 1981 to 2001, the number of adult and adolescent AIDS deaths in the United States may be modeled by a piecewise function f, whose graph is shown in the following figure along with a scatter plot of the number of deaths as a function of years since the end of 1981.
39. Let f be any continuous, smooth function (no breaks or sharp points) that is defined on the interval ( 2 ` , ` ) . If f ( a ) # f ( x ) for all x, what is the value of f r( a ) ? 40. Let f be any continuous, smooth function (no breaks or sharp points) that is defined on the interval ( 2 ` , ` ) . If f ( a ) $ f ( x ) for all x, what is the value of f r( a ) ?

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What does the difference quotient represent? (3.1) 2. The units of any rate of change are the units of the output divided by the units of the input. Why? (3.1) 3. How are tangent lines and secant lines similar and how are they different? Appropriately refer to the difference quotient in your explanation. (3.1, 3.2) 4. What is the difference between an average rate of change and an instantaneous rate of change? (3.2) 5. What does lim f ( x ) 5 L mean? Write your response as if you were explaining xSc the concept to a classmate who had missed class on the day limits were covered. (3.2) 6. What does the derivative of a function represent? (3.2, 3.4) 7. Why can a tangent line be used to approximate the value of a function? (3.4)
250
8. Given a table of data, what is an effective method for estimating the instantaneous rate of change in the data at a given value? Explain why the method you identified works. (3.4) 9. How are the derivative and the difference quotient related? (3.2, 3.4, 3.5) 10. Why is it important to know how to accurately interpret the meaning of the derivative in a real-world context? (3.6) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
Definitions
Secant Line: A line connecting any two points on a graph, whose slope is calculated by using the difference quotient Tangent Line: The line passing through a point ( a, f ( a )) that has slope f r( a ) Average Rate of Change: The average of the rates of change of a function over an interval Instantaneous Rate of Change: The rate of change of a function at a single point
Limit of a Function The limit of a function f exists at x 5 c if f is defined for all values of x near c and lim f ( x ) 5 L for some finite number L. Limit Operations Given that c and k are real numbers and n is a positive integer, the following properties and operations are valid provided that lim f ( x ) and xSc lim g ( x ) exist.
xSc
xSc
Constant: Constant Multiple: Addition/Subtraction: Multiplication:
lim k 5 k
xSc xSc xSc xSc
lim k f ( x ) 5 k( lim f ( x ))
xSc
lim ( f ( x ) 6 g ( x )) 5 lim f ( x ) 6 lim g ( x )

xSc xSc
lim ( f ( x ) ? g ( x )) 5 ( lim f ( x )) ? ( lim g ( x ))

xSc xSc
Division: Power: Rational Exponent:
lim
xSc
lim f ( x ) f (x) 5 xSc , if lim g ( x ) 2 0 xSc g (x) lim g ( x )

xSc
n
lim 3 f ( x ) 4 5 3lim f ( x ) 4 n
xSc xSc
lim 3 f ( x ) 4 1>n 5 3lim f ( x ) 4 1>n (If n is even, lim f ( x ) must be positive.)

xSc xSc xSc
Chapter 3
Review Exercises
251
Continuity at a Point A function f ( x ) is said to be continuous at x 5 c if all of the following criteria are satisfied. 1. lim f ( x ) exists. 2. f ( c ) exists. 3. lim f ( x ) 5 f ( c )
xSc
Limits at Infinity A function f ( x ) has a limit at 2 ` if there exists some constant value L to which f ( x ) gets infinitely close for negative x of sufficiently large absolute value. Symbolically, we write
x S 2`
xSc
lim f ( x ) 5 L
Similarly, a function f ( x ) has a limit at ` if there exists some constant value L to which f ( x ) gets infinitely close for sufficiently large x. Symbolically, we write
xS`
lim f ( x ) 5 L
Formulas
f (b) 2 f (a) f (a 1 h) 2 f (a) or b2a h f (a 1 h) 2 f (a) Derivative at a Point ( a, f ( a )) : f r( a ) 5 lim hS0 h Numerical Estimate of the Derivative at a Point ( a, f ( a )) : f (a 1 h) 2 f (a 2 h) f r( a ) < 2h f (x 1 h) 2 f (x) Derivative Function: f r( x ) 5 lim hS0 h Difference Quotient:

Section 3.1 In Exercises 12, calculate the average rate of change of the function over the given interval.
1. f ( x ) 5 6x 2 12; 31, 84 2. g ( x ) 5 x2 2 2x 1 1; 30, 24 In Exercises 34, calculate the average rate of change in the designated quantity over the given interval. 3. Change in per capita income between 1969 and 1989.
United States Per Capita Income, 19591989
Year (t) 1959 1969 1979 1989
Income (dollars) (P) 7,259 9,816 12,229 14,420
252
4. Change in per capita income between 1959 and 1989.

Washington Per Capita Income, 19591989
In Exercises 1112, use the difference quotient (with h 5 0.1, h 5 0.01, and h 5 0.001) to estimate the instantaneous rate of change of the function at the given input value. You may find it helpful to apply the Technology Tip demonstrated in Section 3.2. 11. f ( x ) 5 2x23 1 x; x 5 4 12. P( t ) 5 16( 1.04 ) t; t 5 20 In Exercises 1314, determine the instantaneous rate of change of the function at the indicated input value. (You may find it helpful to apply the Technology Tip demonstrated in Section 3.2.) Explain the real-life meaning of the result. 13.
Olympic Runners Based on the run
Year Since 1959 (t) 0 10 20 30
Income (dollars) (P) 7,978 10,565 13,528 14,923
In Exercises 56, use the difference quotient to calculate the slope of the secant line through the points ( 1, f ( 1 )) and ( 3, f ( 3 )) . 5. f ( x ) 5 3x 6. The graph of f is shown.
y 2 1 0 1 2 3 4 1 2 3 x
times of the top seven runners of the 100-meter mens race in the 2000 Olympics, a runners time may be modeled by T ( p ) 5 0.003056p3 2 0.04274p2 1 0.2178p 1 9.694 seconds, where p is the place (rank) of the runner. (Source: Modeled from
www.sporting-heroes.net data.)
Find and interpret the meaning of T r( 7 ) .
14.
Presidential Election Funding Based on data from 1980 to 1996, the amount of federal funds used for the presidential election may be modeled by
F ( t ) 5 0.03867t 3 2 0.6804t 2 1 6.604t 1 62.62 million dollars, where t is the number of years since 1980. (Source: Modeled from Statistical
Find and interpret the meaning of F r( 16 ) . In Exercises 1516, calculate the
Section 3.2 In Exercises 78, use the difference quotient (with h 5 0.1, h 5 0.01, and h 5 0.001) to estimate the instantaneous rate of change of the function at the given input value.
7. f ( x ) 5 2x2 1 3; x 5 1 8. h( x ) 5 2 x2 1 5x 2 2; x 5 4 In Exercises 910, use the derivative to calculate the instantaneous rate of change of the function at the given input value. (In each exercise, you can eliminate h algebraically.) Compare your answers to the solutions of Exercises 78. 9. f ( x ) 5 2x2 1 3; x 5 1 10. h( x ) 5 2 x2 1 5x 2 2; x 5 4
limit, if it exists. 15. lim

xS2
Section 3.3
2x 2 4 x23 x2 2 9 x13
16. lim
xS 23
In Exercises 1718, determine the limit at 1` or 2 ` , as indicated in the exercise. 17. lim x23
x S 2`
18. lim
xS`
4x2 2 1 2x2 1 8
Chapter 3
Review Exercises
253
In Exercises 1920, determine if the function is continuous. 19. g ( t ) 5 t13 t2 1 9
In Exercises 2526, estimate the specified derivative by using the data in the table. Then interpret the result. 25.
Highway Fatalities
t2 1 4 20. f ( t ) 5 2 t 24
Highway Fatalities
Section 3.4 In Exercises 2122, determine the equation of the line tangent to the function at the given point. Then graph the tangent line and the function together.
21. f ( x ) 5 2x2 2 2x; ( 1, 0 ) 22. f ( x ) 5 2 x2 1 6x; ( 2, 8 ) In Exercises 2324, answer the questions by calculating the slope of the tangent line and the tangent-line equation, as appropriate. 23.
Social Security Benefit Based on data
Years Since 1970 (t) 0 10 20 25 30
Fatalities (F) 52,627 51,091 44,599 41,817 41,821
Source: U.S. Department of Transportation, Bureau of Transportation Statistics.
Estimate F r( 15 ) . 26.
Aviation Fatalities
from 1940 to 2000, the average monthly Social Security benefit for a retired man may be modeled by P ( t ) 5 0.3486t 2 2 5.505t 1 35.13 dollars, where t is the number of years since 1940. How quickly was the average monthly Social Security benefit for a retired man increasing in 2000? What was the average monthly Social Security benefit for a retired man in 2001? (Use a tangent-line approximation.) 24. The weight of a 2-year-old to 13-year-old girl may be modeled by W( t ) 5 0.289a2 1 2.464a 1 23.10
Weight of an Average Girl
General Aviation Fatalities
Years Since 1970 (t) 0 10 20 25 30
Fatalities (F) 1,310 1,239 767 734 592
pounds, where a is the age of the girl. (Source:

Source: U.S. Department of Transportation, Bureau of Transportation Statistics.
How quickly is the average girl gaining weight when she is 12 years old? What is the estimated weight of a 13-year-old girl? (Use a tangent-line approximation.)
Estimate F r( 25 ) .
254
Section 3.5 In Exercises 2728, find the derivative of the function.

27. f ( x ) 5 2x2 2 4 28. f ( x ) 5 2 x2 1 3x
thousand people, where t is the number of years since 1985. (Source: Modeled from Statistical
Abstract of the United States, 2001, Table 1241, p. 759).
In Exercises 2930, find the slope of the line tangent to the function at x 5 1 , x 5 3 , and x 5 5 . 29. g ( x ) 5 3x2 1 2x 2 5 30. g ( x ) 5 x2 2 6x 1 9 In Exercises 3132, estimate the derivative of the function, using h 5 0.001 if you are unable to eliminate the h in the denominator of the derivative formula algebraically. 31. P( x ) 5 2x 32. R( x ) 5 3 ? 4x
Was attendance increasing faster at the end of 1995 or the end of 1999?
Section 3.6 In Exercises 3536, interpret the real-life meaning of the function and of the derivative evaluated at the given point.
35.
McDonalds Sales Based on data from 1990 to 2001, franchised sales at McDonalds fast-food restaurants may be modeled by S( t ) 5 1233.5t 1 12,213
In Exercises 3334, use the derivative function to answer the questions. 33.
Major League Baseball Salaries Based
million dollars, where t is the number of years since 1990. (Source: Modeled from McDonalds July
26, 2002, Financial Report.)
on data from 1980 to 2000, the average annual salary of a major league baseball player may be modeled by B( t ) 5 2.699t 2 1 20.76t 1 158.4 thousand dollars, where t is the number of years since 1980. (Source: Modeled from Statistical
Interpret the meaning of S( 12 ) 5 27,015 and S r( 12 ) 5 1233.5 .
36.
Was the average player salary growing faster at the end of 1989 or the end of 1999? 34.
Womens NCAA Basketball Based on
Church Membership Based on data from 1978 to 2000, the number of congregations (wards and branches) in the United States of the Church of Jesus Christ of Latter-Day Saints may be modeled by C( t ) 5 8679( 1.023 ) t congregations, where t is the number of years since 1978. (Source: Modeled from data compiled by
Mark Davies, Brigham Young University.)
data from 1985 to 1999, the number of people attending womens NCAA basketball games may be modeled by W( t ) 5 17.1t 2 1 359t 1 135
Interpret the meaning of C ( 14 ) 5 11,932 and C r( 14 ) 5 271 .
P R O J E C T
What to do
1. Collect a minimum of six data points related to a public or privately owned business (for example, the profit or revenue of a company over a six-year period). 2. Model the data with a polynomial function. 3. Find the derivative of the model function. 4. Evaluate the derivative at three different points. 5. Interpret the real-life meaning of the results of Step 4. 6. Discuss the benefits and the drawbacks of using the model. 7. Explain how company management could use the results of your analysis.

Company Web Sites
Many company web site addresses are of the form www.companyname.com, where companyname is the name of the company. Look for the area of the web site with investor information. Some web sites to try are www.mcdonalds.com www.levistrauss.com www.albertsons.com www.gm.com www.kelloggs.com
Company Financial Reports

Financial reports are typically mailed annually to investors who own shares in a company.
Financial Newspapers and Magazines

The Wall Street Journal, BusinessWeek, the New York Times, Forbes magazine, and others contain a wealth of business information.
255
Chapter
Differentiation Techniques
A
mathematical model may be used to forecast a companys revenue at a given point in time. However, business executives and investors arent interested only in the dollar amount of a companys revenue; they are also interested in the direction in which revenue is heading. As a rate of change, the derivative of the revenue function shows whether revenues are increasing or decreasing.
4.1
Basic Derivative Rules
Use various forms of derivative notation Use the Constant Rule, Power Rule, Constant Multiple Rule, and Sum and Difference Rule to find the derivative of a function
4.2
The Product and Quotient Rules
Use the Product Rule and Quotient Rule to find the derivative of a function
4.3
The Chain Rule
Use the Chain Rule to find the derivative of a function
4.4
Exponential and Logarithmic Rules
Use the Exponential Rule and Logarithmic Rule to find the derivative of a function
4.5
Implicit Differentiation
Use implicit differentiation to differentiate functions and nonfunctions
256
4.1 Basic Derivative Rules
257
Use various forms of derivative notation Use the Constant Rule, Power Rule, Constant Multiple Rule, and Sum and Difference Rule to find the derivative of a function
Home video-game systems have become increasingly popular over the last two decades. Based on data from 1990 to 1999, electronic game system factory sales may be modeled by S( t ) 5 8.451t 3 2 111.9t2 1 470.0t 1 965.0 million dollars where t is the number of years since 1990. How quickly were game system sales increasing in 1999? How quickly were sales expected to increase in 2000? In the previous chapter, you learned how to find the derivative function by using the limit definition of the derivative. Unfortunately, calculating the difference quotient for a complex function (such as the electronic gaming function) is cumbersome and prone to error. In this section, we will introduce some alternative forms of derivative notation and then demonstrate several shortcuts that will greatly enhance our efficiency in calculating derivatives.
Derivative Notation
Up to this point, we have used f r( x ) to represent the derivative of a function f ( x ) . Due in part to the fact that calculus was first developed by two people working independently (Newton and Leibniz), there are multiple ways to represent the same concept.
DERIVATIVE NOTATION
The derivative of a function y 5 f ( x ) may be represented by any of the following: f r ( x ) , read f prime of x y r , read y prime
dy , read dee why dee ex or the derivative of y with respect to x dx
d dy means find the derivative with respect to x, whereas is the dx dx dy derivative of y with respect to x. The form is referred to as Leibniz notation dx for the derivative. Throughout this section, we will use various forms of notation for the derivative. Additionally, we will use the term differentiate to mean find the derivative of and the term differentiation to refer to the process of finding the derivative. A function whose derivative exists for all values of x in its domain is said to be differentiable. Note that
258
CHAPTER 4 Differentiation Techniques
The Constant Rule

Recall that the derivative represents the slope of the tangent line of the function. Consider the horizontal line, f ( x ) 5 c. The slope of the tangent line of f at ( a, c ) is f r( a ) 5 lim
hS0
f (a 1 h) 2 f (a) h c2c h 0 h
Since dividing 0 by any nonzero number yields 0
5 lim
hS0
5 lim 50
hS0
CONSTANT RULE
The derivative of a constant function y 5 c is yr 5 0
EXAMPLE
Finding the Derivative of a Constant Function

Find the derivative of y 5 5.
SOLUTION
yr 5 0
The Power Rule

Recall that a power function is a function of the form f ( x ) 5 xn, where x is a variable and n is a constant. In Exercises 3.5, you found that the power function y 5 x3 had the derivative yr 5 3x2. Similarly, the derivative of y 5 x4 is yr 5 4x3, and the derivative of y 5 x5 is yr 5 5x4. Although the derivative methods previously introduced allow us to find the derivative of any power function, the Power Rule provides a remarkably quick and easy way to calculate the derivative.
POWER RULE
The derivative of a function y 5 xn, where x is a variable and n is a nonzero constant, is y r 5 nx n 2 1
259
EXAMPLE
Finding the Derivative of a Power Function

Find dy given y 5 x3. dx
SOLUTION
dy 5 3x3 2 1 dx 5 3x2
EXAMPLE

Differentiate g( x ) 5 x25.
SOLUTION
gr( x ) 5 2 5x25 2 1 5 2 5x26
EXAMPLE

Find yr given y 5 x .
SOLUTION Initially, this function doesnt look like a power function. However,
1
recall that y 5
1 x
is equivalent to y 5 x21, which is of the form y 5 xn. yr 5 2 1x21 2 1 5 2 x22 52 1 x2
EXAMPLE
Finding the Derivative of a Power Function with a Rational Exponent

3 Find f r( x ) given f ( x ) 5 "x.
However, any function of the form f ( x ) 5 "x may be rewritten as f ( x ) 5 x1>n.

n 3 Therefore, f ( x ) 5 "x is equivalent to f ( x ) 5 x1>3, which is of the form y 5 xn.
SOLUTION Again, f ( x ) 5 "x doesnt appear to be of the form y 5 xn.

3
f r( x ) 5 5
1 1>3 2 1 x 3 1 22>3 x 3
260
This result may be written in a variety of alternative forms, such as the following: f r( x ) 5 f r( x ) 5 1 3x2>3 3"x2
3
When working with functions with rational exponents, it is helpful to recall that y 5 "xm
n
5 xm>n
5 ( xm ) 1>n Each of these forms is a correct way to represent the function y 5 "xm.
n
5 ( x1>n ) m
Constant Multiple Rule

The Constant Multiple Rule is used when a constant is multiplied by a function. This rule allows us to factor out the constant, take the derivative of the function, and then multiply the result by the factored-out constant.
CONSTANT MULTIPLE RULE
The derivative of a differentiable function g( x ) 5 kf ( x ) with constant k is given by gr( x ) 5 kf r( x )
EXAMPLE
Finding a Derivative of a Power Function

Differentiate g( x ) 5 3x2.
SOLUTION
gr( x ) 5
d ( 3x2 ) dx d 2 (x ) dx
Constant Multiple Rule Power Rule
53?
5 3( 2x ) 5 6x
261
EXAMPLE

Find the derivative of y 5 4x3.
SOLUTION
yr 5
d ( 4x3 ) dx d 3 (x ) dx
2
54?
5 4 ( 3x2 ) 5 12x
Sum and Difference Rule

The Sum and Difference Rule allows us to calculate the derivative of a function with multiple terms by adding together the derivatives of each of the terms.
SUM AND DIFFERENCE RULE
The derivative of a differentiable function h( x ) 5 f ( x ) 1 2 g( x ) is given by hr( x ) 5 f r( x ) 1 2 gr( x )
EXAMPLE
Finding the Derivative of a Sum of Functions

Differentiate h( x ) 5 3x22 1 5x.
SOLUTION
hr( x ) 5 5
d ( 3x22 1 5x ) dx d d ( 3x22 ) 1 ( 5x ) dx dx d 22 d (x ) 1 5 ? (x) dx dx

23
Sum and Difference Rule Constant Multiple Rule Power Rule
53?
5 3( 2 2x23 ) 1 5( 1 ) 5 2 6x 15
Although we showed multiple steps in this example, as you become skilled in using the rules, you will frequently be able to calculate the derivative in one or two steps.
262
EXAMPLE
Finding the Derivative of a Sum of Functions

Find dy given y 5 5x3 2 6x2 1 1. dx
SOLUTION
dy d 5 ( 5x3 2 6x2 1 1 ) dx dx 5 d d d ( 5x3 ) 2 ( 6x2 ) 1 (1) dx dx dx

By the Sum and Difference Rule By the Constant Multiple, Constant, and Power Rules
5 15x2 2 12x 1 0 5 15x2 2 12x
We will now return to the electronic games sales function introduced at the beginning of this section.
EXAMPLE 10
Using Derivative Rules to Quickly Calculate an Instantaneous Rate of Change

Based on data from 1990 to 1999, electronic game system factory sales may be modeled by S( t ) 5 8.451t 3 2 111.9t 2 1 470.0t 1 965.0 million dollars where t is the number of years since 1990. (Source: Modeled from Statistical Abstract
How quickly were game system sales increasing in 1999? How quickly were sales expected to be increasing in 2000? t 5 10. Sr( t ) 5 5
SOLUTION We are asked to evaluate the derivative of the function at t 5 9 and
d ( 8.451t 3 2 111.9t 2 1 470.0t 1 965.0 ) dt d d d d ( 8.451t 3) 2 ( 111.9t 2) 1 ( 470.0t ) 1 ( 965.0 ) dt dt dt dt

Sum and Difference Rule Power, Constant Multiple, and Constant Rules
5 3( 8.451 ) t2 2 2( 111.9 ) t 1 470.0 1 0 5 25.353t2 2 223.8t 1 470.0 Sr( 9 ) 5 25.353( 9 ) 2 2 223.8( 9 ) 1 470.0 5 2053.593 2 2014.2 1 470.0 < 509.4 Sr( 10 ) 5 25.353( 10 ) 2 2 223.8( 10 ) 1 470.0 5 2535.3 2 2238.0 1 470.0 5 767.3
In 1999, game system sales were increasing by $509.4 million per year. In 2000, game system sales were expected to be increasing at a rate of $767.3 million per year.
263
EXAMPLE 11
Using Derivative Rules to Quickly Calculate an Instantaneous Rate of Change

In Example 10, we saw that game system sales were increasing rapidly between 1990 and 1999. Game software sales were also increasing rapidly over the same time period. Based on data from 1990 to 1999, electronic game software factory sales may be modeled by S( t ) 5 39.79t 2 2 48.41t 1 2378 million dollars where t is the number of years since 1990. (Source: Modeled from Statistical Abstract
Were electronic game software sales increasing faster in 1995 or 1999?

SOLUTION We need to evaluate Sr( t ) at t 5 5 and t 5 9.
Sr( t ) 5
d ( 39.79t 2 2 48.41t 1 2378 ) dt

Constant Multiple, Power, Constant, and Sum and Difference Rules
5 2( 39.79 ) t 2 48.41 1 0 5 79.58t 2 48.41 Sr( 5 ) 5 79.58( 5 ) 2 48.41 5 397.9 2 48.41 5 349.49 Sr( 9 ) 5 79.58( 9 ) 2 48.41 5 716.22 2 48.41 5 667.81
In 1995, game software sales were increasing by $349.5 million per year. In 1999, sales were increasing by $667.8 million per year. Sales were increasing much faster in 1999 than in 1995.
4.1 Summary
In this section you learned how to use the Constant Rule, the Power Rule, the Constant Multiple Rule, and the Sum and Difference Rule. You also learned alternative ways of representing the derivative.

1. Simplify the expression 5( 3x2 ) 1 4( 2x ) . 2. Simplify the expression 2 5( 4x3 ) 1 20 32 4( x ) 254 1 7. 6. Rewrite the expression with a rational 5 exponent: "x3 7. Rewrite the expression with a rational 5 exponent: 3 x 2 8. Rewrite the expression as the sum of two terms 17x2 1 1 without a denominator: x4
!
3. Simplify the expression 16( 2 x22 ) 1 11( 3x2 ) 1 4( 2x ) . 4. Simplify the expression 2 14( 2x ) 2 19( 2 2x23 ) . 5. Rewrite the expression with a rational exponent: 3 4 " x
264
4.1 Exercises
In Exercises 110, use the Constant Rule, the Power Rule, the Constant Multiple Rule, and the Sum and Difference Rule (as appropriate) to find the derivative of the function. 1. f ( x ) 5 5 3. v( t ) 5 5t 3 2 10t22 4. s( t ) 5 7t 4 2 12t 2 1 2t 1 43 5. g( x ) 5 2 2x 1 4x 3 6. s( n ) 5 44n 2 2 n 7. f ( t ) 5 4t21 8. h( t ) 5 15t 5 2 75t 4 1 300t 3 9. g( t ) 5 t 3 1 3t 2 1 6t 1 6 10. f ( x ) 5 0 In Exercises 1120, find the equation of the tangent line at the indicated domain value. 11. y 5 " x; x 5 1 12. y 5 6.9x 4 2 3.2x 2 1 9.1; x 5 0 13. y 5 2 2.6x 5 1 3x22; x 5 2 1 14. y 5 2 7.62x 2 5.21x 1 6.11; x 5 1
2
thousand acres, where t is the number of years since 1992. (Source: Modeled from Statistical
2. g( x ) 5 2x 1 4
How quickly was the amount of organic cropland increasing in 1996? 22. Based on data from 1970 to 2000, the number of highway traffic fatalities may be modeled by F( t ) 5 1.554t 3 2 72.90t 2 1 427.0t 1 52,615
Traffic Fatalities
fatalities, where t is the number of years since 1970. (Source: Modeled from U.S. Department of
Transportation, Bureau of Transportation Statistics data.)
At what rate was the number of highway fatalities changing in 1995? 23.
Average Height of a Girl The average height of a 2- to 13-year-old girl may be modeled by H( a ) 5 2 0.0392a 2 1 2.987a 1 29.69
inches, where a is the age of the girl. (Source:

Does the average height of a 4-year-old girl increase faster than the average height of an 11year-old girl? 24.
Average Height of a Boy The average
15. y 5 1235.3t 3 1 551.23t 2 1203.9; t 5 0 16. y 5 2 t21 1 2t22 2 3t23; t 5 4 17. f ( n ) 5 9n1.3 1 5n2.1 1 92n20.2; n 5 1 18. w( t ) 5 3.2t 2 t 3; t 5 5 19. g( x ) 5 x23 1 2.22x 3 2 12.3; x 5 1 20. r ( x ) 5
1 1.5 x ; 2
height of a 2- to 13-year-old boy may be modeled by H( a ) 5 2 0.0507a 2 1 2.997a 1 30.44 inches, where a is the age of the boy. (Source:
x50 25.
In Exercises 2130, answer the question by evaluating the derivative using the derivative rules introduced in this section. 21.
Organic Cropland Based on data from 1992 to 1996, the amount of certified organic cropland in the United States may be modeled by A( t ) 5 20.57t 2 2 24.51t 1 410.3
Does the average height of a 4-year-old boy increase faster than the average height of an 11year-old boy? Using the models in Exercises 23 and 24, does a 9-year-old boy or a 9-year-old girl grow at a quicker rate?
Average Weight of a Girl The average weight of a 2- to 13-year-old girl may be modeled by W( a ) 5 0.289a 2 1 2.464a 1 23.10
26.
265
pounds, where a is the age of the girl. (Source:

At what rate was television advertising expected to increase in 2000?
Does the average weight of a 4-year-old girl increase faster than that of an 11-year-old girl? 27.
Average Weight of a Boy The average
weight of a 2- to 13-year-old boy may be modeled by W( a ) 5 0.215a2 1 2.993a 1 23.78 pounds, where a is the age of the boy. (Source:
31. What is the relationship between the equation yr 5 2x and the graph of y 5 x2? 32. Write three different ways to express the derivative of y 5 f ( x ) . d 33. What is the difference in meaning between dx dy and ? dx 34. Explain why the derivative of a constant function y 5 c is equal to zero. 35. What is the degree of the derivative of a polynomial of degree n?
Does the average weight of a 4-year-old boy increase faster than the average weight of an 11year-old boy? 28. Using the models in Exercises 26 and 27, does a 9-year-old boy or a 9-year-old girl gain weight at a quicker rate?
Billboard Advertising Based on data
29.
from 1990 to 2000, the amount of money spent on billboard advertising may be modeled by A( t ) 5 34.99t 2 1 94.91t 1 8963 million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical
36. Let f, g, and h be differentiable functions with h( x ) 5 f ( x ) ? g( x ) . May the Constant Multiple Rule be used to find hr( x ) ? Explain. 37. Let f ( x ) 5 2x. Is f r( x ) 5 x ? 2x 2 1 ? Defend your conclusions. 38. Use the limit definition of the derivative to prove that if s is differentiable and s( x ) 5 f ( x ) 1 g( x ) , then sr( x ) 5 f r( x ) 1 gr( x ) . 39. Use the limit definition of the derivative to prove that if f is differentiable, k is a constant, and f ( x ) 5 k ? g( x ) , then f r( x ) 5 k ? gr( x ) . 40. Given f ( x ) 5 x1>3 , for what values of x is f r( x ) undefined?
Was billboard advertising expected to increase more rapidly in 1999 or in 2000? 30.
Television Advertising Based on data from 1990 to 2000, the amount of money spent on broadcast TV advertising may be modeled by A( t ) 5 10.22t 4 2 221.5t 3 1 1616t 2 2 2488t 1 26,671
million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical Abstract
266
4.2 The Product and Quotient Rules
Use the Product Rule and Quotient Rule to find the derivative of a function
Based on data from 1985 to 1999, the per capita consumption of chicken may be modeled by C( t ) 5 0.002623t 4 2 0.07545t 3 1 0.6606t2 2 0.4688t 1 36.92 pounds of boneless, trimmed chicken, where t is the number of years since 1985. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 202, p. 129.) Based on data from 1985 to 2000, the average retail price of fresh whole chicken may be modeled by P( t ) 5 0.0007346t 2 2 0.009701t 1 0.7803 dollars per pound, where t is the number of years since 1985. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 706, p. 468.) A whole chicken loses about 30 percent of its weight when it is boned and trimmed. Was per capita spending on chicken increasing or decreasing in 2000? By how much? We will use the Product Rule to answer this question in Example 4 of this section. In this section, we will demonstrate how to find the derivative of a product by using the Product Rule and the derivative of a quotient by using the Quotient Rule.
PRODUCT RULE
The derivative of a function h( x ) 5 f ( x ) ? g( x ) is given by hr( x ) 5 f r( x ) ? g( x ) 1 gr( x ) ? f ( x )
A common error among beginning calculus students is to assume that d 3 f ( x ) ? g( x ) 4 5 f r( x ) ? gr( x ) . We can easily convince ourselves that this is dx not true by calculating the derivative of f ( x ) 5 x2 using the Power Rule and then by using the erroneous rule. Using the Power Rule, d 2 ( x ) 5 2x dx Using the erroneous rule, d 2 d (x ) 5 (x ? x) dx dx 5 d d (x) ? (x) dx dx
51?1 51
267
Since 2x 2 1, we know that the erroneous rule is not valid. Lets calculate the derivative using the Product Rule. Using the Product Rule, d 2 d (x ) 5 (x ? x) dx dx 5 d d (x) ? x 1 (x) ? x dx dx
51?x11?x 5 2x The Product Rule yielded the same result as the Power Rule.
EXAMPLE
Finding the Derivative of a Product of Functions

Find the derivative of h( x ) 5 ( 2x 1 4 )( 5x3 2 3x ) .
SOLUTION
hr( x ) 5
d 3( 2x 1 4 )( 5x3 2 3x ) 4 dx d d ( 2x 1 4 ) d ? ( 5x3 2 3x ) 1 c ( 5x3 2 3x ) d ? ( 2x 1 4 ) dx dx

3 3 2
5 c
Product Rule
5 ( 2 ) ? ( 5x3 2 3x ) 1 ( 15x2 2 3 ) ? ( 2x 1 4 ) 5 ( 10x 2 6x ) 1 ( 30x 1 60x 2 6x 2 12 ) 5 40x3 1 60x2 2 12x 2 12
Power, Constant Multiple, and Constant Rules
We can check our work by multiplying out the factors of h( x ) and differentiating the result using the Power Rule. h( x ) 5 ( 2x 1 4 )( 5x3 2 3x ) 5 10x4 1 20x3 2 6x2 2 12x hr( x ) 5 40x3 1 60x2 2 12x 2 12 The result is the same as that obtained by using the Product Rule. We are confident that we have done the problem correctly.
In the previous two examples, we saw that there are multiple ways to calculate the derivative of a function. In general, we will use the simplest techniques possible to find the derivative. However, in order to demonstrate the Product Rule, in this section we may use the Product Rule in places where other methods are typically preferred.
268
Just In Time
Multiplying Polynomials
A binomial is a two-term polynomial and a trinomial is a three-term polynomial. Although the FOIL method works well when multiplying two binomials, it is not useful for multiplying two trinomials or a binomial and a trinomial. When multiplying any two polynomials, it is important to remember that each term of the first polynomial must be multiplied by each term of the second polynomial. The resultant products are then summed. The computations often are easier if we first write the product of the first term of the first polynomial times the entire second polynomial, then we write the product of the second term of the first polynomial times the entire second polynomial, and so on. For example, ( 3x2 1 2x 2 4 )( 6x3 2 5x 1 7 ) 5 3x2( 6x3 ) 1 3x2( 2 5x ) 1 3x2( 7 ) 1 2x( 6x3 ) 1 2x( 2 5x ) 1 2x( 7 ) 2 4( 6x3 ) 2 4( 2 5x ) 2 4( 7 ) 5 18x5 2 15x3 1 21x2 1 12x4 2 10x2 1 14x 2 24x3 1 20x 2 28 5 18x5 1 12x4 2 39x3 1 11x2 1 34x 2 28
EXAMPLE
Finding the Derivative of a Product of Functions

Differentiate y 5 ( 2x3 2 6x 1 5 )( x4 2 2x2 1 1 ) .
SOLUTION
dy d 5 (y) dx dx 5 d 3( 2x3 2 6x 1 5 )( x4 2 2x2 1 1 ) 4 dx d d ( 2x3 2 6x 1 5 ) d ( x4 2 2x2 1 1 ) 1 c ( x4 2 2x2 1 1 ) d ( 2x3 2 6x 1 5 ) dx dx
5 c
5 ( 6x2 2 6 )( x4 2 2x2 1 1 ) 1 ( 4x3 2 4x )( 2x3 2 6x 1 5 ) 5 ( 6x6 2 12x4 1 6x2 2 6x4 1 12x2 2 6 ) 1 ( 8x6 2 24x4 1 20x3 2 8x4 1 24x2 2 20x ) 5 14x6 2 50x4 1 20x3 1 42x2 2 20x 2 6
EXAMPLE
Finding the Derivative of a Product of Three Functions

Differentiate R( t ) 5 ( 5t 2 1 )( 4t 1 3 )( 2t 1 6 ) .
SOLUTION This function is different because it is the product of three factors instead of two. Nevertheless, the Product Rule may be generalized to work here. If we have a function of the form w( x ) 5 f ( x ) ? g( x ) ? h( x ) , then wr( x ) 5 f r( x ) g( x ) h( x ) 1 gr( x ) f ( x ) h( x ) 1 hr( x ) f ( x ) g( x ) . To find each term
269
of the derivative, we calculate the derivative of each factor and then multiply it by the remaining factors. The derivative of w( x ) is the sum of the individual terms. Rr( t ) 5 5( 4t 1 3 )( 2t 1 6 ) 1 4( 5t 2 1 )( 2t 1 6 ) 1 2( 5t 2 1 )( 4t 1 3 ) 5 5( 8t 2 1 24t 1 6t 1 18 ) 1 4( 10t 2 1 30t 2 2t 2 6 ) 1 2( 20t 2 1 15t 2 4t 2 3 ) 5 5( 8t 2 1 30t 1 18 ) 1 4( 10t 2 1 28t 2 6 ) 1 2( 20t 2 1 11t 2 3 ) 5 ( 40t 2 1 150t 1 90 ) 1 ( 40t 2 1 112t 2 24 ) 1 ( 40t 2 1 22t 2 6 ) 5 120t2 1 284t 1 60
EXAMPLE
Applying the Product Rule in a Real-World Context

Based on data from 1985 to 1999, the per capita consumption of chicken may be modeled by C( t ) 5 0.0026t4 2 0.075t3 1 0.66t2 2 0.47t 1 37 pounds of boneless, trimmed chicken, where t is the number of years since 1985.
Based on data from 1985 to 2000, the average retail price of fresh whole chicken may be modeled by P( t ) 5 0.00073t2 2 0.0097t 1 0.78
dollars per pound, where t is the number of years since 1985. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 706, p. 468.) A whole chicken loses about 30 percent of its weight when it is boned and trimmed. Was per capita spending on chicken increasing or decreasing in 2000? At what rate?
SOLUTION We first need to find a function to model per capita spending on chicken. The units of C( t ) are pounds of boneless, trimmed chicken and the units of P( t ) are dollars per pound of fresh whole chicken. Since 30% of a whole chickens weight is lost when it is boned and trimmed, 1 pound of fresh whole chicken yields 0.7 pound of boneless, trimmed chicken. We define a new function S( t ) 5 C( t ) P( t ) . What are the units of S( t ) ? We have
( C pounds of boneless chicken ) a
P dollars b pound of whole chicken
Dividing the pounds of boneless chicken by 0.7 will yield the number of pounds of whole chicken that were used to get C pounds of boneless chicken and allow us to cancel units. Thus a C P dollars 10 pounds of whole chicken b a b 5 CP dollars 0.7 pound of whole chicken 7
Therefore, the units of S( t ) are dollars. S( t ) 5 5 10 C( t ) ? P( t ) 7 10 ( 0.0026t4 2 0.075t 3 1 0.66t 2 2 0.47t 1 37 )( 0.00073t 2 2 0.0097t 1 0.78 ) 7 We are asked to calculate the instantaneous rate of change of S( t ) in 2000. That is, we are to find Sr( 15 ) . We know from the Product Rule that
270
S r( t ) 5
10 dollars 3Cr( t ) ? P( t ) 1 Pr( t ) ? C( t ) 4 7 year
Although we could calculate Sr( 15 ) algebraically, the complexity of S( t ) would make the process extremely tedious. In cases such as this, it is appropriate to use the power of a graphing calculator, as demonstrated in the following Technology Tip. Using the Technology Tip, we determine that Sr( 15 ) < 5.33 dollars year
In 2000, per capita spending on chicken was increasing by $5.33 per year.
Calculating a Derivative with nDeriv( 1. Enter the function equation using the Y editor. (In this case, we entered C in Y1, P in Y2, and S in Y3. We input the variables Y1 and Y2 into equation Y3 by using the VARS Y-VARS; 1:Function key sequence and selecting the respective functions.) 2. Press 2ND MODE to quit and return to the home screen. Then press MATH and select 8:nDeriv(. Then press ENTER .
3. The nDeriv( function requires three input values: the function name, the variable of differentiation, and the value of the variable. Enter the appropriate values.
4. Press
ENTER
to display the result.
Although any quotient of two functions may be rewritten as a product, using the Quotient Rule to find the derivative of a quotient is often easier than using the Product Rule.
271
QUOTIENT RULE
The derivative of a function h( x ) 5 hr( x ) 5
f (x) is given by g(x)
f r( x ) ? g( x ) 2 gr( x ) ? f ( x ) 3g( x ) 4 2
EXAMPLE
Finding the Derivative of a Quotient of Functions

Find dy for y 5 dx
x2 1 1 . x
SOLUTION
dy ( 2x )( x ) 2 ( 1 )( x2 1 1 ) 5 dx (x)2 5 5 2x2 2 x2 2 1 x2 x2 2 1 x2
EXAMPLE
Finding the Derivative of a Quotient of Functions

Differentiate y 5
SOLUTION
6x 1 1 . 3x 2 1
yr 5 5 5
( 6 )( 3x 2 1 ) 2 ( 3 )( 6x 1 1 ) ( 3x 2 1 ) 2 18x 2 6 2 18x 2 3 ( 3x 2 1 ) 2 29 ( 3x 2 1 ) 2
272
Calculating a Derivative Numerically Follow the steps for creating a difference quotient. Then select a very small value of h (e.g., h 5 0.00001). For small values of h, the difference quotient approximates the derivative.
4.2 Summary
In this section, you learned how to use the Product and Quotient Rules to find the derivative of a function that is written as a product or quotient of factors. You also learned how technology may be used to differentiate functions with complex factors.

In Exercises 18, simplify the expression. 1. 4x( 5x2 1 2x 1 7 ) 1 ( 10x 1 2 )( 2x2 ) 2. 2 x22( 3x2 2 9 ) 1 ( 6x )( x21 ) 3. ( 14x 1 12 )( x3 2 1 ) 1 ( 7x2 1 12x 2 4 )( 3x2 ) 4. 4x( 6x2 2 7 )( x4 1 9 ) 1 ( 12x )( 2x2 )( x4 1 9 ) 1 4x3( 2x2 )( 6x2 2 7 ) 5. 6x( x2 2 1 ) 2 2x( 3x2 ) ( x2 2 1 ) 2 8. 6. 7. 4x( 5x2 1 2x 1 7 ) 2 ( 10x 1 2 )( 2x2 ) ( 5x2 1 2x 1 7 ) 2 2 x22( x2 1 5x 1 1 ) 2 ( 2x 1 5 )( x21 ) x2 1 5x 1 1 9x2( x2 2 17x 1 2 ) 2 ( 2x 2 17 )( 3x3 1 2 ) ( x2 2 17x 1 2 ) 2
4.2 Exercises
In Exercises 110, use the Product or the Quotient Rule to find the derivative of the function. Dont simplify the result. 1. f ( x ) 5 2x( 3x 1 4 ) 4. s( t ) 5 2t3( 3t2 1 7t ) 5. w( n ) 5 ( 3n2 1 8 )( n3 2 n ) 2. g( x ) 5 5x( 9x 2 2 ) 3. f ( t ) 5 ( 2t 2 6 )( 10t 1 5 ) 10. g( x ) 5 6. D( p ) 5 ( p2 2 7p )( 1 2 p3 ) 7. Q( t ) 5 ( t2 1 2t 1 1 )( t2 1 4t 1 4 ) 8. P( t ) 5 t 2 1 5t 1 6 t 2 1 5t 1 4 x3 1 5x2 x 2 4x 1 2
2
9. f ( x ) 5
x3 2 1 x4 2 x3
273
In Exercises 1120, determine the slope of the graph at the indicated domain value. 11. h( t ) 5 12. g( t ) 5 13. s( t ) 5 14. p( t ) 5 t 2 1 5t 1 6 ; t 5 22 t t 3 2 12t 2 ;t51 t t 2 6t 1 8 ; t 5 21 2t
4 2
thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
In 1999, at what rate was employer spending on rubber and plastics manufacturing industry employee earnings increasing? 23.
Employer Labor Costs Based on data
7t 2 2 2t 1 9 ;t52 t
from 1995 to 1999, the average annual earnings in the paper and allied products manufacturing industry may be modeled by E( t ) 5 1335t 1 39,408 dollars per employee, and the average number of employees in the same industry may be modeled by N( t ) 5 2 t3 1 5.571t2 2 12.29t 1 684.9 thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
15. f ( x ) 5 2x( 3x 1 4 )( 5x 1 7 ) ; x 5 0 16. R( p ) 5 5p( 3p 1 1 )( 4p 2 1 ) ; p 5 1 17. r( x ) 5 ( 2x 1 11 )( x 2 4 )( 7x 1 3 ) ; x 5 4 18. T( n ) 5 ( n2 1 4 )( 2n 2 8 )( n 1 9 ) ; n 5 3 19. f ( x ) 5 ( x 1 1 )( x 2 1 )( x2 1 1 ) ; x 5 2 1 20. g( x ) 5 ( x 2 2 )( x 1 2 )( x2 1 4 ) ; x 5 2 In Exercises 2125, use the Technology Tip to answer the questions from the real-life scenarios. 21.
In 1999, at what rate was employer spending on paper and allied products manufacturing industry employee earnings increasing? 24.
from 1995 to 1999, the average annual earnings in the lumber and wood products manufacturing industry may be modeled by S( t ) 5 2 63.57t 2 1 1253t 1 25,066 dollars per employee, and the average number of employees in the lumber and wood products manufacturing industry may be modeled by E( t ) 5 3.143t2 1 5.029t 1 772.5 thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
from 1995 to 1999, the average annual earnings in the apparel and other textile products manufacturing industry may be modeled by E( t ) 5 1118t 1 18,729 dollars per employee, and the average number of employees in the same industry may be modeled by N( t ) 5 2 3.5t3 1 20.21t2 2 85.86t 1 918.2 thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
In 1999, at what rate was employer spending on apparel and other textile products manufacturing industry employee earnings changing? 25.
In 1999, at what rate was employer spending on lumber and wood products manufacturing industry employee earnings increasing? 22.
from 1995 to 1999, the average annual earnings in the rubber and plastics manufacturing industry may be modeled by E( t ) 5 1196t 1 29,825 dollars per employee, and the average number of employees in the rubber and plastics manufacturing industry may be modeled by N( t ) 5 2 2.917t3 1 15.29t2 2 10.73t 1 962.9
from 1995 to 1999, the average annual earnings in the printing and publishing industry may be modeled by E( t ) 5 88.36t 2 1 1291t 1 34,528 dollars per employee, and the average number of employees in the printing and publishing industry may be modeled by N( t ) 5 2 4.417t3 1 24.93t2 2 25.30t 1 1450 thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
274
In 1999, at what rate was employer spending on printing and publishing industry employee earnings increasing?
26. A classmate erroneously claims that d 3 f ( x ) ? g( x ) 4 5 f r( x ) ? gr( x ) dx Explain, using examples as appropriate, why her claim is incorrect. 27. A classmate claims that d f (x) c d 5 f r( x ) 3g( x ) 4 21 2 3g( x ) 4 22gr( x ) f ( x ) dx g( x ) Does the Quotient Rule support or refute this claim? 28. When constructing a product of functions in a real-world context, why is it important to analyze the units of each function in the product? 29. Given the function y 5 ( 3x 1 5 )( 2x 2 1 ) , do you think it is more efficient to use the Product Rule to find the derivative or to multiply out the product and then use the basic derivative rules? Explain. 30. Is the expression f r( x ) g( x ) 1 f ( x ) gr( x ) equivalent to the expression f ( x ) gr( x ) 1 g( x ) f r( x ) ? Explain.
Suppose that a farmer has an apple orchard with 40 trees per acre. The orchard yields 10 bushels per tree. The farmer estimates that for each additional tree planted per acre, the average yield per tree is reduced by 0.1 bushel. If y 5 f ( x ) is the total number of bushels of apples produced per acre when an additional x trees per acre are planted, calculate and interpret the meaning of f r( 25 ) , f r( 30 ) , and f r( 35 ) . 32. Apple Farming Suppose that a farmer has an apple orchard with 30 trees per acre. The orchard yields 12 bushels per tree. The farmer estimates that for each additional tree planted per acre, the average yield per tree is reduced by 0.1 bushel. If y 5 f ( x ) is the total number of bushels of apples produced per acre when an additional x trees per acre are planted, calculate and interpret the meaning of f r( 40 ) , f r( 45 ) , and f r( 50 ) . 33. Based on the results of Exercises 31 and 32, do you think that planting a particular number of trees will maximize the yield per acre? Justify your answer. 34. Apple Supplier Prices A fruit farmer sells apples to a grocery store chain. The amount of apples the store buys depends linearly upon the price per pound that the farmer charges. The farmer estimates that for every $0.02 per pound increase in the price, the store will reduce its order by 44 pounds. The store currently orders 440 pounds per week and pays $0.18 per pound. What price should the farmer charge in order to maximize her revenue from apple sales? 35. Apple Retailer Prices A grocery store has priced apples at $0.65 per pound and sells 1000 pounds per week. The amount of apples the store sells depends linearly upon the price per pound that the store charges. The store manager estimates that for every $0.05 per pound increase in the price, the store will reduce its sales by 100 pounds. What price should the store charge for its apples in order to maximize revenue from apple sales?
31.
Apple Farming Historically, many
apple farmers spaced trees 40 feet by 40 feet apart (27 trees per acre). Trees typically took 25 years to reach their maximum production of 500 bushels per acre. (A bushel is about 44 pounds.) In recent years, agriculturists have created dwarf and semidwarf varieties that allow trees to be spaced 10 feet by 10 feet apart.
(Source: USDA.)
4.3 The Chain Rule
275
4.3 The Chain Rule
Use the Chain Rule to find the derivative of a function modeled by
Based on data from 1995 to 1999, the average annual earnings of an employee in the lumber and wood products industry may be
S( n ) 5 2 0.7685n2 1 1295n 2 516,596 dollars where n is the number of employees in thousands. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 979, p. 622.) Based on data from 1995 to 1999, the number of employees in the lumber and wood products industry may be modeled by n( t ) 5 3.143t 2 1 5.029t 1 772.5 thousand employees where t is the number of years since 1995. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 979, p. 622.) How quickly was the average annual earnings of an employee increasing in 1998? We will answer this question using the Chain Rule. In this section, we will demonstrate how to find the composition of two functions. We will then demonstrate how to use the Chain Rule to find the derivative of a composition of functions.
Composition of Functions
A composition of two functions occurs when the outputs of one function are the inputs of a second function. Lets consider a common example. Let f represent the freezer function and b represent the blender function.
f (w)
b(f(w))
FIGURE 4.1
b( f ( w )) means that first w is plugged into the function f, and then the result is substituted into b. In our example, water was placed in the freezer, and the result, ice, was put in the blender.
276
EXAMPLE
Finding a Composition of Functions

Let f ( t ) 5 3t 1 2 and g( t ) 5 t 2 1 1. Find f ( g( t )) .
SOLUTION
f ( g( t )) 5 f ( t 2 1 1 ) 5 3( t 1 1 ) 1 2
2
The output of the function g is the input for f The variable in f is replaced with the output of g
5 3t 1 3 1 2
2
5 3t 2 1 5
Is f ( g( t )) the same as g( f ( t )) ? Lets return to the freezer and blender example and see.
b (w)
f(b(w))
FIGURE 4.2
When we reversed the order of the functions, we got a different result. Ice is not the same as chopped ice. There are certain functions for which the two different compositions are equal, but, in general, f ( g( t )) 2 g( f ( t )) .
EXAMPLE
Finding a Composition of Functions

Let f ( t ) 5 3t 1 2 and g( t ) 5 t 2 1 1. Find g( f ( t )) .
SOLUTION
g( f ( t )) 5 g( 3t 1 2 ) 5 ( 3t 1 2 ) 2 1 1 5 9t 2 1 12t 1 4 1 1 5 9t 2 1 12t 1 5
Note that although Examples 1 and 2 used the same functions f and g, f ( g( t )) 2 g( f ( t )) .
4.3 The Chain Rule
277
The Chain Rule

The Chain Rule is used whenever we have to find the derivative of a composition of functions.
CHAIN RULE
The derivative of a function h( x ) 5 f ( g( x )) is given by hr( x ) 5 f r( g( x )) ? gr( x )
The Chain Rule tells us to find f r( x ) , substitute g( x ) in for the variable x, and then multiply the result by gr( x ) .
EXAMPLE
Finding a Derivative by Using the Chain Rule

Given h( x ) 5 ( 2x 1 1 ) 2 2 4, find hr( x ) .
SOLUTION Because there is a nonconstant function ( 2x 1 1 ) inside the parentheses, we know that we will need the Chain Rule. The function h( x ) 5 ( 2x 1 1 ) 2 2 4 is a function of the form h( x ) 5 f ( g( x )) . We must determine the functions f and g. Often, selecting the function inside the parentheses as g gives the desired result. In this case, we pick g( x ) 5 2x 1 1. Then
f ( g( x )) 5 ( 2x 1 1 ) 2 2 4 To find the function f, we simply replace the g( x ) in the equation with the variable x. Thus we have f ( x ) 5 x2 2 4 Well first find f r( x ) and then use it to calculate f r( g( x )) . f r( x ) 5 d 2 (x 2 4) dx
By substitution
f ( g( x )) 5 3g( x ) 4 2 2 4
5 2x f r( g( x )) 5 2 ? g( x ) 5 2( 2x 1 1 ) Next well find gr( x ) . gr( x ) 5 d ( 2x 1 1 ) dx
52 The derivative of f ( g( x )) is f r( g( x )) ? gr( x ) 5 32( 2x 1 1 ) 4 ? ( 2 ) 5 4( 2x 1 1 ) 5 8x 1 4
278
Well rework the problem using a slightly different but somewhat easier approach. h( x ) 5 ( 2x 1 1 ) 2 2 4 dh d d 5 3( 2x 1 1 ) 24 2 (4) dx dx dx
Sum and Difference Rule
To use the Chain Rule to differentiate ( 2x 1 1 ) 2, we treat the expression 2x 1 1 inside the parentheses as if it were a single variable and use the Power Rule. Then we multiply the result by the derivative of 2x 1 1, the expression inside the parentheses. dh d 5 2( 2x 1 1 ) 2 2 1 ? ( 2x 1 1 ) 2 0 dx dx 5 2( 2x 1 1 ) ? ( 2 ) 5 4( 2x 1 1 ) 5 8x 1 4
Chain, Power, and Constant Rules
The second technique shown in Example 3 is a demonstration of the Generalized Power Rule.
GENERALIZED POWER RULE
Let u 5 g( x ) . The derivative of a function f ( x ) 5 un is given by f r( x ) 5 nu n 2 1ur
The Generalized Power Rule is a special application of the Chain Rule. Well use this rule in the next two examples.
EXAMPLE
Finding a Derivative by Using the Generalized Power Rule

Differentiate f ( x ) 5 4( 2x2 1 x ) 3. as f ( x ) 5 4u3. From the Generalized Power Rule, we know that d d ( f ( x )) 5 ( 4u3 ) dx dx 54 d 3 (u ) dx
Constant Multiple Rule Generalized Power Rule SOLUTION We pick u 5 2x2 1 x. Then f ( x ) 5 4( 2x2 1 x ) 3 may be rewritten
f r( x ) 5 12u2ur
4.3 The Chain Rule
279
We need to write the result in terms of x, not u. Since u 5 2x2 1 x and ur 5 4x 1 1, the derivative may be rewritten as f r( x ) 5 12u2ur 5 12 3x( 2x 1 1 ) 4 2( 4x 1 1 ) 5 12x2( 2x 1 1 ) 2( 4x 1 1 ) 5 12( 2x2 1 x ) 2( 4x 1 1 )
5 12( x2 )( 2x 1 1 ) 2( 4x 1 1 )
We will often write derivatives in factored form. In real-life optimization problems, we are often interested in the values of x that make the derivative equal to zero. These values are easiest to find if the function is written in factored form.
EXAMPLE
Finding a Derivative by Using the Generalized Power Rule

Find dy given y 5 dx
1 . 2x3 1 4x 1 1 1 2x3 1 4x 1 1
SOLUTION We first rewrite y 5
as y 5 ( 2x3 1 4x 1 1 ) 21. If
we pick u 5 2x3 1 4x 1 1, then y 5 ( 2x3 1 4x 1 1 ) 21 may be rewritten as y 5 u21. From the Generalized Power Rule, we know that d d 21 ( y) 5 (u ) dx dx dy 5 2 u22ur dx 52 ur u2
Generalized Power Rule Since u22 5 1 u2
Since u 5 2x3 1 4x 1 1, ur 5 6x2 1 4. Thus dy ur 52 2 dx u 52 52 6x2 1 4 ( 2x3 1 4x 1 1 ) 2 2( 3x2 1 2 ) ( 2x3 1 4x 1 1 ) 2
As demonstrated in the next example, it is necessary to apply the Chain Rule repeatedly for some functions.
EXAMPLE
Finding a Derivative by Using the Chain Rule Repeatedly

Find the derivative of S( t ) 5 33( 2t 1 1 ) 2 1 54 3.
280
an expression inside parentheses. If we let v 5 2t 1 1, then u 5 3v2 1 5. S( t ) 5 u3 dS 5 3u2ur dt 5 3u2 d (u) dt d ( 3v2 1 5 ) dt d (v) dt d ( 2t 1 1 ) dt
SOLUTION We let u 5 3( 2t 1 1 ) 2 1 5. Notice that this function also contains
Generalized Power Rule Since ur 5 d (u) dt
5 3( 3v2 1 5 ) 2 ?
Since u 5 3v2 1 5 Generalized Power Rule Since vr 5 d (v) dt
5 3( 3v2 1 5 ) 2 ? 6vvr 5 3( 3v2 1 5 ) 2 ? 6v ?
5 3 33( 2t 1 1 ) 2 1 54 2 ? 6( 2t 1 1 ) ?
Since v 5 2t 1 1
5 3 33( 2t 1 1 ) 2 1 54 2 ? 6( 2t 1 1 ) ? ( 2 )
At this point, we have calculated the derivative. If we want to rewrite the derivative in its simplest factored form, we continue as follows: dS 5 3 33( 2t 1 1 ) 2 1 54 2 ? 6( 2t 1 1 ) ? ( 2 ) dt 5 36 33( 2t 1 1 ) 2 1 54 2 ( 2t 1 1 ) 5 36 33( 4t 2 1 4t 1 1 ) 1 54 2 ( 2t 1 1 ) 5 36 ( 12t 2 1 12t 1 3 1 5 ) 2 ( 2t 1 1 ) 5 36 ( 12t 2 1 12t 1 8 ) 2 ( 2t 1 1 ) 5 36 34( 3t 2 1 3t 1 2 ) 4 2 ( 2t 1 1 ) 5 576 ( 3t 2 1 3t 1 2 ) 2 ( 2t 1 1 )
5 36 ( 4 ) 2 ( 3t 2 1 3t 1 2 ) 2 ( 2t 1 1 )
Writing the derivative in factored form required a significant number of additional steps. Although there are some instances that require us to write derivatives in factored form, we will consider the additional steps optional at this point.
Chain Rule: Alternative Form

The Chain Rule may also be expressed using Leibniz notation.
CHAIN RULE: ALTERNATIVE FORM
If y 5 f ( x ) and x 5 g( t ) , then y 5 f ( g( t )) and dy dx dy 5 ? dt dx dt The units of dx dy dy are obtained by multiplying the units of by the units of . dt dx dt
4.3 The Chain Rule
281
EXAMPLE
Finding a Derivative by Using the Alternative Form of the Chain Rule

Based on data from 1989 to 1998, restaurant sales in the United States may be modeled by S( r ) 5 0.9291r 2 160.0 billion dollars where r is the number of restaurants (in thousands). The number of restaurants may be modeled by r( t ) 5 0.3405t 2 1 6.910t 1 330.9 thousand restaurants where t is the number of years since 1989. (Source: Modeled from Statistical Abstract
How quickly were restaurant sales increasing in 1998? dS . Recall that the alternative form of the dt dS dS dr dS dr 5 ? . We will first find Chain Rule states that and . dt dr dt dr dt
SOLUTION We are asked to find
S( r ) 5 0.9291r 2 160.0 dS billion dollars 5 0.9291 dr thousand restaurants r( t ) 5 0.3405t2 1 6.910t 1 330.9 dr thousand restaurants 5 0.681t 1 6.910 dt year dS dS dr 5 ? dt dr dt 5 a 0.9291 billion dollars thousand restaurants b ? a 0.681t 1 6.910 b thousand restaurants year billion dollars thousand restaurants thousand restaurants year
5 ( 0.9291 ) ? ( 0.681t 1 6.910 )
5 ( 0.9291 ) ? ( 0.681t 1 6.910 ) billion dollars per year < 0.6327t 1 6.420 billion dollars per year We need to determine the rate of change in sales for 1998 ( t 5 9 ) . dS 5 0.6327( 9 ) 1 6.420 ` dt t 5 9 < 12 . 11 billion dollars per year dS dS ` means the value of when t 5 9.) dt t 5 9 dt In 1998, restaurant sales in the United States were increasing at an estimated rate of 12.11 billion dollars per year. (Note: The notation
Well now return to the problem introduced at the beginning of this section.
282
EXAMPLE
Finding a Derivative by Using the Alternative Form of the Chain Rule

Based on data from 1995 to 1999, the average annual earnings of an employee in the lumber and wood products industry may be modeled by S( n ) 5 2 0.7685n2 1 1295n 2 516,596 dollars where n is the number of employees in thousands. (Source: Modeled from
Based on data from 1995 to 1999, the number of employees in the lumber and wood products industry may be modeled by n( t ) 5 3.143t 2 1 5.029t 1 772.5 thousand employees where t is the number of years since 1995. (Source: Modeled from Statistical
How quickly was the average annual earnings of an employee increasing in 1998?
SOLUTION We are asked to find
will first find
dS dn and . dn dt
dS dS dS dn 5 ? . We . By the Chain Rule, dt dt dn dt
S( n ) 5 2 0.7685n2 1 1295n 2 516,596 dS dollars 5 2 1.537n 1 1295 dn thousand employees n( t ) 5 3.143t 2 1 5.029t 1 772.5 dn thousand employees 5 6.286t 1 5.029 dt year dS dS dn 5 ? dt dn dt 5 a 2 1.537n 1 1295 thousand employees dollars b ? a 6.286t 1 5.029 b thousand employees year dollars thousand employees thousand employees year
5 ( 2 1.537n 1 1295 ) ? ( 6.286t 1 5.029 )
< ( 2 4.8308t 2 2 7.7296t 2 1187.3325 1 1295 ) ? ( 6.286t 1 5.029 ) dS ` , since t 5 3 in the year 1998. dt t 5 3
5 32 1.537( 3.143t 2 1 5.029t 1 772.5 ) 1 12954 ? ( 6.286t 1 5.029 ) dollars per year
5 ( 2 4.8308t 2 2 7.7296t 1 107.6675 ) ? ( 6.286t 1 5.029 ) dollars per year We must find
dS 5 32 4.8308( 3 ) 2 2 7.7296( 3 ) 1 107.66754 ? 36.286( 3 ) 1 5.0294 ` dt t 5 3 5 ( 2 43.4772 2 23.1888 1 107.6675 ) ? ( 23.887 ) 5 ( 41.0015 ) ? ( 23.887 ) < 979.40 dollars per year
4.3 The Chain Rule
283
According to the model, the average annual earnings of a lumber and wood products industry employee was increasing at a rate of roughly $979 per year in 1998.
4.3 Summary
In this section you learned how to compose two functions. You also discovered how to use the Chain Rule to find the derivative of a composition of functions.

In Exercises 14, simplify the expression. 1. 4( 2x 1 1 ) ( 2 )
3
In Exercises 58, evaluate the expression. 5. 4 32( 2 2 ) 1 14 3 ( 2 ) 6. 2 3 36( 3 ) 2 2 4( 3 ) 4 24 312( 3 ) 2 44
2. 2 3( 6x2 2 4x ) 24 ( 12x 2 4 ) 3. 3( 2x2 1 7x 2 11 ) 2 ( 4x 1 7 ) 4. 2 2( 4x2 1 6x 2 5 ) 23 ( 8x 1 6 )
7. 3 32( 5 ) 2 1 7( 5 ) 2 114 2 34( 5 ) 1 74
8. 2 2 34( 2 1 ) 2 1 6( 2 1 ) 2 54 23 38( 2 1 ) 1 64
4.3 Exercises
In Exercises 15, find the function f ( g( x )) . Then simplify the result. 1. f ( x ) 5 2x 1 3, g( x ) 5 x 1 4
2
12. g( t ) 5 4( 6t 2 7 ) 3 1 2 13. h( t ) 5 3( t 3 2 t 2 1 1 ) 4 14. S( n ) 5 ( n3 2 3n2 ) 2 15. f ( n ) 5 5( n2 1 2n 1 1 ) 2 1 6 16. g( x ) 5 34( 3x 1 5 ) 2 1 x4 2 18. s( t ) 5 ( 2t 1 1 ) 2( 6t ) 17. f ( x ) 5 32 5( 4x 2 1 ) 2 1 2x4 2 19. f ( t ) 5 ( 5t 2 1 t )( 3t 1 2 ) 2 20. h( x ) 5 ( 4x 1 1 ) 2( 2 x 1 2 ) 2 In Exercises 2125, determine the slope of the graph at the indicated domain value. 21. g( x ) 5 22. r( x ) 5 23. R( x ) 5 x2 1 3 ;x52 x2 2 x 2x2 2 8 ;x50 2x2 1 8 3x 1 1 ;x52 2x 1 5
2. f ( x ) 5 2 2x 1 7x, g( x ) 5 x 2 10
2
3. f ( x ) 5 7x 2 4, g( x ) 5 x 2 2x 1 1
2
4. f ( x ) 5 x 2 1, g( x ) 5 x 1 1
2 2
5. f ( x ) 5 3x 2 11x, g( x ) 5 x
2
In Exercises 610, the function is of the form h( x ) 5 f ( g( x )) . Determine f ( x ) and g( x ) . 6. h( x ) 5 5( x 2 4 ) 1 1

2 3
7. h( x ) 5 ( x 1 1 ) 2 2 6( x 1 1 ) 1 9 8. h( x ) 5 e2x 2 4
3 2 3
9. h( x ) 5 2 ln( x2 1 4 )
10. h( x ) 5 4( x 2 1 ) 2 x 1 1 In Exercises 1120, use the Chain Rule in finding the derivative of the function. 11. f ( x ) 5 5( 3x 1 1 ) 2
284
24. P( t ) 5 ( 3t 2 1 ) 23( 5t 2 2 ) ; t 5 0.4 25. f ( x ) 5 35( 4x 1 1 ) 2 1 24 2; x 5 2 1 33. An individuals gross pay g (in dollars) is a function of the number of hours worked x. The individuals take-home pay t (in dollars) is a function of the gross pay g. In this context, what do each of the following derivatives represent: dg dt dt , , ? dx dg dx 34. A classmate claims that the Quotient Rule is unnecessary because any quotient may be rewritten as a product, and then the Product and Chain Rules may be used to find the derivative of the resultant function. Check the accuracy of your f (x) classmates statement by rewriting as a g( x ) product and using the Product and Chain Rules to find the derivative. 35. Explain how function composition is related to the Chain Rule. 36. Given y 5 f ( x ) and x 5 g( t ) , show that the following two expressions are equivalent: dy dx f r( g( t )) gr( t ) and ? . dx dt 37. The Leibniz notation form of the Chain Rule is dy dy dx 5 ? . Explain why this notation is often dt dx dt useful when calculating a derivative in a realworld context.
In Exercises 2630, find the equation of the tangent line at the indicated domain value. 26. R( n ) 5 35( 2 n 1 5 ) 3 2 2n4 2; n 5 5 27. s( t ) 5 ( 3t 1 1 ) 2 32( 2 4t 2 1 ) 1 64 22; t 5 0 28. g( x ) 5 35x( x2 1 1 ) 3 1 ( x 1 1 ) 34 2; x 5 2 1 29. p( x ) 5 53( x 1 1 ) 2 1 24 3 1 36 4; x 5 2 1 30. r( t ) 5 53( 2t 1 1 ) 2 2 14 3 1 26 4; t 5 0
In Exercises 3132, use the Chain Rule to answer the questions. 31.
Average Height of a Girl The average height of a 2- to 13-year-old girl may be modeled by F( m ) 5 1.071m 2 3.554 inches where m is the average height of a boy of the same age. The average height of a 2- to 13-year-old boy may be modeled by m( a ) 5 2 0.0507a2 1 2.997a 1 30.44 inches where a is the age of the boy in years. (Source:
According to the models, how quickly is the height of a girl increasing when she is 10 years old? 32.
Average Weight of a Boy The average
weight of a 2- to 13-year-old boy may be modeled by M( f ) 5 0.9099f 1 5.065 pounds where f is the average weight of a girl of the same age. The average weight of a 2- to 13-year-old girl may be modeled by f ( a ) 5 0.289a2 1 2.464a 1 23.10 pounds where a is the age of the girl in years. (Source:
38. Find the derivative of f ( x ) 5 3( x2 1 1 ) 2 1 14 2 1 1. 40. Given that
39. Differentiate g( t ) 5 ( t 2 1 1 ) 3( t 2 1 3 ) 4. dy 5 5 people per day, dt
According to the models, how quickly is the average weight of a boy increasing when he is 10 years old?
dt 5 2 days per year, and dx dp 5 4 sales per person, calculate dy dp , including units. dx
4.4 Exponential and Logarithmic Rules
285
Use the Exponential Rule and Logarithmic Rule to find the derivative of a function
Despite the infamous Got Milk? advertising campaign, the per capita milk beverage consumption decreased from nearly 28 gallons per year in 1980 to less than 24 gallons per year in 1999. The annual per capita milk beverage consumption may be modeled by M( t ) 5 27.76( 0.9914 ) t gallons where t is the number of years since 1980. Does the four-gallon decrease over the 19-year period indicate that consumer consumption of milk will continue to drop? According to the model, at what rate was consumption decreasing in 2000? Questions such as these may be answered using the Exponential Rule for derivatives. In this section, we will introduce the Exponential Rule and the Logarithmic Rule for derivatives. These rules will allow us to calculate the exact derivative of these types of functions quickly and accurately instead of relying upon a numerical estimate, as we have done previously.
Exponential Rule
Recall that an exponential function may be written in the form y 5 abx, where a and b are constants and b is positive and not equal to 1. Since the variable of an exponential function is in the exponent, the Power Rule cannot be used to find the derivative of this type of function.
EXPONENTIAL RULE
The derivative of a function f ( x ) 5 bx is given by f r( x ) 5 ( ln b ) bx
This rule is not entirely obvious. To better understand its origin, lets consider the limit definition of the derivative of an exponential function. f r( x ) 5 lim 5 lim 5 lim 5 lim
hS0
f (x 1 h) 2 f (x) h bx 1 h 2 bx h bxbh 2 bx h bx( bh 2 1 ) h ( bh 2 1 ) h

Since bx is independent of the value of h Since f ( x ) 5 bx Algebraic rules of exponents
hS0
hS0
hS0
5 bx lim
hS0
286
CHAPTER 4 Differentiation Techniques ( bh 2 1 ) h hS0
Although we cannot simplify lim

hS0
algebraically, it turns out that
lim
( bh 2 1 ) h
5 ln( b ) . Consequently, f r( x ) 5 bx lim

hS0
( bh 2 1 ) h
5 ( ln b ) bx Exponential functions frequently use the number e as the base. Recall that e < 2.7182818.
EXAMPLE
Finding the Derivative of an Exponential Function

Differentiate y 5 e x.
SOLUTION
dy 5 ( ln e ) e x dx 5 1 ? ex 5e
x
Since ln e 5 1
The derivative of e x is e x! This quirky result makes the function y 5 ex one of the most popular functions in calculus. For the function f ( x ) 5 ex, the instantaneous rate of change in the function at a point ( a, f ( a )) is the same as f ( a ) . In other words, the slope of the tangent line to the graph of f ( x ) 5 ex at x 5 a is f (a).
EXPONENTIAL RULE: SPECIAL CASE
The derivative of a function f ( x ) 5 ex is given by f r( x ) 5 ex
EXAMPLE

Find the derivative of g( x ) 5 2x.
SOLUTION
gr( x ) 5 ( ln 2 ) 2x
287
EXAMPLE

Calculate d ( 3 ? 4x ) . dx
SOLUTION
d d ( 3 ? 4x ) 5 3 ? ( 4x ) dx dx 5 3( ln 4 ) 4x
Constant Multiple Rule Exponential Rule
EXAMPLE
Finding a Derivative by Using the Exponential and Chain Rules

Find yr given y 5 3x x.
2
SOLUTION This function is different because there is a function in the exponent rather than a single variable. Consequently, well have to use the Chain Rule in conjunction with the Exponential Rule to find the derivative.
yr 5 ( ln 3 ) 3 x x ?
2 2
d 2 (x 1 x) dx
x 2x
Exponential and Chain Rules Power Rule
5 ( ln 3 ) 3 x x ? ( 2x 1 1 ) 5 ( 2x 1 1 )( ln 3 ) 3
EXAMPLE 5
Finding a Derivative by Using the Exponential and Chain Rules

Differentiate y 5 e5x 2 6.
SOLUTION
dy d 5 ( ln e ) e5x 2 6 ? ( 5x 2 6 ) dx dx 5 ( 1 ) e5x 2 6 ? ( 5 ) 5 5e5x 2 6
Exponential and Chain Rules Constant, Constant Multiple, and Power Rules
In general, a function of the form y 5 eu, where u is some function of x, will have the derivative yr 5 eu ? ur. This is referred to as the Generalized Exponential Rule; it is a special application of the Chain Rule together with the Exponential Rule.
GENERALIZED EXPONENTIAL RULE
Let u 5 g( x ) . The derivative of a function f ( x ) 5 bu is given by f r( x ) 5 ( ln b ) buu r If b 5 e, the rule simplifies to f r( x ) 5 euu r
288
EXAMPLE
Finding a Derivative by Using Multiple Derivative Rules

Find the derivative of P( t ) 5 2te3t.
SOLUTION Although this function looks relatively simple, there is a lot going on. To find the derivative, we must use the Constant Multiple Rule, the Power Rule, the Product Rule, and the Generalized Exponential Rule.
dP d d 5 ( 2t ) ? e3t 1 ( e3t ) ? ( 2t ) dt dt dt 5 2e3t 1 ( e3t ? 3 ) ? ( 2t ) 5 2e3t 1 6te3t 5 e3t( 2 1 6t ) 5 2e3t( 3t 1 1 )
Product Rule Constant Multiple, Generalized Exponential, and Power Rules
EXAMPLE
Finding a Derivative by Using the Generalized Exponential Rule

The annual per capita milk beverage consumption may be modeled by M( t ) 5 27.76( 0.9914 ) t gallons where t is the number of years since 1980. According to the model, at what rate was consumption decreasing in 2000?
SOLUTION
dM 5 27.76 3ln( 0.9914 ) 4 ( 0.9914 ) t dt < 2 0.2398( 0.9914 ) t gallons per year In 2000, t 5 20, so dM ` 5 2 0.2398( 0.9914 ) 20 gallons per year dt t 5 20 < 2 0.2398( 0.8414 ) < 2 0.2018 In 2000, the per capita milk beverage consumption was decreasing by about twotenths (0.2) of a gallon per year. It is important to note that although the per capita consumption of milk was decreasing, overall milk beverage sales still could have increased as a result of an increase in the U.S. population.
Evaluate ln(0.9914) and multiply by 27.76
Logarithmic Rule
What is the derivative of a function of the form y 5 logb x? The answer is not obvious. We will state the rule here and show how the formula was determined when we discuss implicit differentiation in a subsequent section.
289
LOGARITHMIC RULE
The derivative of a function f ( x ) 5 logb x is given by f r( x ) 5 1 1 ? ln b x
EXAMPLE
Finding the Derivative of a Logarithmic Function

Differentiate y 5 ln x.
SOLUTION Recall that ln x 5 loge x.
yr 5 5 5
1 1 ? ln e x 1 1 ? 1 x 1 x
EXAMPLE
Finding the Derivative of a Logarithmic Function

Find the derivative of y 5 3 log t.
SOLUTION Recall that log t means log10 t.
dy d 5 ( 3 log t ) dt dt 53? 53? 5 d ( log10 t ) dt 1 1 ? ln 10 t

Constant Multiple Rule Logarithmic Rule
3 1 ? ln 10 t
GENERALIZED LOGARITHMIC RULE
The derivative of a function f ( x ) 5 ln u where u is a function of x is given by f r( x ) 5 1 ur. u
The Generalized Logarithmic Rule is simply a combination of the Logarithmic Rule and the Chain Rule.
290
EXAMPLE 10
Finding the Derivative by Using the Generalized Logarithmic Rule

Find dy for y 5 log( x2 ) . dx
SOLUTION Since y is the composition of f ( x ) 5 log x and g( x ) 5 x2, we
must use the Chain Rule in conjunction with the Logarithmic Rule. 1 1 dy d 5a ? b ? ( x2 ) dx ln 10 x2 dx 5a 5 5 1 1 ? 2 b ? ( 2x ) ln 10 x
Generalized Logarithmic Rule Power Rule
2x 1 ? ln 10 x2 2 ( ln 10 ) x
EXAMPLE 11
Finding a Derivative by Using Multiple Rules

Differentiate P( t ) 5 e2t ln( 3t ) .
SOLUTION We must use the Product Rule, the Chain Rule, the Exponential Rule, the Constant Multiple Rule, and the Power Rule in finding the derivative.
Pr( t ) 5
d 2t d ( e ) ln( 3t ) 1 3ln( 3t ) 4 e2t dt dt 1 ? 3 b e2t 3t
Product Rule
5 ( e2t ? 2 ) ln( 3t ) 1 a 5 2e2t ln( 3t ) 1 a
Generalized Exponential, Logarithmic, Constant Multiple, and Power Rules
3 2t be 3t
1 5 e2t c 2 ln( 3t ) 1 d t
EXAMPLE 12
Applying the Logarithmic Rule in a Real-World Context

The year in which the per capita consumption of bottled water will reach w gallons may be modeled by T( w ) 5 2 9.249 1 9.839( ln w ) years since 1980 interpret the meaning of T r( 20 ) .
(Source: Statistical Abstract of the United States, 2001, Table 204, p. 130.) Calculate and
291
SOLUTION
T r( w ) 5 9.839 ? T r( 20 ) 5
1 w
Constant and Logarithmic Rules
9.839 years since 1980 20 gallon
5 0.4920 year per gallon When the annual per capita consumption of bottled water is 20 gallons, it will take about 0.4920 year (roughly 6 months) for the annual per capita consumption to increase to 21 gallons.
EXAMPLE 13
Applying the Logarithmic Rule in a Real-World Context

Based on data from 1989 to 1999, the percentage of auto fatalities that are not alcohol-related may be modeled by F( t ) 5 49.96 1 5.112( ln t ) percent where t is the number of years since 1989. (Source: Statistical Abstract of the United States, 2001, Table 1099, p. 688.) According to the model, how quickly was the percentage of non-alcohol-related fatalities increasing in 2000?
SOLUTION
dF d 5 349.96 1 5.112( ln t ) 4 dt dt 5 0 1 5.112 ? 5 1 t

Constant, Constant Multiple, and Logarithmic Rules
5.112 percentage points t year 5.112 percentage points dF ` 5 dt t 5 11 11 year < 0.4647 percentage point per year
In 2000, t 5 11, so
In 2000, the percentage of non-alcohol-related auto fatalities was increasing by 0.46 percentage point per year. This means that alcohol-related auto fatalities were declining by about 0.46 percentage point per year in 2000, a good sign.
4.4 Summary
In this section, you learned how to use the Exponential Rule and the Logarithmic Rule for derivatives. You discovered that these rules allow you to calculate the exact derivative of exponential and logarithmic functions instead of having to rely on a numerical estimate.
292

In Exercises 16, simplify the expression. 1. 1 1 ? 2 ? 2x ln 5 x 1 4 5. 2 ln 3( 32t )( t 2 2 1 ) 1 ( 2t )( 32t ) 6. ln 4( 4t 1 1 )( 2t )( 4t ) 1 ln 4( 4t )( 4t 1 1 )
2 2
1 1 ? ?4 2. ln 3 4x 1 8 1 1 ? 3 ? ( 3x 2 1 4 ) 3. ln 10 x 1 4x 2 7 4. 2e2t 3ln( 3t ) 4 1 a 1 b ( 3 )( e 2t ) 3t
In Exercises 78, evaluate the expression.
7. 2 ln 3( 32(0) ) 3( 0 ) 2 2 14 1 32( 0 ) 4 ( 32(0) ) 8. 2 ln 3( 32(1) ) 3( 1 ) 2 2 14 1 32( 1 ) 4 ( 32(1) )
4.4 Exercises
In Exercises 110, use the Exponential Rule or the Logarithmic Rule, as appropriate, to find the derivative of the function. 1. y 5 4x 3. s( t ) 5 4 ln t 5. y 5 5e 1 ln x
x
23. g( x ) 5 33 log( 4x ) 4 ( 3x ) ; x 5 1 24. h( x ) 5 ln( 3e5x ) ; x 5 2 1 25. f ( x ) 5 ln x ;x5e x
2. g( t ) 5 3( 5 ) t 4. f ( t ) 5 3 log t 6. P( n ) 5 4n2( 2n ) 8. g( x ) 5 e2x 2 1 10. h( x ) 5 ln( x3 1 3x )
7. f ( x ) 5 5e x ln x 9. y 5 2x25x
In Exercises 2630, answer the questions by using the Exponential Rule or the Logarithmic Rule, as appropriate. 26. Based on data from 1990 to 2000, the average salary of a public elementary or secondary school teacher may be modeled by S( t ) 5 32.05( 1.027 ) t thousand dollars where t is the number of years since 1990.
Teacher Salaries
In Exercises 1120, determine the slope of the function at the indicated domain value. 11. S( n ) 5 log( n2 2 4 ) ; n 5 3 12. P( x ) 5 xe x; x 5 0 13. w( t ) 5 log ( 2t ) ; t 5 2 1 14. f ( x ) 5 4 log( 3x ) ; x 5 0 15. H( t ) 5 3ln t; t 5 e 17. f ( t ) 5 3 log( 5t ) ; t 5 0 18. y 5 log( 5x ) 1 42x 1 5; x 5 1 19. C( n ) 5 3nen22n1; n 5 1 20. g( t ) 5 5t( et 2 2 ) ; t 5 3 In Exercises 2125, determine the equation of the tangent line at the indicated domain value. 21. f ( t ) 5 2
3 ln t
16. y 5 2log 5t; t 5 0.2
At what rate (in thousands of dollars per year) were salaries increasing in 2000? 27. Based on data from 1994 to 2001, the quarterly tuition of a resident student at Green River Community College may be modeled by E( t ) 5 430.6( 1.042 ) t dollars where t is the number of years since 1994.
College Tuition
(Source: Modeled from Green River Community College data.)
How quickly was tuition increasing in 2001?
;t51
22. y 5 5x ( 2 ) ; x 5 0
2 3x
4.5 Implicit Differentiation
293
28.
Based on data from 1980 to 1998, the average salary of a National Basketball Association (NBA) player may be modeled by P( t ) 5 161( 1.17 ) t thousand dollars where t is the number of years since 1980.
NBA Player Salaries
proportionality related to the derivative of an exponential function? 32. The function f ( x ) 5 e x passes through the points (0, 1), (1, e), and (a, b). Use mental math and your knowledge of the derivative to determine the slope of the graph of f ( x ) at x 5 0, x 5 1, and x 5 a. 33. The function g( x ) 5 ln x passes through the points (1, 0), (e, l), and (a, b). Use mental math and your knowledge of the derivative to determine the slope of g( x ) at x 5 1, x 5 e, and x 5 a. 34. Under what conditions, if any, is the derivative of an exponential function equal to zero? 35. A classmate claims that the graph of a logarithmic function f ( t ) 5 logb t for b . 1 is concave down and increasing. Explain how the derivative of the function may be used to support or refute this claim.
Calculate and interpret the meaning of Pr( 18 ) .

VCR Usage
29.
Based on data from 1985 to 1999, the percentage of homes with a VCR may be modeled by V( t ) 5 22.24 1 24.47 ln t percent where t is the number of years since 1984.
Calculate and interpret the meaning of V r( 15 ) . 30.

Army Personnel Based on data from 1980 to 2000, the number of active-duty Army personnel may be modeled by
A( t ) 5
302.2 1 480 thousand people 1 1 0.0001047e0.7731t where t is the number of years since 1980.
36. Differentiate f ( x ) 5 2ln(x2). 37. Find dy given y 5 2x ln( 2x ) . dx
38. Find gr( t ) given g( t ) 5 22t ln( 22t ) . 39. Differentiate h( t ) 5 324t. 40. Find dx given x( t ) 5 ln 3ln( 32t ) 4 . dt
At what rate was the number of active-duty army personnel changing in 2000?
31. A quantity y is said to be exponentially proportional to x if y 5 ka x for some nonzero constant k. How is the concept of exponential
Use implicit differentiation to differentiate functions and nonfunctions
The intent of this section is to give you the skills necessary to work the related-rate problems in the next chapter. We will show you how to use implicit differentiation to find the derivative of functions and nonfunctions. Knowing how to do implicit differentiation is a key skill that is necessary for our discussion of related rates.
294
The equation of a circle may be written as x 2 1 y 2 5 r 2, where r is the radius of the circle. A circle is not a function because it fails the Vertical Line Test. However, we can draw tangent lines to the graph of a circle. Since the derivative is the slope of the tangent line, we should be able to find the derivative at any point on the circle. However, since most x values have two different y values, the derivative will be a function of both x and y. Consider x 2 1 y 2 5 1 and its associated graph (Figure 4.3). What is the slope of each of the tangent lines shown? In other words, what is the derivative at the indicated points?
y ( 0.8, 0.6) (0.5, 0.866)
( 0.8, 0.6)
x2 + y2 = 1
FIGURE 4.3
We will find the derivative by using implicit differentiation. We begin by taking the derivative of both sides of the equation with respect to x. Since y is a d dy function of x, ( y ) 5 . This fact will become important as we move through dx dx the problem. x2 1 y2 5 1 d 2 d ( x 1 y 2) 5 ( 1 ) dx dx d 2 d (x ) 1 ( y2) 5 0 dx dx 2x 1 2y ? d ( y) 5 0 dx dy 50 dx dy 5 2 2x dx dy 2 2x 5 dx 2y x dy 52 y dx Note that the derivative is a function of both x and y. We can now calculate the slope of the tangent line at each of the points shown.
Sum and Difference and Constant Rules Power and Chain Rules
2x 1 2y ? 2y
295
dy 2 0.8 52 < 1.333. dx 0.6 dy 2 0.8 52 < 2 1.333. At ( 2 0.8, 2 0.6 ) , dx 2 0.6 dy 0.5 52 < 2 0.5774. At ( 0.5, 0.866 ) , dx 0.866 At ( 2 0.8, 0.6 ) ,
Steps of Implicit Differentiation

1. Differentiate both sides of the equation with respect to x.
a Recall
d dy ( y) 5 . b dx dx
2. Algebraically isolate the
dy term. dx
EXAMPLE
Finding
dy by Using Implicit Differentiation dx
Differentiate xy 5 1 using implicit differentiation.

SOLUTION
d d ( xy ) 5 (1) dx dx d d (x) ? y 1 (y) ? x 5 0 dx dx 1?y1 dy ?x50 dx dy 50 dx dy 5 2y dx dy y 52 x dx If we first solve the equation for y and then differentiate, we get a very differentlooking result. xy 5 1 y5 1 x
Product Rule Power Rule
y1x x
d d 1 (y) 5 a b dx dx x d dy 5 ( x21 ) dx dx
296
5 2 x22 52 1 x2
Power Rule
Is this answer equivalent to our implicit differentiation result? Lets see. dy y 52 x dx 1 52 y x 1 1 52 ? x x 52 1 x2

Since y 5 x
1
The two equations are equivalent. When using implicit differentiation, it is common for correct answers to look very different from each other.
EXAMPLE
Finding
Differentiate ( x 1 y ) 2 5 9.
SOLUTION
d d (x 1 y)2 5 (9) dx dx 2( x 1 y ) ? d (x 1 y) 5 0 dx dy b 50 dx dy 0 b 5 dx 2( x 1 y ) dy 50 dx dy 5 21 dx For all values of x and y on the graph of ( x 1 y ) 2 5 9, the slope of the tangent line is 21. Lets solve the original equation for y. (x 1 y)2 5 9 "( x 1 y ) 2 5 "9 ( x 1 y ) 5 63 y 5 2x 6 3
Constant and Chain Rules Power Rule
2( x 1 y ) ? a 1 1 a1 1
11
4.5 Implicit Differentiation y 6 5 4 3 2 1 3 2 1 1 2 3 4 5 6 1
297
When we graph y 5 2 x 2 3 and y 5 2 x 1 3 simultaneously, we see a pair of parallel lines with slope 21 (Figure 4.4). This is consistent with the implicit differentiation result.
y = x + 3
y = x 3
FIGURE 4.4
EXAMPLE
Finding
Differentiate x 5 y 2 .
SOLUTION
d d (x) 5 (y2) dx dx 1 5 2y ? 1 5 2y ? dy 1 5 dx 2y d (y) dx dy dx

Power and Chain Rules
EXAMPLE
Finding
Find the derivative of y 2 1 2y 1 1 5 x with respect to x.

SOLUTION
d 2 d ( y 1 2y 1 1 ) 5 ( x ) dx dx 2y ? d d (y) 1 2 ? (y) 1 0 5 1 dx dx 2y ? dy dy 12? 51 dx dx dy ( 2y 1 2 ) 5 1 dx dy 1 5 dx ( 2y 1 2 )

Power, Chain, and Constant Rules
298
EXAMPLE
Proving
d 1 ( logb( x )) dx ln( b ) ? x
1 . ln( b ) ? x
Prove that the derivative of y 5 logb( x ) is yr 5
SOLUTION We will do some clever manipulations with the function and then use the Exponential Rule to find the derivative. Recall that according to the definition of logarithms, the following two equations are equivalent:
y 5 logb( x ) b 5x
y
and
Differentiating the second equation with respect to x yields d y d (b ) 5 (x) dx dx ( ln b ) b y ? d (y) 5 1 dx dy 51 dx dy 1 5 dx ( ln b ) b y 5 1 ( ln b ) x

Since b y 5 x Chain, Exponential, and Power Rules Chain Rule
( ln b ) b y ?
EXAMPLE
Finding
Find
dy given 3xy 2 5 4xy . dx
SOLUTION
d d ( 3xy 2 ) 5 ( 4xy ) dx dx d d d d ( 3x ) ? y 2 1 ( y 2 ) ? 3x 5 ( 4x ) ? y 1 ( y ) ? 4x dx dx dx dx ( 3 ) y 2 1 a 2y dy dy b ? 3x 5 ( 4 ) y 1 ? 4x dx dx dy dy 5 4y 1 4x dx dx
Product Rule Constant Multiple, Power, and Chain Rules
3y 2 1 6xy 6xy
dy dy 2 4x 5 4y 2 3y 2 dx dx
dy ( 6xy 2 4x ) 5 4y 2 3y 2 dx dy 4y 2 3y 2 5 dx 6xy 2 4x
299
4.5 Summary
In this section, you learned how to use implicit differentiation to find the derivative of functions and nonfunctions. Knowing how to do implicit differentiation will help you understand related rates in the next chapter.

In Exercises 18, solve the equation for d. 1. 2( x 1 y )( 1 1 d ) 5 0 2. 2yd 1 2d 5 1 3. ln 3( 3 y ) d 5 1 4. 3y 2 1 6xyd 5 4y 1 4xd 5. 2x 1 4yd 5 d 6. 2x 1 y 1 xd 5 2yd 7. de x 1 e xy 5 y 2 1 2xyd 8. 1 d 5 xd 1 y y
4.5 Exercises
In Exercises 110, use implicit differentiation to find dy . Then evaluate the derivative function at the dx designated point. 1. 2x 2 1 y 2 5 32; ( 4, 0 ) 3. x2y 2 3 5 y 2; ( 2, 1 ) 5. xy 5 6; ( 2, 3 ) 6. x 2y 2 xy 2 5 20; ( 5, 4 ) 7. 9x 2 y3 5 0; ( 3, 3 ) 8. xy2 1 xy 1 y 5 2 1; ( 3, 2 1 ) 9. 20y 2 x 2y 2 5 75; ( 1, 5 ) 10. xy 3 2 x 3y 5 30; ( 2 2, 2 3 ) In Exercises 1120, use implicit differentiation to find dy . dx 11. y2 5 x 13. ye y 5 2 15. ln y 5 xy 17. 2xy 2 y2 5 9 19. ln( xy2 ) 5 y 12. xe y 5 y 14. 2y 5 ln x 16. x2y 2 y2 1 y 5 x 18. x( x2 2 y2 ) 5 y 20. ye x 5 xy2 26. What is implicit differentiation? 27. Under what circumstances is it advisable, or even necessary, to use implicit differentiation to find a derivative? 28. Your group is asked to find determine that dy given x 2 y 5 y 2 . You dx 2. x2 2 y2 5 9; ( 5, 4 ) 4. 4y2 5 x2; ( 4, 2 ) In Exercises 2125, determine at what points, if any, dy 5 0. dx 21. x2 1 y2 5 1 23. x2 1 2y2 5 y 25. x2 1 xy 5 y2 2 5 22. 3x2 5 y ln y 24. x3 2 y2 5 xy
dy 5 2x while another member dx dy 2xy 5 of your group determines that . dx 2y 2 x2 Who is correct? 29. What is one of the drawbacks of implicit differentiation?
30. The volume of a cylinder is V 5 2pr 2h, where r is the radius and h is the height of the cylinder. Note that the volume is a function of two variables.
300
dV dV and and interpret what each derivative dh dr represents. Find
32. Given V 5 pr2h, write
dr dV dh in terms of , , h, dt dt dt and r. (Hint: Differentiate both sides with respect to t.) dy given ye x 5 xe y. dx
33. Find 31. Given x2 5 y2 , calculate dy dx and . What is the dx dy dy dx relationship between and ? dx dy
34. Given the equation ye x 5 xe y, find the solution to dy 5 0, if it exists. the equation dx 35. Find dy , if possible, given e xy 5 ln( xy ) . dx

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. Why is the derivative of a constant equal to zero? (4.1) 2. Why is the derivative of a linear function equal to a constant? (4.1) 3. Which of the derivative rules is most closely related to the concept of composition of functions? Explain. (4.3) 4. Why is it that
d ( f ( x ) # g( x )) dx d ( f (x) dx
1 g( x )) 5 f r( x ) 1 gr( x ) but
2 f r( x ) gr( x ) ? (4.1, 4.2)
5. What strategy should you use to determine which derivative rule to apply? (4.1, 4.2, 4.3, 4.4) 6. Why cant you use the Power Rule to find the derivative of y 5 2x? (4.1, 4.4) 7. Since there are so many derivative rules that make finding a derivative relatively easy, why is it important to know how to use the limit definition of the derivative, f r( x ) 5 lim 8. How are the following terms related: derivative, differentiate, differentiation, and differentiable? (4.1) 9. What are the advantages of each of the different forms of derivative notation? (4.1) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
f (x 1 h) 2 f (x) ? h hS0
(4.1, 4.2, 4.3, 4.4)
Definitions
Derivative Notation: The derivative of a function y 5 f ( x ) may be represented by any of the following: f r( x ) yr
Chapter 4
Study Sheet
301
dy (Leibniz notation) dx Differentiate: A verb meaning to find the derivative. Symbolically repd resented by . dx Differentiation: The process of finding the derivative Differentiable Function: A function whose derivative exists for all values of x in its domain
Steps of Implicit Differentiation 1. Differentiate both sides of the equation with respect to x (recall that d dy (y) 5 ). dx dx dy 2. Algebraically isolate . dx d ( c ) 5 0 for any constant c dx
Formulas
Constant Rule: Power Rule:
d n ( x ) 5 nxn 2 1 dx d Constant Multiple Rule: ( c ? f ( x )) 5 cf r( x ) for any constant c dx d Sum and Difference Rule: ( f ( x ) 6 g( x )) 5 f r( x ) 6 gr( x ) dx d Product Rule: ( f ( x ) ? g( x )) 5 f r( x ) g( x ) 6 gr( x ) f ( x ) dx d f (x) f ( x ) g( x ) 2 gr( x ) f ( x ) Quotient Rule: a b 5 dx g( x ) 3g( x ) 4 2 d Chain Rule: ( f ( g( x ))) 5 f r( g( x )) gr( x ) dx dy dy dx Chain Rule, Alternative Form: 5 ? dt dx dt d Generalized Power Rule: ( u n ) 5 nu n 2 1ur dx d x Exponential Rule: ( b ) 5 ( ln b ) bx dx d Exponential Rule, Special Case: ( e x ) 5 e x dx d u Generalized Exponential Rule: ( b ) 5 ( ln b ) buur dx d 1 Logarithmic Rule: ( logb x ) 5 dx ( ln b ) x d 1 Logarithmic Rule, Special Case: ( ln x ) 5 x dx d 1 Generalized Logarithmic Rule: ( log b u ) 5 ur dx ( ln b )
302

Section 4.1 In Exercises 14, use the Constant Rule, the Power Rule, the Constant Multiple Rule, and the Sum and Difference Rule (as appropriate) to find the derivative of the function.
1. f ( x ) 5 9 3. v( t ) 5 7t 3 2 4t23 4. s( t ) 5 2 2t 1 t 2 t 1 17
4 2
In Exercises 1112, use the Product Rule in answering the questions from the real-life scenarios. You may find it helpful to use your calculator. 11. Based on data from 1980 to 1998, the federal and state prison rate may be modeled by
Prison Rate
2. g( x ) 5 2 12x 1 14
In Exercises 56, answer the question by evaluating the derivative. 5.

Military Personnel Based on data from 1990 to 1999, the number of active-duty military personnel may be modeled by
R 5 0.3750t 2 1 11.70t 1 138.7 prisoners per 100,000 people, where t is the number of years since 1980. (Source: Modeled from
The number of people in the United States may be modeled by P 5 0.525t 2 1 1.695t 1 226.5 hundred thousand people, where t is the number of years since 1980. (Source: Modeled from
www.census.gov.)
A 5 2.5626t 2 58.90t 1 1321 thousand people, where t is the number of years since 1990. (Source: Modeled from Statistical
2
According to the model, at what rate was the number of active-duty military personnel changing in 1999 and in 2000? 6. Based on data from 1980 to 1998, the federal and state prison rate may be modeled by
Prison Rate
In 2000, at what rate was the number of prisoners increasing (in prisoners per year)? 12. Based on data from 1980 to 1999, the per capita milk beverage consumption in the United States may be modeled by
Milk Consumption
R 5 0.3750t 2 1 11.70t 1 138.7 prisoners per 100,000 people, where t is the number of years since 1980. (Source: Modeled
from Statistical Abstract of the United States, 2001, Table 332, p. 200.)
M 5 2 0.219t 1 27.7 gallons per person, where t is the number of years since 1980. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 202, p. 129.)
The number of people in the United States may be modeled by P 5 0.525t 2 1 1.695t 1 226.5 hundred thousand people, where t is the number of years since 1980. (Source: Modeled from
www.census.gov.)
Was the prison rate increasing more rapidly in 1988 or in 1998?
Section 4.2 In Exercises 710, use the Product Rule to find the derivative of the function. Simplify the result.
7. f ( x ) 5 3x2( 2x 2 4 ) 8. g( x ) 5 2 15x( 5x2 2 2 ) 9. f ( t ) 5 ( 2t 2 4 )( 2 t 1 4 )
2
In 1999, at what rate (in thousands of gallons per year) was milk beverage consumption changing?
Section 4.3 In Exercises 1314, find the function f ( g( x )) . Then simplify the result.
13. f ( x ) 5 2 3x 1 4, g( x ) 5 2x2 2 x 14. f ( x ) 5 2 4x2 1 x, g( x ) 5 2x 1 3
10. s( t ) 5 t 3( 3t 2 1 7t )( t 2 3 )
Chapter 4
Review Exercises
303
In Exercises 1516, use the Chain Rule to find the derivative of the function. Then write the result in factored form. 15. f ( x ) 5 2( 9x 1 5 ) 3 16. g( x ) 5 3( x2 2 2x ) 2
Section 4.4 In Exercises 1922, use the Exponential Rule or the Logarithmic Rule, as appropriate, to find the derivative of the function.
19. y 5 9 x 21. f ( x ) 5 ln( x2 1 3 ) 20. g( t ) 5 20( 0.5 ) t 22. s( t ) 5 ln( 23t 2 4 )
In Exercises 1718, use a composition of functions and the Chain Rule to answer the questions. 17. Based on data from 1990 to 2000, the amount of money spent on advertising on cable television may be modeled by C( m ) 5 0.00009027m2 2 0.0261m 2 1130
In Exercises 2324, answer the questions by using the Exponential Rule or the Logarithmic Rule, as appropriate. 23. The projected population of Hawaii may be modeled by P 5 1175( 1.014 ) t
Hawaii Population
million dollars, where m is the amount of money (in millions of dollars) spent on advertising in magazines. The amount of money spent on advertising in magazines may be modeled by m( t ) 5 42.29t 2 1 158.3t 1 6575 million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical
thousand people, where t is the number of years since 1995. (Source: Modeled from www.census.gov.) In 2010, at what rate is the population of Hawaii expected to be changing? 24.
Church Growth Based on data from 1990 and 2001, the number of Lutherans in the United States may be modeled by L 5 9110( 1.005 ) t
In 2000, at what rate was cable television advertising spending changing? 18.
Advertising Expenditures Based on data from 1990 to 2000, the amount of money spent on advertising on radio may be modeled by R( b ) 5 12.92b 2 4984
thousand people, where t is the number of years since 1990. (Source: Modeled from American
Religious Identification Survey data.)
In 2001, at what rate was the number of Lutherans in the United States increasing? In Exercises 2529, use implicit dy differentiation to find . Then evaluate the derivative dx function at the designated point.
million dollars, where b is the amount of money (in millions of dollars) spent on advertising on billboards. The amount of money spent on advertising on billboards may be modeled by b( t ) 5 9.396t 2 2 11.58t 1 1063 million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical
Section 4.5
25. xe3y 5 7; ( 7, 0 ) 26. log( xy ) 5 2; ( 4, 25 ) 27. x 2y 2 2 1 5 0; ( 0.5, 2 ) 28. 2xy2 2 y 5 10; ( 1, 2 2 ) 29. 4xy 2 2xy 5 4; ( 2 3, 2 1 )
In 2000, at what rate was radio advertising spending changing?
P R O J E C T
What to do
1. Collect a minimum of five data points of something that is reported on a national basis in the United States each year. (For example, the total amount of vegetables and melons consumed in a given year.) 2. For each year identified in (1), record the U.S. population. 3. Model each of the data sets in (1) and (2) with a polynomial, exponential, or logarithmic function. 4. Write a third model defined as g( t ) 5 5. Find the derivative of the third model. 6. Evaluate and interpret the derivative at two different points. If the results dont make sense in their real-world context, choose a new data set for the first model.
first model . population model
Example
Years (Since 2000) 0 1 2 3 4 Vegetable and Melon Consumption (million pounds) 48,293 48,311 49,064 49,704 51,231 Years (Since 2000) 0 1 2 3 4 People (thousands) 282,402 285,329 288,173 291,028 293,907
Source: Statistical Abstract of the United States, 2006, Table 830. v( t ) 5 207.50t 2 2 103.10t 1 48,282 million pounds t years after 2000
Source: Statistical Abstract of the United States, 2006, Table 830. p( t ) 5 2870.9t 1 282,426 thousand people t years after 2000 (Continued)
304
Example (Continued)
The third model is g( t ) 5 5 v( t ) million pounds p( t ) thousand people
207.50t2 2 103.10t 1 48,282 thousand pounds per person 2870.9t 1 282,426
The derivative of the third model is found using the quotient rule. gr( t ) 5 ( 415t 2 103.10 )( 2870.9t 1 282,426 ) 2 ( 2870.9 )( 207.50t2 2 103.10t 1 48,282 ) ( 2870.9t 1 282,426 ) 2 thousand pounds per person per year We evaluate the derivative at t 5 0 and t 5 5. gr( 0 ) 5 ( 415( 0 ) 2 103.10 )( 2870.9( 0 ) 1 282,426 ) 2 ( 2870.9 )( 207.50( 0 ) 2 2 103.10( 0 ) 1 48,282 ) ( 2870.9( 0 ) 1 282,426 ) 2 5 20.0021028 thousand pounds per person per year < 22.1 pounds per person per year In 2000, the consumption of vegetables and melons was decreasing at a rate of 2.1 pounds per person per year. gr( 5 ) 5 ( 415( 5 ) 2 103.10 )( 2870.9( 5 ) 1 282,426 ) 2 ( 2870.9 )( 207.50( 5 ) 2 2 103.10( 5 ) 1 48,282 ) ( 2870.9( 5 ) 1 282,426 ) 2 5 0.0049183 thousand pounds per person per year < 4.9 pounds per person per year In 2005, the consumption of vegetables and melons was increasing at a rate of 4.9 pounds per person per year.
305
Chapter
Derivative Applications
B
usinesses survive by being profitable. A savvy business owner effectively analyzes the factors that contribute to the financial success or failure of her business. Using mathematical models, she may forecast what prices, production levels, shipment schedules, and other such elements will result in maximum profits. Although no mathematical model is perfect at forecasting future results, a model can help a business owner make informed decisions.
5.1
Maxima and Minima

Find relative and absolute extrema Use the First Derivative Test to find relative extrema
5.2
Applications of Maxima and Minima
Analyze and interpret revenue, cost, and profit functions
5.3
Concavity and the Second Derivative
Use the Second Derivative Test to find relative extrema Determine the concavity of the graph of a function
5.4
Related Rates
Solve related-rate problems
306
5.1 Maxima and Minima
307
Find relative and absolute extrema Use the First Derivative Test to find relative extrema
Based on data from 1984 to 2001, the rate at which tuition and fees at community colleges in Washington state are increasing may be modeled by R( t ) 5 2 0.7473t2 1 14.89t 1 9.024 dollars per year where t is the number of years since the 19841985 school year. In what year were tuition and fees at Washington state community colleges increasing the fastest? (Source: Modeled from Washington State Higher Education Coordinating Board, Higher Education Statistics, September 2001.) Questions such as these may be answered using the concepts of relative and absolute extrema. In this section, we will informally discuss continuity. In addition, we will discuss how to find relative extrema and absolute extrema.
Continuity
y 10 9 8 7 6 5 4 3 2 1 3 2 1 0 1 2 3 x
The notion of continuity is best understood graphically. Loosely speaking, a function is said to be continuous if its graph can be drawn by a pencil without lifting the pencil from the page. If there is a break in the graph, the graph is said to be discontinuous. This loose definition of continuity will be sufficient for our purposes. (The formal definition of continuity, which relies heavily on the concept of limits, may be obtained from a traditional calculus text.) Many functions are continuous. For example, linear, polynomial, exponential, and logarithmic functions are all continuous. Frequently, functions that are discontinuous have domain restrictions or are defined piecewise. Consider the 2x 1 6 if x , 1 piecewise function f ( x ) 5 e shown in Figure 5.1. Since there 3x 1 2 if x $ 1 is a break in the graph at x 5 1, the graph is said to be discontinuous.
FIGURE 5.1
Relative and Absolute Extrema

An extremum is a maximum or minimum value of a function. The plural of extremum is extrema.
y 12 8 4 2 1 4 8 12 16
f ( x ) = x 3 3x 2
RELATIVE EXTREMA
A relative maximum of a continuous function f occurs at a point ( c, f ( c )) if f ( x ) # f ( c ) for all x in some interval ( a, b ) containing c. (That is, a , c , b.) A relative minimum of a continuous function f occurs at a point ( c, f ( c )) if f ( x ) $ f ( c ) for all x in some interval ( a, b ) containing c. (That is, a , c , b.)
FIGURE 5.2
According to the definition, a relative extremum may not occur at an endpoint. We will explore the concept of relative extrema by looking at the graph of f ( x ) 5 x3 2 3x2 on the domain 32 2, 44 (Figure 5.2).
308
CHAPTER 5 Derivative Applications
Graphically, we are looking for the peaks and valleys of the graph. In this case, a peak occurs at ( 0, 0 ) and a valley occurs at ( 2, 2 4 ). A relative maximum occurs at the point ( 0, 0 ) , since y 5 0 is larger than all y values of the function nearby x 5 0 . (When we use the term nearby, we mean an arbitrarily small open interval ( a, b ) surrounding x 5 0.) A relative minimum occurs at the point ( 2, 2 4 ) , since y 5 2 4 is smaller than all y values of the function nearby x 5 2.
ABSOLUTE EXTREMA
An absolute maximum of a function f occurs at a point ( c, f ( c )) if f ( x ) # f ( c ) for all x in the domain of f. An absolute minimum of a function f occurs at a point ( c, f ( c )) if f ( x ) $ f ( c ) for all x in the domain of f.
y 12 8 Relative maximum 4 2 1 4 8 12 16
Absolute maximum f ( x ) = x 3 3x 2
Relative minimum Absolute minimum
FIGURE 5.3
It is important to note that unlike relative extrema, absolute extrema may occur at endpoints. To find the absolute maximum of the function f ( x ) 5 x3 2 3x2 on the domain 32 2, 44 , we look for the largest y value. Since f is continuous (there are no breaks in the graph of f ), we need only check the relative maxima and the y values of the endpoints. The graph of f has a relative maximum at ( 0, 0 ) and endpoints at ( 2 2, 2 20 ) and ( 4, 16 ) . Since ( 4, 16 ) has the largest y value, y 5 16 is the absolute maximum. To find the absolute minimum, we are looking for the smallest y value. Since f is continuous, we need only check the relative minima and the endpoints. The graph of f has a relative minimum at ( 2, 2 4 ) and endpoints at ( 2 2, 2 20 ) and ( 4, 16 ) . Since ( 2 2, 2 20 ) has the smallest (most negative) y value, y 5 2 20 is the absolute minimum (Figure 5.3). Precisely determining relative extrema is difficult to do graphically, especially when extrema occur at irrational points such as A p, "2 B . Fortunately, by using the derivative, we can quickly find relative extrema. The derivative of f ( x ) 5 x 3 2 3x 2 is f r( x ) 5 3x2 2 6x. Recall that the derivative at a point represents the slope of the tangent line (or slope of the graph) at that point. If the slope is positive, the graph is increasing. If the slope is negative, the graph is decreasing. If the slope is zero, the graph is neither increasing nor decreasing but remains flat. In other words, the graph has a horizontal tangent line if and only if f r( x ) 5 0. For f ( x ) 5 x3 2 3x2, we have the results shown in Table 5.1.
TABLE 5.1
x 2 1 0 1 2 3 4
f (x) 20 4 0 2 4 0 16
f r( x ) 24 9 0 3 0 9 24
Slope of Tangent Line Positive Positive Zero Negative Zero Positive Positive
Graph of f Increasing Increasing Flat Decreasing Flat Increasing Increasing
309
Observe that f r( x ) 5 0 where the relative extrema occurred [( 0, 0 ) and ( 2, 2 4 ) ].
THE VALUE OF THE DERIVATIVE AT A RELATIVE EXTREMUM
Let f be a function with a continuous graph. If a relative extremum of f occurs at ( c, f ( c )) , then f r( c ) 5 0 or f r( c ) is undefined.
The converse is not true. That is, if f r( c ) 5 0 or f r( c ) is undefined, then a relative extremum is not guaranteed to occur at ( c, f ( c )) . We will demonstrate this in Example 2.
EXAMPLE
Determining from a Graph Where f r( x ) 5 0
The graph of f ( x ) 5 x3 2 12x on the interval 32 3, 34 is shown (Figure 5.4). Determine the points on the graph where f r( x ) 5 0.
y 16 12 8 4 3 2 1 4 8 12 16 1 2 3 x f ( x ) = x 3 12x
FIGURE 5.4
SOLUTION Since the graph is a smooth curve (no sharp points), the derivative will equal zero at the relative extrema. It appears that a relative maximum occurs when x 5 2 2 and a relative minimum occurs when x 5 2. We will now use the derivative to confirm our observation.
f ( x ) 5 x3 2 12x f r( x ) 5 3x2 2 12 f r( 2 ) 5 3( 2 ) 2 2 12 50 f r( 2 2 ) 5 3( 2 2 ) 2 2 12 50 The results confirm our graphical estimate. At the point ( 2 2, 16 ) and the point ( 2, 2 16 ) , f r( x ) 5 0.
310
EXAMPLE
Determining from a Graph Where f r( x ) 5 0
y 110 100 90 80 70 60 50 40 30 20 10 3 2 10 20
The graph of f ( x ) 5 x4 2 4x3 on the interval 32 3, 54 is shown in Figure 5.5. Determine the points on the graph where f r( x ) 5 0.
f ( x ) = x 4 4x 3
SOLUTION The graph is a continuous, smooth curve. We are looking for places on the graph where the function has a horizontal tangent line 3 f r( x ) 5 04 . It looks like the graph has a horizontal tangent line at x 5 0 and x 5 3. We will confirm this result by evaluating f r( x ) at x 5 0 and x 5 3.
f ( x ) 5 x4 2 4x3 f r( x ) 5 4x3 2 12x2

1 2 3 4 5 x
5 4x2( x 2 3 ) f r( 0 ) 5 4( 0 ) 2( 0 2 3 ) 50 f r( 3 ) 5 4( 3 ) 2( 3 2 3 ) 50 So f r( x ) 5 0 at ( 0, 0 ) and ( 3, 2 27 ) . As discussed previously, the fact that f r( c ) 5 0 does not guarantee that a relative extremum occurs at ( c, f ( c )) . In this example, a relative minimum occurred at ( 3, 2 27 ) , but a relative extremum did not occur at ( 0, 0 ) .
FIGURE 5.5
In Example 3, we will use the absolute value function. Recall that the absolute value function is formally defined as a piecewise function.
y 2 1 x
f (x) = |x|
Zx Z 5 e
x if x $ 0 2 x if x , 0
1 1 2
The graph of the absolute value function looks like a V (Figure 5.6). Since the graph is decreasing for x , 0 , the slope of the tangent line of f is negative for x , 0 . Since the graph is increasing for x . 0 , the slope of the tangent line of f is positive for x . 0 . But what is the slope of the tangent line at x 5 0 ? Recall that the derivative of a function f at x 5 a is formally defined as f r( a ) 5 lim
hS0
f (a 1 h) 2 f (a) h Z0 1 h Z 2 Z0 Z h Zh Z h
For f ( x ) 5 Z x Z , the derivative of the function at x 5 0 is f r( 0 ) 5 lim

hS0
FIGURE 5.6
5 lim
hS0
5.1 Maxima and Minima |h| h
311
As h gets close to 0, what value does
approach? We first observe that h 2 0
because division by 0 is not defined. From the piecewise definition of the absolute value function, we know h if h . 0 h Zh Z 5 e undefined if h 5 0 h 2h if h , 0 h Simplifying the equation yields 1 if h . 0 Zh Z 5 c undefined if h 5 0 h 21 if h , 0 For positive values of h,
|h| h
5 1. For negative values of h,

hS0
doesnt approach a constant value as h nears 0, lim
f ( x ) 5 Z x Z , f r( 0 ) is undefined. In other words, the derivative of the absolute value function does not exist at x 5 0 . In general, the derivative of any function does not exist wherever the graph of a function has a sharp point.
|h| h
|h| h
5 2 1. Since
|h| h
does not exist. That is, for
EXAMPLE
Finding the Relative Extrema of a Function

The graph of f ( x ) 5 2 Z x Z 1 2 is shown in Figure 5.7. Find the relative extrema of f on the interval 32 3, 34 .
y 2 1 x f ( x ) = | x | + 2
1 1 2
FIGURE 5.7
SOLUTION From the graph, we can see that the function has a relative maximum at ( 0, 2 ) . Therefore, f r( x ) 5 0 or f r( x ) is undefined at that point. At any sharp point on a graph, the derivative is undefined. Therefore, f r( 0 ) is undefined. Does f r( x ) 5 0 anywhere? From the graph, we see that f is increasing for all x , 0 . Therefore, f r( x ) is positive for x , 0 . Similarly, f is decreasing for all x . 0 . Therefore, f r( x ) is negative for x . 0 . Hence, f r( x ) 2 0 anywhere. Thus the only relative extremum occurs at ( 0, 2 ) .
312
Critical Values
In Example 2, we saw that relative extrema may occur where f r( x ) 5 0. In Example 3, we saw that relative extrema may also occur where f r( x ) is undefined. These two situations represent two different types of critical values of a function.
CRITICAL VALUES
Let f be a continuous function on the interval ( a, b ) with a , c , b. If f r ( c ) 5 0 or f r ( c ) is undefined then c is called a critical value of f. The two types of critical values may be further classified as stationary values 3 f r( c ) 5 04 and singular values 3 f r( c ) is undefined4 . The corresponding point ( c, f ( c )) is called a critical point, stationary point, or singular point of f, respectively.
EXISTENCE OF RELATIVE EXTREMA
Let f be a continuous function on the interval ( a, b ) . All relative extrema of f occur at critical values of f.
EXAMPLE
Finding Critical Values of a Function

Find the critical values of f ( x ) 5 x3 2 3x on the interval 32 2, 44 .
SOLUTION To find the stationary values, we set the derivative equal to zero and solve for x.
f r( x ) 5 3x2 2 3 0 5 3x2 2 3 0 5 3( x 2 1 )
2
Set f r( x ) 5 0
0 5 3( x 2 1 )( x 1 1 ) x2150 x51 x1150 x 5 21

Set each factor equal to 0 and solve for x
The stationary values are x 5 1 and x 5 2 1 . Since the derivative is defined for all values of x, there are no singular values.
EXAMPLE
Finding Critical Points of a Function

Find the critical points of f ( x ) 5 x1>3 on the interval 32 2, 24 .
313
SOLUTION
f r( x ) 5 5
1 22>3 x 3 1 3x2>3
f r( x ) 2 0 for all values of x, since a fraction can equal zero only if its numerator is zero. Therefore, f does not have any stationary points. g( x ) A rational expression is undefined when h( x ) is equal to zero. What h( x ) value of x makes the denominator of f r( x ) equal zero? 0 5 3x2>3 05x
3 2>3
Set the denominator equal to 0 Divide by 3 Cube both sides of the equation ( x2>3 ) 3 5 x2>3 ? 3 5 x2 Take the square root
(0) 5 (x ) 05x
2
2>3 3
x50
Therefore, f r( 0 ) is undefined. A singular point may occur at x 5 0. We need only confirm that the function f is defined at x 5 0. Since f ( 0 ) 5 0, x 5 0 is in the domain of f and is indeed a singular value of f. The point ( 0, 0 ) is a singular point of f.
The First Derivative Test

Is there a nongraphical way to tell whether a relative maximum or a relative minimum occurs at a critical value? Fortunately, yes. Recall that when a function is increasing, its derivative is positive. When a function is decreasing, its derivative is negative. If the derivative changes sign at a critical value, a relative extremum occurs at that value. The First Derivative Test will allow us to determine where relative extrema occur for nonconstant functions. If f is a constant function 3 f ( x ) 5 k for some constant k4 , then f r( x ) 5 0 for all x and all points of the function are critical points. Furthermore, since k # f ( x ) # k for all x , both relative maxima and relative minima occur at all points of f. Since most real-life data models are not constant functions, we will focus our discussion on nonconstant functions.
FIRST DERIVATIVE TEST
Let ( c, f ( c )) be a critical point of a nonconstant function f. If f r changes from positive to negative at x 5 c , then a relative maximum occurs at ( c, f ( c )) . If f r changes from negative to positive at x 5 c , then a relative minimum occurs at ( c, f ( c )) . If f r doesnt change sign at x 5 c , then a relative extremum does not occur at ( c, f ( c )) .
314
It is often helpful to develop a sign chart when determining where relative extrema occur. We will demonstrate this with an example and then detail the specific steps of the process.
EXAMPLE 6
Finding Relative and Absolute Extrema

Find the relative and absolute extrema of f ( x ) 5 x4 2 4x3 on the interval 32 1, 54 .
SOLUTION We will first find the relative extrema by using the First Derivative
Test. f r( x ) 5 4x3 2 12x2 5 4x2 ( x 2 3 ) 4x2 5 0 x50 x2350 x53

Set each factor equal to zero Solve for x
Setting the derivative equal to zero, we see that x 5 0 and x 5 3 are critical values of f. We begin by drawing a number line with the critical values clearly marked.
f' 0 3
We will evaluate f r( x ) at an x value in each of the following three intervals: 32 1, 0 ) , ( 0, 3 ) , and ( 3, 54 . (These intervals may be equivalently expressed as 2 1 # x , 0, 0 , x , 3, and 3 , x # 5, respectively.) We are not concerned with the actual value of the derivative. We simply want to know if the derivative is positive or negative. From the interval 32 1, 0 ) , we pick x 5 2 1. f r( 2 1 ) 5 4( 2 1 ) 2 ( 2 12 3 ) f r( 2 1 ) , 0
The product is equivalent to (4)(1)(4) Since there is an odd number of negative factors
On the interval 32 1, 0 ) , the derivative is negative. We update the chart to reflect our finding.
f' 0 3
From the interval ( 0, 3 ) , we pick x 5 1. f r( 1 ) 5 4( 1 ) 2 ( 12 3 ) f r( 1 ) , 0

The product is equivalent to (4)(1)(2) Since there is an odd number of negative factors
On the interval ( 0, 3 ) , the derivative is negative. We update the chart to reflect our finding.
f' 0 3
From the interval ( 3, 54 , we pick x 5 4. f r( 4 ) 5 4( 4 ) 2 ( 42 3 )

The product is equivalent to (4)(16)(1) Since there are no negative factors
On the interval ( 3, 54 , the derivative is positive. We update the chart to reflect our finding.
f r( 4 ) . 0
5.1 Maxima and Minima 0 3 +
315
f'
At the critical value x 5 0 , the derivative did not change sign, so a relative extremum does not occur there. Graphically speaking, the graph of f is decreasing through the whole interval 32 1, 3 ) . At the critical value x 5 3 , the derivative changed sign from negative to positive, so a relative minimum occurs there.
f' 0 min + 3
Graphically speaking, the graph of f changed from decreasing to increasing at x 5 3. To find absolute extrema, we consider the relative extrema and evaluate the function f at the endpoints (see Table 5.2).
TABLE 5.2
x 1 3 5
f (x) 5 27 125
Relative minimum, absolute minimum Absolute maximum
The largest value of f ( x ) in the table is the absolute maximum, and the smallest value is the absolute minimum. In this case, the relative and absolute minimum occurs at ( 3, 2 27 ) and is equal to 2 27. The absolute maximum occurs at ( 5, 125 ) and is equal to 125. There is no relative maximum. (Recall from our definition of a relative extremum that a relative extremum may not occur at an endpoint. This definition allows us to use critical values to find all relative extrema.)
Constructing a Derivative Sign Chart

1. Label a number line with the critical values of the function f. 2. Write f r next to the number line. 3. Evaluate f r at a value in each of the number line intervals. 4. Record a 1 if the derivative is positive and a 2 if the derivative is
negative. 5. Use the First Derivative Test to determine where relative maxima and minima occur and record the results on the chart.
EXAMPLE
Finding Relative and Absolute Extrema

Find the absolute and relative extrema of g( x ) 5 x3 2 2x2 2 5x 1 6 on the interval 32 3, 44 .
316
SOLUTION
gr( x ) 5 3x2 2 4x 2 5 3x2 2 4x 2 5 5 0
We set gr( x ) 5 0 . This is a quadratic equation in the form ax2 1 bx 1 c 5 0 . Since the derivative doesnt factor, we must use the Quadratic Formula to find the stationary points (or use a calculator or computer). x5 5 5 5 5 Since
2 3
2 b 6 "b2 2 4ac 2a
4 6 "76 6
2 ( 2 4 ) 6 "( 2 4 ) 2 2 4( 3 )( 2 5 ) 2( 3 )
2 "19 6 3 3
2 3
4 6 2 "19 6
x < 2 0.7863 and x < 2.120 . The derivative is defined for all values of x, so there are no singular values. We now construct a sign chart.
g' 0.7863 2.120
"19 3
< 2.120 and
"19 3
< 2 0.7863, the stationary values are
To determine the sign of the derivative gr( x ) 5 3x2 2 4x 2 5 in each interval, we evaluate the derivative at x 5 2 1 , x 5 0 , and x 5 3 . gr( 2 1 ) 5 3( 2 1 ) 2 2 4( 2 1 ) 2 5 531425 gr( 2 1 ) . 0 gr( 0 ) 5 3( 0 ) 2 2 4( 0 ) 2 5 502025 gr( 0 ) , 0 gr( 3 ) 5 3( 3 ) 2 2 4( 3 ) 2 5 5 27 2 12 2 5 gr( 3 ) . 0 We record the sign of the derivative in each interval on the sign chart.
g' + max 0.7863 min + 2.120 2 3
A relative maximum occurs at x 5 minimum occurs at x 5

2 3
"19 3
"19 3
< 2 0.7863 and a relative
< 2.120.
317
To find the absolute extrema, we consider the relative extrema and evaluate the function g at the endpoints, as shown in Table 5.3.
TABLE 5.3
x 3 0.7863 2.120 4
g( x ) 24 8.209 4.061 18
Absolute minimum Relative maximum Relative minimum Absolute maximum
The absolute minimum occurs at ( 2 3, 2 24 ) , although there is a relative minimum at ( 2.120, 2 4.061 ) . The absolute maximum occurs at ( 4, 18 ) , although there is a relative maximum at ( 2 0.7863, 8.209 ) .
It is often helpful to use a graphing calculator to calculate values of the derivative, especially when evaluating the derivative at other than whole-number values. The following Technology Tip will detail how to use your calculator to evaluate the derivative at a point quickly.
Evaluating the Derivative at a Point 1. Enter the function f ( x ) as Y1 by pressing the Y button and typing the equation.
2. Graph the function over the specified domain. (You may need to press WINDOW and adjust the Xmin and Xmax settings. Press Zoom and 0:ZoomFit then ENTER to automatically adjust the y values so that the entire graph will appear on screen.) 3. Press then TRACE to bring up the menu. Select item 6:dy/dx and press ENTER .
2nd CALCULATE
(Continued)
318
4. Type in the x value of the point where you want to evaluate the derivative or use the blue arrows to select the point graphically. Then press ENTER .
5. The value of the derivative is displayed at the bottom of the screen. (You can assume that a value like 1E-6 is equal to zero. 1E-6 = 1 3 1026 5 0.000001. The error occurs because the calculator uses numerical methods to differentiate.)
EXAMPLE
Interpreting Extrema in a Real-World Context

Based on data from 1980 to 1999, the per capita consumption of breakfast cereal may be modeled by C( t ) 5 2 0.004718t 3 1 0.1165t 2 2 0.3585t 1 12.17 pounds per person where t is the number of years since 1980. (Source: Statistical Abstract of the United States, 2001, Table 202, p. 129.) Find the relative and absolute extrema of C( t ) on the interval 30, 194 and interpret the meaning of the results.
SOLUTION We must first find the critical values of C( t ) .
Cr( t ) 5 2 0.014154t 2 1 0.2330t 2 0.3585 0 5 2 0.014154t 2 1 0.2330t 2 0.3585

Set C(t) equal to 0
Recall that the quadratic formula states that the solutions to a quadratic equation of the form at 2 1 bt 1 c 5 0 are given by t5 5 5 t5 2 b 6 " b2 2 4ac 2a
2 ( 0.2330 ) 6 " ( 0.2330 ) 2 2 4( 2 0.014154 )( 2 0.3585 ) 2( 2 0.014154 ) 2 0.2330 6 0.1844 2 0.028308 t5 2 0.2330 2 0.1844 2 0.028308 < 14.7
2 0.2330 1 0.1844 2 0.028308 < 1.7
C' (t )
min of C + max of C 1.7 14.7
The critical values are t < 1.7 and t < 14.7. Using these values together with the interval test values of t 5 0, t 5 10, and t 5 18, we construct the first derivative sign chart. C has a relative minimum at t < 1.7 and a relative maximum at t < 14.7. Evaluating C( t ) 5 2 0.004718t 3 1 0.1165t 2 2 0.3585t 1 12.17 at the critical values and the endpoints, we construct a table of values, as shown in Table 5.4.
319
TABLE 5.4
t 0 1.7 14.7 19
C( t ) 12.2 11.9 17.1 15.1

Relative and absolute minimum Relative and absolute maximum
According to the model, the per capita consumption of breakfast cereal was decreasing between the end of 1980 ( t 5 0 ) and mid-September 1982 ( t 5 1.7 ) . It reached a relative low of 11.9 pounds per person in mid-September 1982. From mid-September 1982 to mid-September 1995, the per capita consumption of cereal was increasing. It reached a relative high of 17.1 pounds per person in midSeptember 1995. Between mid-September 1995 and the end of 1999, the per capita consumption of cereal decreased. The highest per capita consumption of cereal between 1980 and 1999 was 17.1 pounds per person in the one-year period ending in mid-September 1995, and the lowest per capita consumption of cereal between 1980 and 1999 was 11.9 pounds per person in the one-year period ending in mid-September 1982.
EXAMPLE
Interpreting Extrema in a Real-World Context

Based on data from 1980 to 1999, the annual per capita bottled water consumption may be modeled by W( t ) 5 2.593( 1.106 ) t gallons where t is the number of years since the end of 1980. (Source: Statistical Abstract of the United States, 2001, Table 204, p. 130.) In what year between 1980 and 1999 was per capita bottled water consumption the greatest? on the interval 30, 194 . We first observe that W is an exponential function. Recall that the derivative of an exponential function y 5 ab x is yr 5 a ln( b ) b x. Since a 2 0 and ln( b ) 2 0, yr 5 0 only when b x 5 0. But no value of x can make b x 5 0. Similarly, no value of x can make b x undefined. Therefore, an exponential function y 5 ab x doesnt have any critical values. Thus exponential functions do not have any relative extrema. So to determine when the per capita bottled water consumption was the greatest, we only need to evaluate W( t ) 5 2.593( 1.106 ) t at the endpoints, as shown in Table 5.5.
TABLE 5.5
SOLUTION We are asked to find the absolute maximum of W( t ) 5 2.593( 1.106 ) t
t 0 19
W( t ) 2.6 17.6
Absolute minimum Absolute maximum
According to the model, per capita bottled water consumption reached a maximum of 17.6 gallons at the end of 1999.
320
EXAMPLE 10
Forecasting Maximum Revenue

Based on data from 1985 to 1999, the number of movie theater admissions purchased may be modeled by A( p ) 5 2 1036.5p3 1 12,915p2 2 53,497p 1 74,871 million admissions where p is the price of a movie ticket in constant 1990 dollars. (That is, prices have been adjusted for inflation.) (Source: Modeled from Statistical Abstract of the
United States, 2001, Table 1244, p. 761 and Table 694, p. 455.)
Assuming that prices (in constant 1990 dollars) will continue to remain between $3.45 and $4.76, at what price will the revenue from movie ticket sales be the greatest?
SOLUTION Revenue is the product of the price and the number of admissions
sold. R( p ) 5 pA 5 p( 2 1,036.5p3 1 12,915p2 2 53,497p 1 74,871 ) 5 2 1036.5p4 1 12,915p3 2 53,497p2 1 74,871p million dollars of revenue The change in revenue is the derivative of the revenue function. Rr( p ) 5 2 4146p3 1 38,745p2 2 106,994p 1 74,871
R' ( p) 1000 (3.77, 0) 0 3.45 1000 2000 3000 (4.52, 0) p 4.76
revenue in millions of dollars price in dollars
The graph of Rr( p ) on the interval 33.45, 4.764 is shown in Figure 5.8. Using the Technology Tip following this example, we determined that the horizontal intercepts of Rr( p ) occur at p 5 3.77 and p 5 4.52. In other words, Rr( 3.77 ) 5 0 and Rr( 4.52 ) 5 0 . That is, the horizontal intercepts of Rr( p ) are the critical points of R( p ) . From the graph of Rr( p ) , we can readily construct the sign chart of Rr( p ) . When the graph of Rr( p ) is below the horizontal axis, Rr( p ) , 0 . When the graph of Rr( p ) is above the horizontal axis, Rr( p ) . 0 . A relative minimum occurs at p 5 3.77 , and a relative maximum occurs at p 5 4.52 . We evaluate the function R at the endpoints and the relative extrema, as shown in Table 5.6.
TABLE 5.6
FIGURE 5.8
p
R' ( p) min 3.77 + max 4.52
R( p ) 5,053 4,556 5,456 5,053

3.45 3.77 4.52 4.76
According to the model, the maximum revenue of $5,456 million is achieved when the ticket price is $4.52. Recall that the price and revenue are in constant 1990 dollars. A movie theater would need to adjust these figures for inflation if it were to use the model as a guide in setting ticket prices today.
321
Finding the x-intercepts of a Function 1. Enter the function f ( x ) as Y1 by pressing the Y button and typing the equation.
2. Graph the function over the specified domain. (Press WINDOW and adjust the Xmin and Xmax settings. Press ZOOM then 0:ZoomFit then ENTER to automatically adjust the y values so that the entire graph will fit on the screen.) 3. Press then TRACE to bring up the menu. Select item 2:zero and press ENTER .
2nd
CALCULATE
4. The calculator asks Left Bound?. Enter an x value that is smaller than the x-intercept. Or, if you prefer, use the blue arrows to select a point to the left of the x-intercept visually.
5. The calculator asks Right Bound?. Enter an x value that is larger than the x-intercept. Or, if you prefer, use the blue arrows to select a point to the right of the x-intercept visually.
6. The calculator asks Guess?. You may enter a guess or simply press ENTER . The calculator returns the coordinates of the x-intercept. Repeat Steps 3 through 6 to find additional x-intercepts.
322
Evaluating the Derivative at a Point Follow the steps for creating a difference quotient. Then select a very small value for h (e.g., h 5 0.00000001). For small values of h, the difference quotient approximates the derivative at the indicated point [in this case, (2 0.7863, 8.209 )4 .
Finding the x-intercepts of a Function 1. Create a cell for the x value and an adjacent cell with the formula for f ( x ) as shown.
2. Select Tools on the menu bar. Then select Solver.
(Continued)
323
3. In the Solver Parameters dialog box, change Set Target Cell to B2, Value of to 0, and By Changing Cells to A2. The $ symbol tells Excel to keep the cell reference fixed.
4. Click Solve. The program will adjust the value of cell A2 until the target cell B2 equals the indicated value (0). The solver shows that f ( 1.061 ) < 0. (Assume that small values such as 2 4.9 3 1027 are equal to zero.)
5. To find additional x-intercepts, change the value of x to a value distant from the first intercept. Then repeat steps 2 through 4. Some trial and error may be necessary to find all values.
5.1 Summary
In the section, you learned how to find the relative and absolute extrema of a function. You discovered that the First Derivative Test is an excellent tool for finding the location of relative extrema algebraically.
324

1. Solve the equation ( x 1 2 )( x 2 5 ) 5 0 . 2. Solve the equation ( 2 x 1 4 )( x 2 2 1 ) 5 0. 3. Solve the equation ( x 2 1 12 x )( x 2 15 ) 5 0. 4. Solve the equation x 3 2 5x 2 1 6x 5 0. 5. For what values of x is y 5 ( x 1 2 )( x 2 5 ) positive? 6. For what values of x is y 5 ( 2 x 1 4 )( x 2 2 1 ) negative? 7. For what values of x is y 5 x( x 1 12 )( x 2 15 ) positive? 8. For what values of x is y 5 x 3 2 5x 2 1 6x negative?
5.1 Exercises
In Exercises 110, determine the points on the graph where f r( x ) 5 0 or f r( x ) is undefined. 1. f ( x ) 5 x2 2 2x
y 3 2 1 1 1 2 3 x
4. f ( x ) 5 2 x3 1 3x2 2 3x 1 5
y 20 1 20 40 60 80 100 1 2 3 4 5 x
5. f ( x ) 5 2 x2 1 4x 2 4
y 1 1 2 3 4 5 6 7 8 x 1 2 3 4 5 x
2. f ( x ) 5 Z x 2 1 Z 1 2x
y 8 7 6 5 4 3 2 1 1 0 1 2 3
6. f ( x ) 5 x3 2 9x2 1 24x 1 3
y 30
3. f ( x ) 5 x3 1 1
y 6 4 2 1 2 4 6 1 x
20 10 1 10 20 1 2 3 4 5 x
325
7. f ( x ) 5 2 Z x 2 2 Z 1 4
y 4 3 2 1 2 1 1 1 2 3 4 5 6 x
16. R( p ) 5 p4 2 2p2 1 1 17. h( x ) 5 x x 11

2
18. g( t ) 5
1 t2
19. C( p ) 5 4p3 2 12p 1 7 20. f ( x ) 5 2 x 1 " x
8. f ( x ) 5 Z x 1 1 Z 2 Z 2x Z
y 1 2 11 2 3 4 5 1 2 3 4 5 6 x
In Exercises 2135, construct a sign chart for the derivative. Then determine the relative and absolute extrema of the function on the specified domain. Use the First Derivative Test as appropriate. (Note: Exercises 2130 use the same functions as Exercises 1120.) 21. f ( x ) 5 x2 2 4x; 32 1, 54 22. g( t ) 5 t 3 2 1; 32 2, 24
23. h( t ) 5 t 3 2 6t 2; 32 1, 74
24. f ( x ) 5 x1>5 1 4; 32 1, 14
9. f ( x ) 5 x5 2 5x4 1 200
y 200 100 2 1 100 200 300 1 2 3 4 x
25. g( x ) 5 3x1>3 2 x; 32 3, 34 27. h( x ) 5 28. g( t ) 5 x ; 32 2, 24 x 11

2
26. R( p ) 5 p4 2 2p2 1 1; 30, 44
29. C( p ) 5 4p3 2 12p 1 7; 32 3, 34 30. f ( x ) 5 2 x 1 "x; 30, 24 31. f ( x ) 5 x3 2 4x; 32 3, 34 32. g( t ) 5 e t 2 t; 32 2, 24 33. h( t ) 5 t ln( t ) ; 30.1, 34
1 ; 32 2, 24 t2
10. f ( x ) 5 5
y 5 4 3 2 1 2 1 1 1 2 3 4 x
34. p( n ) 5 n( 2n ) ; 32 2, 14
35. g( x ) 5 e x 2 ln( x ) ; 30.1, 34 In Exercises 3640, the number of tickets sold, N( p ) , is given as a function of the ticket price p. Determine what ticket price will maximize the revenue from ticket sales. 36. N( p ) 5 2 200p 1 1000 37. N( p ) 5 2 60p 1 3000 38. N( p ) 5 2500e20.1p 39. N( p ) 5 625e20.025p 40. N( p ) 5 8500e20.04p
In Exercises 1120, find the critical values of the function. Then classify the critical values as stationary values or singular values. 11. f ( x ) 5 x2 2 4x 13. h( t ) 5 t 3 2 6t 2 15. g( x ) 5 3x1>3 2 x 12. g( t ) 5 t 3 2 1 14. f ( x ) 5 x1>5 1 4
326
In Exercises 4145, use the methods discussed in this section to answer the questions. 41.
Margarine Consumption Based on data from 1980 to 1999, the per capita consumption of margarine may be modeled by M( t ) 5 0.0002868t 4 2 0.01149t 3 1 0.1329t 2 2 0.5098t 1 11.30
of DVDs shipped by manufacturers may be modeled by V( x ) 5 2 0.2173x2 1 25.84x 2 0.02345 million dollars, where x is the number of DVDs shipped in millions. (Source: Modeled from
Recording Industry Association of America data.)
pounds per year, where t is the number of years since 1980. (Source: Modeled from Statistical
According to the model, how many DVDs should be shipped in order to maximize the value of the manufacturers DVD shipments?
Between 1980 and 1999, when was the per capita consumption of margarine increasing? 42. Based on data from 1990 to 1999, electronic gaming software sales may be modeled by S( t ) 5 39.79t 2 2 48.41t 1 2378
Gaming Software Sales
46. What is the difference between a relative maximum and an absolute maximum? 47. How are relative extrema related to critical values? 48. A function h ( t ) has the property that h ( 3 ) is undefined. Is t 5 3 a critical value for h ( t ) ? 49. According to the First Derivative Test, a relative maximum occurs when the sign of the derivative changes from positive to negative at a critical value x 5 a. Explain what a change in the sign of the derivative means in terms of the graph of the original function. 50. Why is it important to label a sign chart with the name of the function represented in the chart?
million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical
In what year (between 1990 and 1999) did gaming software sales reach their highest point and in what year did they reach their lowest point? 43. Based on data from 1990 to 1999, electronic gaming hardware sales may be modeled by S( t ) 5 8.451t 3 2 111.9t 2 1 470.0t 1 965.0
Gaming Hardware Sales
million dollars, where t is the number of years since 1990. (Source: Modeled from Statistical
In what year (between 1990 and 1999) did gaming hardware sales reach their highest point, and in what year did they reach their lowest point? 44. Based on data from 1995 to 1999, the average earnings of a lumber and wood products industry employee may be modeled by S( n ) 5 2 0.7685n2 1 1295n 2 516,596
Lumber Industry Earnings
51. A continuous function f has the property that f r( 2 ) . 0, f r( 3 ) 5 0, and f r( 4 ) , 0. Does f have a relative maximum at x 5 3 ? Justify your answer. 52. Give an example of a nonlinear function that does not have any critical points. 53. Give an example of a continuous function that has a singular point but no relative extrema. 54. The graph of a polynomial function of degree n changes from increasing to decreasing (or vice versa) at most n 2 1 times. What is the maximum number of stationary points the function can have? Defend your conclusions. 55. A continuous function f r has two x-intercepts: x 5 2 and x 5 5 . The function f has exactly d two relative extrema. (Note: ( f ) 5 f r .) dx If f r( 3 ) 5 1, determine the x values of the relative maximum and relative minimum of f.
dollars, where n is the number of employees in thousands. (Source: Modeled from Statistical Abstract
Assuming that the employee union wants to maximize the average wage of its members, what is the optimal number of employees to have working in the lumber and wood products industry (according to the model)? 45.
DVD Market Value Based on data from 1998 to 2001, the total dollar value
5.2 Applications of Maxima and Minima
327
Analyze and interpret revenue, cost, and profit functions
The equation that expresses the price of an item as a function of the number of items sold is referred to as a demand equation. Based on data from 1992 to 2000, the wholesale demand equation for audio compact disks may be modeled by p 5 2 0.000002549x 2 1 0.001036x 1 13.12 where x is the number of CDs shipped by manufacturers (in millions) and p is the price per CD (in constant 1992 dollars). The financial data have been adjusted for inflation. (Source: Modeled from Recording Industry Association of America data.) How many CDs should be sold in order to maximize revenue? Questions such as these can be answered using the notion of the derivative. In this section, we will further illustrate applications of maxima and minima. Specifically, we will look at the business concepts of revenue, cost, and profit as well as marginal revenue, marginal cost, and marginal profit. We will also explore area and volume problems.
Revenue, Cost, and Profit

A companys revenue is the total amount of money it brings in. The companys cost is the total amount of money it spends; this includes both fixed costs and variable costs. Fixed costs are those costs that the company incurs regardless of production. Variable costs are typically expenses that vary based on the level of production. Profit is the difference between the companys revenue and its cost. The break-even point is the production level that results in revenue equaling cost. Marginal revenue is an approximation of the additional revenue generated if one more unit is produced and sold. Marginal cost is the approximate cost incurred by producing one more unit, and marginal profit is the approximate profit resulting from the production and sale of one more unit. In order to maximize profits, product manufacturers should continue to increase production as long as the marginal profit is positive. If they continue production when the marginal profit turns negative, they will actually lose money by producing more items. Marginal profit is determined by taking the derivative of the profit function. This relationship between the derivative and marginal profit is seen by looking at the difference quotient of the profit function P( x ) . Suppose we wanted to estimate the additional profit earned by selling one more item after 100 items had been sold. We know that P(101) P(100) is the additional profit earned by selling the 101st item. Note that P( 101 ) 2 P( 100 ) 5 < lim P( 101 ) 2 P( 100 ) 101 2 100 P( 100 1 h ) 2 P( 100 ) ( 100 1 h ) 2 100 P( 100 1 h ) 2 P( 100 ) h
hS0
< lim
hS0
< P r( 100 )
328
It may seem illogical to evaluate the derivative at x 5 100 when we could calculate the exact amount of the additional profit earned by producing and selling the 101st item by evaluating P( 101 ) 2 P( 100 ) . However, if P is a complicated function, it is often easier to evaluate the derivative at a single value instead of calculating P at two different values. In a similar manner, we can see that marginal cost is the derivative of the cost function and marginal revenue is the derivative of the revenue function. Well begin by working a straightforward example to demonstrate the concepts of marginal revenue and marginal profit.
EXAMPLE
Forecasting Maximum Profit

The demand equation for a certain brand of squirt gun is given by p 5 2 x 1 15 where x is the number of squirt guns sold (in thousands) and p is the price per squirt gun (in dollars). The cost to produce x thousand squirt guns is C( x ) 5 5x thousand dollars (a) If the company is currently producing 7000 squirt guns per year, should it increase or decrease squirt gun production? Explain using the concepts of marginal revenue and marginal profit. (b) Determine the production level that will maximize profit.
SOLUTION
(a) The revenue function R( x ) is the product of the price p and the number of squirt guns sold. R( x ) 5 a p dollars b ? ( x thousand squirt guns ) squirt gun ( dollars )( thousand squirt guns ) squirt gun
5 ( 2 x 1 15 ) x
5 2 x2 1 15x thousand dollars The marginal revenue is the derivative of the revenue function. Rr( x ) 5 2 2x 1 15 thousand dollars thousand squirt guns
Evaluating the marginal revenue at a production level of 7 thousand squirt guns yields Rr( 7 ) 5 2 2( 7 ) 1 15 51 thousand dollars thousand squirt guns
Increasing production from 7 thousand squirt guns to 8 thousand squirt guns will increase revenue by about $1000.
329
Profit is the difference between revenue and cost. The profit function is P( x ) 5 R( x ) 2 C( x ) 5 ( 2 x2 1 15x ) 2 ( 5x ) 5 2 x2 1 10x thousand dollars The marginal profit function is Pr( x ) 5 2 2x 1 10 thousand dollars thousand squirt guns
Evaluating the marginal profit at a production level of 7 thousand squirt guns yields Pr( 7 ) 5 2 2( 7 ) 1 10 5 2 4 thousand dollars per thousand squirt guns Increasing production from 7 thousand squirt guns to 8 thousand squirt guns will decrease profit by about $4000. Although marginal revenue is positive at a production level of 7 thousand squirt guns, marginal profit is negative at that production level. This means that any additional revenue brought in wont be enough to cover the cost of producing the extra squirt guns. The company should reduce production. (b) The company wants to maximize profit. When profit is maximized, the marginal profit will be equal to 0. Recall that the profit function is P( x ) 5 2 x2 1 10x thousand dollars and the marginal profit function is the derivative Pr( x ) 5 2 2x 1 10 thousand dollars per thousand squirt guns Setting the marginal profit to zero yields 0 5 2 2x 1 10 2x 5 10 x 5 5 thousand squirt guns Does a maximum or a minimum of the profit function occur at the critical value x 5 5? Observe that the graph of P( x ) 5 2 x2 1 10x is a concave down parabola. A concave down parabola has one relative maximum and no relative minima. The relative maximum is also an absolute maximum. Consequently, an absolute maximum of the profit function P( x ) 5 2 x2 1 10x occurs at the critical value x 5 5. At a production level of five thousand squirt guns, the profit will be maximized. The maximum profit is P( 5 ) 5 2 ( 5 ) 2 1 10( 5 ) 5 25 thousand dollars We can verify the accuracy of our result by evaluating the profit function at other production levels on either side of x 5 5. P( 4 ) 5 2 ( 4 ) 2 1 10( 4 ) 5 24 thousand dollars
330
P( 7 ) 5 2 ( 7 ) 2 1 10( 7 ) 5 21 thousand dollars Increasing or decreasing the production level from 5000 squirt guns will decrease profit.
EXAMPLE
Forecasting Maximum Revenue

Based on data from 1992 to 2000, the wholesale demand equation for audio compact disks may be modeled by p 5 2 0.000002549x2 1 0.001036x 1 13.12 where x is the number of CDs shipped by manufacturers (in millions) and p is the price per CD (in constant 1992 dollars). The financial data have been adjusted for inflation. (Source: Modeled from Recording Industry Association of America data.) (a) What is the revenue function for the wholesale shipment of CDs? (b) At a production level of 1 billion CDs, what is the marginal revenue? (c) At a production level of 2 billion CDs, what is the marginal revenue? (d) How many CDs should be shipped in order to maximize revenue?
SOLUTION
(a) Revenue is the product of an items price and the number of items sold. R( x ) 5 ( 2 0.000002549x2 1 0.001036x 1 13.12 ) x 5 2 0.000002549x3 1 0.001036x2 1 13.12x million dollars (b) Marginal revenue is the derivative of the revenue function. Rr( x ) 5 d ( 2 0.000002549x3 1 0.001036x2 1 13.12x ) dx millions of dollars millions of CDs
5 2 0.000007647x2 1 0.002072x 1 13.12
Rr( 1000 ) 5 2 0.000007647( 1000 ) 2 1 0.002072( 1000 ) 1 13.12 5 7 . 55 million dollars per million CDs Increasing production from 1000 million to 1001 million CDs will increase revenue by about 7.55 million dollars. (c) Evaluating marginal revenue at a production level of 2000 million CDs yields Rr( 2000 ) 5 2 0.000007647( 2000 ) 2 1 0.002072( 2000 ) 1 13.12 5 2 13.32 million dollars per million CDs Increasing production from 2000 million to 2001 million CDs will decrease revenue by about 13.32 million dollars. (d) We will use a sign chart and the First Derivative Test to determine the production level that will maximize revenue. Rr( x ) 5 2 0.000007647x2 1 0.002072x 1 13.12
331
The derivative is defined for all values of x, so there are no singular values. Well find the stationary values by setting the derivative equal to zero and solving the equation with the Quadratic Formula. 0 5 2 0.000007647x2 1 0.002072x 1 13.12 0 5 2 7.647x2 1 2072x 1 13,120,000 x5 5 5 2 b 6 "b 2 4ac 2a
2
Multiply equation by 1,000,000
2 2072 6 "405,607,744 2(27.647 ) 5
2 2072 6 "( 2072 ) 2 2 4( 2 7.647 )( 13,120,000 ) 2( 2 7.647 )
2 2072 6 20,140 2 15.294 < 2 1181; 1452
R'
max 1452 R(x)
10,000 5000 2000 1000 5000 10,000 x
1000 2000
FIGURE 5.9
R(x) 15,000
The revenue function has two critical values: x < 2 1181 and x < 1452. In the context of the problem, it doesnt make sense to talk about a production level of 1181 million CDs. Consequently, well throw out that critical value. Were only interested in what happens when 1452 million CDs are produced. Using the data from parts (b) and (c) of this problem, we can construct the sign chart. A relative maximum exists at x 5 1452. Is this an absolute maximum? The domain of the revenue function is not explicitly stated; however, it is implied. Clearly, x $ 0, since it doesnt make sense to talk about a negative quantity of CDs being shipped. To determine the maximum domain value, we analyze the graph of the cubic function R( x ) 5 2 0.000002549x3 1 0.001036x2 1 13.12x (Figure 5.9). Using a graphing calculator, we determined that the revenue function has an x-intercept at x 5 2481.03. As seen on the graph, the revenue function R is negative for values of x greater than 2481.03. Since it doesnt make sense to have negative revenue, x # 2481.03. Therefore, the implied domain of the function is 30, 2481.034 . Redrawing R( x ) on the implied domain yields Figure 5.10. From the graph and equation of R( x ) , we confirm the results of our earlier analysis, as shown in Table 5.7.
TABLE 5.7
x 0 1452 2481.03
R( x ) 0 13,431 0
Absolute minimum Relative and absolute maximum Absolute minimum
10,000
5000 x 0 2481.03
FIGURE 5.10
According to the model, 1452 million CDs should be shipped in order to maximize revenue. The maximum revenue is forecasted to be 13,431 million dollars (in constant 1992 dollars).
332
EXAMPLE
Forecasting Maximum Profit

Based on data from 1992 to 2000, the wholesale demand for cassette tapes may be modeled by p 5 8.954 2 156.2 dollars per cassette (in constant 1992 dollars) x
where x is the number of cassettes shipped by manufacturers (in millions). The financial data have been adjusted for inflation. In order for the price to be positive, x must equal at least 17.5 million cassettes. (Source: Modeled from Recording
Industry Association of America data.)
The cassette cost function for cassette tape retailers is the same as the revenue function for cassette tape manufacturers, since retailers purchase their cassettes from the manufacturers. (a) Find the product cost function for cassette tape retailers. (b) Assuming that the average retail price of a cassette is p 5 11 2 400, find the revenue function for cassette tape retailers. (c) Find the profit function for cassette tape retailers. (d) Determine what quantity of cassettes will maximize the profit of cassette tape retailers.
SOLUTION
x
(a) The cost function for cassette tape retailers is C( x ) 5 a 8.954 2 156.2 x x b
with the domain 317 . 5, ` ). (b) The revenue function for cassette tape retailers is R( x ) 5 a 11 2 5 11x 2 x bx 400
5 8.954x 2 156.2 million dollars
x2 million dollars 400
(c) The profit function for cassette tape retailers is P( x ) 5 R( x ) 2 C( x ) 5 a 11x 2 5 2.046x 2 x2 b 2 ( 8.954x 2 156.2 ) 400
x2 1 156.2 with domain 317 . 5, ` ) 400 2x 400 x with domain 317 . 5, ` ) 200
(d) The marginal profit function is Pr( x ) 5 2.046 2 5 2.046 2
333
We want to know when the marginal profit is zero. 0 5 2.046 2 x 200
x 5 2.046 200 x 5 409.2 Evaluating the derivative on either side of the critical value x 5 409.2, we get Pr( 0 ) 5 2.046 Pr( 500 ) 5 2.046 2
P' + max 409.2
500 200
5 2 0.454 The profit function has a relative maximum when 409.2 million cassettes are sold, as shown in the sign chart and Table 5.8.
TABLE 5.8
x 17.5 409.2 1,000
P( x ) 191.2 574.8 2297.8

Relative and absolute maximum
We do not have a right-hand endpoint, since the domain of the function is 317 . 5, ` ) . However, we can calculate the profit at a large value of x greater than 409.2 and compare it to P( 409.2 ) . Additionally, since the graph of the profit function is a concave down parabola, we know that there will be an absolute maximum at the vertex. The vertex is ( 409.2, 574.8 ) . According to the model, the maximum profit of $574.8 million occurs when 409.2 million cassettes are sold.
EXAMPLE
Minimizing Inventory Costs

Some authors turn to self-publishing when they are unable to convince a publisher to produce their work. Authors such as Deepak Chopra, Virginia Woolf, and James Joyce have self-published. (Source: The Self Publishing Manual by Dan Poynter.) An author anticipates that she will sell 10,000 copies of her book annually. The setup cost for each print run is $300. The storage cost for each printed book is $1.50 per year. Assuming that customer demand for the book is constant for the 250 business days of a year, determine the number and size of each print run that will minimize her inventory cost. (Inventory cost is the sum of the setup and storage costs.)
SOLUTION The author expects to sell 10,000 copies in the 250 business days
10,000 250
of the year. The number of books sold per day is
5 40. She could print
334
40 books a day; however, her annual setup cost would be $75,000! She could print all 10,000 books at once; however, her annual storage cost would be $7500. We seek to determine the size of the print run that will minimize her annual inventory cost. Let x be the number of books in a print run. Since the demand is 40 books per day, the number of books in storage will decrease linearly from x books to 0 books at a rate of 40 books per day. Because the number of books in storage decreases linearly, the average number of books in storage is equal to the average of the starting and ending quantities. That is, the average number of books in storage is given by
x10 2
5 2. On the same day that the number of books from the first
print run reaches zero, the x books from the second print run are delivered. For example, if 400 books were produced in a print run, then the number of books in inventory would be as shown in Figure 5.11.
400 Number of books in inventory 350 300 250 200 150 100 50 0 0 5 10 15 Number of business days 20
FIGURE 5.11
It takes 10 days to sell all of the books from a print run. The average number of 5 200 books. Since this books in inventory over each 10-day period is 2 value is the same for every 10-day period, the average inventory over the course of the year is 200 books when 400 books are produced per print run. Since the average number of books in storage over the course of the year is x and the storage cost is $1.50 per book per year, the annual storage cost is 2 x S( x ) 5 1.5 a b 2 5 0.75x dollars Since she needs to produce 10,000 books annually, the total number of print runs 10,000 is equal to x , where x is the number of books in a print run. Since the setup cost of each print run is $300, her annual setup cost is C( x ) 5 300 a 10,000 x b
400 1 0
5 3,000,000x21 dollars
335
The authors annual inventory cost is the sum of the storage and the setup costs. I( x ) 5 S( x ) 1 C( x ) 5 0.75x 1 3,000,000x21 Differentiating I and setting it equal to 0 yields 0 5 0.75 2 3,000,000x22 3,000,000 5 0.75 x2 3,000,000 5 0.75x2 4,000,000 5 x2 x 5 2000
I' (x) min of I 2000 +
The critical value of I( x ) is x 5 2000. We construct a sign chart for Ir( x ) . The inventory cost function is minimized when 2000 books are ordered. The author will need to make five 2000-book orders during the year. Her minimum annual inventory cost is given by I( 2000 ) 5 0.75( 2000 ) 1 3,000,000( 2000 ) 21 5 $3000
Area and Volume

Shipping companies such as United Parcel Service (UPS) and the U.S. Postal Service (USPS) classify packages by size and weight. USPS advises customers that Parcel PostTM is the best value when shipping a single package or a small number of packages. A Parcel Post package can weigh up to 70 pounds and measure up to 130 inches in length and girth combined. (Source: U.S. Postal Service.) The length of a package is the length of the longest side, and the girth of a package is the distance around its thickest part. Businesses are often interested in fitting the maximum amount of their product into a box whose dimensions meet the postal service guidelines. They can save a substantial amount of money in shipping costs by conforming to the postal service standard.
EXAMPLE
Maximizing Volume
A compact disk in a plastic jewel case weighs 3.75 ounces and has dimensions 5.6 s 3 4.9 s 3 0.41 s . A box 22 inches long can fit four CDs placed end to end with an additional 2.4 s of space to ensure ease of packing. The box will be shipped Parcel Post. (a) What are the dimensions of the 22-inch rectangular box that conforms to the Parcel Post criteria that has maximum volume? (b) How many CDs could be packed into the box without exceeding the 70-pound weight restriction?
SOLUTION To help us visualize what is going on in the problem, we draw a picture of the box (Figure 5.12).
336
w h h w
FIGURE 5.12
l=22 inches
(a) The volume of a box is V 5 lwh. We want to maximize V 5 22wh; however, we cant take the derivative of V because it is a function of two variables, w and h. We must find a secondary equation that will allow us to write one of the variables in terms of the other. The girth of the package is g5w1h1w1h 5 2w 1 2h A Parcel Post package can measure up to 130 inches in length and girth combined ( l 1 g 5 130 ) . l 1 g 5 130 22 1 ( 2w 1 2h ) 5 130 2w 1 2h 5 108 w 1 h 5 54 w 5 54 2 h Now that we have w in terms of h, we may write V in terms of a single variable. V 5 22wh 5 22( 54 2 h ) h 5 1188h 2 22h2 Do we have any domain restrictions on h? Yes. Clearly, h $ 0 and w $ 0, since neither the height nor the width of the package can be negative. Recall that w 5 54 2 h. Thus we may write w $ 0 in terms of h. w$0 54 2 h $ 0 54 $ h The domain of the volume function V 5 1188h 2 22h2 is 30, 544 . We must maximize V on this interval. V r( h ) 5 1188 2 44h 0 5 1188 2 44h 44h 5 1188 h 5 27
Set V r( h ) 5 0
h # 54
337
The critical value of the volume function is h 5 27. Since V is a quadratic equation with a negative coefficient on h2, the graph of V is a concave down parabola, and a relative and absolute maximum occurs at h 5 27 (see Table 5.9).
TABLE 5.9
h 0 27 54
V( h ) 0 16,038 0
Relative and absolute maximum
We need to find the value of w in order to determine the dimensions of the box. w 5 54 2 h 5 54 2 27 5 27 A rectangular box of size 22 s 3 27 s 3 27 s will have a volume of 16,038 cubic inches. This is the maximum volume of a box 22 inches long that meets the Parcel Post criteria. (b) Well first determine how many CDs can fit across the width of the box. 27 inches Width of box 5 Width of CD 5.6 inches per CD 5 4.8 CDs Since it doesnt make sense to talk about a fraction of a CD, well round this figure to 4 CDs. The width of a row 4 CDs wide is 5.6( 4 ) 5 22.4 inches. There is an additional 4.6 inches of space available for packing material. Well then determine how many CDs can fit across the length of the box. Length of box 22 inches 5 Length of CD 4.9 inches per CD 5 4.5 CDs Since it doesnt make sense to talk about a fraction of a CD, well round this figure to 4 CDs. The length of a row 4 CDs long is 4.9( 4 ) 5 19.6 inches. There is an additional 2.4 inches of space available for packing material. (This confirms the information given at the start of the problem.) Well next determine how high we can stack the CDs. Height of box 27 inches 5 Height of CD 0.41 inch per CD < 65 CDs
338
We can stack the CDs 65 units high. Visually, a stack 4 CDs wide, 4 CDs long, and 24 CDs high in the box would look like Figure 5.13.
FIGURE 5.13
A total of 4 ? 4 ? 65 5 1040 CDs can fit in the box. The total weight of the CDs will be 1040 ? 3.75 5 3900 ounces. There are 16 ounces in a pound. The weight of the CDs is 3900 ounces ? 1 pound < 244 pounds 16 ounces
Since the weight of the package exceeds the 70-pound limit, we will need to put fewer CDs in the box. Since 70 pounds is equivalent to 1120 ounces, the combined weight of the CDs and packing material can be at most 1120 ounces. 1120 ounces 2 5 298 3 CDs ounces 3 . 75 CD We can pack at most 298 CDs into the box. Depending upon the weight of the packing materials, we may have to remove additional CDs to remain within the weight limit.
EXAMPLE
Minimizing Use of Resources

Moving companies are hired by businesses and private individuals to pack, load, move, and unload household goods. As part of their service, movers bring cardboard boxes that can be easily constructed and easily broken down. Boxes are purchased by the moving company from a box manufacturer. In an effort to cut costs, a box manufacturer wants to design a closable box with a square base that has a volume of 8 cubic feet and uses the least amount of cardboard. The box must be at least 24 inches in height and will be constructed according to the design shown in Figure 5.14. What are the dimensions of the box that uses the least material?
339
Top flap 1
2h
Side 1 w w
Top flap 1
Top flap 2
Top flap 3
Top flap 4
2h
Side 1
Side 2
Side 3
Side 4
Bottom flap 1 w
Bottom flap 2 w
Bottom flap 3 w
Bottom flap 4 w
1 4
FIGURE 5.14
SOLUTION The surface area of the cardboard material is
A 5 ( h 1 2h 1 h )( w 1 w 1 w 1 w ) 1 5 ( 4h )( 4w ) 1 5 16hw 1 1 hw 2
1 w( 2h ) 4
1 hw 2 5 16.5hw square feet Since we want to minimize the amount of cardboard used, we must minimize the surface area. We are unable to find Ar because it is a function of two variables. We must use a secondary equation to write one variable in terms of the other. The volume of the box is V 5 w ? w ? 2h 5 2hw 2 Since the box has a volume of 8 cubic feet, we have 8 5 2hw 2 h5 8 2w 2
5 4w 22 Substituting this result into the surface area equation, we get A 5 16.5hw 5 16.5( 4w22 ) w 5 66w 21 Do we have any restrictions on w? Clearly, w . 0, since the box must have a positive width. (If w 5 0, the surface area function is undefined.) Additionally, the box is required to have a height of at least 24 inches (2 feet). That is, 2h $ 2 h$1
340
Since h 5 4w22, we can write h $ 1 in terms of w.

f (w) 4 3 2 1 2 1 1 2 3 4 1 2 w
h$1 4w
22
$1
4 $ w2 4 2 w2 $ 0 Graphically speaking, we want to know when the graph of f ( w ) 5 4 2 w 2 is on or above the horizontal axis. [This is when f ( w ) $ 0.] Rewriting f ( w ) in the standard form of a quadratic equation 3 f ( w ) 5 aw 2 1 bw 1 c4 yields f ( w ) 5 2 w 2 1 4. Since a , 0, the parabola is concave down. Since c 5 4, the graph has a vertical intercept of ( 0, 4 ) . Since f ( 2 2 ) 5 0 and f ( 2 ) 5 0, ( 2 2, 0 ) and ( 2, 0 ) are the horizontal intercepts of f ( w ) 5 2 w 2 1 4 (Figure 5.15). From the graph, we see that when 2 2 # w # 2, f ( w ) 5 2 w 2 1 4 $ 0. Recall that since w is the width, we also required w . 0. Thus the domain of the surface area function is ( 0, 24 . To minimize the surface area, we will differentiate A. A( w ) 5 66w 21 Ar( w ) 5 2 66w 22 52 66 w2
FIGURE 5.15
Ar( w ) is negative for all values of w in the domain of A, so there are no critical values. Since the derivative is negative, the surface area function is decreasing over its entire domain. Consequently, the smallest area will occur at the rightmost endpoint, w 5 2. This may be easily confirmed by looking at a table of values for A (Table 5.10).
TABLE 5.10
w 0 0.5 1.0 1.5 2.0 When w 5 2, h 5 1, since
A( w ) Undefined 132 66 44 33
Absolute minimum
h5 5
4 w2 4 (2)2 4 4
51 The box has dimensions w 3 w 3 2h, so the optimal dimensions are 2r 3 2r 3 2r.
341
EXAMPLE
7
0.125in.
Minimizing Use of Resources

Many bulk food commodities are packaged in #10 cans. These cylindrical metal cans are 6.875 inches tall and have a 7.25-inch diameter. There is a 0.125-inch lip on the top and on the bottom of the can, which reduces the height of the interior of the can (from lid to lid) to 6.625 inches (Figure 5.16). The volume of a cylindrical can is V 5 pr2h, where r is the radius of the can and h is the interior height of the can. A cylindrical can may be constructed from a flat sheet of metal of length 2pr and two circular lids each with surface area A 5 pr2. The height of the sheet of metal is h 1 0.5 (Figure 5.17).
7.25in.
6.875in. #10 Can
FIGURE 5.16
h+0.5
Side
2r Side of can
Top
Bottom
FIGURE 5.17
The additional 0.5 inch of metal is required to create the lip at the top and the bottom of the can. (The metal initially extends upward about 0.25 inch but is folded in half when the can is sealed, creating the 0.125-inch top lip. We assume that the 0.125-inch bottom lip was created in a similar manner.) (a) What is the volume of a #10 can? (b) How much metal is required to construct a #10 can? (c) Is it possible to construct a can with the same volume as a #10 can that requires less metal? If so, what is the minimum amount of metal required?
SOLUTION
(a) The diameter is 7.25 inches. Since the radius r is equal to half of the diameter, r 5 0.5( 7.25 ) 5 3.625 The volume V of a cylinder with radius r and height h is given by V 5 pr2h. V 5 pr 2h 5 p ( 3.625 ) 2 ( 6.625 ) 5 273.5 cubic inches The volume of the can is 273.5 cubic inches.
342
(b) The amount of metal required to construct the sides of the can is S 5 2pr( h 1 0.5 ) 5 2p( 3.625 )( 6.625 1 0.5 ) 5 162.3 square inches The amount of metal required to construct each lid is L 5 pr 2 5 p ( 3.625 ) 2 5 41.28 square inches The total amount of metal required is A 5 2pr ( h 1 0.5 ) 1 pr 2 1 pr 2 5 162.3 1 41.28 1 41.28 5 244.9 square inches (c) The surface area of a cylindrical can with a 0.125-inch lip at the top and bottom of the can is A 5 2pr ( h 1 0.5 ) 1 pr 2 1 pr 2 5 2prh 1 pr 1 2pr 2 5 pr ( 2h 1 1 1 2r ) In order to minimize A, we must find Ar. However, A is presently a function of two variables. We must use a secondary equation to solve for one variable in terms of the other. The volume of the can is V 5 pr2h. We want the can to have the same volume as a #10 can. So V 5 pr 2h 273.5 5 pr 2h h5 273.5 pr2
The volume of a #10 can is 273.5 cubic inches
We may now rewrite the surface area equation in terms of a single variable, r. A 5 pr ( 2h 1 1 1 2r ) 5 pr c 2 a 5 pr a 5 273.5 b 1 1 1 2r d pr 2
547 1 1 1 2r b pr 2
547 1 pr 1 2pr 2 r
5 547r 21 1 pr 1 2pr 2 Ar( r ) 5 2 547r 22 1 p 1 4pr 052 547 1 p 1 4pr r2
343
We will use the graphing calculator to find the x-intercept(s) of the graph of Ar( r ) (Figure 5.18).
FIGURE 5.18
The only critical value is r 5 3.436. We can visually see that Ar( r ) , 0 for values of r , 3.436. Likewise, Ar( r ) . 0 for values of r . 3.436. Since the sign of Ar( r ) changes from negative to positive at r 5 3.436, a relative minimum of A occurs at r 5 3.436. The corresponding value of h is h5 273.5 p( 3.436 ) 2
5 7.374 inches Using these values of h and r, we calculate the surface area. A( r ) 5 547r 21 1 pr 1 2pr 2 A( 3.436 ) 5 547( 3.436 ) 21 1 p( 3.436 ) 1 2p( 3.436 ) 2 5 244.2 square inches Remarkably, the #10 can very nearly minimizes the amount of metal required (244.9 square inches versus the minimal 244.2 square inches).
Just In Time
Extraneous Solutions
When modeling real-world situations with mathematical equations, sometimes a mathematical solution to the equation doesnt make sense in the realworld context. For this reason, it is critical to check that a solution makes practical sense before declaring it a solution to the problem. For example, a 1600-square-foot rectangular garden is to be installed in a backyard that is 80 feet long and 45 feet wide. The entire garden will be fenced to keep out small animals. What is the minimum amount of fencing that will be needed? We let l represent the length of the garden and w represent the width of the garden. We know that lw 5 1600. The perimeter of the garden is P 5 2l 1 2w. We are to minimize the perimeter of the garden. Solving 1600 lw 5 1600 for l yields l 5 w . (Continued)
344
Rewriting the perimeter equation in terms of a single variable yields 1600 b 1 2w w 5 3200w21 1 2w Pr 5 2 3200w22 1 2 2 3200 05 12 w2 3200 52 w2 3200 5 2w2 w2 5 1600 w 5 640 It doesnt make sense to talk about a width of 2 40 feet. The only solution that makes sense is w 5 40 feet. The length and width are 40 feet and the perimeter is 160 feet. P 5 2a
EXAMPLE
Minimizing Landscaping Costs

A landscape designer offers an Economy garden package to her clients. The rectangular garden has flowers along the front of the garden and shrubs along the back and sides (Figure 5.19).
FIGURE 5.19
The designer charges $25 per linear foot for the shrubs and $6 per linear foot for the flowers. A client wants to install a 400-square-foot garden. What will be the dimensions of the least expensive garden?
345
SOLUTION We begin by labeling the figure. We let l represent the length of the garden and w represent the width of the garden (Figure 5.20). The area of the garden is given by
A 5 lw 400 5 lw
l
A=lw lw=400
FIGURE 5.20
Assuming that the shrubs extend across the back of the garden and fully down both sides, the cost of the shrubs (in dollars) is S 5 25( w 1 l 1 w ) 5 25l 1 50w Assuming that the flowers extend across the front of the garden, the cost of the flowers (in dollars) is F 5 6l The combined cost for the garden is C 5 ( 25l 1 50w ) 1 6l 5 31l 1 50w We want to minimize this cost. We cant differentiate C because it contains two input variables, l and w. However, recall that lw 5 400 l5 400 w
Therefore, the cost function may be rewritten as C 5 31l 1 50w 5 31 a 400 b 1 50w w
5 12,400w 21 1 50w
346
Differentiating the cost function and setting it equal to zero yields Cr( w ) 5 2 12,400w 22 1 50 0 5 2 12,400w 22 1 50 12,400 5 50 w2 12,400 5 50w 2 w 2 5 248 w < 15.75
Since w . 0, we omit the solution w < 2 15.75 Set Cr( w ) 5 0
Applying the First Derivative Test, we determine that a relative minimum occurs at w 5 15.75. (Note: Cr( 15 ) , 0 and Cr( 16 ) . 0.) To find the corresponding value of l, we substitute the value of w into the area equation. l5 5 400 w
400 15.75
5 25.40 The dimensions that will minimize the cost are 25.40 feet by 15.75 feet. The minimum cost is C 5 31l 1 50w 5 31( 25.40 ) 1 50( 15.75 ) 5 $1574.90 The shape of the garden (drawn to scale) is shown in Figure 5.21. The client may want to change the dimensions; however, any change in dimensions will result in a higher cost.
FIGURE 5.21 25.4 feet 3 15.75 feet
347
5.2 Summary
In this section, you learned how to find marginal revenue, marginal cost, and marginal profit. You also learned how to use the derivative to optimize specified areas and volumes.

1. In the equation for the area of a circle, A 5 pr2, what does each of the variables represent? 2. The volume of a cylinder is equal to its height times the surface area of one of its ends. Write the equation for the volume of a cylinder as a function of its radius and its height. 3. What is the equation for the volume of a rectangular box? Be sure to state the meaning of any variables you use. 4. Given the equations V 5 xy2 and x 1 4y 5 60, write V as a function of x. 5. Given the equations V 5 xy2 and x 1 4y 5 60, write V as a function of y. 6. The volume of a rectangular box is 900 cubic inches, and the length plus the girth of the box is 49 inches. Write the equation that you would need to solve to determine the dimensions of the box. Make sure the equation contains only one independent variable. 7. What is the domain of the function for the area of a circle, A 5 pr2? 8. Does the function for the area of a circle have an inverse function? Explain.
5.2 Exercises
In Exercises 130, use the derivative techniques demonstrated in this section to determine the answer to each question. 1.
Company Profit In 2001, the Kellogg
(b) Between 1999 and 2001, in which year was gross profit minimized? 2. Based on data from 2003 to 2005, a demand function model for the Apple iPodTM is 52.1 p5 dollars 0.00484q 1 0.152 where q is the number of iPods sold (in millions) in a year. (Source: Modeled from Apple Computer
Corporation 2005 Annual Report, pp. 3132.)
Company introduced a new breakfast cereal: Special K Red Berries cereal. It quickly achieved an impressive 1 percent market share during its first six months on the market and helped to boost corporate sales. Based on data from 1999 to 2001, the net sales of the Kellogg Company may be modeled by R( t ) 5 964.1t 2 2 993.6t 1 6984 million dollars, where t is the number of years since 1999. The cost of goods sold may be modeled by C( t ) 5 399.8t 2 2 379.9t 1 3325.1 million dollars, where t is the number of years since 1999. (Source: Modeled from Kellogg Company
2001 Annual Report, pp. 7, 27.)
(a) Find the equation for the gross profit. (This is net sales minus the cost of goods sold.)
Assume that between 0 and 200 million iPods will be sold in a given year. According to the model, (a) What is the revenue function for the iPod? (b) What is the marginal revenue function for the iPod? (c) What is the maximum revenue the company will earn from iPod sales in a single year? (d) What will be the price per iPod when the revenue is maximized?
348 3. iPod is
Based on data from 2003 to 2005, a revenue function model for the Apple R( q ) 5 52.1q million dollars 0.00484q 1 0.152
(b) Based on data from 1999 to 2001, the net operating revenue of the Coca-Cola Company may be modeled by R( s ) 5 2 0.4029s 2 1 14.44s 2 109.2 billion dollars, where s is the number of unit cases sold (in billions). Find the marginal revenue function. (c) According to the model, at what unit case production level will revenue be maximized? (d) Rewrite the revenue function as a function of t (years since 1999) instead of as a function of s (billions of unit cases sold). (Hint: Use a composition of functions.) (e) According to the model, in what year will revenue be maximized? 6.
Company Profit Based on data from 1999 to 2001, the net operating revenues of the Coca-Cola Company may be modeled by R( t ) 5 2 201t 2 1 806t 1 19,284
where q is the number of iPods sold (in millions) in a year. (Source: Modeled from Apple Computer
Corporation 2005 Annual Report, pp. 3132.)
Suppose that the cost to produce q million iPods is given by C( q ) 5 35q million dollars Assume that between 0 and 200 million iPods will be sold in a given year. According to the model, (a) What is the maximum profit? (b) What price yields the maximum profit?
4.
Company Revenue Uline Shipping
Supply Specialists sell 40 s 3 48 s hardwood pallets. The price they charge per pallet is reduced as the order quantity increases. The price of hardwood pallets is given by the piecewise function 2 5.5x 1 124.5 1 # x # 2 105 3 # x # 10 dollars, where x is the number of groups of 10 pallets. (Source: www.uline.com.) (a) Write the equation for revenue as a function of the number of 10-pallet groups sold. (b) Determine the marginal revenue when 10, 20, 30, and 40 pallets are sold. (c) Explain the real-world meaning of the results of (b). p5 e 5.
Company Revenue In its 2001 Annual Report, the Coca-Cola Company reported that Our worldwide unit case volume increased 4 percent in 2001, on top of a 4 percent increase in 2000. The increase in unit case volume reflects consistent performance across certain key operations despite difficult global economic conditions. Our business system sold 17.8 billion unit cases in 2001. (Source:
million dollars and the cost of goods sold may be modeled by C( t ) 5 2 177.5t 2 1 372.5t 1 6009 million dollars, where t is the number of years since 1999. (a) Find the gross profit function. (b) According to the model, in what year is the gross profit projected to reach a maximum? (c) Does the result of part (b) seem reasonable? Explain. 7.
Company Costs Frito-Lay North America, a subsidiary of PepsiCo, produces Doritos, Cheetos, Fritos corn chips, and a variety of other salty, sweet, or grain-based snacks. Based on data from 1999 to 2001, the net sales (revenue) of Frito-Lay North America may be modeled by R( t ) 5 2 168t 2 1 907t 1 8232
million dollars and the operating profit (earnings before interest and taxes) may be modeled by P( t ) 5 2 47.5t 2 1 283.5t 1 1679 million dollars, where t is the number of years since 1999. (Source: Modeled from 2001 PepsiCo
Annual Report, pp. 23, 44.)
Coca-Cola Company 2001 Annual Report, p. 46.)
A unit case is equivalent to 24 eight-ounce servings of finished beverage. (a) Find a model for the number of unit cases sold, s, as a function of t, where t is the number of years since 1999.
(a) In what year are net sales projected to reach a maximum?
349
(b) Find the cost function for Frito-Lay North America. (c) According to the model, in what year are costs expected to reach a maximum? (d) Compare the results of (a) and (c). Do the results seem reasonable? 8.
Company Revenue Versus Profit Based on data from 1999 to
2001, the net sales (revenue) of Pepsi-Cola North America may be modeled by R( t ) 5 2 65.5t 2 2 749.5t 1 2605 million dollars and the operating profit may be modeled by P( t ) 5 6t 2 1 76t 1 751 million dollars, where t is the number of years since 1999. (Source: Modeled from PepsiCo 2001
Annual Report, p. 44.)
(b) Shortly before construction, the homeowner decides to increase the 100-foot distance around the pool by 1 foot. For both shapes of pool, use the derivative to approximate how much the surface area of the pool will increase by increasing the perimeter (circumference) by 1 foot. (c) Does your result in (b) confirm your conclusion in (a)? Explain. 11.
Apple Farming Historically, many
(a) Find the marginal revenue and marginal profit functions. (b) At what rate was revenue increasing in 2001? (c) At what rate was profit increasing in 2001? (d) Project the rate of change in revenue and in profit in 2014. Does this projection seem reasonable? Explain. 9.
Employee Wages Based on data from
1991 to 2001, the average wage of a Ford Motor Company employee may be modeled by W( t ) 5 2 0.003931t 4 1 0.1005t 3 2 0.8295t 2 1 3.188t 1 16.48 dollars per hour, where t is the number of years since 1990. (Source: Modeled from Ford Motor
Company 2001 Annual Report, p. 71.)
apple farmers spaced trees 40 feet by 40 feet apart (27 per acre). Trees typically took 25 years to reach their maximum production of 500 bushels per acre. (A bushel is about 44 pounds.) In recent years, agriculturists have created dwarf and semi-dwarf varieties that allow trees to be spaced 10 feet by 10 feet apart. (Source: USDA.) Suppose that a farmer has an apple orchard with 40 trees per acre. The orchard yields 10 bushels per tree. The farmer estimates that for each additional tree planted (per acre), the average yield per tree is reduced by 0.1 bushel. If y 5 f ( x ) is the total number of bushels of apples produced per acre when an additional x trees per acre are planted, determine how many additional trees should be planted in order to maximize the number of bushels of apples produced. 12. Apple Farming Suppose that a farmer has an apple orchard with 30 trees per acre. The orchard yields 12 bushels per tree. The farmer estimates that for each additional tree planted (per acre), the average yield per tree is reduced by 0.1 bushel. If y 5 f ( x ) is the total number of bushels of apples produced per acre when an additional x trees per acre are planted, determine how many additional trees should be planted in order to maximize the number of bushels of apples produced. 13. Apple Supplier Prices A fruit farmer sells apples to a grocery store chain. The amount of apples the store buys depends linearly upon the price per pound the farmer charges. The farmer estimates that for every $0.02 per pound increase in the price, the store will reduce its order by 50 pounds. The store presently orders 500 pounds per week and pays $0.18 per pound. What price should the farmer charge in order to maximize her revenue from apple sales?
Was the average wage changing more rapidly in 1995 or 2000? 10. Pool Size Optimization A pool builder makes two types of economy-priced pools: square and circular. He estimates the price of the job by multiplying the perimeter or circumference of the pool (in linear feet) by a fixed price per linear foot. A homeowner wants her pool to have the maximum amount of water surface area for the lowest possible price. (a) Should the homeowner have a square or a circular pool built? Explain. (The area of a 1 square with a perimeter p is S 5 16 p2. The area of a circle with circumference c is 1 A 5 4p c2.)
350
14. Apple Retailer Prices A grocery store has priced apples at $0.70 per pound and sells 1000 pounds per week. The amount of apples the store sells depends linearly upon the price per pound the store charges. The store manager estimates that for every $0.04 per pound increase in the price, the store will reduce its sales by 100 pounds. What price should the store charge for its apples in order to maximize revenue from apple sales? 15.
Company Sales Based on data from 1993 to 2002, the annual sales of Starbucks Corporation may be modeled by S( t ) 5 29.23t2 1 79.33t 1 177.4
The Rectangular Garden has shrubs along the back and sides of the garden and flowers along the front of the garden. (See figure.)
The designer charges $20 per linear foot for the shrubs and $5 per linear foot for the flowers. A client wants to install a 250-square-foot garden. What will be the dimensions of the least expensive garden? 19. Inventory Cost An author anticipates that she will sell 6,000 copies of her book annually. The setup cost for each print run is $300. The storage cost for each printed book is $2.50 per year. Assuming that customer demand for the book is constant for the 250 business days of a year, determine the size of each print run that will minimize her inventory cost. 20. Inventory Cost An author anticipates that she will sell 5,000 copies of her book annually. The setup cost for each print run is $300. The storage cost for each printed book is $1.92 per year. Assuming that customer demand for the book is constant for the 250 business days of a year, determine the size of each print run that will minimize her inventory cost. 21. Inventory Cost An author anticipates that he will sell 500 copies of his book annually. The setup cost for each print run is $300. The storage cost for each printed book is $1.20 per year. Assuming that customer demand for the book is constant for the 250 business days of a year, determine the size of each print run that will minimize his inventory cost. 22. Inventory Cost An author anticipates that he will sell 12,000 copies of his book annually. The setup cost for each print run is $300. The storage cost for each printed book is $1.25 per year. Assuming that customer demand for the book is constant for the 250 business days of a year, determine the size of each print run that will minimize his inventory cost. 23.
Fleet Vehicle Sales
million dollars, where t is the number of years since 1993. According to the model, how much more rapidly were sales increasing in 2001 than in 1999? 16.
Company Sales Based on data from 1991 to 2001, the franchised sales of McDonalds Corporation may be modeled by S( t ) 5 2 41.293t2 1 1729.0t 1 11,139
million dollars, where t is the number of years since 1990. Calculate the instantaneous rate of change of sales in 1999 and 2001. Explain the financial significance of the result. 17. Landscape Design A landscaping company offers its clients the Rectangular Garden package at a reduced rate. The Rectangular Garden has shrubs along the back and sides of the garden and flowers along the front of the garden. (See figure.)
The designer charges $25 per linear foot for the shrubs and $8 per linear foot for the flowers. A client wants to install a 500-square-foot garden. What will be the dimensions of the least expensive garden? 18. Landscape Design A landscaping company offers its clients the Rectangular Garden package at a reduced rate.
A business that owns or leases 10 or more vehicles may
351
qualify for auto manufacturer fleet purchase incentives. Incentives vary from $300 to $7000 on new vehicles. In August 2003, the manufacturers fleet incentive for a 2004 Chevrolet Cavalier was $1400. (Source: www.fleet-central.com.) An auto dealer offers an additional discount to fleet buyers who purchase one or more new Cavaliers. To encourage sales, the dealer reduces the after-incentive price of each car by x%, where x is the total number of cars purchased. Assuming that the pre-incentive price of a 2004 Chevrolet Cavalier is $14,400, how many vehicles would the dealer need to sell in order to maximize revenue? 24. In August 2003, the manufacturers fleet incentive for a 2004 Cadillac Seville was $2000. (Source: www
Fleet Vehicle Sales
how many vehicles would the dealer need to sell in order to maximize revenue? 27. Pricing Analysis Based on the results of Exercises 23 to 26, what additional restrictions (if any) should the dealer place on his advertised discount in order to ensure that selling additional vehicles wont reduce his revenue? 28. Profit Lost Due to Waste A fruit vendor purchases x pounds of fruit from her supplier. She estimates that 0.2x% of each pound of produce she buys spoils before it is purchased by a consumer. For each pound of fruit she sells, she makes a profit of $0.40. How many pounds of fruit should she buy in order to maximize profit? 29. Profit Lost Due to Waste A fruit vendor purchases x pounds of fruit from his supplier. He estimates that 0.2x% of each pound of produce he buys spoils before it is purchased by a consumer. For each pound of fruit he sells, he makes a profit of $0.50. How many pounds of fruit should he buy in order to maximize profit? 30. Profit Analysis For many commodities, the marginal profit fluctuates. Based on the results of Exercises 28 and 29, what effect does the profit per pound have on the quantity of fruit that maximizes profit?
.fleet-central.com.)
An auto dealer offers an additional discount to fleet buyers who purchase one or more new Cadillac Sevilles. To encourage sales, the dealer reduces the after-incentive price of each car by 2x%, where x is the total number of cars purchased. Assuming that the pre-incentive price of a 2004 Cadillac Seville is $41,000, how many vehicles would the dealer need to sell in order to maximize revenue? 25. In August 2003, the manufacturers fleet incentive for a 2004 Buick Century was $1800. (Source: www
Fleet Vehicle Sales
An auto dealer offers an additional discount to fleet buyers who purchase one or more new Buick Century vehicles. To encourage sales, the dealer reduces the after-incentive price of each car by 200x dollars, where x is the total number of cars purchased. Assuming that the pre-incentive price of a 2004 Buick Century is $19,000, how many vehicles would the dealer need to sell in order to maximize revenue? 26. In August 2003, the manufacturers fleet incentive for a 2004 Pontiac Bonneville was $900. (Source: www
Fleet Vehicle Sales
31. How can calculus be used to solve maximization and minimization problems? 32. Setting up a maximization or minimization problem is often the most difficult part of finding a solution. What steps do you recommend students follow in order to be successful at setting up maximization or minimization problems? 33. The volume of a cylindrical can is the product of its height and pi times the square of its radius. The surface area of a cylindrical can is the product of its height and pi times twice its radius plus 2 times the square of its radius times pi. If the surface area of the can is 360p square inches, what is the equation for the volume of the can as a function of its radius? State the meaning of any variables you use. 34. When solving optimization problems in a realworld context, why is it especially important to identify the domains of the functions involved?
An auto dealer offers an additional discount to fleet buyers who purchase one or more new Pontiac Bonneville vehicles. To encourage sales, the dealer reduces the after-incentive price of each car by 250x dollars, where x is the total number of cars purchased. Assuming that the pre-incentive price of a 2004 Pontiac Bonneville is $24,000,
352
35. Your group is asked to work together to solve the following problem: A rectangular corral is to be built alongside a barn using 120 feet of fencing. The side of the barn adjacent to the corral is 50 feet long. The barn is to be used as one side of the corral. Determine the dimensions of the corral with maximum area. Your group accurately constructs and correctly labels the following diagram to graphically represent the problem.
50 ft
Each of the questions below deals with a parametric function model of the position of a roller coaster train over a six-second interval. The x value of the model represents the horizontal position of a roller coaster train (in meters) at time t seconds. The y value of the model represents the vertical position of the roller coaster train (in meters) at time t seconds. The model is x 5 t 3 2 9t 2 1 23t y 5 6t 2 t 2 36. Calculate and interpret the meaning of dx dy and . dt dt
Barn
37. Determine when

x Corral 120 2x x
dy dx 5 0 and 5 0. dt dt
38. Given that the domain of the function is 0 # t # 6, find the relative and absolute extrema of x and y. 39. Over the six-second interval, what was the slowest horizontal speed of the train? 40. Over the six-second interval, what was the slowest vertical speed of the train? 41. Do you think that the roller coaster would be safe to ride? Defend your conclusion.
One member of your group proposes the following solution. A 5 x( 120 2 2x ) 5 120x 2 2x2 Ar 5 120 2 4x 0 5 120 2 4x 4x 5 120 x 5 30 The group member asserts that the dimensions of the corral are 30 feet by 60 feet. Is this solution correct? Explain.
5.3 Concavity and the Second Derivative
Use the Second Derivative Test to find relative extrema Determine the concavity of the graph of a function
Based on data from 1992 to 1998, the average annual earnings of a prepackaged software retailer may be modeled by W( t ) 5 49.62t 4 2 637.12t 3 1 2895t 2 2 3768t 1 32,150 dollars where t is the number of years since the end of 1992. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 1123, p. 703.) In what year were annual earnings increasing the fastest, and in what year were they increasing the slowest? Did the rate of increase in earnings continually grow, or did the rate of increase in earnings shrink at some point? Questions such as these may be answered using the second derivative.
353
In this section, we will introduce the second derivative and discuss the graphical concepts of concavity and inflection points. We will use these concepts to investigate the relationship between position, velocity, and acceleration. We will also introduce the Second Derivative Test as an alternative means of finding relative extrema. We will conclude with curve-sketching techniques.
Concavity
f (x) 15 12 9 6 3 3 2 1 3 6 9 12 15 1 2 3 x
The term concavity refers to the curvature of a graph. A graph is said to be concave up if it curves upward and concave down if it curves downward. A simple rhyme is helpful in remembering the meaning of the terms. Concave up is like a cup. Concave down is like a frown. We will explore the concept of concavity by looking at the graph of the function f ( x ) 5 x3 2 12x on the interval 32 3, 34 (Figure 5.22). From x 5 2 3 to x 5 0 , the graph is curved downward (looks like a frown). At x 5 0 , the concavity appears to change. From x 5 0 to x 5 3 , the graph is curved upward (looks like a cup). Is there a way to determine the concavity of a graph algebraically? Lets see. Well begin by finding the derivative of the function and then the derivative of the derivative. The derivative of the derivative is called the second derivative and is commonly denoted by y s (read y double prime) or f s ( x ) (read f double prime of x). f ( x ) 5 x3 2 12x f r ( x ) 5 3x2 2 12 f s ( x ) 5 6x We will generate a table of values for each of the functions (see Table 5.11).
TABLE 5.11
FIGURE 5.22
x 23 22 21 0 1 2 3
f (x) 9 16 11 0 211 216 29
f r( x ) 15 0 29 212 29 0 15
f s(x) 218 212 26 0 6 12 18
We can learn much about the graph of the function from this table. The graph of the function has x-intercepts when f ( x ) 5 0 . The x-intercept of the function on the domain 32 3, 34 occurs at x 5 0, since f ( 0 ) 5 0 . The graph of the function has relative extrema when f r( x ) 5 0 and the derivative changes sign. A relative maximum occurs at x 5 2 2, since f r( 2 2 ) 5 0 and the derivative changes from positive to negative at x 5 2 2 . A relative minimum occurs at x 5 2 , since f r( 2 ) 5 0 and the derivative changes from negative to positive at x 5 2 .
354
We observe that on the interval 32 3, 0 ) , f s ( x ) , 0 , and on the interval ( 0, 34 , f s ( x ) . 0 . At x 5 0, f s ( x ) 5 0 . Observe that the graph of f is concave down when f s ( x ) , 0 and is concave up when f s ( x ) . 0 . The graph changes concavity from down to up at x 5 0. The point on the graph where a function changes concavity is called an inflection point.
CONCAVITY AND INFLECTION POINTS OF THE GRAPH OF A FUNCTION
1. A continuous function f is concave up at a point ( c, f ( c )) if f s ( c ) . 0. 2. A continuous function f is concave down at a point ( c, f ( c )) if f s ( c ) , 0. 3. A continuous function f has an inflection point at ( c, f ( c )) if f s ( c ) 5 0 or f s ( c ) is undefined and f s ( c ) changes sign at x 5 c
y 14 12 10 8 6 4 2 1 0 1 x f (x) = x 4
By referring back to the earlier rhyme, we can come up with a clever strategy to remember how to determine the concavity of a graph. When were positive, we are up (like the cup). When were negative, were down and we frown. Admittedly, the rhyme is a bit silly; however, its sheer wackiness will make the concept easier to remember. A common error made by learners in their search for inflection points is to assume that if f s ( c ) 5 0 , then ( c, f ( c )) is an inflection point. Although f s ( c ) 5 0 is a necessary condition for an inflection point, it is not a sufficient condition. Consider the function f ( x ) 5 x4 with its associated second derivative f s ( x ) 5 12x2 . Although f s ( 0 ) 5 0 , the point ( 0, 0 ) is not an inflection point of f, since the graph of f is concave up on both sides of the point (Figure 5.23).
FIGURE 5.23
EXAMPLE
Determining the Concavity of a Graph

The graph of f ( x ) 5 x4 2 4x3 on the interval 32 3, 54 is shown in Figure 5.24. Determine where the graph is concave up and where the graph is concave down. Identify all inflection points.
y 110 100 90 80 70 60 50 40 30 20 10 3 2 1 20 f (x) = x 4 4x 3
FIGURE 5.24
355
SOLUTION The graph appears to be concave up on 32 3, 04 , concave down on 30, 24 , and concave up again on 32, 54 . It is difficult to determine the exact location of inflection points visually, so we will proceed algebraically to verify our visual conclusions.
f ( x ) 5 x4 2 4x3 f r( x ) 5 4x3 2 12x2 f s ( x ) 5 12x2 2 24x In order to identify potential inflection points, we must set f s ( x ) equal to zero. 0 5 12x2 2 24x 0 5 12x( x 2 2 ) 12x 5 0 x50 or x2250 x52
Inflection points may occur at x 5 0 or x 5 2; however, we must verify that f s ( x ) changes sign at each of these points. We will do this using a sign chart for f s. We evaluate f s ( x ) using points on either side of the potential inflection points. f s ( x ) 5 12x( x 2 2 ) f s ( 2 1 ) 5 12( 2 1 )( 2 1 2 2 ) . 0 f s ( 1 ) 5 12( 1 )( 1 2 2 ) , 0 f s ( 3 ) 5 12( 3 )( 3 2 2 ) . 0 We update the sign chart for f s with the results. Since f s ( x ) changes sign at x 5 0 and x 5 2 , inflection points occur at each of these points. Since f s ( x ) . 0 on 32 3, 0 ) and on ( 2, 54 , it is concave up on those intervals. Since f s ( x ) , 0 on ( 0, 2 ) , it is concave down on that interval. Our algebraic analysis confirms our graphical estimation.
f''
+ 0
EXAMPLE
Determining the Concavity of a Graph

Determine the concavity of the graph f ( x ) 5 x3 1 3x2 1 3x 1 1 on the interval 32 2, 24 . Identify the location of any inflection points.
SOLUTION
f ( x ) 5 x3 1 3x2 1 3x 1 1 f r( x ) 5 3x2 1 6x 1 3 f s ( x ) 5 6x 1 6 We set f s ( x ) 5 0 and solve. 0 5 6x 1 6 2 6x 5 6 x 5 21 An inflection point may occur at x 5 2 1. Well evaluate the second derivative on either side of x 5 2 1 and construct a sign chart for f s .
f'' 1 +
f s ( 2 2 ) 5 6( 2 2 ) 1 6 5 26
356
f s ( 0 ) 5 6( 0 ) 1 6 f ( x ) is concave down on the interval 32 2, 2 1 ) and concave up on the interval ( 2 1, 24 . An inflection point occurs at the point ( 2 1, 0 ) (Figure 5.25).
f (x) = x 3 + 3x 2 + 3x + 1 y 4 3 2 1 4 3 2 1 1 2 3 4 1 2 3 4 x
56
FIGURE 5.25
EXAMPLE
Using the Second Derivative to Determine Maximum and Minimum Rates of Change
Based on data from 1992 to 1998, the average annual earnings of a prepackaged software retailer employee may be modeled by W( t ) 5 49.62t4 2 637.12t 3 1 2895t 2 2 3768t 1 32,150 dollars where t is the number of years since 1992. (Source: Modeled from Statistical Abstract
In what year were annual earnings increasing the fastest, and in what year were they increasing the slowest? Did the rate of increase in earnings continually grow, or did the rate of increase in earnings shrink at some point?
SOLUTION The rate of change of the earnings function is W r( t ) .
W r( t ) 5 198.48t 3 2 1911.36t 2 1 5790t 2 3768
dollars year
This function tells us how fast or how slowly earnings are increasing. We want to determine the extrema of W r( t ) . To do this, we differentiate W r( t ) . W s ( t ) 5 595.44t 2 2 3822.72t 1 5790 2 b 6 "b2 2 4ac 2a dollars per year year
We set W s ( t ) equal to zero and solve using the quadratic formula. t5 5 5
3822.72 6 "( 2 3822.72 ) 2 2 4( 595.44 )( 5790 ) 2( 595.44 ) 3822.72 6 907.08 1190.88
5 2.448; 3.972
357
W''
max of W' 2.448
min of W' 3.972
We evaluate W s ( t ) on either side of the critical points of W r( t ) and construct a sign chart for W s ( t ) . The relative extrema of W r( t ) are in fact the inflection points of W( t ) . Inflection points tell us when the growth rate of W( t ) is accelerating or slowing down. Lets interpret the financial meaning of our results. W r( 2.448 ) 5 1863 dollars per year W r( 3.972 ) 5 1513 dollars per year Lets look again at the sign chart of W s . Recall that W r is the growth rate of employee earnings (in dollars per year). Between the end of 1992 ( t 5 0 ) and the middle of 1995 ( t 5 2.448 ) , the growth rate of employee earnings was increasing. However, between the middle of 1995 and the end of 1996 ( t 5 3.972 ) , the growth rate of employee earnings was decreasing. That is, although earnings were increasing, they were increasing at a slower rate. In the middle of 1995 ( t 5 2.448 ) , the earnings growth rate was $1863 per year, but by the end of 1996 ( t 5 3.972 ) , the earnings growth rate had fallen to $1513 per year. Between the end of 1996 and the end of 1998, the growth rate of employee earnings was again increasing. We can verify our conclusions by generating a table of values for W, W r, and W s . Tables such as Table 5.12 may be quickly generated using the Technology Tip following this example.
TABLE 5.12
Years Since 1992 (t) 0 1 2 3 4 5 6
W( t ) (dollars) 32,150 30,690 31,891 33,718 35,325 37,058 40,452
W r( t ) (dollars per year) 23,768 309 1,754 1,759 1,513 2,208 5,035
W s(t) (dollars per year per year) 5,790 2,563 526 2319 26 1,562 4,290
In what year were annual earnings increasing the fastest, and in what year were they increasing the slowest? Earnings were diminishing at a rate of $3768 per year in 1992. This was the slowest rate of growth between 1992 and 1998. Earnings were increasing at a rate of $5035 per year in 1998. This was the fastest rate of growth between 1992 and 1998. Did the rate of increase in earnings grow continually, or did the rate of increase in earnings shrink at some point? Notice that the values of W r( t ) increased every year except between 1995 ( t 5 3 ) and 1996 ( t 5 4 ) . The fact that W s ( 3 ) was negative told us that the rate of increase in earnings was decreasing in 1995.
358
Creating a Table of Values 1. Use the Y editor to enter the function(s) to be evaluated. In this case we enter W, W r, and W s .
2. Press 2nd WINDOW to open Table Setup. TblStart= is the x value where we want the table to begin, and Tbl is the distance between consecutive x values. We will start the table at x 5 0 and space values one unit apart. 3. Press
2nd GRAPH
to display the table.
4. To see additional function values, use the blue arrows to navigate between rows and columns.
5. If you prefer to select the x values of the function(s) manually, press 2nd WINDOW and select Indpnt:Ask.
6. Press 2nd GRAPH to bring up the table, then enter the desired x values.
359
The Second Derivative Test

As weve seen, the second derivative may be used to determine the concavity of the graph and find the location of inflection points. The second derivative may also be used to find relative extrema. Consider the function f ( x ) 5 x3 2 12x. Differentiating the function yields f r( x ) 5 3x2 2 12 0 5 3( x2 2 4 ) 0 5 3( x 2 2 )( x 1 2 ) x2250 x52 or x1250 x 5 22
Set f r( x ) 5 0 to find critical values
The function f has critical values x 5 2 2 and x 5 2 , since f r( 2 2 ) 5 0 and f r( 2 ) 5 0 . Graphically speaking, the graph of f ( x ) 5 x3 2 12x is flat (has a horizontal tangent line) at x 5 2 2 and x 5 2 . The second derivative of f ( x ) 5 x3 2 12x is f s ( x ) 5 6x. We evaluate f s at each of the critical values of f. f s ( x ) 5 6x f s ( 2 2 ) 5 6( 2 2 ) 5 2 12 f s ( 2 ) 5 6( 2 ) 5 12
y 16 12 8 4 3 2 1 4 8 12 16 1 2 3 x f (x) = x 3 12x
Since f s ( 2 2 ) , 0, the graph of f is concave down at x 5 2 2. However, since f also has a horizontal tangent line at x 5 2 2, a relative maximum occurs there. Since f s ( 2 ) . 0, the graph of f is concave up at x 5 2. However, since f also has a horizontal tangent line at x 5 2, a relative minimum occurs there (Figure 5.26). These results are summarized in the Second Derivative Test.
THE SECOND DERIVATIVE TEST
FIGURE 5.26
Let f be a continuous function with f r( c ) 5 0. If f s ( c ) . 0 , then a relative minimum of f occurs at x 5 c . f s ( c ) , 0 , then a relative maximum of f occurs at x 5 c . f s ( c ) 5 0 , then the test is inconclusive.
EXAMPLE
Finding Relative Extrema with the Second Derivative Test

Use the Second Derivative Test to find the relative extrema of the function f ( x ) 5 x5 2 5x4.
SOLUTION
f r( x ) 5 5x4 2 20x3 0 5 5x3( x 2 4 )

Set f r( x ) 5 0 and find critical values
360
5x3 5 0 x50
or
x2450 x54
The critical values are x 5 0 and x 5 4. To apply the Second Derivative Test, we calculate f s ( x ) and evaluate it at the critical values, x 5 0 and x 5 4. Writing the second derivative in factored form will make it easier to evaluate. f s ( x ) 5 20x3 2 60x2 5 20x2( x 2 3 ) f s ( 0 ) 5 20( 0 ) 2( 0 2 3 ) 50 f s ( 4 ) 5 20( 4 ) 2( 4 2 3 ) 5 320 Since f s ( 4 ) . 0, a relative minimum occurs at x 5 4 . The relative minimum is ( 4, 2 256 ) . At x 5 0 , the Second Derivative Test is inconclusive. What is happening at x 5 0 ? Lets evaluate f r( x ) at points on either side of the critical value x 5 0 to see if the slope of f changes sign at x 5 0 . (This is the First Derivative Test.)
y 50 1 50 100 150 200 250 1 2 3 4 5 x f (x) = x 5 5x 4
f r( 2 1 ) 5 5( 2 1 ) 3( 2 1 2 4 ) 5 25 Since f r( 2 1 ) . 0, the graph of f is increasing at x 5 2 1 . f r( 1 ) 5 5( 1 ) 3( 1 2 4 ) 5 2 15 Since f r( 1 ) , 0, the graph of f is decreasing at x 5 1 . Since f r changes from positive to negative at x 5 0 , a relative maximum occurs at x 5 0 . The relative maximum is ( 0, 0 ) . The graph of the function (Figure 5.27) confirms our conclusion. Although calculations are often simpler when we use the Second Derivative Test, we may have to revert back to the First Derivative Test if the Second Derivative Test yields an inconclusive result.
FIGURE 5.27
Sometimes we want to maximize the value of the first derivative or the second derivative instead of the value of the original function. In Example 5, we will maximize the derivative using the First Derivative Test. The notation gets a bit tricky because the critical points of f r are found by setting f s equal to zero.
EXAMPLE
Using the Second Derivative in a Real-World Context

Based on data from 1991 to 2001, the average wage of a Ford Motor Company employee may be modeled by W( t ) 5 2 0.003931t 4 1 0.1005t 3 2 0.8295t 2 1 3.188t 1 16.48 dollars per hour where t is the number of years since 1990. (Source: Modeled from Ford Motor Company 2001 Annual Report, p. 71.) According to the model, did the rate of increase in employee hourly wages (in dollars) ever diminish between 1991 and 2001? What
361
was the minimum and what was the maximum annual increase in the average hourly wage?
SOLUTION The rate of increase in hourly wages is given by
W r( t ) 5 2 0.015724t 3 1 0.3015t 2 2 1.659t 1 3.188
dollars year
We want to know if W r ever decreased between 1991 and 2001. That is, we want to know when W s , 0. W s ( t ) 5 2 0.047172t 2 1 0.603t 2 1.659 dollars per year year
We set W s ( t ) 5 0 and solve using the Quadratic Formula. 0 5 2 0.047172t 2 1 0.603t 2 1.659 t5 5 2 0.603 6 "( 0.603 ) 2 2 4( 2 0.047172 )( 2 1.659 ) 2( 2 0.047172 ) 2 0.603 6 0.2249 2 0.094344
5 4.008, 8.775
W'' min of W' 4.008 + max of W' 8.775
The critical values of W r are t 5 4.008 and t 5 8.775. Well evaluate W s ( t ) on either side of the critical values of W r. W s ( 0 ) 5 2 1.659 W s ( 5 ) 5 0.1767 W s ( 10 ) 5 2 0.3462 From 1991 ( t 5 1 ) to early 1995 ( t 5 4.008 ) , the annual rate of increase in hourly wages was decreasing (since W s , 0). Although the hourly wage continued to rise during those years, each annual increase was less than that of the year before. From early 1995 ( t 5 4.008 ) to late 1999 ( t 5 8.775 ) , the annual rate of increase in hourly wages was increasing (since W s . 0). The hourly wage continued to rise during those years, and each annual increase was more than that of the year before. From late 1999 ( t 5 8.775 ) to the end of 2001 ( t 5 11 ) , the annual rate of increase in hourly wages was decreasing (since W s , 0). Although the hourly wage continued to rise during those years, each annual increase was less than that of the year before. The maximum annual increase in hourly wage will occur when W r reaches its maximum value. From the sign chart for W s , we see that a relative maximum of W r occurs when t 5 8.775. Well need to check the endpoints to see if this is an absolute maximum. The minimum annual increase in hourly wage will occur when W r reaches its minimum value. From the sign chart for W s , we see that a relative minimum of W r occurs when t 5 4.008. Well need to check the endpoints to see if this is an absolute minimum. (See Table 5.13.)
362
TABLE 5.13
Years Since 1990 (t) 1 4.008 8.775 11
Dollars per Year [W r( t ) ] 1.81 0.37 1.22 0.49

Absolute maximum of W r( t ) Relative and absolute minimum of W r( t ) Relative maximum of W r( t )
According to the model, the average hourly wage was increasing most slowly in early 1995 ( t 5 4.008 ) . At that time, it was increasing at a rate of $0.37 per year. According to the model, the average hourly wage was increasing most rapidly at the end of 1991 ( t 5 1 ) . At that time, it was increasing at a rate of $1.81 per year.
Point of Diminishing Returns

In business, we are often interested in knowing when the rate of change in revenue or profit from the sale of an item reaches a maximum value. The point at which this occurs is called the point of diminishing returns. In terms of calculus, we want to know when the derivative Rr( x ) of a revenue function R( x ) attains a maximum.
EXAMPLE
Finding the Point of Diminishing Returns

Based on data from 1990 to 2000, the revenue from sales of newspaper advertising may be modeled by A( t ) 5 2 32.08t 3 1 656.2t 2 2 1641t 1 31,914 million dollars where t is the number of years since the end of 1990. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 1272, p. 777.) Find the point of diminishing returns for newspaper advertising revenue and determine the rate at which advertising revenue is changing at that point.
SOLUTION The rate of change in newspaper advertising is Ar( t ) .
A( t ) 5 2 32.08t 3 1 656.2t 2 2 1641t 1 31,914 Ar( t ) 5 2 96.24t 2 1 1312.4t 2 1641 million dollars per year To determine when Ar( t ) attains a maximum, we must find A s ( t ) . A s ( t ) 5 2 192.48t 1 1312.4 million dollars per year per year To find the critical values of Ar( t ) , we set A s ( t ) 5 0 . 0 5 2 192.48t 1 1312.4 192.48t 5 1312.4 t5 1312.4 192.48
t < 6.8
363
A''(t)
maximum of A'(t) 6.8
At t 5 6.8, Ar( t ) attains a relative maximum. Thus the point of diminishing returns occurs at t 5 6.8. The point of diminishing returns is ( 6.8, A( 6.8 )) 5 ( 6.8, 41,011 ) . According to the model, revenue from newspaper advertising was $41,011 million in late 1997 ( t 5 6.8 ) . To determine the rate of change in revenue at the point of diminishing returns, we evaluate Ar( t ) at the critical value. (We also evaluate Ar( t ) at the implied endpoints as one strategy for verifying the accuracy of our work; see Table 5.14.)
TABLE 5.14
t 0 6.8 10
Ar( t ) 5 2 96.24t 2 1 1,312.4t 2 1641 21,641 2,833 1,859 Newspaper advertising revenue was increasing most rapidly in late 1997 ( t 5 6.8 ) . At that time, revenue was increasing at a rate of $2833 million per year.
Absolute minimum of Ar( t ) Relative and absolute maximum of Ar( t )
Position, Velocity, and Acceleration

The velocity of an object is the rate of change in its position over time. The acceleration of an object is the rate of change in its velocity over time. These relationships between position, velocity, and acceleration are nicely captured using the derivative concept.
RELATIONSHIP BETWEEN POSITION, VELOCITY, AND ACCELERATION
Let s( t ) be the function that describes the position of an object at time t. Then v( t ) 5 s r ( t ) , where v( t ) is the velocity of the object at time t. a( t ) 5 v r ( t ) 5 s s ( t ) , where a( t ) is the acceleration of the object at time t.
EXAMPLE
Using a Position Function to Determine Velocity and Acceleration

The author tracked his mileage as he drove through a residential area and into a cemetery. Every 15 seconds he recorded the mileage (accurate to 0.05 mile). Based on 1.5 minutes of data, his distance from a stoplight at the bottom of a hill is given by the position function s( t ) 5 2 0.5333t 4 1 1.333t 3 2 0.7667t 2 1 0.3167t miles where t is in minutes. Determine the velocity and acceleration functions. Then calculate the authors velocity and acceleration at 1 minute into his timed trip and at 1.25 minutes into his trip.
364
SOLUTION
v( t ) 5 sr( t ) v( t ) 5 d ( 2 0.5333t 4 1 1.333t 3 2 0.7667t 2 1 0.3167t ) dt miles minute
v( t ) 5 2 2.1332t 3 1 3.999t 2 2 1.5334t 1 0.3167 a( t ) 5 vr( t ) a( t ) 5
d ( 2 2.1332t 3 1 3.999t 2 2 1.5334t 1 0.3167 ) dt miles per minute minute
a( t ) 5 2 6.3996t 2 1 7.998t 2 1.5334
To calculate the velocity and acceleration one minute into the trip, we evaluate v( 1 ) and a( 1 ) . v( 1 ) 5 2 2.1332( 1 ) 3 1 3.999( 1 ) 2 2 1.5334( 1 ) 1 0.3167 5 0.6491 5 0.6491 miles minute miles 60 minutes ? minute 1 hour miles minute
5 38.946 miles per hour < 39 miles per hour a( 1 ) 5 2 6.3996( 1 ) 2 1 7.998( 1 ) 2 1.5334 5 0.065 5 0.065 5 0.065 5 3.9 miles per minute minute miles 1 ? minute minute miles 1 60 minutes ? ? minute minute 1 hour miles per minute minute
miles 1 ? hour minute
< 4 miles per hour per minute At 1 minute into his trip, the author was traveling at approximately 39 miles per hour and was accelerating at a rate of 4 miles per hour per minute. That is, if he maintained his current rate of acceleration for the next minute, his speed would increase by about 4 miles per hour. To calculate the velocity and acceleration 1.25 minutes into the trip, we evaluate v( 1.25 ) and a( 1.25 ) . v( 1.25 ) 5 2 2.1332( 1.25 ) 3 1 3.999( 1.25 ) 2 2 1.5334( 1.25 ) 1 0.3167 5 0.4820 miles minute miles minute
365
5 0.4820
miles 60 minutes ? minute 1 hour miles per minute minute
< 29 miles per hour a( 1.25 ) 5 2 6.3996( 1.25 ) 2 1 7.998( 1.25 ) 2 1.5334 5 2 1.535 5 2 1.535 5 2 92.10 miles per minute minute miles 1 60 minutes ? ? minute minute 1 hour miles 1 ? hour minute
< 2 92 miles per hour per minute At 1 minute and 15 seconds into his trip, he was traveling at approximately 29 miles per hour and was decelerating at a rate of 92 miles per hour per minute. That is, if he maintained his current rate of deceleration for the next minute, his speed would decrease by about 92 miles per hour. The rapid rate of deceleration indicates that he is braking, probably in preparation for turning into the cemetery. (Since he was traveling at 29 mph, we know that he wont be able to maintain the high rate of deceleration for an entire minute. If he did, his velocity would turn negative, indicating that he was moving away from the cemetery and back toward the stoplight.)
Curve Sketching
Although nowadays it is customary to graph functions using a graphing calculator, knowing how to graph functions by hand will greatly increase your understanding of calculus. If you know the x- and y-intercepts, the relative and absolute extrema, and the concavity and inflection points of a function, it is relatively easy to come up with a good sketch of the curve. We will demonstrate the curve-sketching process in Example 8.
EXAMPLE
Sketching a Polynomial Function

Sketch the graph of f ( x ) 5 x3 2 2x2 2 4x 1 8 on the interval 32 3, 44 .
SOLUTION We will set f ( x ) 5 0 to find x-intercepts, f r( x ) 5 0 to find critical values, and f s ( x ) 5 0 to determine where inflection points may occur.
x- and y-intercepts
f ( x ) 5 x3 2 2x2 2 4x 1 8 0 5 ( x3 2 2x2 ) 2 ( 4x 2 8 ) 0 5 x ( x 2 2 ) 2 4( x 2 2 )
2
Set f ( x ) 5 0 and group terms Factor each group Factor ( x 2 2 ) out of each term Factor ( x2 2 4 ) Group like terms
0 5 ( x 2 2 )( x 2 4 )
2
0 5 ( x 2 2 )( x 2 2 )( x 1 2 ) 0 5 ( x 2 2 ) 2( x 1 2 ) x2250 x52 or x1250 x 5 22
366
y 25 20 15 10 5 3 2 15 10 15 20 25 1 2 3 4 x
The graph of f ( x ) 5 x3 2 2x2 2 4x 1 8 has x-intercepts at ( 2 2, 0 ) and ( 2, 0 ) . To find the y-intercept, we evaluate f ( x ) 5 x3 2 2x2 2 4x 1 8 at x 5 0 . f ( 0 ) 5 ( 0 ) 3 2 2( 0 ) 2 2 4( 0 ) 1 8 58 The y-intercept is ( 0, 8 ) . We plot these points (Figure 5.28).
Relative and Absolute Extrema To find critical values, we find f r( x ) and set it equal to zero.
f ( x ) 5 x3 2 2x2 2 4x 1 8 f r( x ) 5 3x2 2 4x 2 4 0 5 3x 2 4x 2 4
2
FIGURE 5.28
Differentiate f ( x ) Set f r ( x ) 5 0 Factor
0 5 ( 3x 1 2 )( x 2 2 ) 3x 1 2 5 0 3x 5 2 2 x52
f ' (x) + max of f
2 3
or
x2250 x52
min of f 2
2 3
2
f ( x ) has critical values at x 5 2 3 and x 5 2 . We construct a sign chart for f r( x ) to determine where the relative extrema occur. We evaluate f ( x ) at the critical values and the endpoints (see Table 5.15).
TABLE 5.15
x 23
23
2
f ( x ) 5 x3 2 2x2 2 4x 1 8 225
9 27 < 9.5
13
2 4
y 25 20 15 10 5 3 2 15 10 15 20 25 1 2 3 4 x
0 24
We plot the corresponding points on the graph of f (Figure 5.29). From the sign chart of f r( x ) , we can also determine the intervals on which f is on those intervals. Similarly, since f r( x ) , 0 on the interval C2 , 2 B , f is 3 decreasing on that interval. We wont update the graph of f with this information yet, since we still dont know the concavity of the graph on these intervals.
2
increasing or decreasing. Since f r( x ) . 0 on C2 3, 2 B and ( 2, 44 , f is increasing 3

2
Inflection Points To find the inflection points of f, we set f s ( x ) 5 0.

f r( x ) 5 3x2 2 4x 2 4 f s ( x ) 5 6x 2 4 0 5 6x 2 4 4 5 6x x5 2 3
Simplify to
4 6 2 3
FIGURE 5.29
Differentiate f r ( x ) Set f s ( x ) 5 0
367
f '' (x)
inflection point of f
2 3
An inflection point may occur at x 5 . Well construct a sign chart for f s ( x ) and 3 see if the concavity of f changes at x 5 . 3
2
that interval. Since f s ( x ) . 0 on the interval
Since f s ( x ) , 0 on the interval C2 3, B , the graph of f is concave down on 3

2 2 3
on that interval. An inflection point occurs at x 5 .

TABLE 5.16
, 4 D , the graph of f is concave up A2 3
x
2 3
f ( x ) 5 x3 2 2x2 2 4x 1 8 4 27 < 4.7

20
Inflection point
y 25 20 15 10 5 3 2 15 10 15 20 25 1 2 3 4 x
We update the graph with the inflection point (Figure 5.30). Referring to our earlier observations regarding the increasing/decreasing behavior and concavity of f, we finish the graph (Figure 5.31).
y 25 20 15 10 5 3 2 15 10 15 20 25 1 2 3 4 x
FIGURE 5.30
FIGURE 5.31
We summarize the curve-sketching steps in the following box.
Curve Sketching To graph a function f ( x ) on an interval 3a, b4 , complete the following steps. After each step, graph the corresponding point(s). 1. Find the x-intercepts of f ( x ) by setting f ( x ) 5 0 and solving for x. 2. Find the y-intercept of f ( x ) by evaluating f ( 0 ) . 3. Find the relative extrema and increasing/decreasing behavior of f ( x ) by constructing a sign chart for f r( x ) . (Continued)
368
4. Find the absolute extrema of f ( x ) by evaluating f ( x ) at each critical value
and at the endpoints x 5 a and x 5 b .
5. Find the inflection points and concavity of f ( x ) by constructing a sign
chart for f s ( x ) . 6. Connect the points, paying attention to the increasing/decreasing behavior and concavity of f ( x ).
EXAMPLE
Sketching a Function
Using the curve-sketching techniques of calculus, graph the function f (x) 5
ln( x ) 2x
on the domain 31, 84 .
SOLUTION
x- and y-intercepts
f (x) 5 05 ln( x ) 2x ln( x ) 2x
Set f ( x ) 5 0 Multiply both sides by 2x Rewrite as an exponential function Since e0 5 1
0 5 ln( x ) e 5x
0
x51
The function has an x-intercept at ( 1, 0 ) . Since x 5 0 is not in the domain of the function, the function does not have a y-intercept.
Relative Extrema and Increasing/Decreasing Behavior

f (x) 5 ln( x ) 2x 1
f r( x ) 5 5 05
A x B ( 2x ) 2 ( 2 ) 3ln( x ) 4
( 2x ) 2
Apply the Quotient Rule
2 2 2 ln( x ) 4x2 2 2 2 ln( x ) 4x2

Set f r( x ) 5 0 Set numerator equal to 0
0 5 2 2 2 ln( x ) 2 5 2 ln( x ) 1 5 ln( x ) e1 5 eln(x) e5x
Exponentiate both sides Since eln(x) 5 x
369
f ' (x)
max of f e
The critical value of the function is x 5 e . [Although the value x 5 0 makes f r( x ) undefined, x 5 0 is not a critical value because it is not in the domain of f ( x ) .] We construct a sign chart to determine the relative extrema and the increasing/decreasing behavior of f . The graph of f is increasing on the interval 31, e ) and decreasing on the interval ( e, 84 . A relative maximum occurs at x 5 e . We evaluate f at the critical value and endpoints (see Table 5.17).
TABLE 5.17
x 1
e
f (x) 5 0
ln( x ) 2x
Absolute minimum Relative maximum, absolute maximum
1 < 0.18 2e ln( 8 ) < 0.13 16
8
y 0.20 0.15 0.10 0.05 1 0.05 x
We create a graph of the points (Figure 5.32).
Inflection Points and Concavity

f r( x ) 5 2 2 2 ln( x ) 4x2
1 2 3
2 a 2 b ( 4x2 ) 2 8x 32 2 2 ln( x ) 4 x f s(x) 5 ( 4x2 ) 2 5 5 5 05 2 8x 2 16x 1 16x ln( x ) 16x4 2 24x 1 16x ln( x ) 16x4 2 8x 33 2 2 ln( x ) 4 16x4 2 8x 33 2 2 ln( x ) 4 16x4 3 2 2ln( x ) 5 0 3 5 2ln( x ) 3 5 ln( x ) 2 x 5 e3>2
Apply the Quotient Rule
FIGURE 5.32
Factor the numerator Set f r ( x ) 5 0 Multiply both sides by 16x4
2 8x 5 0 x50
0 5 2 8x 33 2 2 ln( x ) 4 or
Since x 5 0 is not in the domain of the function, x 5 0 is not a point of inflection. We construct a sign chart for f s ( x ) .
370
f '' (x)
inflection point of f e3/2
The graph of f is concave down on 31, e3>2 ) and concave up on ( e3>2, 84 . An inflection point occurs at x 5 e3>2 < 4.5 (see Table 5.18).
TABLE 5.18
x e3>2
f (x) 5
ln( x ) 2x
Inflection point
3 < 0.17 4e3>2
y 0.20 0.15 0.10 0.05 1 0.05 x
We update the graph with the inflection point (Figure 5.33). Paying attention to the increasing/decreasing behavior and concavity of f, we finish the graph (Figure 5.34).
y 0.20 0.15 0.10 0.05 1 0.05 x
1 2 3
4 5
FIGURE 5.33
1 2 3
4 5
FIGURE 5.34
Creating a Table of Values 1. Create column headings for the functions to be evaluated. Also create a heading for the starting x-value and distance between consecutive x-values (delta x). 2. Enter a starting table value and a D x, enter the appropriate function equations, and set A2 5 E2. (Continued)
371
3. In cell A3, type A2 1 $F$2. The $ symbol tells Excel not to change the cell reference F2 when the equation is pasted into other cells.
4. Click the lower right-hand corner of the box in cell A3 and drag down several cells. This pastes the formula in cell A3 to the remainder of the cells. The reference to A2 automatically updates to reference the vertically adjacent cell. The reference to F2 remains unchanged.
5. Select the cells containing the function formulas, click the black box in the lower righthand corner of the rightmost cell, and drag down to copy the formulas. The resultant function values for the varying values of x are displayed.
5.3 Summary
In this section, you learned how to use the second derivative in discussing the graphical concepts of concavity and inflection points. You saw how the Second Derivative Test may be used as an alternative means of finding relative extrema. You also discovered that velocity is the derivative of the position function and that acceleration is the second derivative of the position function. You also learned some curve-sketching techniques.
372

1. If a function is increasing and the rate of change of the function is decreasing, what do you know about the concavity of the function? 2. If a function is increasing and the rate of change of the function is increasing, what do you know about the concavity of the function? 3. If a function is decreasing and the rate of change of the function is decreasing, what do you know about the concavity of the function? 4. If a function is decreasing and the rate of change of the function is increasing, what do you know about the concavity of the function? 5. Calculate the first and second differences for the function given in the table. x 0 1 3 4 2 f (x) 2 3 11 18 6 First Difference Second Difference
2 50 100 150 200 250 2 4 2 1
7. Describe the concavity of the function shown in the figure.

y 6 4 2 2 4 6 1 2 x
8. Describe the concavity of the function shown in the figure.

y 50 x
6. Based on the results given in the table, what do you know about the concavity of the function? x 0 1 2 3 4 f( x ) 22 21 2 7 14 First Difference 1 3 5 7 Second Difference 2 2 2
373
5.3 Exercises
In Exercises 110, indicate on the graph where the function is concave up and where the function is concave down. Mark the graph where you think inflection points occur. 1. f ( x ) 5 x3 2 3x2
y 10 5 2 1 5 10 15 1 2 3 4 x
5. y 5 x4 2 8x3 1 18x2
y 80 70 60 50 40 30 20 10 2 10 1 2 3 4 5 x
6. f ( x ) 5 x4 1 2x3 2 12x
y 40 30 20 10
2. y 5
ln( x ) x
y 0.4 0.3 0.2 0.1 0 0.1 1 2 3 4 5 6 7 x
2 1 10
7. y 5 2 x4 2 2x3 1 12x2
y 100 50 3 2 1 50 y 50 40 30 20 10 100 150 1 2 3 x
3. f ( x ) 5 x4 2 12x2
8. f ( x ) 5 x5 2 5x4
1 2 3 4 x y 100 2 1 100 x
4 3 2 1 20 30
4. g( x ) 5 x2
y 20 15 10 5 4 3 2 1 0 1 2 3 4 x
1 2 3 4 5
200
374
9. f ( x ) 5
x2 x 11
2
y 0.8 0.6 0.4 0.2 2 1 0.2 1 2 x
In Exercises 2130, find the stationary points of the function. Then use the Second Derivative Test to determine where relative extrema occur. If the Second Derivative Test is inconclusive, use the First Derivative Test. You may find it helpful to refer to the graphs of the functions in Exercises 110 and your work in Exercises 1120. 21. f ( x ) 5 x3 2 3x2 22. y 5 ln( x ) x
23. f ( r ) 5 r4 2 12r 2 10. g( x ) 5 3 x2 1 1

y 3.0 2.5 2.0 1.5 1.0 0.5 2 1 0 1 2 x
24. g( x ) 5 x2 25. y 5 x4 2 8x3 1 18x2 26. s( n ) 5 n4 1 2n3 2 12n 27. y 5 2 x4 2 2x3 1 12x2 28. f ( t ) 5 t 5 2 5t4 29. f ( x ) 5 30. g( x ) 5 x2 x 11
2
3 x2 1 1
In Exercises 1120, use the second derivative to find the inflection points of the function algebraically. You may find it helpful to refer to the graphs of the functions in Exercises 110. 11. f ( x ) 5 x3 2 3x2 12. y 5 ln( x ) x
In Exercises 3140, use the curve-sketching techniques introduced in the section to graph each function by hand. 31. f ( x ) 5 1 3 x 2 x2 on 32 2, 44 3
32. g( x ) 5 x4 1 4x3 on 32 2, 54
13. f ( x ) 5 x4 2 12x2 14. g( x ) 5 x2 15. y 5 x4 2 8x3 1 18x2 16. f ( t ) 5 t4 1 2t3 2 12t 17. y 5 2 x4 2 2x3 1 12x2 18. f ( x ) 5 x5 2 5x4 t2 19. h( t ) 5 2 t 11 3 20. g( x ) 5 2 x 11
33. f ( x ) 5 2 x3 1 3x 1 1 on 32 2, 24 34. f ( x ) 5 2 x4 1 2x2 on 32 2, 24 35. g( x ) 5 2 x3 1 12x on 32 4, 44 1 37. g( x ) 5 x5 2 x4 on 32 3, 54 5 38. f ( x ) 5 ln( x ) on 31, 104 x
36. f ( x ) 5 x4 2 2x2 1 1 on 32 2, 24
39. g( x ) 5 2x 2 x ln( 2 ) on 32 2, 24 40. h( x ) 5 22x 2 1 on 32 2, 44
Hint: This function doesnt have any x-intercepts.
375
In Exercises 4150, answer the questions by using the Second Derivative Test, as appropriate. In determining where absolute extrema occur, remember to check the endpoints. 41. Based on data from 1975 to 2000, the average monthly wage (in addition to room and board) of a sailor working on the West Coast may be modeled by W( t ) 5 0.1446t 2 2 5.765t 1 883.7
West Coast Sailor Wages
that the average person spends on the Internet annually may be modeled by L( t ) 5 2 0.4514t 3 1 6.125t 2 1 6.171t 1 1.994 hours, where t is the number of years since 1995.
In what year (between 1995 and 2004) was Internet usage expected to increase most slowly, and in what year was it expected to increase most rapidly? 46. Based on the results of Exercise 45, do you think that the rate of change in annual Internet usage will ever exceed 35 hours per year? Justify your answer.
Manufacturing Employees Based on data from 1995 to 1999, the number of employees in the leather and leather products manufacturing industry may be modeled by A( t ) 5 2 0.6667t 3 1 4.5t 2 2 14.83t 1 106
What was the maximum and what was the minimum increase in monthly wages, and in what year (between 1975 and 2000) did each occur? (Round to the nearest year.) Hint: A decrease in wages is a negative increase. 42.
East Coast Sailor Wages Based on data
47.
from 1975 to 2000, the average monthly wage (in addition to room and board) of a sailor working on the East Coast may be modeled by E( t ) 5 0.1643t 2 2 5.658t 1 112.3 dollars, where t is the number of years since 1975.
thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
What was the maximum and what was the minimum increase in monthly wages, and in what year (between 1975 and 2000) did each occur? (Round to the nearest year.) 43. How could an East Coast sailors labor union use the information in Exercises 41 and 42 in negotiating future wage increases?
Salary Negotiations Homicide Rate Based on data from 1990 to 2000, the homicide rate may be modeled by H( t ) 5 0.01129t 3 2 0.2002t 2 1 0.4826t 1 9.390
In what year did the number of employees in the leather and leather products industry drop most rapidly? In what year did the number of employees increase most rapidly (or decrease most slowly)? 48.
Political Influence People for the Ethical Treatment of Animals (PETA) is a nonprofit organization that believes that animals are not ours to eat or wear. (Source: www.peta.org.) Based on the results of Exercise 47, do you think PETA is having an influence on the leather and leather products industry? Justify your answer. Love Triangles Based on data from 1991 to 2000, the cumulative number of homicides since the start of 1991 and the end of year t that are due to a romantic triangle may be modeled by R( t ) 5 2 20.45t 2 1 471.8t 2 152.5
44.
49.
deaths per 100,000 people, where t is the number of years since 1990. (Source: Modeled from FBI
Uniform Crime Reports, 2000.)
Between 1990 and 2000, in what year did the homicide rate increase most rapidly, and in what year did it decrease most rapidly? 45. Based on usage data from 1995 to 1999 and Census Bureau projections of future usage, the number of hours
Internet Usage
homicides, where t is the number of years since the end of 1990. (Source: Modeled from Crime in the
United States 2000, Uniform Crime Report, FBI.)
In what year between 1991 and 2000 did the cumulative number of such homicides increase most rapidly? How many homicides due to a romantic triangle occurred that year?
376 50.
Based on data from 1980 to 1999, the per capita consumption of breakfast cereal may be modeled by
Breakfast Cereal Consumption
57.
C( t ) 5 2 0.004718t 3 1 0.1165t 2 2 0.3585t 1 12.17 pounds, where t is the number of years since 1980. (Source: Modeled from Statistical Abstract of
Box Office Movie Sales The movie My Big Fat Greek Wedding was first released on April 19, 2002. Based on the first 52 weeks of sales, the cumulative gross box office sales of the movie may be modeled by
In what year (between 1980 and 1999) was the per capita breakfast cereal consumption increasing most rapidly, and in what year was it decreasing most rapidly? Exercises 5155 deal with position, velocity, and acceleration. The vertical position of a free-falling object on Earth may be modeled by s( t ) 5 2 16t 2 1 v0t 1 s0 feet above the earth after being airborne for t seconds. v0 is the velocity of the object, and s0 is the position of the object at time t 5 0. Use this information and the techniques shown in the section to answer the questions. 51. A rock is thrown upward into the air at a rate of 32 feet per second. After how many seconds of flight will it attain its maximum height above the ground? 52. When the rock in Exercise 51 attains its maximum height, what will be its velocity? 53. Given s( t ) 5 2 16t 2 1 v0t 1 s0 , calculate s s ( t ) and interpret the meaning of the result. 54. A pebble is dropped into the ocean from the top of a 240-foot cliff. What will be the pebbles velocity when it hits the surface of the ocean? 55. When will the pebble in Exercise 54 attain its maximum speed? In Exercises 5657, find the point of diminishing returns for the given function. 56.
Electronic Game Unit Sales Based on data from 1996 to 2000, the annual sales of computer and video game units may be modeled by 121.8 S( t ) 5 1 100 1 1 17.23e21.794t units, where t is the number of years since the end of 1996. (Source: Modeled from Interactive Digital
240,800,000 1 1 125.4e20.2086t dollars, where t is in weeks with t 5 1 corresponding to the first week the movie was released. Find the point of diminishing returns and interpret its real-life meaning. S( t ) 5
58. What is the relationship between the second derivative of a function and the concavity of the graph of the original function? 59. Suppose that the graph of the revenue function for a company is concave down and increasing. Explain what concave down and increasing means in practical terms for the business. 60. Suppose that an increasing function f has a single inflection point. Draw a possible graph of f and a graph of its derivative f r on the same set of axes. 61. Explain why a relative minimum of a function g occurs at a critical value x 5 c if g s ( c ) . 0. 62. A classmate observes that f s ( c ) . 0 and f r( c ) , 0 for a function f. He concludes that the graph of f is increasing and has an inflection point at x 5 c. Do you agree or disagree? Explain.
Each of the questions below deals with a parametric function model of the position of a roller-coaster train over a six-second interval 30, 64 . The x value of the model represents the horizontal position of a roller-coaster train (in meters) at time t seconds. The y value of the model represents the vertical position of the roller-coaster train (in meters) at time t seconds. The model is x 5 t 3 2 9t 2 1 23t y 5 6t 2 t 2 63. Find the second derivative of x and of y with respect to t. (These may be represented as x s and y s .) 64. Determine when x s 5 0 and y s 5 0 .
Software Association State of the Industry Report, 2000 2001.)
Find the point of diminishing returns and interpret its real-life meaning.
5.4 Related Rates
377
65. Given that the domain of the function is 0 # t # 6, find the points of inflection of x and y. 66. Over the six-second interval 30, 64 , what was the quickest horizontal acceleration of the train?
67. Over the same six-second interval, what was the slowest horizontal acceleration of the train? 68. Do you think that the roller coaster would be safe to ride? Defend your conclusion.
5.4 Related Rates
Solve related-rate problems
On March 24, 1989, the oil tanker Exxon Valdez ran aground on Bligh Reef in Prince William Sound, Alaska. In the ensuing spill, 11 million gallons of crude oil flowed into the water. About 250,000 seabirds; 2,650 sea otters; 1,000 cormorants; and hundreds of loons, harbor seals, and bald eagles were killed. Although much larger spills have occurred internationally, the Exxon Valdez spill was one of the most environmentally destructive. Because of the location of the accident, approximately 1,300 miles of the 9,000 miles of shoreline in the region were contaminated. Despite cleanup efforts, there were still pockets of oil remaining a decade after the accident.
When an oil spill occurs, many questions immediately arise. How quickly is the oil spilling into the water? How rapidly is the area covered by the oil expanding? If the oil continues to spill at its current rate, how much area will be covered an hour from now? These questions may be addressed using the concept of related rates. In this section, we will demonstrate how related rates can be used to measure how the rate of change in one variable affects the rate of change in another variable. We will use the implicit differentiation techniques covered in Section 4.5 extensively in this section.
The area of a circular oil spill with radius r feet is given by the formula A 5 pr 2. If the area of the circle is increasing at a rate of 100 square feet per minute, how quickly is the radius changing? Since the radius rate of change function is the
378
derivative of the radius function, we are looking for of change in the area is given by
dA dA square feet . The units of are minute . Notice that both dt dt A and r are differentiated with respect to t. Using implicit differentiation, we have A 5 pr2 d d ( A ) 5 ( pr2 ) dt dt dA d 5 p ( r2 ) dt dt dA dr 5 p a 2r b dt dt dA dr 5 2pr dt dt dr 1 dA 5 dt 2pr dt Thus the rate of change in the radius is the rate of change in the area divided by the product of 2p and the current radius. For this problem, we indicated that the dA 5 100. area was increasing at a rate of 100 square feet/minute. That is, dt Therefore, the rate of change in the radius is dr 1 dA 5 dt 2pr dt 5 5 When the radius is 1 foot, radius is 2 feet, 1 ( 100 ) 2pr 50 pr
dr feet with units minute. The rate dt
50 dr 5 < 0.01592 foot per minute . Assuming that the shape of the spill dt p( 1000 ) remains circular, we can calculate the rate of change in the radius for any value of the radius. Related rates are most often used when comparing the changes in two related quantities over time. In the next several examples, we will illustrate how related rates are used.
50 dr 5 < 7.96 feet per minute . When the radius is 1000 feet, dt p( 2 )
dr 50 5 < 15.9 feet per minute . When the dt p( 1 )
5.4 Related Rates
379
EXAMPLE
Interpreting the Meaning of
dV dr dh , and , dt dt dt
The volume of space occupied by a liquid in a cylindrical can is given by V 5 pr2h , where r is the radius of the can and h is the depth of the liquid in the can. dV dr dh Differentiate V with respect to t and interpret the meaning of , , and . dt dt dt Assume that the radius and height are measured in centimeters and that time is measured in seconds. Assuming that the radius of the can is fixed, simplify the dV result for . dt
SOLUTION
d d ( V ) 5 ( pr2h ) dt dt dV d 5 p ( r2h ) dt dt 5 p a 2r dr dh 2 ?h1 ?rb dt dt dr dh 1r b dt dt

Constant Multiple Rule Product and Chain Rules
5 pr a 2h
dr is the rate of change in the radius in centimeters per second. dt dh is the rate of change in the depth of the liquid in centimeters per second. dt dV is the rate of change in the volume of the liquid in the can in cubic dt centimeters per second. It is a function of the radius, the depth, and the rates of change in the radius and the depth. Notice that if the can has a fixed dr radius, then 5 0 and dt dV dh 5 pr c 2h( 0 ) 1 r d dt dt 5 pr 2 dh dt
EXAMPLE
Using Related Rates in a Real-World Context

The volume of a cylindrical can is given by V 5 pr2h, where r is the radius and h is the height of the interior of the can from lid to lid. In Section 5.2, we learned that a #10 can has a radius of 3.625 inches (9.208 centimeters). Suppose that the can is being filled with rice and that the rice is being poured into the can at a constant rate of 20 cups per minute. At what rate is the height of the rice in the can changing? Will it take more or less than a minute to fill the can?
380
SOLUTION You may immediately ask, Why do we care? Admittedly, we may not care if we have to fill only one can. However, if we worked in a cannery and wanted to program a machine to fill thousands of cans, the problem would immediately become meaningful. Failure to program the machine with the correct fill rate would result in overfilled cans or insufficiently filled cans. Both of these would hamper the cannerys effort to create a uniform product with minimum waste. Since we are filling #10 cans, the volume equation becomes
V 5 p( 9.208 ) 2h 5 84.79ph cubic centimeters and the rate of change in the volume is given by dV dh cubic centimeters 5 84.79p dt dt minute One of the challenges of the standard measurement system (inches, feet, miles, cups, gallons, etc.) is that conversion between quantities (e.g., cups to cubic inches) is difficult. For this reason, we will convert standard measurements to their metric equivalents. One cup is equivalent to 236.6 cubic centimeters. Since the volume is changing at a rate of 20 cups per minute, we have dV cups 236.6 cubic centimeters 5 20 ? dt minute 1 cup 5 4732 We must find dh . dt dV dh 5 84.79p dt dt 4732 5 266.4 dh 4732 5 dt 266.4 5 17.76 centimeters per minute The height of the rice in the can is increasing by about 17.76 centimeters per minute. Since the interior height of a #10 can is 16.83 centimeters (6.625 inches), it will take less than a minute to fill the can. (To be precise, it will take 57 seconds.) dh dt cubic centimeters minute
EXAMPLE
Using Related Rates in a Real-World Context

A circular above-ground swimming pool has a 12-foot radius and may be filled to a maximum depth of 4 feet. By measuring the amount of time it took to fill a bucket, the author determined that water leaves his outside faucet at a rate of 9 gallons per minute. How quickly is the depth of the water in the pool changing when the pool is being filled from a hose connected to the faucet? How long will it take to fill the pool?
5.4 Related Rates
381
SOLUTION We will first convert the dimensions to metric measurements. We know that 1 foot < 30.48 centimeters and 1 gallon < 3785 cubic centimeters. The radius of the pool is 12( 30.48 ) 5 365.76 centimeters . The rate at which the water is leaving the faucet is 9( 3785 ) 5 34,065 cubic centimeters per minute. The pool is cylindrical, so its volume is V 5 pr2h, where r is the radius and h is the depth of the water in the pool. Since the radius of the pool will not change, we have
V 5 p( 365.76 ) 2h 5 420,283.45h cubic centimeters Differentiating with respect to t, we get dV dh cubic centimeters 5 420,283.45 dt dt minute dV But we know that 5 34,065 cubic centimeters per minute, so dt 34,065 5 420,283.45 dh dt
dh centimeters 5 0.08105 dt minute 5 0.08105 centimeters 60 minutes ? minute 1 hour
5 4.863 centimeters per hour The pool depth is increasing at a rate of 4.863 centimeters per hour. Since 4 feet 5 121.92 centimeters, the amount of time it will take to fill the pool is 121.92 centimeters 5 25.07 hours 4 . 863 centimeters per hour It will take about 25 hours to fill the pool.
EXAMPLE
Calculating a Change in Wind Chill Temperature

Wind chill temperature is the temperature in calm air that has the same chilling effect on a person as that of a particular combination of temperature and wind. For example, when the temperature is 35 Fahrenheit (F) and the wind speed is 40 miles per hour, it feels like it is 3F. The wind chill temperature may be calculated according to the following formula: where w is the wind speed (in mph) and T is the temperature (in F). (Source: If the temperature is currently 20 and is dropping at a rate of 3 per hour, and if the wind is currently blowing at 16 mph and is increasing at a rate of 2 mph per hour, how quickly is the wind chill temperature changing?
F 5 91.42 A 0.474677 2 0.020425w 1 0.303107"w B ( 91.4 2 T ) degrees Fahrenheit

National Climatic Data Center.)
382
SOLUTION We must first differentiate the wind chill equation with respect to
time. d d ( F ) 5 C 91.4 2 A 0.474677 2 0.020425w 1 0.303107"w B ( 91.4 2 T ) D dt dt dF dw dw 5 0 2 c 2 0.020425 1 0.5( 0.303107w20.5 ) d ( 91.4 2 T ) dt dt dt 1 a2 dT b 32 A 0.474677 2 0.020425w 1 0.303107"w B 4 dt
Product Rule
dF dw dw 5 c 0.020425 2 0.5( 0.303107w20.5 ) d ( 91.4 2 T ) dt dt dt 1 a
dT b A 0.474677 2 0.020425w 1 0.303107"w B dt
The current wind speed is 16 mph, so w 5 16. dw is the rate of change in the wind and equals 2 mph per hour. dt The current temperature is 20, so T 5 20. dT is the rate of change in the temperature and equals 2 3 degrees per hour. dt Substituting these values into the derivative equation, we get dF 5 50.020425( 2 ) 2 0.5 30.303107( 16 ) 20.54 ( 2 ) 6391.4 2 ( 20 ) 4 dt 1 ( 2 3 ) C 0.474677 2 0.020425( 16 ) 1 0.303107"16D 5 ( 2 0.03493 )( 71.4 ) 2 4.081 5 2 2.494 2 4.081 5 2 6.575 The wind chill index is dropping by about 6.6 degrees per hour. That is, although the temperature is dropping by only 3 degrees per hour, it feels like it is dropping by 6.6 degrees per hour because of the wind. Because of the number of terms and the complexity of the coefficients, the type of calculation shown in Example 4 is time-consuming and prone to error. For this reason, we recommend using the following Technology Tip to speed up calculations and increase accuracy.
Using the Equation Solver 1. Press MATH 0:Solver... to bring up the Equation Solver. Immediately press the blue up arrow. Press CLEAR to delete any equation that may be visible.
(Continued)
5.4 Related Rates
383
2. Enter the equation you need to solve, using the alpha keys for variables. Since the calculator requires the equation to be dF set equal to zero, we must subtract dt from both sides of our equation. For our dF dT 5 A, 5 B, equation, we let dt dt dw 5 C. and dt 3. Press ENTER to display the variable input menu. Edit the variables to reflect the known values from our problem.
4. Move the cursor to the line containing ENTER to solve A= and press ALPHA the equation for A.
EXAMPLE
Calculating a Change in Wind Chill Temperature

If the temperature is currently 30 and dropping at a rate of 5 per hour, and the wind is currently blowing at 20 mph and is decreasing at a rate of 10 mph per hour, how quickly is the wind chill temperature changing? Use the Technology dF Tip and the equation for from Example 4 to answer the question. dt
SOLUTION
The current wind speed is 20 mph, so w 5 20. dw C5 is the rate of change in the wind and equals 10 mph per hour. dt The current temperature is 30, so T 5 30. dT B5 is the rate of change in the temperature and equals 5 degrees per hour. dt dF . Since we have already entered the equation (from dt Example 4) in our calculator, well use the Technology Tip to solve the problem (see Figure 5.35). The wind chill temperature is increasing by 1.1579 < 1.2 degrees per hour. Although the temperature is dropping 5 degrees per hour, it feels like it is warming by 1.2 degrees per hour because the wind speed is decreasing. We need to solve for A 5
FIGURE 5.35
384
Calculating a Value of a Multivariable Function 1. Create a cell for each domain variable and an adjacent cell with the formula for the function as shown. 2. Enter the appropriate domain values for the function. The function value is displayed.
5.4 Summary
In this section, you learned how related rates can be used to measure how the rate of change in one variable affects the rate of change in another variable. You discovered that technology can simplify the arithmetic computations of related-rate problems.

1. Write an equation that relates the area of a circle to its radius. 2. Write an equation that relates the volume of a cube to the length of one of its sides. 3. If the radius of a circle increases from 3 inches to 5 inches, by how much does the area of the circle change? 4. If the side of a cube doubles in length, what happens to the volume of the cube? 5. Give a real-world example that relates two different quantities. 6. Distance traveled is a function of the average speed (rate) and the time spent traveling. If the average speed remains constant, what effect does a change in the time traveled have on the distance traveled? 7. What is a rate? 8. If the radius of a circle increases by 1 unit, by how much does the circumference of the circle increase?
5.4 Exercises
In Exercises 110, differentiate the function with respect to t. Explain the physical meaning of each of the rates. (For each of the problems, A is area, V is volume, r is radius, h is height, l is length, w is width, b is base, and t is time. Use inches, square inches, cubic inches, and minutes as the units of length, area, volume, and time, respectively.) 1. A 5 lw (area of a rectangle) 2. A 5 2prh 1 2pr 2 (surface area of a cylinder including top and bottom) 3. V 5 lwh (volume of a box) 4. V 5 3 pr 3 (volume of a sphere) 5. A 5 2lw 1 2wh 1 2lh (surface area of a box) 6. V 5 pr 2h (volume of a cone) 3 7. A 5 4pr 2 (surface area of a sphere) 8. A 5 2 bh (area of a triangle) 9. A 5 pr 2 (area of a circle) 10. A 5 2pr 2 (surface area of a hemisphere)
1 1 4
5.4 Related Rates
385
In Exercises 1115, solve the related-rate problem using the techniques demonstrated in this section. 11. Swimming Pool Depth A circular above-ground swimming pool has a 9-foot radius and may be filled to a maximum depth of 4 feet. The pool is being filled with a faucet that releases water at a rate of 10 gallons per minute. How quickly is the depth of the water in the pool changing when the pool is being filled from a hose connected to the faucet? How long will it take to fill the pool? 12. Water Pressure A circular above-ground swimming pool has a 9-foot radius and may be filled to a maximum depth of 4 feet. If the depth of the water in the pool is rising at a rate of 0.5 inch per minute, at what rate is the water leaving the hose that is filling the pool? 13. Water Depth A cylindrical can has a 5-inch radius and an 8-inch height and is to be filled with water. At what constant rate must the depth of the water increase if the can is to be filled in exactly 30 seconds? 14. Pizza Size The owner of a pizzeria is contemplating changing the size of his large pizza. He intends to decrease the size of the large pizza from a 16-inch diameter to a 14-inch diameter by gradually reducing the diameter by 1 inch each month. (He figures that consumers 2 wont notice the change if it is gradual.) At what rate will the surface area of the pizza be changing two months into the reduction period? 15. Oil Spill Spread The surface area of a circular oil spill is increasing at a rate of 100 square feet per hour. How quickly is the radius increasing when the radius is 157 feet? In Exercises 1620, solve the related-rate problems using the Technology Tip demonstrated in this section. The wind chill temperature may be estimated using the following model equation: F 5 91.4 2 ( 0.474677 2 0.020425w 1 0.303107 ! w ) ? ( 91.4 2 T ) degrees Fahrenheit where w is the wind speed (in mph) and T is the temperature (in F). (Source: National Climatic Data
Center.)
hour, and the wind is currently blowing at 18 mph and is increasing at a rate of 2 mph per hour, how quickly is the wind chill temperature changing? 17.
Wind Chill If the temperature is currently 10F and is warming at a rate of 8F per hour, and the wind is currently blowing at 25 mph and is increasing at a rate of 3 mph per hour, how quickly is the wind chill temperature changing?
18.
If the temperature is currently 2F and is decreasing at a rate of 4F per hour, and the wind is currently blowing at 55 mph and is decreasing at a rate of 5 mph per hour, how quickly is the wind chill temperature changing?
Wind Chill
19.
If the temperature is currently 40F and is increasing at a rate of 4F per hour, and the wind is currently blowing at 5 mph and is not increasing its speed, how quickly is the wind chill temperature changing?
Wind Chill
20.
If the temperature is currently 32F and is not changing, and the wind is currently blowing at 25 mph and is increasing its speed by 4 mph per hour, how quickly is the wind chill temperature changing?
Wind Chill
21. What do we mean when we say that one rate is related to another rate? 22. The equation w 5 f ( h ) gives a persons monthly gross wages as a function of the number of hours worked. The equation n 5 g( w ) gives a persons net monthly earnings as a function of monthly gross dw dn dn , , and represent, and dh dw dh how are they related to one another? wages. What do 23. Explain why Leibniz notation for the derivative is especially useful when working with related rates. 24. Can heart rate and running speed be considered related rates for an individual? Explain. 25. Why is it especially important to keep track of units when working related-rate problems?
16.
Wind Chill If the temperature is currently
0F and is dropping at a rate of 5F per
386

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What is the difference between a relative maximum and an absolute maximum? (5.1) 2. What is going on graphically at a stationary critical value? (5.1) 3. In terms of a rate of change, what does the second derivative represent? (5.3) 4. For which derivatives of a function is it helpful to construct a sign chart, and why? (5.1, 5.3) 5. If the value of the second derivative is a positive constant, what do you know about the graph of the first derivative and the graph of the original function? (5.3) 6. Is it true that if f s ( c ) . 0, the function f has a relative minimum at x 5 c? Explain. (5.3) 7. Why is it that a relative maximum occurs at a point where the derivative changes from negative to positive? (5.1) 8. For a continuous, smooth function f, what are the relationships among the inflection points of f, the relative extrema of f r, and the x-intercepts of f s ? (5.3) 9. Why can we say that acceleration represents the second derivative of position? (5.3) 10. What does it mean when we say that one rate is related to another rate? (5.4) You should also be familiar with the following definitions and procedures, properties, and tests that were emphasized in this chapter.
Definitions

Extremum: A maximum or a minimum of a function f (plural extrema) Relative Maximum: A relative maximum of a function f occurs at a point ( c, f ( c )) if f ( c ) $ f ( x ) for all x in some interval (a, b) containing c. That is, a , c , b. Relative Minimum: A relative minimum of a function f occurs at a point ( c, f ( c )) if f ( c ) # f ( x ) for all x in some interval (a, b) containing c. That is, a , c , b. Absolute Maximum: An absolute maximum of a function f occurs at a point ( c, f ( c )) if f ( c ) $ f ( x ) for all x. Absolute Minimum: An absolute minimum of a function f occurs at a point ( c, f ( c )) if f ( c ) # f ( x ) for all x. Critical Values: A critical value of a continuous function f on the interval ( a, b ) is a value c with a , c , b, where f r( c ) 5 0 or f r( c ) is undefined. Second Derivative: The derivative of the derivative. The notation f s ( x ) or y s is commonly used. Inflection Point: A point on the graph of a continuous function f at which the graph of f changes concavity.
Chapter 5
Study Sheet
387
Value of the Derivative at a Relative Extremum: If f is a continuous function with a relative extremum at ( c, f ( c )) , then f r( c ) 5 0 or f r( c ) is undefined. Existence of Relative Extrema: If f is a continuous function on an interval (a, b), then all relative extrema of f occur at critical values of f. First Derivative Test A relative maximum of f occurs at a critical value x 5 c if f r changes from positive to negative at ( c, f ( c )) . A relative minimum of f occurs at a critical value x 5 c if f r changes from negative to positive at ( c, f ( c )) . A relative extremum of f does not occur at a critical value x 5 c if f r does not change sign at ( c, f ( c )) . Concavity of a Continuous Function A continuous function f is concave up at ( c, f ( c )) if f s ( c ) . 0. A continuous function f is concave down at ( c, f ( c )) if f s ( c ) , 0. Location of Inflection Points: A continuous function f has an inflection point at ( c, f ( c )) if f s ( c ) 5 0 or f s ( c ) is undefined and f s changes sign at x 5 c. Second Derivative Test: If f is a continuous function with f r( c ) 5 0, then A relative minimum occurs at x 5 c if f s ( c ) . 0. A relative maximium occurs at x 5 c if f s ( c ) , 0. The test is inconclusive if f s ( c ) 5 0. Relationship Between Position, Velocity, and Acceleration Velocity is the derivative of position. Acceleration is the derivative of velocity and the second derivative of position.

Section 5.1 In Exercises 12, determine the points on the graph where f r( x ) 5 0 or f r( x ) is undefined.
1. f ( x ) 5 Z x 2 2 Z 2 ( x 2 2 ) 2 1 2 2. f ( x ) 5 x5 2 4x3 1 10
388
In Exercises 36, find the critical points of the function. 3. f ( x ) 5 x5 2 4x3 1 10 4. g( x ) 5 5x1>5 6. y 5 2 2x 1 4 In Exercises 710, construct a sign chart for the derivative. Then determine the relative and absolute extrema of the function on the specified domain. Use the First Derivative Test as appropriate. (These functions are the same as those in Exercises 36.) 7. f ( x ) 5 x 5 2 4x 3 1 10; 32 2, 24 8. g( x ) 5 5x1>5; 32 2, 24 10. y 5 2 2x 1 4; 30, 44 9. y 5 2 4x2 1 4; 32 3, 34 5. y 5 2 4x2 1 4
Section 5.2 In Exercises 1314, use the derivative and other optimization techniques to find the optimal value requested.
13.
Company Revenue and Cost
Gatorade/Tropicana North America, a subsidiary of PepsiCo, produces fruit juices and other flavored beverages. Based on data from 1999 to 2001, the net sales (revenue) of Gatorade/Tropicana North America may be modeled by R( t ) 5 2 107t 2 1 496t 1 3452 million dollars and the operating profit (earnings before interest and taxes) may be modeled by P( t ) 5 2 18.5t 2 1 85.5t 1 433 million dollars, where t is the number of years since 1999. (Source: Modeled from 2001 PepsiCo
Annual Report, p. 44.)
In Exercises 1112, use the concepts of relative and absolute extrema to find the answer to the question. 11.
Exxon Valdez Oil Spill Based on data from the fourth to the fifty-sixth day after the accident, the spread of the oil from the site of the Exxon Valdez oil spill may be modeled by D( t ) 5 0.004108t 3 2 0.4432t 2 1 21.07t 2 42.07
(a) In what year are net sales projected to reach a maximum? (b) Find the cost function for Gatorade/Tropicana North America. (c) According to the model, in what year are costs expected to reach a maximum? (d) Compare the results of (a) and (c). Do the results seem reasonable? 14. Ice Cream Cone Design The volume of an ice cream cone (minus the ice cream) is given by 1 2 pr h 3 cubic inches, where r is the radius of the top of the cone and h is the height of the cone (both in inches). The surface area of the exterior of the cone is given by A 5 pr"r 2 1 h2 square inches. For a cone with a surface area of 47.12(15p) square inches, what are the dimensions of the cone with maximum volume? Should ice cream cones be constructed with these dimensions? Explain. V5
miles from the spill site t days after the accident. Note that the model is invalid for t , 4. (Source:
Modeled from www.oilspill.state.ak.us/facts/ spillmap.html.)
According to the model, what was the greatest distance the oil had spread between the fourth and the fifty-sixth day? 12.
Industry Work Force Size Based on
data from 1995 to 1999, the number of employees in the printing and publishing industry may be modeled by E( t ) 5 2 4.417t 3 1 24.93t 2 2 25.30t 1 1450 thousand employees, where t is the number of years since 1995. (Source: Modeled from Statistical
According to the model, when did the industry have its maximum and minimum number of employees? How many people were employed at these times?
Chapter 5
Review Exercises
389
Section 5.3 In Exercises 1516, indicate on the graph where the function is concave up and where the function is concave down. Mark the graph where you think inflection points occur.
15. f ( x ) 5 x2 1 9 x2 1 10
y 0.96 0.95 0.94 0.93 0.92 0.91 0.90 0.89 3 2 1 0 1 2 3 x
In Exercises 2122, use the second derivative to determine where the maximum and minimum rates of change occur for the specified function. Be sure to evaluate the rate of change function at the endpoints to ensure that you have found the absolute extrema. 21.
Exxon Valdez Oil Spill Based on data from the fourth to the fifty-sixth day after the accident, the spread of the oil from the site of the Exxon Valdez oil spill may be modeled by D( t ) 5 0.004108t 3 2 0.4432t 2 1 21.07t 2 42.07 miles from the spill site t days after the accident.
(Source: Modeled from www.oilspill.state.ak.us/facts/ spillmap.html.)
On what day was the oil spreading at the quickest rate? At what rate was it spreading? 22.
Minivan Speed Based on 2.5 minutes
16. y 5 x3 2 x2 2 8x 1 12
y 20 10 3 2 1 10 20 1 2 3 x
of data, the distance of the authors minivan from a stoplight at the beginning of a highway may be modeled by M( t ) 5 2 0.0591t 3 1 0.3399t 2 1 0.5019t 1 0.005594 miles, where t is the amount of time (in minutes) that the minivan has been moving away from the stoplight. Rounded to the nearest mile per hour, what were the highest and lowest speeds of the minivan?
In Exercises 1718, use the second derivative to find the inflection points of the function algebraically. You may find it helpful to refer to the graphs of the functions in Exercises 1516. 17. f ( x ) 5 x2 1 9 x2 1 10
18. y 5 x3 2 x2 2 8x 1 12 In Exercises 1920, find the stationary points of the function. Then use the Second Derivative Test to determine where relative extrema occur. If the Second Derivative Test is inconclusive, use the First Derivative Test. You may find it helpful to refer to the graphs of the functions in Exercises 1516 and your work in Exercises 1718. 19. f ( x ) 5 x2 1 9 x2 1 10
Section 5.4 In Exercises 2324, differentiate the function with respect to t. Explain the physical meaning of each of the rates. ( A is area and s is the length of a side. Lengths are measured in centimeters, areas in square centimeters, and time in seconds.)
23. A 5 5s2 (area of an equilateral cross)
24. A 5
"3 2 s (area of an equilateral triangle) 4
20. y 5 x3 2 x2 2 8x 1 12
390
In Exercises 2530, solve the related-rate problem. 25. Swimming Pool Depth A circular above-ground swimming pool has a 4-foot radius and may be filled to a maximum depth of 3 feet. A woman is filling the pool with a faucet that releases water at a rate of 10 gallons per minute. How quickly is the depth of the water in the pool changing 2 minutes after she begins to fill the pool? 26. Swimming Pool Depth How rapidly is the depth of the water in the pool in Exercise 25 increasing 5 minutes, 10 minutes, and 15 minutes after the woman begins to fill the pool? 27. Swimming Pool Depth A rectangular aboveground swimming pool has a 12-foot length and a 6-foot width and may be filled to a maximum depth of 4 feet. A teen is filling the pool with a
faucet that releases water at a rate of 9 gallons per minute. How quickly is the depth of the water in the pool changing 2 minutes after he begins to fill the pool? 28. Swimming Pool Depth How rapidly is the depth of the water in the pool in Exercise 27 increasing 5 minutes, 10 minutes, and 15 minutes after the teen begins to fill the pool? 29. Oil Spill Spread The surface area of a circular oil spill is increasing at a rate of 250 square feet per hour. How quickly is the radius increasing when it is 2500 feet? 30. Oil Spill Spread The radius of a circular oil spill is increasing at a rate of 10 feet per hour. How quickly is the surface area changing when the radius is 200 feet?
P R O J E C T
What to do
1. Recruit a friend, neighbor, or family member to go on a drive with you. 2. Drive to the base of a freeway (or major highway) on-ramp. 3. As you drive up the on-ramp and onto the freeway, have your friend record the odometer reading every 15 seconds for at least 2 minutes. If the odometer is between tenths of a mile, round to the nearest twentieth. (That is, if it is between 2.0 and 2.1, round to 2.05.) 4. Find the equation of the function that best fits the data. 5. Use differentiation to find the velocity and acceleration functions. 6. Determine your maximum velocity and maximum acceleration.
How to get a good data set

1. As youre driving, try to avoid any rapid changes in your acceleration. (This will make a smoother scatter plot.) 2. Collect several minutes of data. You can eliminate extra points at the end of your data set if the model doesnt fit those points well. 3. Use as many data points as possible, yet aim for a model coefficient of determination (r 2) of 0.99 or better. 4. Avoid excessive speeds, erratic driving, and driving under the influence. 5. If you see flashing lights behind you, pull over to the right side of the road and abort the project!.
391
Chapter
The Integral
O
n March 24, 1989, the Exxon Valdez oil tanker ran aground in Prince William Sound. The ensuing oil spill was one of the most widely publicized environmental disasters in history. If a function modeling the variable rate at which oil is spilling from a tanker is known, the Fundamental Theorem of Calculus may be used to calculate the total amount of oil spilled over a specified time period.
6.1
Indefinite Integrals
Use integration rules to find the antiderivative of a function
6.2
Integration by Substitution
Use the method of substitution to integrate functions
6.3
Using Sums to Approximate Area
Estimate the area between a curve and the x-axis by using left- and right-hand sums
6.4
The Definite Integral

Apply definite integral properties Calculate the exact area between a curve and the x-axis by using definite integrals
6.5
The Fundamental Theorem of Calculus
Apply the Fundamental Theorem of Calculus Find the function for accumulated change given a rate of change function
392
6.1 Indefinite Integrals
393
Use integration rules to find the antiderivative of a function
A speedometer measures a cars velocity. If we record our cars velocity at 15-second intervals, we will generate a table of data that may be used to find a model for the velocity of the car. Can we determine how far the car traveled from the velocity equation alone? Yes. By using a process called integration, we can determine the distance equation if we know the velocity equation. In this section, we will introduce integration: the process of finding the equation of a function that has a given derivative. We will also define and demonstrate how to use basic rules of integration.
We will begin by looking at three functions: f ( x ) 5 x 3 2 x 2 2 8x 1 17, g( x ) 5 x 3 2 x 2 2 8x 1 12, and h( x ) 5 x 3 2 x 2 2 8x 1 7. The equations of these functions look remarkably similar. In fact, their equations differ only by a constant. Similarly, the basic shape of the graphs is the same; they differ only in their vertical placement on the coordinate system (see Figure 6.1).
f (x) = x3 x2 8x + 17 g (x) = x3 x2 8x + 12 y 20 10 x
2 1 10 20
h (x) = x3 x2 8x + 7
FIGURE 6.1
Lets differentiate each of the functions. f r( x ) 5 3x2 2 2x 2 8 gr( x ) 5 3x2 2 2x 2 8 hr( x ) 5 3x2 2 2x 2 8 The functions all have the same derivative! In fact, any function of the form f ( x ) 5 x 3 2 x 2 2 8x 1 C (C is any constant) will have the derivative f r( x ) 5 3x2 2 2x 2 8 Thus, an infinite number of functions share the same derivative. The graphs of all of these functions will have the same basic shape, but their vertical placement on the coordinate system will differ.
394
CHAPTER 6 The Integral
Often we are given the rate of change function (derivative) and asked to find the equation of the function that has the given derivative. As previously illustrated, there is an infinite number of functions that share the same derivative. To represent all such functions, we will write the letter C in the place of the constant term.
EXAMPLE
Finding the Antiderivative of a Function

Find the general form of the function that has the derivative f ( x ) 5 2x.
SOLUTION We are looking for a function F( x ) such that
Lets consider the function F( x ) 5 x2 1 C . d d 3F( x ) 4 5 ( x2 1 C ) dx dx 5 2x 1 0 5 2x
d 3F( x ) 4 5 f ( x ) . dx
Therefore, any function of the form F( x ) 5 x2 1 C has the derivative f ( x ) 5 2x. Specific functions include F( x ) 5 x2 1 2 , F( x ) 5 x2 2 12 , F( x ) 5 x2 1 256 , and so on.
DEFINITION OF TERMS
Integration: The process of finding the general form of a function that has a given derivative. Also referred to as antidifferentiation. Indefinite integral: The general form of a function that has a given derivative. Also referred to as the antiderivative. Integral sign: The symbol 3, which indicates that integration is to be performed. Integrand: The function written to the right of the integral sign. This is the function that is to be integrated.
The notation 3 f ( x ) dx is read the integral of f ( x ) with respect to x or the d 3 f ( x ) 4 means find the deantiderivative of f ( x ) with respect to x. Just as dx rivative of f ( x ) with respect to x, 3 f ( x ) dx means find the general form of the function that has derivative f ( x ) with respect to x. The integral sign 3 must always be written together with a differential (e.g., dx) to indicate the variable of integration. (We discuss differentials in more detail in the next section.) A function f is said to be integrable if there exists a function F such that F r 5 f.
395
EXAMPLE
Finding the Antiderivative of a Function

Find 33x2 dx . d 3 ( x ) 5 3x2 . Therefore, the indefinite integral dx of f ( x ) 5 3x2 is F( x ) 5 x3 1 C . We can check our work by differentiating the function F with respect to x. tive is f ( x ) 5 3x2 . Recall that d d 3F( x ) 4 5 ( x3 1 C ) dx dx 5 3x2 1 0 5 3x2
SOLUTION We are to find the general equation of the function whose deriva-
You may have noticed that we have shifted our notation slightly from our earlier treatment of derivatives. Since we are given the rate of change function as the original function, we are using the notation f ( x ) instead of f r( x ) to represent the rate of change function. We use the notation F( x ) to represent the function whose derivative is f ( x ) . That is, F r( x ) 5 f ( x ) . Admittedly, this shift in notation is a bit confusing. However, since the notation shift is so widely adopted by people who use calculus, we will also use this commonly accepted notation. Making the transition may be difficult at first, but by the end of the chapter, you will be using the new notation with ease.
Basic Integration Rules

As you might expect, the rules for derivatives and the rules for integrals are intimately related. In fact, many of the rules even share the same names.
POWER RULE FOR INTEGRALS
Let f ( x ) 5 x n, where x is a variable and n is a constant with n 2 2 1. Then

n 3x dx 5
xn11 1 C, n 2 2 1 n11
The rule can be easily verified by differentiating the resultant function. d xn11 1 d n11 a 1 Cb 5 (x ) 1 0 dx n 1 1 n 1 1 dx 5 n 1 1 n1121 x n11
5 xn Notice that n 2 2 1. If n 5 2 1, the denominator of

xn11 n11
1 C is equal to 0,
which makes the expression undefined. We will deal with the case of n 5 2 1 later.
396
CONSTANT MULTIPLE RULE FOR INTEGRALS
Let f ( x ) 5 a ? g( x ) , where x is a variable and a is a constant. Then 3 3a ? g( x ) 4 dx 5 a ? 3g( x ) dx 5 a ? G( x ) 1 C where g( x ) 5 Gr( x ) .
EXAMPLE
Integrating a Function
Integrate f ( x ) 5 6x2 .
SOLUTION
2 2 36x dx 5 6 3x dx
5 6a 5
x3 1 C1 b 3
6x3 1 6C1 3
5 2x3 1 C Notice that we rewrote 6C1 as C. Since C1 is a constant, multiplying it by 6 will still result in a constant. The same would be true if we were to add two different constants C1 and C2. The result would still be a constant, which we could write as C. Therefore, when doing integration, we will write a single constant C to represent a multiple of a constant or a sum of constants.
Remember that the derivative represents the slope of the tangent line. The derivative of a linear function f ( x ) 5 mx 1 b is f r( x ) 5 m. That is, for a linear function, the derivative of the function is the slope of the line. Similarly, the antiderivative of a constant function f ( x ) 5 m is the linear function F( x ) 5 mx 1 b . This relationship is summarized in the Constant Rule for Integrals.
CONSTANT RULE FOR INTEGRALS
Let k be a constant. Then 3k dx 5 kx 1 C .
The Sum and Difference Rule is another extremely useful rule to use when working with integrals.
397
SUM AND DIFFERENCE RULE FOR INTEGRALS
Let f ( x ) and g( x ) be integrable functions of x. Then 3 3 f ( x ) 6 g( x ) 4 dx 5 3 f ( x ) dx 6 3g( x ) dx
EXAMPLE
Antidifferentiating a Function
Find the antiderivative of h( x ) 5 2x 1 4.
SOLUTION
3( 2x 1 4 ) dx 5 32x dx 1 34 dx 5 2 3x dx 1 4 31 dx x2 5 2 a b 1 4( x ) 1 C 2 5 x2 1 4x 1 C
Sum and Difference Rule Constant Multiple Rule Power and Constant Rules
EXPONENTIAL RULE FOR INTEGRALS
Let a . 0 and a 2 1. Then 1 x x 3a dx 5 ln( a ) a 1 C
The Exponential Rule may be easily verified by differentiating y 5 ln( a ) a x 1 C. d 1 d 1 d a x 1 Cb 5 a x b 1 (C) a a dx ln( a ) dx ln( a ) dx 5 5 5 1 d c (a x) d 1 0 ln( a ) dx 1 3ln( a ) a x4 ln( a ) ln( a ) x (a ) ln( a )
Sum and Difference Rule Constant and Constant Multiple Rules Exponential Rule
5 ax It immediately follows from the Exponential Rule for Integrals that 1 x x 3e dx 5 ln( e ) e 1 C 5 1 x e 1C 1
5 ex 1 C
398
EXAMPLE
Integrating an Exponential Function

Find 32 ? 3x dx.
SOLUTION
x x 32 ? 3 dx 5 2 33 dx 1 52? ( 3x ) 1 C ln( 3 )
Constant Multiple Rule for Integrals Exponential Rule for Integrals
2 ( 3x ) 1 C ln( 3 )
for Integrals to integrate f ( x ) 5 x21. However, the function is integrable.
Recall that x21 5 x . We saw previously that we cannot use the Power Rule
INTEGRAL RULE FOR
1 x
If x 2 0 , then 1 3 x dx 5 ln Z x Z 1 C
x . 0 , but ln Z x Z is defined for all x 2 04 . We can convince ourselves that the rule holds true by differentiating ln Z x Z 1 C . d d d ( ln Z x Z 1 C ) 5 ( ln Z x Z ) 1 ( C ) dx dx dx 5 ln( x ) x . 0 d ae b 10 ln( 2 x ) x , 0 dx
Sum and Difference Rule Definition of absolute value Constant Rule Logarithm Rule Logarithm and Chain Rules
The absolute value of x is necessary to ensure that the two functions 1 f ( x ) 5 and F( x ) 5 ln Z x Z 1 C have the same domain 3ln( x ) is defined only for
x
1 ,x.0 x 5 1 (21), x , 0 2x 1 ,x.0 x 5 1 ,x,0 x 1 5 ,x 2 0 x We see that for all nonzero values of x,
d 1 ( ln Z x Z 1 C ) 5 . x dx
399
EXAMPLE
Integrating a Function of the Form y 5

Integrate x with respect to x.
SOLUTION
3
c x
3 1 3 a x b dx 5 3 3 a x b dx 5 3 ln Z x Z 1 C
Constant Multiple Rule for Integrals Integral Rule for x

1
Indefinite Integral Applications

Recall that the units of dy are the units of y over the units of x. Since dx dy dy 3 a dx b dx 5 y 1 C , the units of the antiderivative of dx are the units of y. In real-world applications, we are often given enough information to figure out the specific value of C, as demonstrated in Example 7.
EXAMPLE
Using Integration to Find a Position Function

Based on 2.5 minutes of data, the velocity of the authors minivan as he drove from a stoplight onto a highway may be modeled by V( t ) 5 2 0.1773t 2 1 0.6798t 1 0.5019 miles per minute, where t is the number of minutes since he left the stoplight. Find the position function D( t ) that shows his distance from the stoplight at time t. Find the total distance traveled during the first minute and during the first two minutes. function V( t ) .
SOLUTION The position function D( t ) is the antiderivative of the velocity
2 3 3V( t ) 4 dt 5 3( 2 0.1773t 1 0.6798t 1 0.5019 ) dt t3 t2 D( t ) 5 2 0.1773 1 0.6798 1 0.5019t 1 C 3 2
Constant Multiple, Power, and Sum and Difference Rules
5 2 0.0591t 3 1 0.3399t 2 1 0.5019t 1 C miles You may ask, What is the meaning of C in this context? Lets see. D( 0 ) represents the distance traveled after 0 minutes. Intuitively, we know that after 0 minutes, the minivan hasnt traveled at all, so D( 0 ) 5 0 . D( 0 ) 5 2 0.0591( 0 ) 3 1 0.3399( 0 ) 2 1 0.5019( 0 ) 1 C 5C But D( 0 ) 5 0 , so C 5 0. D( t ) 5 2 0.0591t 3 1 0.3399t 2 1 0.5019t is the distance function for the minivan. According to our distance model, D( 1 ) 5 0.78 and D( 2 ) 5 1.89 . That is, the total distance traveled during the first minute is 0.78 mile. The total distance traveled during the first two minutes is 1.89 miles.
400
EXAMPLE
Using Antidifferentiation to Find a Cost Function

Have you ever dreamed of being a published author? Golden Pillar Publishing (www.goldenpillarpublishing.com) publishes books for aspiring authors. In 2003, the marginal cost for printing an 8.25 s 3 11 s soft-cover book was c( p) 5 0.018 dollar per page where p is the number of pages in the book. Including the setup costs, a 1000page book has a $19.30 production cost. Find the book production cost function.
SOLUTION Recall that marginal cost is the derivative of the cost function. Therefore, the antiderivative of the marginal cost is the cost function.
3c ( p) dp 5 30.018 dp C( p) 5 0.018p 1 k
Constant Rule for Integrals
(Note: We used k as a constant, since the function was named C.) The cost of a 1000-page book is $19.30, so C( 1000 ) 5 0.018( 1000 ) 1 k 19.30 5 18 1 k k 5 1.30 The constant k 5 1.30 represents the fixed setup cost. The cost to produce a book with p pages is C( p) 5 0.018p 1 1.30 dollars.
6.1 Summary
In this section you learned how to find the equation of some functions that have a given derivative through the process of integration. You also discovered how to use basic integration rules to integrate a variety of functions.

1. Write the expression 6x without a denominator. 2. Write the expression x 2 without a denominator. 3. Write the expression 4. Write the expression
3x4 4 5x7 7 3 3
6. If a positive constant c is added to any function, what effect does it have on that functions graph? 7. Give an example of two different functions whose graphs have equal instantaneous rates of change (slopes). 8. Give an example of two different nonlinear functions whose equations differ by only a constant.
as a constant times x4 . as a constant times x7 .
5. What do parallel lines have in common?
401
6.1 Exercises
In Exercises 120, find the general antiderivative of the function. 1. f ( x ) 5 2 3. r( t ) 5 t21 2 3t 5. v( t ) 5 0.5t 1 20 2. g( x ) 5 2 2x 1 9 4. f ( t ) 5 1.21 t 30. 3( 4t22 2 2t ) dt 200 400 31. 3 a 2 1 1 50 b dx x x 32. 3( 9t 2 3t ) dt 34. 3 a 20 2 5u 1 2 b du u 4 5 33. 3 a 2 2 b dx x x
6. a( t ) 5 2 32
7. s( t ) 5 t 3 2 3t 2 1 3t 2 1 8. s( x ) 5 2x 10. f ( x ) 5 0 12. r( x ) 5 200 2 225 x 9. h( x ) 5 2.3 11 x

x 3
11. p( x ) 5 3 2 x
35. 3( 4t 1 4t ) dt In Exercises 3639, apply the concept of integration in solving the real-world applications. 36.
Publishing Costs Golden Pillar Publishing publishes books for aspiring authors. In 2003, the marginal printing cost for a 7.5 s 3 9.25 s soft-cover book was
13. f ( t ) 5 4t22 1 2t21 1 1 14. g( x ) 5 2x 2 x 2 15. h( x ) 5 2( 3x ) 2 3( 2x ) 1000 16. q 5 2500 2 2 p 4x 3 2 15 20. h( x ) 5 x2 In Exercises 2135, calculate the indefinite integral. 21. 3( 3x 2 5 ) dx 23. 3 22. 3( 2 x 2 1 4x ) dx 2000 500 2 2 17. q 5 p p
18. f ( x ) 5 3( 2x ) 1 4x21 19. y 5 5t22 1 2t 37.
c( p ) 5 0.013 dollar per page, where p is the number of pages in the book. Including setup costs, a 1000-page book has a $13.90 production cost. (Source:
www.goldenpillarpublishing.com.)
Find the book production cost function.

Publishing Costs Golden Pillar Publishing publishes books for aspiring authors. In 2003, the marginal printing cost for a 6 s 3 9 s hard-cover book was
t22 dt (Hint: Divide each term in the t numerator by the denominator before integrating.) 25u 2 1 du u 38.
c( p) 5 0.013 dollar per page, where p is the number of pages in the book. Including setup costs, a 1000-page book has an $18.00 production cost. (Source:
www.goldenpillarpublishing.com.)
24. 3
Find the book production cost function.

Cost Analysis The marginal printing cost of the soft-cover and hard-cover books in Exercises 36 and 37 was the same, yet their cost functions were different. Why?
25. 3( 3x22 2 4x 3 1 2 ) dx 26. 3( 4t21 2 12 ) dt 27. 3( 5t22 2 16t 2 2 9 ) dt 28. 3 35( 2t ) 2 3t 2 4 dt 29. 3 5u2 2 4u 1 1 du u2
39.
Based on 1.5 minutes of data, the velocity of the authors minivan as he drove from a stoplight into a sparsely populated residential area may be modeled by
Minivan Position
402
V( t ) 5 2 2.133t 3 1 3.999t 2 2 0.1533t 1 0.3167 miles per minute, where t is the number of minutes since he left the stoplight. Find the position function D( t ) that shows his distance from the stoplight at time t.
45. Let F1( x ) and F2( x ) be specific antiderivatives of f ( x ) . What is the graphical interpretation of Z F1( x ) 2 F2( x ) Z ? 46. Let F1( x ) and F2( x ) be specific antiderivatives of f ( x ) , with F1( x ) 2 F2( x ) . Do the graphs of F1( x ) and F2( x ) ever intersect?
40. Explain why a single function may have an infinite number of antiderivatives. 41. Can the Power Rule for Integrals, 3x ndx 5 be used to solve 3x21dx? Explain. 42. We know that d 1 1 3ln( x ) 4 5 , so why does 3 dx x x dx equal ln Z x Z 1 C rather than ln( x ) 1 C ?
xn 1 1 , n11
47. Let f ( x ) be a polynomial of degree n with f ( x ) 5 Fr( x ) . At most, how many x-intercepts does F( x ) have? 48. Let Fr( x ) 5 3 1 x 2. Is F( x ) an increasing or a decreasing function? Explain. 49. Let Fr( x ) 5 3x 2 and F( 0 ) 5 0. On what interval(s) is F( x ) $ 0? 50. Let Fr( x ) 5 4 2 2x and F( 0 ) 5 2 4. On what interval(s) is F( x ) $ 0?
1
43. We know that the derivative of any linear function is equal to its slope. Explain how this is related to the Constant Rule for Integrals. 44. If the units of dy are dollars per year, what will be dx dy the units of 3 a b dx? dx
6.2 Integration by Substitution
Use the method of substitution to integrate functions
The process of integration by substitution is a clever way to integrate a function that initially doesnt appear to be integrable. (Recall that a function f is said to be integrable if there exists a function F such that Fr 5 f.) Because of the relatively complex nature of the process, we will defer illustrations of how this process may be used in real-life applications to later sections. In this section, we will introduce the concept of differentials. We will then demonstrate how to do integration by substitution.
Differentials
d dy dy is a single term, not the quotient of dy ( y ) 5 . The derivative dx dx dx dy and dx. However, it would be useful for us to be able to treat as if it were a dx fraction. For this purpose, we introduce the concept of differentials. We define the differential dx to be an arbitrarily small real number and Recall that dy 5 yrdx
403
Observe that dividing both sides of the equation dy 5 yrdx by dx yields dy 5 yr dx which is consistent with our earlier notation for the derivative. Thus, by using dy differentials, we can treat dy and dx in as separate entities and use the rules for dx fractions in manipulating the terms.
EXAMPLE
Calculating the Differential dy

Find dy given y 5 x2 .
SOLUTION
dy 5 2x dx
EXAMPLE
Calculating the Differential dy

Find dy given y 5 2x3 2 3x .
SOLUTION
dy 5 ( 6x2 2 3 ) dx
We may also use differentials with other variables, such as u and t, as demonstrated in the next two examples.
EXAMPLE
Calculating the Differential du

Find du given u 5 e 3t 1 2.
SOLUTION
du 5 3e 3tdt
EXAMPLE
Calculating the Differential du

Find du given u 5 ln( t ) .
SOLUTION
du 5
1 dt t
404
Integration by Substitution
The technique of integration by substitution is used to rewrite a function that doesnt appear to be readily integrable in such a way that it can be integrated using the basic integration rules.
EXAMPLE
Integrating by Substitution
Integrate f ( x ) 5
2 ln( x ) . x 2 ln( x ) x
dx . This function cannot be readily integrated using any of the rules covered in Section 6.1. This is because
SOLUTION We must find 3
2 ln( x ) x
is the product of 2 ln( x ) and x , and we havent yet discussed how to integrate a product. (Contrary to what you might think, 3 3 f ( x ) g( x ) 4 dx 2 3 f ( x ) dx ? 3g( x ) dx.) We will let u 5 ln( x ) . We take the derivative of both sides and use the concept of differentials to find du. u 5 ln( x ) d d (u) 5 3ln( x ) 4 dx dx du 1 5 x dx du 5 1 dx x
f (x) 5
We will now rewrite the integral in terms of u. 1 32 3ln( x ) 4 a x dx b 5 32u du 5 5 u du But we know that
2 32u du 5 u 1 C (Although the differential du represents an arbitrarily small number and the du of an integral means with respect to u, the two uses of the notation du may be interchanged. A more in-depth explanation may be obtained from an advanced calculus text.) Since the variable u was something that we introduced into the problem, we must rewrite the solution in terms of x. Recall that u 5 ln( x ) .
u2 1 C 5 3ln( x ) 4 2 1 C
Thus, we have 3 2 ln( x ) dx 5 3ln( x ) 4 2 1 C x
405
It is not always possible to find a substitution of variables that makes a difficult function integrable. However, if the integral can be rewritten in any of the following forms, it may be integrated using the basic integration rules.
n 3u du
3e du 1 3 u du
u
Just In Time
Changing Variables
Oftentimes equations are made easier by changing the variables and their associated meanings. When changing variables, it is critical to keep track of the meaning of each variable and to interpret the solution(s) appropriately. For example, the population of Indonesia may be modeled by the equation P 5 222,781,000 ( 1.012 ) t 2 1990 which gives the number of people living in Indonesia as a function of the year (e.g., 1992). If we let I be the number of people (in thousands) and x be the P number of years since 1990, we have the following relationships: I 5 1000 and x 5 t 2 1990. Dividing each side of the model by 1000, we have P 222,781,00 (1.012 ) t 2 1990 5 1000 1000 I 5 222,781 (1.012 ) x The new equation is simpler than the original equation, making computations easier. If we are asked to predict the number of people living in Indonesia in 2010, we first determine that 2010 is 20 years after 1990. So I( 20 ) 5 222,781( 1.012 ) 20 < 282,806 In 2010, the population of Indonesia is predicted to be 282,806 thousand people.
EXAMPLE
Find the antiderivative of f ( x ) 5 2x( x 2 1 5 ) 10 .
SOLUTION We need to determine 32x( x2 1 5 ) 10 dx . The function is a product
of two functions, so we hope to be able to write it in the form u ndu. Well try u 5 x2 1 5 . Then u 5 x2 1 5 du 5 2x dx du 5 2x dx Lets see if we can rewrite the integral in terms of u and du. Recall that u 5 x2 1 5 and du 5 2x dx .
406

2 10 2 10 32x( x 1 5 ) dx 5 3 3( x 1 5 ) 4 ( 2x dx )
Switch the order of the factors and group Since u 5 x2 1 5 and du 5 2x dx
5 3u10 du 5 u11 1C 11
Rewriting the result in terms of x, we have

2 10 32x( x 1 5 ) dx 5
( x2 1 5 ) 11 1C 11
We can check our result by differentiating the antiderivative and comparing it to f (x). d ( x2 1 5 ) 11 1 d c 1 Cd 5 3( x2 1 5 ) 114 1 0 dx 11 11 dx 1 5 11 ? ( x2 1 5 ) 10 ? 2x 11 5 2x( x2 1 5 ) 10
Constant and Constant Multiple Rules Power and Chain Rules
The derivative of the antiderivative is equal to f ( x ) , so we are confident that we integrated the function correctly.
EXAMPLE
Integrate g( t ) 5 e 2t 2 4.
SOLUTION We must find 3e 2t 2 4 dt . We hope to be able to rewrite the integral
as 3e udu. Well let u 5 2t 2 4. Then du 52 dt du 5 2 dt Looking at the integrand (the function to be integrated), we see that we have dt but not 2 dt . Can we solve du 5 2 dt for dt? Yes! 2 dt 5 du 1 dt 5 du 2 Substituting, we get 3e
2t 2 4
1 dt 5 3e u A du B 2 5 1 u e du 23
Since u 5 2t 2 4 and dt 5 2 du Constant Multiple Rule for Integrals Exponential Rule Rewrite in terms of t
1 5 eu 1 C 2 1 5 e 2t 2 4 1 C 2
407
EXAMPLE
Find the general form of a function that has the derivative f ( x ) 5 x2 1
2x 1 3 2x 1 3 . 3x 1 4
dx . We observe that the numerator is SOLUTION We must find 3 2 x 1 3x 1 4 the derivative of the denominator. Consequently, we hope to be able to rewrite the integral in the form 3 u du. Well let u 5 x2 1 3x 1 4 . Then du 5 2x 1 3 dx du 5 ( 2x 1 3 ) dx Rewriting the integral in terms of u and solving, we get 2x 1 3 ( 2x 1 3 ) dx 3 x2 1 3x 1 4 dx 5 3 x2 1 3x 1 4 du 5 3 u 1 5 3 du u 5 ln Z u Z 1 C 5 ln Z x2 1 3x 1 4 Z 1 C
2
Since u 5 x2 1 3x 1 4 and du 5 ( 2x 1 3 ) dx
Integral Rule for u Rewrite in terms of x
All functions of the form F( x ) 5 ln Z x 1 3x 1 4 Z 1 C have the derivative f ( x ) 5 x2 1

2x 1 3 . 3x 1 4
In each of the previous examples, we selected the correct u each time. However, it is very common to attempt integration by substitution with the wrong function for u. If you rewrite a function in terms of u and the resulting function isnt integrable, go back and select a different function to be u. This is part of the process of integration by substitution and should not be viewed as an error. As you gain experience with these types of problems, you will more easily recognize what is the best choice for u. In Example 9, we will demonstrate a type of problem that requires an especially clever selection of u.
EXAMPLE
Integrating by Substitution a Function Containing a Radical

Find 3x"x 2 1 dx.
SOLUTION We begin by rewriting the integrand with a rational exponent.
1>2 3x( x 2 1 ) dx
Lets pick u 5 x 2 1. Then du 51 dx du 5 dx
408
We make the substitution. 3xu du What do we do with the extra x? The integrand must be written in terms of u. (We cant use the Constant Multiple Rule to move the x to the left-hand side of the integral sign because x is a variable, not a constant.) Recall that u 5 x 2 1; therefore, x 5 u 1 1 . Thus we may rewrite the integral as
1>2 1>2 3xu du 5 3( u 1 1 ) u du Since we have completely rewritten the integral in terms of u, we may continue. We will multiply out the integrand and then integrate each term. 1>2
3 Au
3>2
1 u1>2 B du 5 3u 3>2 du 1 3u1>2 du 5 u5>2

5 2
u3>2
3 2
1C
Power Rule for Integrals
2 2 5 u5>2 1 u3>2 1 C 5 3 2 2 5 ( x 2 1 ) 5>2 1 ( x 2 1 ) 3>2 1 C 5 3 Thus 3x"x 2 1 dx 5 5 ( x 2 1 ) 5>2 1 3 ( x 2 1 ) 3>2 1 C.

2 2
EXAMPLE 10
Determining When Integration by Substitution Wont Work

Integrate f ( x ) 5 xe x.
SOLUTION We must find 3xe x dx . If we pick u 5 x, then du 5 dx and the
integral becomes
u 3ue du
Observe that this is identical to the original function written, except that it is written in terms of u. Using the integral techniques weve covered, we dont know how to integrate a function in this form. Lets try u 5 e x. Then u 5 ex ln( u ) 5 ln( e x ) ln( u ) 5 x x 5 ln( u ) Since du 5 e x, du 5 e xdx. The resultant integral is dx
x 3xe dx 5 3 3ln( u ) 4 du
409
We havent yet learned how to integrate ln( u ) , so we are unable to find the solution using this choice of u. As a final attempt, well try u 5 xe x. Then du 5 ( 1 ? e x 1 e x ? x ) dx, by the Product Rule. If we attempt to make a substitution, well have a mixture of u and x in the integrand, leaving us with a function that we cant integrate The function f ( x ) 5 xe cannot be integrated using integration by substitution. Nevertheless, there is a method that can be used to integrate f ( x ) 5 xe x. In Section 7.1, we will show how to integrate this function using integration by parts.
x
a 3xe xdx 5 3u (1
du # ex 1 ex # x)
b.
It is important to note that, despite the many advances in calculus, there are some functions that we still dont know how to integrate. Nevertheless, we must attempt to integrate a function using all known methods before we can conclude that it is not integrable.
6.2 Summary
In this section you learned how to work with differentials. You used integration by substitution to find the antiderivative of a function that didnt initially appear to be integrable. You also learned that there are functions that we cant integrate simply by reversing the derivative rules.

1. Let u 5 x 2 1 5 and v 5 2x. Rewrite the expression 2x( x 2 1 5 ) 3 in terms of u and v. 2. Let u 5 4x 1 6x and v 5 12x 1 6. Rewrite the expression ( 12x 2 1 6 )( 4x 3 1 6x ) 22 in terms of u and v.
3 2
5. Let u 5 ln( x ) and v 5 x . Rewrite the expression

ln( x ) x
in terms of u and v.
2 16 3 ln( x ) 4 3 x
2
6. Let u 5 2 ln( x ) and v 5 x . Rewrite the expression in terms of u and v.

2
3. Let u 5 x 2 1 7 and v 5 2x. Rewrite the expression x( x 2 1 7 ) 3 in terms of u and v. 4. Let u 5 4x 5 2 x 2 1 1 and v 5 20x4 2 2x. x4 2 4x Rewrite the expression 4x40 in terms of u 5 2 x2 1 1 and v.
7. Let u 5 5x 2 10 and v 5 10x. Rewrite the expression ( 40x ) 25x 2 10 in terms of u and v. 8. Let u 5x x22 and v 5 2 x21. Rewrite the expres3 sion 2 x in terms of u and v.
22
410
6.2 Exercises
In Exercises 140, integrate the functions. Some integrals will require integration by substitution, while others may be integrated using basic integration rules. All of the functions may be integrated using the techniques weve covered. 1. f ( x ) 5 2x( x 1 3 )
2 5
35. a( t ) 5 ( 4t 2 1 )( t2 2 2t ) 36. q 5 37. q 5 1000 2 p"p 2 2 p 4p 2 4p"2p2 1 1 2p2 1 1
2. g( x ) 5 6x ( x 1 7 )
2 3
3. h( x ) 5 2x( 3x 1 5 ) 4. f ( t ) 5 ( 2t 1 3 ) ( t 2 1 3t 1 1 ) 3 5. s( x ) 5 e x( e x 1 1 ) 8. w( t ) 5 10te5t 10. y 5 3e 2x 12. f ( x ) 5 ( 2x 2 9 ) e x 13. g( t ) 5 4e 3t 15. s( x ) 5 17. h( t ) 5 19. f ( x ) 5 21. p( x ) 5 ln( x2 ) x 2t 2 1 2 t 2t12 6x 2 x3 2 9 1 x ln( x )
224
38. s( t ) 5 6t( t2 1 1 ) 3 39. s( t ) 5 ( 4t 2 2 )( t2 2 t ) 21 40. v( x ) 5 x( 2 x

221
6. v( t ) 5 3t 1 t21 9. p( x ) 5 ( x 2 2 ) e x 11. y 5 3xe3x

229x 1 21 226 224x
7. f ( x ) 5 1.2x 2 2 2.4x 1 0.6 41. What is the purpose of integration by substitution? 42. When using integration by substitution, what do you look for when choosing u? 43. If du 5 dx, what is the relationship between u and x? 44. In solving 33x( x2 1 1 ) dx , a student picks u 5 x2 1 1 and du 5 3x dx . What is wrong with this approach? 45. How is the Chain Rule for derivatives related to integration by substitution?
14. y 5 3( 2x ) 16. y 5 4x( x2 1 1 ) 3 18. g( x ) 5 20. h( t ) 5 22. s( x ) 5 x24 2 x 2 8x 1 9 2t 2 1 t 2t12

2
ln( x ) x
23. y 5 3x"x 2 2
25. g( x ) 5 2x"x2 1 1 27. f ( t ) 5 t"t 2 2 1 29. f ( x ) 5 30. g( t ) 5

3
24. f ( x ) 5 x"x 1 2
26. h( x ) 5 ( x 2 1 ) "x2 2 2x
46. Determine if f ( x ) 5 2x3( x2 2 1 ) 4 may be integrated using integration by substitution. 47. Determine if f ( x ) 5 48. Determine if f ( x ) 5 49. Determine if f ( x ) 5
ln( x ) 1 1 x ln( x ) x ln( x ) ln( x2 ) 4x
28. s( x ) 5
3x2 2 2x 1 1 x 2 x2 1 x 2 1 2 "t e"t
"x 2 1
2
may be integrated
using integration by substitution. may be integrated using
integration by substitution. may be integrated using
x 31. f ( x ) 5 2x 2 1
e x 2 ln( 2 )( 2x ) 32. g( x ) 5 e x 2 2x 33. h( x ) 5 2x 4x2 2 5 34. v( t ) 5 ln( 2 )( 3t )
integration by substitution. 50. Determine if f ( x ) 5 3x2 ( x2 2 1 ) 7 may be integrated using integration by substitution.
411
Estimate the area between a curve and the x-axis by using left- and right-hand sums
River rafting or kayaking is an exciting sport that requires both physical skill and an intellectual understanding of the hydraulics of a river. One factor that enthusiasts must consider before navigating a river is the flow rate of the water. The flow rate indicates how much water (in cubic feet) will pass a fixed point in a second. The flow rate fluctuates constantly, ever changing the intensity of the river. For the inexperienced, flow-rate values mean little. One way to add conceptual meaning to the numbers is to determine how much water has passed a given point over a period of time. As will be explained in the section, we can do this by estimating the area between a graph of the flow rate and the horizontal axis. In this section, we will demonstrate how left- and right-hand sums may be used to approximate the area between the graph of a function and the x-axis. This will prepare us to link the notion of area with the concept of the definite integral in Section 6.4. The relationship between the two concepts is truly remarkable.
Suppose that the flow rate f ( x ) of a river is a constant 3000 cubic feet per second. Lets consider the graph of the flow-rate function f ( x ) 5 3 , shown in Figure 6.2. The graph is a horizontal line with y-intercept ( 0, 3 ) . We ask the question, What is the area of the region between the graph of the function and the x-axis on the interval 31, 34 ?
Flow rate (thousand cubic feet per second)
Flow rate (thousand cubic feet per second)
y 4 3 2 1 0 0 1 2 3 Seconds 4 x f (x) = 3
y 4 3 2 1 0 0 1 2 3 Seconds 4 x f (x) = 3
FIGURE 6.2
FIGURE 6.3
That is, what is the area of the shaded region in Figure 6.3? We quickly recognize that the region is a rectangle. To calculate its area, we multiply its height by its width. A53?2 56
412
The area of the region is 6 square units. What does this mean in the context of the problem? Lets recalculate A, this time keeping track of the physical meaning of the length and width of the rectangle. A 5 a3 thousand cubic feet b ( 2 seconds ) second
5 6 thousand cubic feet Between the first and third second after we started timing, 6000 cubic feet of water passed our water flow measurement station. In general, if f ( x ) is a positive rate of change function, then the area between the graph of f ( x ) and the x-axis on the interval 3a, b4 represents the total change in F( x ) [an antiderivative of f ( x ) ] from x 5 a to x 5 b. When the shaded region is a familiar geometric shape (triangle, rectangle, and so on), it is easy to calculate the area using known formulas. However, what do we do if the shaded region is not a familiar geometric shape? Can we still calculate the area? In Example 1, we will estimate the area of an irregularly shaped region by drawing a series of rectangles. As you will see, the accuracy of our estimate will increase as we increase the number of rectangles.
EXAMPLE
Using Riemann Sums to Estimate an Area

Estimate the area of the region between f ( x ) 5 "x 1 2 and the x-axis on the interval 30, 44 .
SOLUTION We begin by drawing the graph of f ( x ) 5 "x 1 2 (see Figure 6.4). We will first estimate the area by drawing a single rectangle whose upper left corner touches the graph of f (Figure 6.5). The area of the rectangle is A 5 2 ? 4 5 8 square units . Since the entire region below the graph is not completely shaded, 8 square units is an underestimate of the actual area. To improve our accuracy, we will draw two rectangles that are each two units wide. The upper left-hand corner of each rectangle will touch the graph (see Figure 6.6).
y 4 3 2 1 0 0 1 2 3 4 x 4 3 2 1 0 0 1 2 3 4 x y 4 3 2 1 0 0 1 2 3 4 A1 x y
A2
FIGURE 6.4
FIGURE 6.5
FIGURE 6.6
The area of the first rectangle is easily calculated. A1 5 2 ? 2 5 4 square units. We can easily see that the second rectangle has a width of 2 units, but it is
413
difficult to tell from the graph what the height of the rectangle is. However, since the upper left-hand corner of the rectangle is on the graph of f , we can use the equation of f ( x ) to calculate the height of the rectangle. Since the rectangle touches the graph when x 5 2 , the height of the rectangle is f ( 2 ) 5 "2 1 2 < 1.414 1 2 < 3.414 units Consequently, the approximate area of the second rectangle is A2 5 3.414 ? 2 5 6.828 We can estimate the area of the region by adding together the areas of the two rectangles. A 5 A1 1 A2 5 4 1 6.828 5 10.828 square units
y 4 3 2 A2 1 0 0 1 2 3 4 A1 x A4
A3
This estimate is substantially better than our first estimate because it has increased the size of the shaded region below the graph. However, it is still an underestimate. How can we increase the accuracy of our estimate? Add more rectangles. This time well use four rectangles of width 1 unit, as shown in Figure 6.7. The width of each rectangle is 1 unit. Since the upper left-hand corner of each rectangle touches the graph of f , the heights of the rectangles are f ( 0 ) , f ( 1 ) , f ( 2 ) , and f ( 3 ) , respectively. The areas of the rectangles are given by A1 5 f ( 0 ) ? 1 52?1 5 2 square units A2 5 f ( 1 ) ? 1 53?1 5 3 square units A3 5 f ( 2 ) ? 1 5 A "2 1 2 B ? 1 5 A "3 1 2 B ? 1
FIGURE 6.7
5 3.414 square units A4 5 f ( 3 ) ? 1
5 3.732 square units The area of the entire shaded region is the sum of the individual areas. A 5 A1 1 A2 1 A3 1 A4 5 2 1 3 1 3.414 1 3.732 5 12.146 square units
414
How can we improve our estimate? Increase the number of rectangles. This time well use eight rectangles of width 0.5 unit (see Figure 6.8). The width of each rectangle is 0.5 unit. The heights of the rectangles are found by evaluating f ( x ) at x 5 0.0 , x 5 0.5 , x 5 1.0 , x 5 1.5 , x 5 2.0 , x 5 2.5 , x 5 3.0 , and x 5 3.5 (see Table 6.1).
y 4 3 2 1 A1 0 0 1 2 3 4 A2 A3 A4 A5 A6 A7 A8
x 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
f ( x ) 5 "x 1 2 2.000 2.707 3.000 3.225 3.414 3.581 3.732 3.871
The area of the shaded region is given by A 5 A1 1 A2 1 A3 1 A4 1 A5 1 A6 1 A7 1 A8 5 2( 0.5 ) 1 2.707( 0.5 ) 1 3( 0.5 ) 1 3.225( 0.5 ) 1 3.414( 0.5 ) 1 3.581( 0.5 ) 1 3.732( 0.5 ) 1 3.871( 0.5 ) 5 ( 2.000 1 2.707 1 3.000 1 3.225 1 3.414 1 3.581 1 3.732 1 3.871 ) ( 0.5 ) 5 ( 25.530 ) ( 0.5 ) 5 12.765 square units We will next estimate the area using 16 rectangles. We know that our area estimate will continue to improve as we increase the number of rectangles; however, as the number of rectangles increases and their width decreases, it becomes increasingly difficult to draw the rectangles. We need a better method.
415
Since the width of the entire region is 4 units, the width of each of the 16 rectangles will be 16 5 0.25 unit. Well make a table for f ( x ) with the x-values starting at x 5 0 and continuing to x 5 4 , with values spaced 0.25 unit apart (see Table 6.2).
TABLE 6.2
4
x 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00
f ( x ) 5 "x 1 2 2.000 2.500 2.707 2.866 3.000 3.118 3.225 3.323 3.414 3.500 3.581 3.658 3.732 3.803 3.871 3.936 4.000
We know that the height of each left-hand rectangle is determined by evaluating f ( x ) at the x-value on the leftmost edge of the rectangle. The leftmost edge of the first rectangle occurs at x 5 0 . The leftmost edge of the second rectangle occurs at x 5 0.25 . Continuing on down, we see that the leftmost edge of the sixteenth rectangle occurs at x 5 3.75 . The heights of the rectangles are highlighted in the red box in Table 6.3.
416
TABLE 6.3
Rectangle Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
x 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00
f ( x ) 5 "x 1 2 2.000 2.500 2.707 2.866 3.000 3.118 3.225 3.323 3.414 3.500 3.581 3.658 3.732 3.803 3.871 3.936 4.000
Since each rectangle has the same width, we may add up all of the heights of the rectangles and multiply them by the width instead of having to multiply the height of each rectangle individually by its width. (Both methods will yield the same result, but the first method requires fewer computations.) A 5 ( 2 1 2.5 1 2.707 1 2.866 1 3 1 3.118 1 3.225 1 3.323 1 3.414 1 3.5 1 3.581 1 3.658 1 3.732 1 3.803 1 3.871 1 3.936 ) ( 0.25 ) 5 ( 52.234 ) ( 0.25 ) < 13.06 square units This method is easier than drawing the rectangles; however, even it can become cumbersome when there are large numbers of rectangles. As it turns out, the exact area of the region is 13 3 square units. Our final estimate was fairly close to the exact value.
1
Each of the estimates in Example 1 is called a left-hand sum, since the upper left corner of each of the rectangles used in the sum touched the graph of f . The notion of using sums is credited to the famous mathematician Bernhard Riemann. Consequently, these sums are commonly called Riemann sums.
417
EXAMPLE

Estimate the area of the region between f ( x ) 5 3x2 and the x-axis on the interval 31, 34 using a left-hand sum with n 5 4.
SOLUTION We are asked to estimate the area of the shaded region in Figure 6.9 by using a left-hand sum. Note that the shaded region is constrained to the interval 31, 34 . Since n 5 4, we are to use 4 rectangles. The width of each rectangle is
y 25 20 15 10 5 0 0 0.5 1.0 1.5 2.0 2.5 3.0 x
321 2 5 4 4 5 0.5 unit (In general, if we want to use n rectangles on a region between x 5 a and x 5 b , the width of each rectangle is We will create a table of data for f ( x ) starting at x 5 1 and continuing to x 5 3 with x-values spaced 0.5 unit apart (see Table 6.4).
TABLE 6.4
b2a .) n
FIGURE 6.9
Rectangle Number 1 2
y 25 20 15 10 5 0 0 0.5 1.0 1.5 2.0 2.5 3.0 x
x 1.0 1.5 2.0 2.5 3.0
f ( x ) 5 3x2 3.00 6.75 12.00 18.75 27.00
3 4
The left-hand sum estimate of the area is A 5 ( 3.00 1 6.75 1 12.00 1 18.75 ) ( 0.5 ) 5 20.25 square units By drawing the rectangles, as shown in Figure 6.10, we can see that this is an underestimate of the actual area.
FIGURE 6.10
EXAMPLE

Estimate the area of the region between f ( x ) 5 31, 24 using a left-hand sum with n 5 5.
1 x
and the x-axis on the interval
SOLUTION We are asked to estimate the area of the shaded region in Figure 6.11 by using a left-hand sum with five rectangles.
418
y 1.0 0.8 0.6 0.4 0.2 0 0 1.00 1.25 1.50 1.75 2.00 x
FIGURE 6.11
The width of each rectangle is 5 0.2 unit. We will create a table of 5 values for f ( x ) starting at x 5 1 and continuing to x 5 2 with x-values spaced 0.2 unit apart (see Table 6.5).
TABLE 6.5
221
Rectangle Number 1 2 3 4 5
x 1.0 1.2 1.4 1.6 1.8 2.0
f (x) 5 1.000 0.833 0.714 0.625 0.556 0.500
1 x
The left-hand sum estimate of the area is A 5 ( 1.000 1 0.833 1 0.714 1 0.625 1 0.556 ) ( 0.2 ) 5 ( 3.728 ) ( 0 . 2 ) 5 0 . 746 square unit By drawing the rectangles as shown in Figure 6.12, we can see that this is an overestimate of the actual area.
y 1.0 0.8 0.6 0.4 0.2 0 0 1.0 1.2 1.4 1.6 1.8 2.0 x
FIGURE 6.12
419
Finding the Left-Hand Sum for Small Values of n ( n # 10 ) Let f be a function defined on the interval 3a, b4 whose graph lies above the x-axis. The left-hand sum estimate for the area between the graph of f and the x-axis may be found using the following steps.
1. Calculate the width of each rectangle, D x 5
D x is read delta x and is the distance between consecutive x values. 2. Create a table of values for f ( x ) starting at x 5 a and ending at x 5 b , with intermediate x-values spaced D x units apart. 3. Add up the first to the penultimate (second to last) values of f ( x ) listed in the table. (This is the sum of the heights of the rectangles.) 4. Multiply the result of Step 3 by D x to get the left-hand sum approximation of the area. (Note: This method may be used for any value of n; however, for values of n larger than 10, the process becomes extremely tedious.)
b2a . n
EXAMPLE
Using a Left-Hand Sum to Estimate an Area

Estimate the area between the graph of f ( x ) 5 2 x2 1 4x and the x-axis on the interval 30, 44 using a left-hand sum with eight rectangles.
SOLUTION We are asked to estimate the area of the region shown in
Figure 6.13.
y 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 x
FIGURE 6.13
The width of each rectangle is Dx 5 5 420 8 1 2
5 0.5
420
We will create a table of values starting at x 5 0 and continuing to x 5 4 with the x-values spaced 0.5 unit apart (see Table 6.6).
TABLE 6.6
Rectangle Number 1 2 3 4 5 6 7 8
x 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
f ( x ) 5 2 x2 1 4x 0.00 1.75 3.00 3.75 4.00 3.75 3.00 1.75 0.00
The sum of the heights of the rectangles is 0 1 1.75 1 3 1 3.75 1 4 1 3.75 1 3 1 1.75 5 21 We now multiply the sum of the heights by the width of a rectangle. A 5 21( 0.5 ) 5 10.5 square units This left-hand sum estimate is close to the exact area (10 3 square units).
2
As you may have guessed, we can also calculate a right-hand sum. The righthand sum is the sum of the areas of rectangles whose upper right-hand corner touches the graph of f . The process for finding the right-hand sum is similar to the process for finding the left-hand sum, as will be shown in Example 5.
EXAMPLE
y 2.0 1.5 1.0 0.5 0 0 0.5 1.0 1.5
Using a Right-Hand Sum to Estimate an Area

Estimate the area between the graph of f ( x ) 5 x3 2 3x2 1 3x and the x-axis on the interval 30, 24 using a right-hand sum with four rectangles.
SOLUTION We are asked to estimate the area of the shaded region in Figure 6.14. For the right-hand sum, the upper right-hand corner of each rectangle will touch the graph of f ( x ) , as shown in Figure 6.15. The width of each rectangle will be
Dx 5
2.0 x
220 4
5 0.5
FIGURE 6.14

y 2.0 1.5 1.0 0.5 0 0 0.5 1.0 1.5 2.0 x
421
We will create a table of values for f ( x ) starting at x 5 0 and continuing to x 5 2 with the x-values spaced 0.5 unit apart (see Table 6.7). For the right-hand sum, we select the second through the last term in the table. These are the respective heights of the rectangles.
TABLE 6.7
Rectangle Number 1 2 3 4
x 0 0.5 1.0 1.5 2.0
f ( x ) 5 x3 2 3x2 1 3x 0.000 0.875 1.000 1.125 2.000
FIGURE 6.15
We will sum the heights together and multiply them by the width of a rectangle. A 5 ( 0.875 1 1.000 1 1.125 1 2.000 ) ( 0.5 ) 5 ( 5 ) ( 0.5 ) 5 2.5 square units The right-hand sum estimate for the area is 2.5 square units (using four rectangles).
The process for finding the right-hand sum is identical to the process for finding the left-hand sum, with the exception of Step 3. For the left-hand sum, you add up the first through the penultimate term in the table. For the right-hand sum, you add up the second through the last term.
Finding the Right-Hand Sum for Small Values of n ( n # 10 ) Let f be a function defined on the interval 3a, b4 whose graph lies above the x-axis. The right-hand sum estimate for the area between the graph of f and the x-axis may be found using the following steps.
1. Calculate the width of each rectangle, D x 5 2. Create a table of values for f ( x ) starting at x 5 a and ending at x 5 b ,
b2a . n
with intermediate x-values spaced D x units apart. 3. Add up the second through the last values of f ( x ) listed in the table. (This is the sum of the heights of the rectangles.) 4. Multiply the result of Step 3 by D x to get the right-hand sum approximation of the area. (Note: This method may be used for any value of n; however, for values of n larger than 10, the process becomes extremely tedious.)
422
The table method for finding the left- and right-hand sums works well for small values of n. However, for large values of n (say n 5 100,000), the table method is not practical. In Section 6.4, we will demonstrate how summation notation may be used for left- and right-hand sums with large values of n. You may wonder, Is the left-hand or the right-hand sum a better estimate of the area? Actually, it varies from function to function. Often you can get one of the best area estimates by averaging the left- and right-hand sums.
EXAMPLE

Estimate the area between the graph of f ( x ) 5 x3 2 5x2 1 6x 1 1 and the x-axis on the interval 30, 44 with n 5 4 using a left-hand sum, a right-hand sum, and the average of the sums.
y 9 8 7 6 5 4 3 2 1 0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
SOLUTION We are asked to estimate the area of the shaded region in Figure 6.16. We generate the table of values for f ( x ) given in Table 6.8.
TABLE 6.8
x 0 1
x
f ( x ) 5 x3 2 5x2 1 6x 1 1 1 3 1 1 9
2 3 4
FIGURE 6.16
The left-hand sum is given by LHS 5 ( 1 1 3 1 1 1 1 )( 1 ) 5 6 square units The right-hand sum is given by RHS 5 ( 3 1 1 1 1 1 9 )( 1 ) 5 14 square units Averaging the two sums, we get LHS 1 RHS 6 1 14 5 2 2 5 10 square units Our estimate is fairly close to the exact area of 9 square units. (In Section 6.5, 3 we will show you how to calculate the exact area.)
1
423
EXAMPLE
Using Riemann Sums to Estimate Distance Traveled

The velocity of the authors minivan as he drove from a traffic light onto a highway may be modeled by V( t ) 5 2 0.1773t2 1 0.6798t 1 0.5019 miles per minute where t is time in minutes. Estimate the area between the velocity graph and the t-axis on the interval 30.0, 2.54 using five rectangles. Then interpret the real-world meaning of the result.
SOLUTION We are asked to estimate the area of the shaded region in Figure 6.17 using five rectangles.
V (t) 1.00 0.75 0.50 0.25 0 0 0.5 1.0 1.5 2.0 2.5 t
FIGURE 6.17
We construct a table of values for V( t ) starting at t 5 0.0 and ending at t 5 2.5 with t-values spaced D t 5
TABLE 6.9
2.5 2 0 5
5 0.5 minute apart (see Table 6.9).
t 0.0 0.5 1.0 1.5 2.0 2.5
V( t ) 5 2 0.1773t2 1 0.6798t 1 0.5019 0.502 0.797 1.004 1.123 1.152 1.093
Since the height of the rectangles represents velocity in miles per minute and the width of the rectangles represents minutes, the units of the area will be miles ? minute 5 miles minute We first find the left-hand sum. LHS 5 ( 0.502 1 0.797 1 1.004 1 1.123 1 1.152 )( 0.5 ) 5 ( 4.578 )( 0.5 ) < 2.289 miles
424
Then we find the right-hand sum. RHS 5 ( 0.797 1 1.004 1 1.123 1 1.152 1 1.093 )( 0.5 ) 5 ( 5.169 )( 0 . 5 ) < 2.585 miles Now we average the sums. 2.289 1 2.585 LHS 1 RHS 5 2 2 5 2.437 miles < 2 . 45 ( rounded to the nearest twentieth of a mile ) We estimate that in the first 2.5 minutes of the highway trip, the author traveled a distance of 2.45 miles. (We rounded to twentieths of a mile because the raw data were recorded accurate to the nearest twentieth of a mile.) According to the raw data, the actual distance traveled by the author was 2.45 miles. In this case, our rounded estimate was right on!
Although the TI-83 Plus doesnt come preloaded with a Riemann sum program, there are several excellent programs available at www.ticalc.org. One outstanding program is Riemann.8xp, written by Mike Miller of Corban College. We transferred the downloaded program from our PC to our calculator using the TICONNECT software. In the following Technology Tip, we will repeat Example 7 using the Riemann.8xp program. (Users of the TI-83 calculator should download the file Rieman83.83P.)
Finding Riemann Sums with Riemann.8xp 1. Press the PRGM key to bring up the list of programs. Select the program RIEMANN. Press ENTER twice.
2. The introduction and credit screen is displayed. Press ENTER .
(Continued)
425
3. To find a Riemann sum of a function f on an interval 3a, b4 with n rectangles, enter the following: f ( x ) : ( Enter the function f .) Lower Bound: a Upper Bound: b Partitions: n 4. The calculator draws the graph of f and draws vertical lines at x 5 a and x 5 b . Press ENTER .
5. The calculator displays an option screen. To calculate the left-hand sum, select 2:LEFT SUM and press ENTER .
6. The calculator graphs the function, draws the left-hand rectangles, and displays the value of the left-hand sum. Press ENTER to return to the option screen.
7. Select 3:RIGHT SUM. The calculator graphs the function, draws the righthand rectangles, and displays the value of the right-hand sum. Press ENTER to return to the option screen.
8. Select 5:TRAPEZOID SUM to calculate the average of the left- and right-hand sums. Press ENTER to return to the option screen.
(continued)
(Continued)
426
9. Select 1:Settings to change the function, interval, or number of rectangles. Otherwise, select 7:QUIT to quit the program.
EXAMPLE
Using Riemann Sums to Estimate Water Volume

Based on data from November 22 to 24, 2002, the flow rate of the Rogue River near Agness, Oregon, at noon may be modeled by f ( t ) 5 25t 2 2 75t 1 1800 cubic feet per second where t is the number of 24-hour periods since noon on November 22, 2002.
(Source: waterdata.usgs.gov.)
How much water (in gallons) passed by this point near Agness, Oregon, between noon on November 22 and noon on November 24, 2002?
SOLUTION We must initially find the area of the shaded region in Figure 6.18.
f (t) 1800 1600 1400 1200 1000 800 600 400 200 0 0 0.5 1.0 1.5 2.0
FIGURE 6.18
We decide to use four rectangles and generate the appropriate table of values (Table 6.10).
TABLE 6.10
t 0.0 0.5 1.0 1.5 2.0
f ( t ) 5 25t 2 2 75t 1 1800 1,800 1,769 1,750 1,744 1,750
427
This problem is a bit tricky, since the flow rate is given in cubic feet per second and the time intervals are 24-hour periods. Additionally, were asked to give our answer in gallons, not cubic feet. Well need to do some unit conversion before reaching our final answer. Since the flow rate is given in cubic feet per second and t is in terms of 24-hour periods, the units of the area of each rectangle will be cubic feet ? 24-hour periods second Well convert these to the appropriate units after finding the left- and right-hand sums. LHS 5 ( 1800 1 1769 1 1750 1 1744 )( 0.5 ) 5 ( 7063 )( 0.5 ) 5 3531 ( accurate to four significant figures ) RHS 5 ( 1769 1 1750 1 1744 1 1750 )( 0.5 ) 5 ( 7013 )( 0.5 ) 5 3506 ( accurate to four significant figures ) LHS 1 RHS 3531 1 3506 5 2 2 5 3519 cubic feet ? 24-hour periods second 24-hour periods second
5 3519 cubic feet ?
How many seconds are there in a 24-hour period? 24-hour periods 5 24 ? 1 hour 5 24 ? 60 minutes 5 1440 ? 1 minute 5 1440 ? 60 seconds 5 86,400 seconds Thus we have 5 3519 cubic feet ? 86,400 seconds second
5 304,041,600 cubic feet Since 1 gallon equals 0.13368 cubic foot, we have 304,041,600 cubic feet ? 1 gallon < 2,274,398,564 gallons 0 . 13368 cubic feet
Between noon on November 22 and noon on November 24, we estimate that more than 2,274 million gallons of water passed by the measurement station on the Rogue River near Agness, Oregon. This is enough water to fill more than 267 Olympic-size swimming pools! As noted earlier in the section, if f ( x ) is a rate of change function, then the area between the graph of f ( x ) and the x-axis on the interval 3a, b4 represents the total change in F( x ) [an antiderivative of f ( x ) ] from x 5 a to x 5 b . We will discuss this in greater detail in Section 6.5.
428
Calculating Riemann Sums 1. Our objective is to calculate Riemann sums for a function f ( x ) on an interval 3a, b4 using n rectangles. We begin by creating cells for b2a a, b, and n. We calculate D x 5 n . 2. In column A, create a list of consecutive whole numbers from 0 through the largest number of rectangles you plan to use. Label columns for x and y.
3. In cell B5, type =IF(A5<=$D$2,$B$2+A5 *$E$2,0). This command tells Excel that if the number in column A is less than or equal to the number of rectangles we want to use, it is to add the appropriate multiple of D x to the initial value a. Otherwise, it is to set the value of x to 0.
(Continued)
429
4. In cell C5, type =IF(A5<=$D$2,20.1773 *B5^2+0.6798*B5+0.5019,0). This command tells Excel that if the number in column A is less than or equal to the number of rectangles we want to use, it is to evaluate f ( x ) 5 2 0.1773x2 1 0.6798x 1 0.5019 at the indicated x value. Otherwise, it is to set the value of y to 0.
5. Copy and paste the formulas in B5 and C5 to the cells below them. Notice that cell values after the indicated number of rectangles (n 5 5) are equal to 0.
6. We need Excel to recognize which cell contains the last y value to be used in any of our sums. In cell D5, type =IF(A5=$D$2,C5,0). This tells Excel that if the index number in column A matches the number of rectangles being used, the corresponding y value is the last y value to be used in a sum. Copy the formula to cells below D5.
(Continued)
430
7. The left-hand sum is the product of D x and the sum of the first y value through the second-to-last y value. In cell E5, type =(SUM(C5:C15)-SUM(D5:D15))*E2. This sums all of the y values and then subtracts the last y value. The result is multiplied by D x. The left-hand sum is 2.021.
8. The right-hand sum is the product of D x and the sum of the second y value through the last y value. In cell E8, type =SUM(C6:C15)*$E$2. This sums the second through the last y values and multiplies the result by D x. The right-hand sum is 2.306.
9. To average the two sums, type average(E5,E8) in cell E11. The average is 2.021.
431
6.3 Summary
In this section, you learned how to use left- and right-hand sums to approximate the area between the graph of a function and the horizontal axis. A solid understanding of these concepts will help you grasp the material presented in the next section.

1. What is the equation for the area of a rectangle? 2. Each rectangle in a group of four rectangles has a width of 2 centimeters. The heights of the four rectangles are 1 centimeter, 4 centimeters, 9 centimeters, and 16 centimeters, respectively. What is the combined area of the group of rectangles? 3. Each rectangle in a group of four rectangles has a width of 0.5 centimeter. The heights of the four rectangles are 3 centimeters, 4 centimeters, 5 centimeters, and 6 centimeters, respectively. What is the combined area of the group of rectangles? 4. The interval 33,54 is to be divided into four equally sized subintervals. What is the width of each subinterval? 5. The interval 31,44 is to be divided into eight equally sized subintervals. What is the width of each subinterval? 6. What is the average of the numbers 210 and 140? 7. Convert cubic feet per second into cubic yards per minute. 8. Convert miles per hour into feet per second. (Hint: There are 5280 feet in a mile.)
6.3 Exercises
In Exercises 15, draw the rectangles used to calculate the left-hand sum estimate of the area between the graph of the function and the horizontal axis on the specified interval. (In each case, use four rectangles.) Then calculate the left-hand sum. 1. f ( x ) 5 2 ; 31, 54 2. g( x ) 5 2 2x 1 9 ; 30, 44 3. h( x ) 5 2x 1 x; 30, 24 4. f ( t ) 5 3 ; 31, 34 t 9. h( x ) 5 2 1 1; 31, 54 x
10. f ( x ) 5 2 2x 1 20 ; 30, 104 11. s( x ) 5 ln( x ) ; 3e, e24 12. y 5 x2 2 2x; 32, 44 14. q 5 20 2 15. f ( t ) 5 13. y 5 4t2 2 1; 32, 44 1000 ; 3100, 2004 p
5. v( t ) 5 0.5t 1 20; 33, 74 In Exercises 615, use the left-hand sum to estimate the area between the graph of the function and the horizontal axis on the specified interval. For each exercise, calculate the sum with n 5 2, n 5 4, and n 5 10 rectangles. 6. s( t ) 5 2 3t2 1 3t ; 30, 14 8. g( x ) 5 2 x2 1 4 ; 30, 24 7. s( t ) 5 t3 2 3t2 1 3t 2 1 ; 31, 24
t12 ; 33, 114 t22
In Exercises 1635, use left- and right-hand sums (with n 5 4) to estimate the area between the graph of the function and the horizontal axis on the specified interval. In each exercise, calculate the left-hand sum, the right-hand sum, and the average of the two sums. The exact area, A, is given so that you can compare your estimates to the actual area. 16. f ( x ) 5 6x 1 1 on 32, 44 ; A 5 38 17. g( x ) 5 x2 2 2x 1 2 on 33, 54 ; A 5 20 3
2
432
18. h( x ) 5 3x2 1 3 on 30, 24 ; A 5 14

3
19. f ( x ) 5 2 ( x 2 2 ) 1 2 on 31, 24 ; A 5 2.25 20. g( x ) 5 x3 2 10x2 1 33x 2 36 on 35, 94 ; A 5 250

2 3
cubic feet per second, where t is the number of hours since 5:45 a.m. (Source: waterdata.usgs.gov.) (a) Convert the units of the flow-rate function from cubic feet per second to cubic feet per hour. (b) Use left- and right-hand sums with n 5 8 to approximate the total amount of water that flowed past the flow-rate measurement station between 5:45 a.m. and 7:45 a.m. (c) Why do you think the flow rate decreased so substantially? 38.
River Flow Based on data from
21. h( x ) 5 x3 2 3x2 1 2x 1 1 on 30, 24 ; A 5 2 22. f ( x ) 5 x4 2 5x2 1 8 on 30, 34 ; A 5 27.6 24. g( x ) 5 x2 2 7x 1 16 on 33, 54 ; A 5 8 3

2
23. f ( x ) 5 x3 2 9x2 1 23x on 32, 64 ; A 5 64 25. h( x ) 5 x3 2 9x2 1 26x on 31, 54 ; A 5 96 26. y 5 ln( 4 )( 4x ) 2 ln( 2 )( 2x ) on 30, 14 ; A52 27. f ( x ) 5 4x 2 x4 on 30, 14 ; A 5 1.8
November 22 to 24, 2002, the flow rate of the Wenatchee River at Peshastin, Washington, at noon may be modeled by f ( t ) 5 2 100t 1 1300 cubic feet per second, where t is the number of 24-hour periods since noon on November 22, 2002. (Source: waterdata.usgs.gov.) How much water (in gallons) passed by Peshastin between noon on November 22 and noon on November 24, 2002? 39. Based on data from November 22 to 24, 2002, the flow rate of the Snake River near Irwin, Idaho, at noon may be modeled by
River Flow
28. g( x ) 5 x4 2 4x3 on 35, 64 ; A 5 259.2 30. f ( t ) 5 e 3t 2 e t on 30, ln( 4 ) 4 ; A 5 18 31. s( t ) 5 ln( t ) on 31, e4 ; A 5 1 32. p( x ) 5 ln( 3 )( 3x ) on 30, 14 ; A 5 2 33. y 5 e x on 3ln( 2 ) , ln( 5 ) 4 ; A 5 3
29. h( x ) 5 e 2x 2 e x on 30, ln( 2 ) 4 ; A 5 0.5
34. y 5 3x2 2 ln( 2 )( 2x ) on 33, 84 ; A 5 237 35. y 5 5x4 2 ln( 5 )( 5x ) on 32, 44 ; A 5 392 In Exercises 3640, estimate the area between the graph of the function and the horizontal axis. Then interpret the physical meaning of the result. 36.
Minivan Position The velocity of the
f ( t ) 5 2 50t 2 1 100t 1 1050 cubic feet per second, where t is the number of 24-hour periods since noon on November 22, 2002. (Source: waterdata.usgs.gov.) How much water (in gallons) passed by this point near Irwin, Idaho, between noon on November 22 and noon on November 24, 2002? (Use n 5 8.) 40. Based on data from November 22 to 24, 2002, the flow rate of the North Snoqualmie River near Snoqualmie Falls, Washington, at noon may be modeled by f ( t ) 5 25t 2 2 145t 1 490 cubic feet per second, where t is the number of 24-hour periods since noon on November 22, 2002. (Source: waterdata.usgs.gov.) How much water (in gallons) passed by this point near Snoqualmie Falls between noon on November 22 and noon on November 24, 2002? (Use n 5 8.)
River Flow
authors minivan as he drove from a traffic light onto a highway may be modeled by V( t ) 5 2 0.1773t2 1 0.6798t 1 0.5019 miles per minute, where t is time in minutes. Use left- and right-hand sums with n 5 8 to estimate how far he traveled in the first 2.0 minutes of his trip. 37. Between 5:45 a.m. and 7:45 a.m. on November 22, 2002, the stream flow rate of the Columbia River below the Priest Rapids Dam dropped dramatically. The flow rate during that time period may be modeled by F( t ) 5 14.50t4 2 80.15t3 1 168.3t2 2 180.8t 1 158.9
River Flow
433
41. Explain why it makes sense to use rectangles to approximate the area between the graph of a function and the horizontal axis. 42. Graphically speaking, what is the difference between a left-hand sum and a right-hand sum? 43. Why does increasing the number of rectangles used in a Riemann sum increase the accuracy of the area estimate? 44. A classmate claims that averaging the left- and right-hand sums always yields a better estimate for the area. Give an example to show that your classmate is incorrect. 45. If a graph gives the velocity of a car (in miles per hour) as a function of the number of hours the car has been moving, what will be the units of the area of the rectangles used to estimate the area between the graph and the horizontal axis? Explain.
47. The left-hand sum and right-hand sum of the function f ( x ) 5 2 x2 1 4x on the interval 30, 44 are both equal to 10 when four rectangles are used. Does this mean that the area between the graph of f ( x ) 5 2 x2 1 4x and the x-axis is equal to 10? Explain. 48. A positive function f is an increasing function. (That is, f ( a ) # f ( b ) whenever a # b.) Which sum will best approximate the area between the graph of f and the x -axis on the interval 3a, b4 : the left-hand sum, the right-hand sum, or the average of the left- and right-hand sums? Explain. 49. Calculate the left-hand sum for the function f ( x ) 5 Z 2 2x 1 2 Z on the interval 30, 24 using n 5 1, n 5 2, n 5 3, n 5 4, n 5 5, and n 5 6. For which value(s) of n does the left-hand sum best approximate the actual area between the graph of f and the x -axis on the interval 3a, b4 ?
46. Give an example of a positive function f on an interval 3a, b4 that has the property that the lefthand sum and right-hand sum approximation of the area between the graph of f and the x-axis are equal for all values of D x.
50. For a nonnegative function f on the interval 3a, b4 , does reducing the width of the rectangles used in a Riemann sum ever worsen the Riemann sum estimate of the area between the graph of f and the x -axis on the interval 3a, b4 ? Explain. (Hint: Consider the function f ( x ) 5 Z 2 2x 1 2 Z using left-hand sums with n 5 2 and n 5 3.)
Apply definite integral properties Calculate the exact area between a curve and the x-axis by using definite integrals
Based on data from 1999 to 2001, the marginal revenue of the Coca-Cola Company may be modeled by M 5 2 0.8058s 1 14.44 dollars per unit case where s is the number of unit cases (in billions). A unit case is equivalent to 24 eightounce servings of finished beverage. In 1999, 16.5 billion unit cases were sold. In 2001, 17.8 billion unit cases were sold. (Source: Modeled from Coca-Cola Company 2001 Annual Report, pp. 46, 57.) By how much did the revenue of the Coca-Cola Company grow as the number of unit cases sold increased from 16.5 billion to 17.8 billion? If you were a financial consultant for Coca-Cola, would you encourage
434
the company to reduce its prices so that it could increase unit case sales? We can address questions such as these using the notion of the definite integral. In this section, we will continue our discussion of area estimates by introducing summation notation and the definite integral. We will also show you several helpful definite integral properties.
Summation Notation
We use summation notation in order to represent the sum of a large number of terms easily. The notation may feel a bit awkward at first; however, as you become skilled in using it, you will come to appreciate its usefulness. Lets return to the table we introduced in Example 5 of Section 6.3 (Table 6.7), reproduced here as Table 6.11.
TABLE 6.11
x 0 0.5 1.0 1.5 2.0
f ( x ) 5 x3 2 3x2 1 3x 0.000 0.875 1.000 1.125 2.000
Recall that for the left-hand sum, the sum of the rectangle heights may be written as f ( 0 ) 1 f ( 0.5 ) 1 f ( 1.0 ) 1 f ( 1.5 ) . Notice that the x values are D x 5 0.5 unit apart. An alternative way to write this sum is a f ( 0 1 0.5i )
3
i50
The Greek letter sigma (S) tells us that we are to sum the values of the form f ( 0 1 0.5i ) . The i is the index of summation. It is a variable, and it is always equal to a whole number. The numbers above and below S are the limits of summation. In this case, the initial value of i is 0 and the ending value is 3. We will first substitute the initial value of i into f ( 0 1 0.5i ) . We will then increase i from 0 to 1 and substitute i 5 1 into f ( 0 1 0.5i ) . f 30 1 0.5( 1 ) 4 5 f ( 0.5 ) f ( 0 ) 1 f ( 0.5 ) The summation symbol S tells us to sum these two values together. So far we have f 30 1 0.5( 0 ) 4 5 f ( 0 )
435
However, the summation tells us that we must continue to increase the index until i 5 3, each time summing the values. f 30 1 0.5( 3 ) 4 5 f ( 1.5 ) f 30 1 0.5( 2 ) 4 5 f ( 1 )
Therefore, a f ( 0 1 0.5i ) 5 f ( 0 ) 1 f ( 0.5 ) 1 f ( 1.0 ) 1 f ( 1.5 ) .

i50
In the next two examples we will temporarily move away from our discussion of heights and areas in order to give you practice using summation notation.
EXAMPLE
Using Summation Notation

Calculate a i 2 .
i51 10
1 through 10 for i.
10 i51
SOLUTION We are to sum terms of the form i 2 by substituting in the values
2 2 2 2 2 2 2 2 2 2 2 a i 5 ( 1 ) 1 ( 2 ) 1 ( 3 ) 1 ( 4 ) 1 ( 5 ) 1 ( 6 ) 1 ( 7 ) 1 ( 8 ) 1 ( 9 ) 1 ( 10 )
5 1 1 4 1 9 1 16 1 25 1 36 1 49 1 64 1 81 1 100 5 385
EXAMPLE

Calculate a ( i 2 3 ) .
i50 6
SOLUTION We must sum terms of the form i 2 3 by substituting in the values 0 through 6 for i.
i50
a (i 2 3) 5 (0 2 3) 1 (1 2 3) 1 (2 2 3) 1 (3 2 3) 1 (4 2 3) 1 (5 2 3) 1 (6 2 3) 5 (23) 1 (22) 1 (21) 1 0 1 1 1 2 1 3 50
We can now describe the left- and right-hand sums in terms of summation notation.
436
SUMMATION NOTATION FOR THE LEFT-HAND SUM
Let f be a function defined on the interval 3a, b4 . If the graph of f is above the x-axis, then the area of the region between the graph of f and the x-axis may be approximated by
n21 i50
c1 f ( xn 2 1 ) D x a f ( xi ) D x 5 f ( x0 ) D x 1 f ( x1 ) D x 1
b2a n
where D x 5
and x0 5 a x1 5 a 1 D x ( xn 2 1 5 a 1 ( n 2 1 ) D x
n21 i50
a f ( xi ) D x is the left-hand sum estimate of the area using n rectangles.
SUMMATION NOTATION FOR THE RIGHT-HAND SUM
Let f be a function defined on the interval 3a, b4 . If the graph of f is above the x-axis, then the area of the region between the graph of f and the x-axis may be approximated by c 1 f ( xn ) D x a f ( xi ) D x 5 f ( x1 ) D x 1 f ( x2 ) D x 1
b2a n
n
i51
where D x 5
and x1 5 a 1 D x x2 5 a 1 2 ? D x ( xn 5 b
i51
a f ( xi ) D x is the right-hand sum estimate of the area using n rectangles.
In general, a Riemann sum does not require that each rectangle have the same width. However, by making the rectangles have the same width, we reduce the number of calculations needed to find the solution. Lets work an area example using summation notation.
EXAMPLE
Using Summation Notation in Calculating a Riemann Sum

Use a left-hand sum with n 5 6 to estimate the area between the graph of f ( x ) 5 x2 1 2 and the x-axis on the interval 31, 34 .
437
SOLUTION We are asked to estimate the area of the shaded region in Figure 6.19 using a left-hand sum with six rectangles.
y 11 10 9 8 7 6 5 4 3 2 1 0 0 0.5 1.0 1.5 2.0 2.5 3.0
FIGURE 6.19
We will first find the width of each rectangle. Dx 5 5 The left-hand sum is given by
n21 i50
321 6 1 unit 3
a f ( xi ) D x 5 a f ( xi ) A 3 B
621 i50 5
5 a 3( xi ) 2 1 24 A 3 B
1
i50 5 i50 5
5 a 3( 1 1 i ? D x ) 2 1 24 A 3 B
1
Since xi 5 a 1 i D x and a 5 1 Since D x 5 3

1 1 1 1
1
5 a 531 1 i A B 4 2 1 26 A B 3 3
1 1
i50
5 531 1 0 A 3 B 4 2 1 26 A 3 B 1 531 1 1 A 3 B 4 2 1 26 A 3 B 1 531 1 2 A 3 B 4 2 1 26 A 3 B

1 1 1 1 1 1 1
5 33 1
5 53( 1 ) 2 1 24 1 3 A 3 B 1 24 1 3 A 3 B 1 24 1 3 A 3 B 1 24 1 3 A 3 B 1 24 1 3 A 3 B 1 246 A 3 B
4
2
1 531 1 3 A 3 B 4 2 1 26 A 3 B 1 531 1 4 A 3 B 4 2 1 26 A 3 B 1 531 1 5 A 3 B 4 2 1 26 A 3 B

1 5
2
< 11.37
43 67 82 1 1 A 9 B 1 (6) 1 A 9 B 1 A 9 B 4 A3 B A 34 9B
Admittedly, formulating the sum using summation notation was more complicated than our table method. However, as we use an increasingly large number of rectangles, using summation notation will be the easiest way to represent the left- and right-hand sums symbolically.
438
The Definite Integral

What would happen if we used infinitely many rectangles to estimate the area between the graph of a function and the horizontal axis? Would the estimate be exact? As we have seen, when we increase the number of rectangles, the width of each rectangle is reduced and our area estimate improves. The definite integral captures the idea of infinitely many rectangles.
THE DEFINITE INTEGRAL
Let f be a continuous function defined on the interval 3a, b4 . The definite integral of f from a to b is given by lim 53 f ( x1 ) 1 f ( x2 ) 1 c 1 f ( xn ) 4 D x6 3 f ( x ) dx 5 n S`
a b
where x1 , x2 , . . . , xn are the right-hand endpoints of subintervals of length Dx 5

b2a . n
The use of infinitely many rectangles may also be represented using summation notation in conjunction with limit notation. We have 3 f ( x ) dx 5 lim a f ( xi ) D x nS` and 3 f ( x ) dx 5 lim a f ( xi ) D x. That is, when we use infinitely many rectangles, nS` the left- and right-hand sums become equal. When we write the symbol lim , we nS` mean that n is made to be infinitely large. For the definite integral 3 f ( x ) dx, a is called the lower limit of integration a and b is called the upper limit of integration. As with indefinite integrals, f ( x ) is called the integrand. If 3 f ( x ) dx is defined, then f is said to be integrable on 3a, b4 .
a b b a
i50
n21
i51
The connection between the indefinite integral 3 f ( x ) dx and the definite integral 3 f ( x ) dx is not immediately obvious. For a nonnegative function f ( x ) ,
a b
3 f ( x ) dx is a family of functions with the same derivative, while 3 f ( x ) dx is the area between the graph of f and the x-axis. However, despite the dramatically different meanings of indefinite and definite integrals, the two are closely related, as will be discussed in Section 6.5. In our interpretation of the definite integral, we required f to be a nonnegative function. In general, the expression 3 f ( x ) dx is the sum of the shaded regions above the x-axis minus the sum of the shaded regions below the x-axis. In other words, 3 f ( x ) dx is the sum of the signed areas.
a b a b a
Consider the graph of f ( x ) 5 x3 2 4x2 1 3x on the interval 30, 34 (Figure 6.20).
6.4 The Definite Integral y 1.0 0.5 0 0.5 1.0 1.5 2.0 2.5 R2 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 x R1
439
FIGURE 6.20
The first region, R1, has area 12 square unit. The second region, R2, has area 23
3
square units. Therefore, 3 ( x3 2 4x2 1 3x ) dx 5 12 2 23 5 2 24 . The fact that the number is negative indicates that more area lies below the x-axis than above it. In considering the relationship between 3 f ( x ) dx and the left- and right-hand sums, it is helpful to think of f ( x ) as the height of an infinitely narrow rectangle and dx as the width of that rectangle. The integral sign 3 is a somewhat distorted S , meaning sum, and tells us to sum the areas of infinitely many adjacent rectangles of width dx between x 5 a and x 5 b . (From our earlier discussion of differentials, we know that we can choose the value of dx to be as small as we like.) The height of each rectangle varies as x moves from a to b in steps of length dx.
b
THE MEANING OF 3 f(x) dx FOR A NONNEGATIVE FUNCTION f

a
Let f be a nonnegative function on 3a, b4 . The definite integral 3 f ( x ) dx is the area of the region between the graph of f and the x-axis.
a b
EXAMPLE
Interpreting the Graphical Meaning of a Definite Integral

3
Interpret the graphical meaning of 3 ( 2 x2 1 3x ) dx 5 4.5 .

0 3
SOLUTION We know that 3 ( 2 x2 1 3x ) dx represents the sum of the signed
areas between the graph of y 5 2 x 2 1 3x and the x-axis. However, since y $ 0
440
for all values of x in the interval 30, 34 , 3 ( 2 x2 1 3x ) dx gives the area of the
0
shaded region in Figure 6.21.

y 2.0 1.5 1.0 0.5 0 0 1 2 3 x
FIGURE 6.21
The area of the shaded region is 4.5 square units.
EXAMPLE
Interpreting the Meaning of a Definite Integral in an Applied Setting

Based on data from 1985 to 1999, the rate of change in the per capita consumption of breakfast cereal may be modeled by R( t ) 5 2 0.014154t2 1 0.233t 2 0.3585 pounds per person per year where t is the number of years since the end of 1980. The graph of R is depicted in Figure 6.22.
y 1.25 1.00 0.75 0.50 0.25 0 0.25 0.50 0.75 1.00 1.25
R2
Pounds per year
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 R1 R3 Years (since 1980)
FIGURE 6.22
The areas of the three shaded regions are 0.296 square unit, 5.214 square units, and 2.034 square units, respectively. Calculate and interpret the meaning of
19
2 3 ( 2 0.014154t 1 0.233t 2 0.3585 ) dt .
0 19
2 3 ( 2 0.014154t 1 0.233t 2 0.3585 ) dt is the sum of the signed 0 areas. Since regions R1 and R3 are below the t-axis, their signed areas are negative.
SOLUTION
441
Therefore,
19
2 3 ( 2 0.014154t 1 0.233t 2 0.3585 ) dt 5 2 0.296 1 5.214 2 2.034 5 2.884
Recall that the units of the integral are found by multiplying the units of the independent variable by the units of the dependent variable. That is, years since 1980 ? pounds 5 pounds year since 1980
So from the end of 1980 through 1999, the annual per capita breakfast cereal consumption increased by a net of 2.884 pounds. That is, taking into account the increases and decreases in cereal consumption rates, annual breakfast cereal consumption increased by 2.884 pounds per person from the end of 1980 through 1999.
Definite Integral Properties

Some of the definite integral properties are directly related to the indefinite integral rules. You will readily note the similarities.
DEFINITE INTEGRAL PROPERTIES
For integrable functions f and g , the following properties hold:
3 f ( x ) dx 5 0
a
3 f ( x ) dx 5 2 3 f ( x ) dx
a b
3 k ? f ( x ) dx 5 k 3 f ( x ) dx for constant k
a a
3 3 f ( x ) 6 g( x ) 4 dx 5 3 f ( x ) dx 6 3 g( x ) dx
a a a
3 f ( x ) dx 5 3 f ( x ) dx 1 3 f ( x ) dx for a # c # b
a a c
EXAMPLE
Applying Integration Properties

0 22 22 0
Given 3 ( 2x 1 3 ) dx 5 2, determine 3 ( 2x 1 3 ) dx.

SOLUTION Switching the limits of integration changes the sign of the definite
22
integral. Therefore, 3 ( 2x 1 3 ) dx 5 2 2.
0
442
EXAMPLE

4 4 4
Given 3 2x dx 5 12 and 3 ( 2x 1 5 ) dx 5 22, find 3 5 dx.

2 2 4 4 4 2
SOLUTION We know that 3 2x dx 1 3 5 dx 5 3 ( 2x 1 5 ) dx. Therefore,

2 4 2 2
12 1 3 5 dx 5 22
2 4
3 5 dx 5 10
2
EXAMPLE

1 21 3 3 21
Given 3 3x2 dx 5 2 and 3 3x2 dx 5 26 , find 3 3x2 dx .

1
SOLUTION The only difference between the three integrals is the limits of
1
integration. Since the upper limit of integration of 3 3x2 dx is the same as the
3 21
lower limit of integration of 3 3x2 dx , the two integrals may be written together
1 3
as a single integral 3 3x2 dx . The value of this integral is found by summing the
21
values of the two integrals from which it was created. That is,
3 21 1 21 3
2 2 2 3 3x dx 5 3 3x dx 1 3 3x dx
5 2 1 26 5 28
EXAMPLE
Interpreting the Meaning of a Definite Integral in an Applied Setting

The rate of change in the annual per capita consumption of chicken from the end of 1985 through 1999 may be modeled by R( t ) 5 0.010492t 3 2 0.22635t 2 1 1.3212t 2 0.4688 pounds per person per year where t is the number of years since the end of 1985. (Source: Modeled from Statistical Abstract of the United States, 2001, Table 202, p. 129.) The sum of the signed areas of the shaded regions of the graph in
15
Figure 6.23 is 3 R( t ) dt and equals 19.75.

0
6.4 The Definite Integral y 3.5 3.0 Pounds per year 2.5 2.0 1.5 1.0 0.5 0 0.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Years (since 1985) t
443
FIGURE 6.23
15 10
Given 3 R( t ) dt 5 7.60, calculate and interpret the meaning of 3 R( t ) dt .

10 10 0
SOLUTION
3 R( t ) dt is the net change in the annual per capita consumption

0 15 15
of chicken from the end of 1985 through the end of 1995 in pounds per person. We know that 3 R( t ) dt 5 19.75 and 3 R( t ) dt 5 7.60, so
0 10 15 10 15
3 R( t ) dt 1 3 R( t ) dt 5 3 R( t ) dt
0 10 10 0 15 15
3 R( t ) dt 5 3 R( t ) dt 2 3 R( t ) dt
0 0 10
5 19.75 2 7.60 5 12.15 From the end of 1985 through 1995, the annual per capita consumption of chicken increased by 12.15 pounds. That is, at the end of 1995, on average, a person was eating 12.15 pounds more chicken per year than he or she was at the end of 1985. For poultry producers, this information is great news. For beef producers, it may be a cause for concern. Over the same time period, annual per capita beef consumption dropped by 10.2 pounds. (Source: Statistical Abstract of the United
States, 2001, Table 202, p. 129.)
EXAMPLE 10
Using a Definite Integral to Calculate an Accumulated Change

Based on data from 1999 to 2001, the marginal revenue of the Coca-Cola Company may be modeled by M 5 2 0.8058s 1 14.44 dollars per unit case where s is the number of unit cases (in billions). A unit case is equivalent to 24 eight-ounce servings of finished beverage. In 1999, 16.5 billion unit cases were sold. In 2001, 17.8 billion unit cases were sold. (Source: Modeled from Coca-Cola Company 2001 Annual Report, pp. 46, 57.) By how much did the revenue of CocaCola Company grow as the number of unit cases sold increased from 16.5 billion to 17.8 billion?
444
SOLUTION We begin by drawing a graph of the marginal revenue function on the interval 316.5, 17.84 and shading the region between the graph and the horizontal axis (see Figure 6.24).
M 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 16.5 16.6 16.7 16.8 16.9 17.0 17.1 17.2 17.3 17.4 17.5 17.6 17.7 17.8 Unit cases (in billions)
Marginal revenue (in dollars per unit case)
FIGURE 6.24
17.8
We are asked to find 3

16.5
( 2 0.8058s 1 14.44 ) ds. Since the marginal revenue

17.8
function is positive on the interval 316.5, 17.84 , 3
( 2 0.8058s 1 14.44 ) ds is
16.5
the area of the shaded region. We recognize that the shaded region is a trapezoid. The area of a trapezoid is given by A 5
1 (b 2 1 b1
1 b2 ) h ,
where h is the height of the trapezoid and b1 and b2 are the respective lengths of the bases. Our graph is a trapezoid turned on its side. We can calculate the length of each base by evaluating M at s 5 16.5 and s 5 17.8. The height of the trapezoid is the length of the interval. We have M( 16.5 ) 5 1.144 and M( 17.8 ) 5 0.097 .
b2
The length of the interval is 17.8 2 16.5 5 1.3. Therefore, the area of the trapezoid is 1 A 5 ( b1 1 b2 ) h 2 1 5 ( 1.144 1 0.097 )( 1.3 ) 2 5 0.8067
445
17.8
Thus 3 ( 2 0.8058s 1 14.44 ) ds < 0.8 billion dollars. (The raw data were 16.5 accurate to the nearest tenth of a billion dollars.) Increasing sales from 16.5 billion to 17.8 billion unit cases increased revenue by $0.8 billion.
When working definite integral problems, it is often helpful to use a graphing calculator to verify the accuracy of our results. The following Technology Tip describes how to find a definite integral graphically.
Finding a Definite Integral Graphically 1. Enter the function f ( x ) as Y1 by pressing the Y button and typing the equation.
2. Graph the function over the specified domain. (You may need to press WINDOW and adjust the Xmin and Xmax settings. Press ZOOM then 0 to automatically adjust the y-values so that the entire graph will appear on the screen.) 3. Press
7:
2nd
CALCULATE
then TRACE to bring up the menu. Select item and press

ENTER
f(x)dx
4. The calculator asks, Lower Limit? Enter the value of the lower limit of integration. Then press ENTER . (Warning: The lower limit must be within the domain of the viewing rectangle.)
(Continued)
446
5. The calculator asks, Upper Limit? Enter the value of the upper limit of integration. Then press ENTER . (Warning: The upper limit must be within the domain of the viewing rectangle.)
6. The calculator shades the region between the graph of the function and the x-axis and then displays the value of the definite integral. In this case, we found
17.8
3
16.5
( 2 0.8058s 1 14.44 ) ds 5 0.8067.
6.4 Summary
In this section, you learned how to use summation notation. You discovered that the definite integral is the sum of the signed areas. You also learned how to manipulate definite integrals using definite integral properties.

1. What is the formula used to calculate the area of a triangle? 2. Calculate the area of the triangle shown in the figure, given that the line that forms the hypotenuse of the triangle has equation y 5 2x 1 4.
y 12 10 8 6 4 2 2 1 2 1 2 3 4 x y 6 5 4 3 2 1 1 1 1 2 3 4 5 6 x
5. Determine the area of the region shown in the graph.

y 4 3
3. What is the formula used to calculate the area of a trapezoid? 4. Calculate the area of the trapezoid shown in the figure, given that the line that forms the top of the trapezoid is y 5 0.5x 1 4.
2 1 2 1 2 4 6 8 x
447
6. How much larger is the portion of the region above the x-axis than the portion of the region below the x-axis?
y 2 1 4 2 1 2 4 6 8 x
8. Is the area of the shaded region above the x-axis greater or less than the combined area of the regions below the x-axis?
y 4 3 2 1 4 2 2 3 4 2 4 6 8 x
7. What is the combined area of the three regions?

y 4 3 2 1 4 2 2 3 4 2 4 6 8 x
6.4 Exercises
In Exercises 110, calculate the sum. 1. a i
i51 5 6
2. a ( i 2 2 i )
i53 4
10. a f ( 2 1 i ? D x )( D x ) given that

i50
f ( x ) 5 3x 2 1 and D x 5 0.25 In Exercises 1120, draw the graph of the integrand function and shade the region corresponding to the definite integral. Then calculate the definite integral using geometric formulas for area.
7 5
3. a ( 2 1 ) ii
i50 3
4. a 3( 3i 2 1 ) ? 44
i50 4
5. a ( 4i 2 2 2i 2 3 )
i52 4
6. a 34( i 2 2 2i ) 4
i51
7. a f ( 1 1 i ? D x )( D x ) given that
i51
11. 3 1 dx
5 4
12. 3 3 dx
2 3
f ( x ) 5 x2 and D x 5 0.25 8. a f ( 1 1 i ? D x )( D x ) given that

i50 3
13. 3 x dx
0 2
14. 3 ( 2 x 1 3 ) dx
0 3
f ( x ) 5 x2 and D x 5 0.25 9. a f ( 2 1 i ? D x )( D x ) given that

i51 4
15. 3 ( 2 2x 1 5 ) dx
1 1
16. 3 ( 4x 2 4 ) dx
1 1
17. 3 ( 2x 1 2 ) dx
21 21
18. 3 ( 3x 1 4 ) dx
21 3
f ( x ) 5 3x 2 1 and D x 5 0.25
19. 3 ( 2 2x 1 1 ) dx
22
20. 3 ( 2 x 1 5 ) dx
22
448
In Exercises 2130, use the definite integral properties to find the numeric value of each definite
4
33. f ( x ) 5 2 x3 1 4x 1 20
y 20
integral given that 3 f ( x ) dx 5 5,

2 6 4 6
3 f ( x ) dx 5 9, 3 g( x ) dx 5 2, and 3 g( x ) dx 5 1.
2 4 2 6 2
10 2 1 10 20 1 2 3 4 x
21. 3 32 f ( x ) 4 dx 23. 3 3 f ( x ) 1 g( x ) 4 dx
2 6 2 6
22. 3 33 f ( x ) 4 dx 24. 3 3 f ( x ) 2 g( x ) 4 dx
2 6 2 6
34. f ( x ) 5
25. 3 f ( x ) dx 27. 3 3 f ( x ) 1 g( x ) 4 dx 29. 3 33 f ( x ) 2 4g( x ) 4 dx 30. 3 3 f ( x ) 1 g( x ) 4 dx

6 6 2 4 2 4 6
26. 3 g( x ) dx 28. 3 32 f ( x ) 2 g( x ) 4 dx
4 4 6
x11 x2 1 1
y 1.25 1.00 0.75 0.50 0.25 20.25 1 2 3 4 x
In Exercises 3135, write the definite integral that represents the sum of the signed areas of the shaded regions of the graph. 31. f ( x ) 5 2 x2 1 5x 2 6
y 1 0 1 2 3 4 5 6
35. f ( x ) 5 2 Z x 2 3 Z 1 3
y 2 1 2 1 1 x
0.5 1.0 1.5 2.0 2.5 3.0
In Exercises 3640, shade the regions of the graph that will be used in calculating the definite integral.
4
2
32. f ( x ) 5 x 2 4x
3
36. 3 ( x 2 3 ) dx
2
y 20 15 10 5 1 5 1 2 3 4 5 x
y 2 1 0 1 2 3 1 2 3 4 5 x
449
37. 3 ( 2 x2 1 4x 2 3 ) dx
1
In Exercises 4145, use the properties of definite integrals to find the solution. 41.
Bottled Water Consumption Based on data from 1980 to 1999, the rate of change in the annual per capita consumption of bottled water may be modeled by R( t ) 5 0.2612( 1.106 ) t
y 1 0 1 2 3 1 2 3 x
gallons per person per year, where t is the number of years since the end of 1980. (Source: Modeled
from Statistical Abstract of the United States, 2001, Table 204, p. 130.) The graph of R( t ) is shown.
y Gallons per person per year
x2 2 1 dx 38. 3 2 0 x 1 1
2
y 0.5 0 0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 x
1.5
1.0
0.5
0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Years (since 1980)
39. 3 e2x 1 1 dx
2
21
The area of the shaded region of the graph is

y 2
t 3 30.2612( 1.106 ) 4 dt and is equal to 15.0.
19
Additionally, 3 30.2612( 1.106 ) t 4 dt 5 4.5.

0
t 3 30.2612( 1.106 ) 4 dt.
10
19
Calculate and interpret the meaning of
2 1
10
42.
1 b dx 40. 3 ln a 2 x 1 1 0
y 0.5 0.5 0.5 1.0 1.5 0.5 1.0 1.5 x
Margarine Consumption Based on data from 1980 to 1999, the rate of change in the annual per capita consumption of margarine may be modeled by R( t ) 5 0.001147t 3 2 0.03445t 2 1 0.2658t 2 0.5098
pounds per person per year, where t is the number of years since the end of 1980. (Source: Modeled
450
y Pounds per person per year 0.1 0 0.1 0.2 0.3 0.4 0.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
0.0009778t 3 2 0.02958t 2 a b dt 5 1.14. 3 1 0.2392t 2 0.06311 15 Calculate and interpret the meaning of 15 0.0009778t 3 2 0.02958t 2 a b dt. 3 1 0.2392t 2 0.06311 0
19
44.
Years (since 1980)
The sum of the signed areas of the shaded 19 0.001147t 3 2 0.03445t 2 regions is 3 a b dt and 1 0.2658t 2 0.5098 0 is equal to 2 3.10. Additionally, 8 0.001147t 3 2 0.03445t 2 a 3 1 0.2658t 2 0.5098 b dt 5 2 0.28 0 Calculate and interpret the meaning of 19 0.001147t 3 2 0.03445t 2 3 a 1 0.2658t 2 0.5098 b dt. 8 43.
Mozzarella Cheese Consumption
Based on data from 1990 to 1999, the rate of change in flounder steak and filet production in the continental United States may be modeled by R( t ) 5 0.2842t 2 4.755 million pounds per year, where t is the number of years since the end of 1990. (Source: Modeled from
Flounder Production
Graph R and shade the region whose area is

9 0
given by 3 ( 0.2842t 2 4.755 ) dt. Then calculate

9
3 ( 0.2842t 2 4.755 ) dt and interpret its real-world meaning. (Hint: Use the area of a trapezoid formula to calculate the area between the graph of the function and the horizontal axis.) 45. Based on data from 1969 to 2002, the rate of change in the number of AP Calculus AB exam participants may be modeled by R( t ) 5 315.70t 2 770.64 people per year, where t is the number of years since the end of 1969. (Source: Modeled from The
AP Calculus Participants
0
Based on data from 1980 to 1999, the rate of change in the annual per capita consumption of mozzarella cheese may be modeled by R( t ) 5 0.0009778t 3 2 0.02958t 2 1 0.2392t 2 0.06311 pounds per person per year, where t is the number of years since the end of 1980. (Source: Modeled
y 0.5 Pounds per person per year 0.4 0.3 0.2 0.1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 0.1 Years (since 1980) x
College Board data.)

33
Graph R and shade the region whose area is

23
given by 3 ( 315.70t 2 770.64 ) dt. Then calculate

33
3 ( 315.70t 2 770.64 ) dt and interpret its real-world meaning. (Hint: Use the area of a trapezoid formula to calculate the value of the definite integral.)
23
46. Explain the relationship between a left-hand sum and a definite integral. 47. What is the difference between an indefinite integral and a definite integral?
The sum of the signed areas of the shaded

19 0.0009778t 3 2 0.02958t 2 regions is 3 a b dt 1 0.2392t 2 0.06311 0 and is equal to 6.20. Additionally,
6.5 The Fundamental Theorem of Calculus

b
451
48. A classmate claims that 3 f ( x ) dx calculates the area between the graph of f and the horizontal axis. In order for your classmates statement to be true, what property must f exhibit? 49. What does the phrase sum of the signed areas mean? 50. What is the relationship between the limits of integration of an integral and the interval over which the integrand is being integrated?
a
52. Suppose that f ( x ) is increasing on the interval 3a, b4 . Is the following statement true for all values of n?
n21 i50
a f ( xi ) D x # 3 f ( x ) dx # a f ( xi ) D x
a
i51
Justify your answer. 53. Suppose that f ( x ) is decreasing on the interval 3a, b4 . Is the following statement true for all values of n?
n21 i50
a f ( xi ) D x # 3 f ( x ) dx # a f ( xi ) D x
a
i51
Justify your answer. 51. A function f on 3a, b4 has the property that a f ( xi ) D x 5 a f ( xi ) D x . (That is, the left-hand i51 sum is equal to the right-hand sum.) Does this
i50 3 4
54. Find three different functions f that have the

4
property that 3 f ( x ) dx 5 0.
2
mean that 3 f ( x ) dx 5 a f ( xi ) D x ? Explain.

a
i50
55. Give an example of an integrable function f that has the property that 3 f ( x ) dx 5 f ( b ) 2 f ( a ) .
a b
Apply the Fundamental Theorem of Calculus Find the function for accumulated change given a rate of change function
Based on data from 1993 to 2002, the rate of change in the sales income of the Starbucks Corporation may be modeled by R( t ) 5 58.46t 1 79.33 million dollars per year where t is the number of years since 1993. According to the model, by how much did the sales income increase between the end of 1996 and the end of 2002? Although we may use the methods previously covered to answer this question, in this section we will demonstrate how the Fundamental Theorem of Calculus can greatly simplify our computations. This theorem is one of the most powerful tools in calculus and gives us a remarkably easy way to calculate the sum of the signed areas between the graph of a function and the x-axis.
y 11 10 9 8 7 6 5 4 3 2 1 0 0 1 2 3
FIGURE 6.25
In Example 3 of Section 6.4, we used a left-hand sum with six rectangles to estimate the area between the graph of f ( x ) 5 x 2 1 2 and the x-axis on the interval 31, 34 (see Figure 6.25). The left-hand sum approximation of the area was approximately 11.37 square units, which we noted was an underestimate of the actual area. The right-hand sum approximation of the area between f ( x ) 5 x 2 1 2 and the x-axis on 31, 34 using six rectangles is 14.04 square units. This is an overestimate of the actual area. Averaging the two sums, we estimate that the shaded region has an approximate area of 12.70 square units.
452
Lets find the antiderivative of the function f .

2 3 f ( x ) dx 5 3( x 1 2 ) dx x3 F( x ) 5 1 2x 1 C 3
Well evaluate the function F at the endpoints of the interval. F( 3 ) 5 (3)3 1 2( 3 ) 1 C 3
5 15 1 C F( 1 ) 5 (1)3 1 2( 1 ) 1 C 3 1 52 1C 3
We next find the difference between the two values. 1 F( 3 ) 2 F( 1 ) 5 15 1 C 2 a 2 1 C b 3 5 12 2 3
< 12.67 Notice that the constant C was eliminated in the computation. This will happen anytime we calculate the difference between two different values of the antiderivative. This value, 12.67, is extremely close to our estimate of 12.70. In fact, 2 the exact area of the region is 12 3 square units. Thus, the definite integral
3
3 f ( x ) dx not only represents the area of the region between f ( x ) and the x-axis 1 on the interval 31, 34 , it also represents the total change in F( x ) from x 5 1 to x 5 3 , where Fr( x ) 5 f ( x ) . This remarkable result is a key component of the Fundamental Theorem of Calculus.
FUNDAMENTAL THEOREM OF CALCULUS
Let f be a continuous function on 3a, b4 . Then

b a
3 f ( x ) dx 5 F( b ) 2 F( a ) where F is any antiderivative of f .
Since F may be any antiderivative of f , we will choose F to be the antiderivative of f that has the constant C 5 0. This choice will make all of our computations easier. Well work several examples to illustrate the power of the Fundamental Theorem of Calculus.
453
EXAMPLE
Using the Fundamental Theorem of Calculus to Calculate a Definite Integral

7
Calculate 3 ( 2x 2 4 ) dx.
3
SOLUTION We must first find an antiderivative of f ( x ) 5 2x 2 4 and then evaluate the antiderivative at x 5 7 and x 5 3.
2 3 ( 2x 2 4 ) dx 5 ( x 2 4x ) ` 3 7
An antiderivative of f ( x ) 5 2x 2 4 is F( x ) 5 x2 2 4x . The vertical bar tells us that we must evaluate this antiderivative at x 5 7 and x 5 3 and then find the difference of the two values. ( x2 2 4x ) ` 5 3( 7 ) 2 2 4( 7 ) 4 2 3( 3 ) 2 2 4( 3 ) 4
7 3
5 ( 49 2 28 ) 2 ( 9 2 12 ) 5 21 2 ( 2 3 ) 5 24
7
Therefore, 3 ( 2x 2 4 ) dx 5 24. Since f ( x ) 5 2x 2 4 is positive on the interval 3 33, 74 , the total area between the graph of f and the x-axis on 33, 74 is 24 square units.
EXAMPLE
Using the Fundamental Theorem of Calculus to Calculate a Definite Integral

1
Calculate 3 ( x3 2 x ) dx .
21
SOLUTION
x4 x2 1 3 ( x 2 x ) dx 5 a 2 b` 3 4 2 21 21
1
5 c
(1)4 (1)2 (21)4 (21)2 2 d 2 c 2 d 4 2 4 2 1 1 1 1 5 a 2 b 2 a 2 b 4 2 4 2 50
This result is due to the fact that the area of the shaded region above the x-axis and the area of the shaded region below the x-axis are equal.
EXAMPLE
Using the Fundamental Theorem of Calculus to Find the Area of a Region

Find the area of the region between the graph of the function f ( x ) 5 x3 2 x and the x-axis over the interval 32 1, 14 .
454
CHAPTER 6 The Integral y 0.4 0.3 0.2 0.1 x
SOLUTION We are asked to find the combined area of the regions enclosed by the graph of f and the x-axis. We begin by graphing f and shading the enclosed regions (see Figure 6.26). The graph of f bordering the first region intersects the x-axis at x 5 2 1 and x 5 0 . The area of the region is given by
0.5 1.0
1.0
0.5
0.1 0.2 0.3 0.4
x4 x2 0 3 3 ( x 2 x ) dx 5 4 2 2 ` 21 21
0
FIGURE 6.26
(0)4 (0)2 (21)4 (21)2 2 d 2 c 2 d 4 2 4 2 1 1 502 a 2 b 4 2 5 1 4

1
5 c
The graph of f bordering the second region intersects the x-axis at x 5 0 and x 5 1 . The region lies below the x-axis, so 3 ( x3 2 x ) dx will be a negative number equal to 21 times the area. x4 x2 1 3 3 ( x 2 x ) dx 5 4 2 2 ` 0 0
1 0
(1)4 (1)2 (0)4 (0)2 2 d 2 c 2 d 4 2 4 2 1 1 5 a 2 b 20 4 2 52 1 4 1 4

1 4
5 c
5 21 ? The area of the second region is also by the graph of f and the x-axis is A5 5
square unit. Therefore, the area bounded
1 1 1 4 4 1 square unit 2
Notice in Example 3 that we used separate definite integrals to calculate the area above the x-axis and the area below the x-axis. Using a single definite integral
1
would have led to an erroneous result. If we calculate 3 ( x3 2 x ) dx, we get the 21 sum of the signed areas, not the actual area. 1 4 1 2 1 3 3 ( x 2 x ) dx 5 4 x 2 2 x ` 21 21 1 1 1 1 5 c (1)4 2 (1)2 d 2 c (21)4 2 (21)2 d 4 2 4 2
1
455
1 1 1 1 5 a 2 b 2 a 2 b 4 2 4 2 50 A common error among beginning calculus students is to assume that the definite integral always yields the area between the graph of the function and the x-axis. As was just illustrated, this is not the case. Although we are often given an interval over which to find the area, we are sometimes required to find the interval ourselves. When we see the phrase Find the area of the region bounded by f ( x ) and the x-axis, we are being asked to calculate the area of the region(s) enclosed by the graph of f and the x-axis. As long as f is a continuous function, we can determine the area by setting up definite integrals with the limits of integration representing each consecutive pair of x-intercepts, as demonstrated in Example 4.
EXAMPLE
Using the Fundamental Theorem of Calculus to Find the Area of a Region

Find the area of the region bounded by f ( x ) 5 4x3 2 2x and the x-axis.
SOLUTION We must first determine where f crosses the x-axis. To do this, we set f ( x ) equal to zero and solve.
0 5 4x3 2 2x 1 0 5 4x a x2 2 b 2 0 5 4x a x 2
y 2.0 1.5 1.0 0.5 1.0 0.5 0.5 1.0 1.5 2.0 0.5 1.0 x
1 1 b ax 1 b 2 2
1 , 2
The function has x-intercepts at x 5 2 On the interval A 2 ` , 2

1 B, 2
x 5 0 , and x 5
function so that we can see the location of the bounded regions (Figure 6.27). a region. The first enclosed region occurs on the interval A 2 enclosed region occurs on the interval A 0, A5 3
1 2
1 . 2
Lets graph the
the graph of f and the x-axis do not enclose , ` B , the B . On the interval A 1 2 ( 4x 2 2x ) dx
3
1 , 2
0 B . The second
graph of f and the x-axis do not enclose a region. The area were looking for is
FIGURE 6.27
0 3 2"1>2
( 4x 2 2x ) dx 1 3
) )
"1>2
Notice that we placed absolute values on each of the definite integrals. This will guarantee that the value returned for each region is that regions area instead of its signed area. A5 3
4 2 "1>2 5 |( x4 2 x2 ) 0 0 0 2"1>2 0 1 0 ( x 2 x ) 0 0
0 3 2"1>2
( 4x 2 2x ) dx 1 3
) )
"1>2
( 4x 2 2x ) dx
3
5 ` b C ( 0 ) 42 ( 0 ) 2 D 2 c a 2
1 4 1 2 1 4 1 2 b 2 a2 b dr ` 1 `bBa b 2 a b R 2 C( 0 ) 4 2 ( 0 ) 2D r ` 2 2 2 2
456
1 1 1 1 5 c (0 2 0) 2 a 2 b d 1 c a 2 b 2 (0 2 0) d 4 2 4 2 1 1 5 ` 0 2 a2 b ` 1 ` a2 2 0b ` 4 4 5 1 square unit 2
) )
The area of the region bounded by f ( x ) 5 4x3 2 2x and the x-axis is square unit. 2 The relationship between a function and its antiderivative becomes especially meaningful if the function represents a rate of change.
ACCUMULATED CHANGE OF A FUNCTION

b
Let f be the rate of change function (derivative) of F on 3a, b4 . Then 3 f ( x ) dx 5 F( b ) 2 F( a ) is the accumulated change in F over the interval 3a, b4 .
a
EXAMPLE
Using the Fundamental Theorem of Calculus to Calculate an Accumulated Change In Revenue

Based on data from 1999 to 2001, the marginal revenue of the Coca-Cola Company may be modeled by m( s ) 5 2 0.8058s 1 14.44 dollars per unit case where s is the number of unit cases (in billions). A unit case is equivalent to 24 eight-ounce servings of finished beverage. In 1999, 16.5 billion unit cases were sold. In 2001, 17.8 billion unit cases were sold. (Source: Modeled from Coca-Cola Company 2001 Annual Report, pp. 46, 57.) By how much did the revenue of CocaCola Company grow as the number of unit cases sold increased from 16.5 billion to 17.8 billion?
SOLUTION Marginal revenue is the rate of change in the revenue function. The accumulated change in revenue is given by
17.8
3
16.5
( 2 0.8058s 1 14.44 ) ds 5 ( 2 0.4029s2 1 14.44s ) `
17.8 16.5
5 ( 2 127.65 1 257.03 ) 2 ( 2 109.69 1 238.26 ) 5 ( 129.38 ) 2 ( 128.57 ) 5 0.81
5 32 0.4029( 17.8 ) 2 1 14.44( 17.8 ) 4 2 32 0.4029( 16.5 ) 2 1 14.44( 16.5 ) 4
Increasing the number of cases sold from 16.5 billion to 17.8 billion increased revenue by 0.8 billion dollars. (The raw data were accurate to the nearest tenth of a billion dollars.)
457
Lets look at the functions m and M graphically. The graph of m is a line, as shown in Figure 6.28. As previously shown in Section 6.4, the area of the shaded region between the graph of m and the horizontal axis is 0.8.
M 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 16.5 16.6 16.7 16.8 16.9 17.0 17.1 17.2 17.3 17.4 17.5 17.6 17.7 17.8 Unit cases (in billions)
Marginal revenue (in dollars per unit case)
FIGURE 6.28
The graph of M is a parabola (Figure 6.29).

y Revenue (in billons of dollars) 129.4 129.2 129.0 128.8 128.6 128.4 128.2 128.0 0 0 16.5 16.6 16.7 16.8 16.9 17.0 17.1 17.2 17.3 17.4 17.5 17.6 17.7 17.8 Unit cases sold (in billions) x M (17.8) M (16.5)
FIGURE 6.29
The vertical distance between the point 317.8, M( 17.8 ) 4 and the point 316.5, M( 16.5 ) 4 is given by M( 17.8 ) 2 M( 16.5 ) . M( 17.8 ) 2 M( 16.5 ) 5 129.38 2 128.57 5 0.81 Amazingly, the area between the graph of m and the horizontal axis on the interval 316.5, 17.84 is easily calculated by finding the difference M( 17.8 ) 2 M( 16.5 ) , where M is an antiderivative of m. A common error of beginning calculus students is to assume that the revenue earned by the company was 0.8 billion dollars. The value 0.8 billion dollars is the accumulated change in revenue, not the revenue value itself.
458
EXAMPLE
Using the Fundamental Theorem of Calculus to Calculate an Accumulated Change in Sales

Based on data from 1993 to 2002, the rate of change in the sales income of the Starbucks Corporation may be modeled by r( t ) 5 58.46t 1 79.33 million dollars per year where t is the number of years since 1993. According to the model, by how much did the sales income increase between the end of 1996 and the end of 2002?
9
SOLUTION We are asked to find 3 ( 58.46t 1 79.33 ) dt .

3
2 3 ( 58.46t 1 79.33 ) dt 5 ( 29.23t 1 79.33t ) ` 9 9 3
5 3081.6 2 501.06
5 329.23( 9 ) 2 1 79.33( 9 ) 4 2 329.23( 3 ) 2 1 79.33( 3 ) 4
< 2580 million dollars Between the end of 1996 and the end of 2002, the sales income increased by 2580 million dollars. (Based on the raw data, the actual change was 2592 million. Our model estimate was off by about $12 million. Although that may seem like a lot, our model estimate was within 0.5 percent of the actual value.)
Changing Limits of Integration

When the integrand function in a definite integral requires integration by substitution, the limits of integration must be changed along with the variable. This technique will be demonstrated in Examples 7 and 8.
EXAMPLE
Changing the Limits of Integration of a Definite Integral

1
Calculate 3 x2( x3 1 1 ) 3 dx .
0
SOLUTION Let u 5 x3 1 1 . Then
d d 3 (u) 5 (x 1 1) dx dx du 5 3x2 dx du 5 3x2dx du 5 x2dx 3
459
Therefore, du 2 3 3 3 3 x ( x 1 1 ) dx 5 3 u a 3 b 0 a
1 b
1 u3du 33 a
What are the values of a and b? Recall that the limits of integration of the original function were x 5 0 and x 5 1 . We will substitute these values of x into the equation relating u and x to find the new limits of integration. Since u 5 x3 1 1 , the new limits of integration are a 5 03 1 1 and b 5 13 1 1. Therefore, 1 2 3 3 3 3 x ( x 1 1 ) dx 5 3 3 u du 0 1 5 5 1 u4 2 ` 3 4 1 24 14 2 12 12
1 2
5 1.25
EXAMPLE
Changing the Limits of Integration of a Definite Integral

3
Calculate 3
1
ln( x ) dx. x
ln( x ) x
SOLUTION At first, the function does not appear to be integrable; however, if
we write f ( x ) 5
u ? ur. Let u 5 ln( x ) . Then
as f ( x ) 5 ln( x ) ? x , we recognize that f is of the form d d ( u ) 5 3ln( x ) 4 dx dx 1 du 5 x dx du 5 1 dx x
that the limits of integration of the original function were x 5 1 and x 5 3 . We will substitute these values of x into the equation relating u and x to find the new limits of integration. Since u 5 ln( x ) , the new limits of integration are a 5 ln( 1 ) and b 5 ln( 3 ) . Since ln( 1 ) 5 0, a 5 0. Therefore,
1 Therefore, 3 c ln( x ) ? d dx 5 3 u du. What are the values of a and b? Recall x 1 a
460
3
1
ln( x ) dx 5 3 x 0 5 5 5
ln (3)
u du
u2 ln (3) ` 2 0
3ln( 3 ) 4 2 (0)2 2 2 2 3ln( 3 ) 4 2 2
< 0.6035
In Section 6.4, we showed how to use the TI-83 Plus to determine the value of a definite integral graphically. In the following Technology Tip, we will demonstrate an alternative way to calculate the definite integral.
Calculating a Definite Integral with fnInt 1. Press MATH to bring up the MATH menu. Select 9:fnInt. Press ENTER .
2. The calculator program fnInt requires four values: the function, the variable of integration, the lower limit, and the upper limit. Enter each of the values and 3 ln( x ) dx, press ENTER . To calculate 3 x 1 we enter fnInt ( ln( x ) > x, x, 1, 3 ) .
3. The calculator displays the value of the definite integral.
461
6.5 Summary
In this section, you learned how to use one of the most powerful tools in calculus: the Fundamental Theorem of Calculus. You used this theorem to calculate a bounded area quickly and to determine the accumulated change of a rate of change function. In future sections, you will continue to use this powerful tool.

1. Given the function F( x ) 5 x 2 1 3x 2 4, calculate F( 2 ) 2 F( 1 ) . 2. Given the function F( x ) 5 x 3, calculate F( 1 ) 2 F( 2 1 ) . 3. Given the function F( x ) 5 4x 2 2 x, calculate F( 0.25 ) 2 F( 0 ) . 4. Given the function F( x ) 5 e , calculate F( 1 ) 2 F( 0 ) .
x
7. Indicate on the graph of the function F( x ) 5 x 3 what F( 1 ) 2 F( 2 1 ) represents.

y 8 6 4 2
5. Given the function F( x ) 5 ln( x ) , calculate F( e ) 2 F( 1 ) . 6. Indicate on the graph of the function F( x ) 5 x 2 1 3x 2 4 what F( 2 ) 2 F( 1 ) represents.
y 12 8 4 1 1 2 x
2 1 2 4 6 8
8. Indicate on the graph of the function F( x ) 5 4x 2 2 x what F( 0.25 ) 2 F( 0 ) represents.

y 0.5 0.4 0.3 0.2 0.1 0.25 0.1 0.25 0.5 x
6.5 Exercises
In Exercises 120, use the Fundamental Theorem of Calculus to calculate the definite integral.
7 5
1 9. 3 a b dx x 1
5
4 10. 3 a b dx x 1
3
1. 3 1 dx
5 4
2. 3 3 dx
2 3
11. 3 ( 2t ) dt
2 4
12. 3 ( 3t ) dt
23 22
3. 3 x dx
0 2
4. 3 ( 2 x 1 3 ) dx
0 3
2
13. 3 ( 5 ? 2 x ) dx 2x 15. 3 a 2 b dx x 11 0
2 22 1
14. 3 ( 7 ? 3 x ) dx ex 16. 3 a x b dx 1 e 1 1
24 4
5. 3 ( 2 3x 1 5 ) dx
1 1
6. 3 ( 4t 2 4t ) dt
2
1 1
7. 3 ( 8x3 1 2x ) dx
21
8. 3 ( 5t 4 2 4t 3 ) dt
21
17. 3 ( x 2 2 1 )( x 3 2 3x ) 3 dx
0
462
1

21
18. 3 ( 3x 1 4 ) dx
21 3
19. 3 ( 2 2x 1 1 ) dx
22
Between the end of 1995 and the end of 1999, by how much did per capita bottled water consumption increase? 33.
Minivan Position The velocity of the
20. 3 ( 2 x 1 5 ) dx
22
In Exercises 2130, calculate the area of the region bounded by the graph of the function and the x-axis. You may find it helpful to graph the function before attempting each exercise. 21. h( x ) 5 x3 1 x2 2 12x 22. g( x ) 5 2 x2 1 4 23. f ( x ) 5 x3 2 x2 24. g( x ) 5 x4 2 4x2 25. f ( x ) 5 x3 2 1 between x 5 2 1 and x 5 1 26. h( x ) 5 x2 2 4 between x 5 2 1 and x 5 1 27. g( x ) 5 1 2 3x between x 5 0 and x 5 2 28. f ( x ) 5 e 2 x between x 5 0 and x 5 2
x 2
authors minivan as he drove from a traffic light onto a highway may be modeled by V( t ) 5 2 0.1773t2 1 0.6798t 1 0.5019 miles per minute, where t is time in minutes. How far did he travel in the first 1.5 minutes of his trip? 34.
McDonalds Sales Based on data from 1990 to 2001, the rate of change of franchised sales of McDonalds Corporation may be modeled by R( t ) 5 2 82.586t 1 1729.0
million dollars per year, where t is the number of years since the end of 1990. (Source: Modeled from
www.mcdonalds.com data.)
29. h( x ) 5 x 2 x2 between x 5 2 1 and x 5 2 30. g( x ) 5 x3 2 1 between x 5 1 and x 5 2 x 35.
Between the end of 1990 and the end of 2001, by how much did McDonalds franchised sales increase?
Mozzarella Cheese Consumption
In Exercises 3135, use the Fundamental Theorem of Calculus to find the answer to the question. 31. Based on data from 1969 to 2002, the rate of change in the number of AP Calculus AB exam participants may be modeled by R( t ) 5 315.70t 2 770.64
AP Calculus Exam
Based on data from 1980 to 1999, the rate of change in the annual per capita consumption of mozzarella cheese may be modeled by R( t ) 5 0.0009778t 3 2 0.02958t 2 1 0.2392t 2 0.06311 pounds per person per year, where t is the number of years since the end of 1980. (Source: Modeled
people per year, where t is the number of years since the end of 1969. (Source: Modeled from The
College Board data.)
Between the end of 1990 and the end of 1999, by how much did per capita mozzarella cheese consumption change?
If the model is an accurate predictor of the future, by how much will the number of people taking the exam each year increase between the end of 2002 and the end of 2012? 32.
Bottled Water Consumption Based on data from 1980 to 1999, the rate of change in the annual per capita consumption of bottled water may be modeled by R( t ) 5 0.2612( 1.106 ) t gallons per person per year, where t is the number of years since the end of 1980. (Source: Modeled
36. What is the relationship between the difference F( b ) 2 F( a ) and the integral 3 f ( x ) dx for a a continuous function f with antiderivative F ? In your explanation, be sure to describe the graphical meaning of each expression. 37. Is it necessary to change the limits of integration when using integration by substitution to determine the value of a definite integral? Explain.
b
Chapter 6
Study Sheet
463
38. According to the Fundamental Theorem of Calculus, x d to what is a 3 f ( t ) dt b equivalent? (Assume a is dx a a constant.) 39. The Fundamental Theorem of Calculus may be rewritten as F( b ) 5 F( a ) 1 3 f ( x ) dx . Given that F( 2 ) 5 1, explain how you could use the graph of f shown below to find F( 4 ) .
y 5 4 3 2 1 1 1 1 2 3 4 5 x
a b
41. A function F( x ) has the properties that Fr( x ) 5 3x 2 1 2x 1 1 and F( 1 ) 5 5 . Find F( 2 ) . 42. A function F( x ) has the properties that Fr( x ) 5 and F( 1 ) 5 1 1 ln( 4 ) . Find F( 4 ) . 43. The velocity of a car is modeled by the function v( t ) 5 66 miles per hour, where t is the number of hours that the car has been traveling. Calculate
1
1 x
3 v( t ) dt and interpret the meaning of the result.

0.5
44. Calculate the area of the region bounded by the graph of g( x ) 5 2x and the graph of h( x ) 5 x2 . (Hint: The graphs of g and h intersect in exactly three places.) 45. Find a function f ( x ) 2 0 such that 3 f ( x ) dx 5 0 on any interval 32 a, a4 . Then discuss the relationship between F( 2 a ) and F( a ) given that Fr( x ) 5 f ( x ) .
2a a
40. The graph of a function f and a graph of its antiderivative F are shown in the following figure. After determining which function is f and which is F, explain how you could use the figure and the Fundamental Theorem of Calculus to find
16 16
3 f ( x ) dx. Then find the value of 3 f ( x ) dx and

0 0
explain what it represents on the graph.

y 12 10 8 6 4 2 0 2 4
12
16

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What is an indefinite integral? (6.1) 2. Why is it important to change the limits of integration when performing integration by substitution? (6.2) 3. Why does increasing the number of rectangles in a Riemann sum tend to increase the accuracy of the area estimate? (6.3) 4. What is the difference between a definite integral and an indefinite integral? (6.4)
464
5. What is the relationship between left- and right-hand Riemann sums and the definite integral? (6.4) 6. The Fundamental Theorem of Calculus states that 3 f ( x ) dx 5 F( b ) 2 F( a ). What is the graphical meaning of 3 f ( x ) dx and of F( b ) 2 F( a ) ? (6.5) 7. What does the phrase sum of the signed areas mean, and how is it related to the definite integral? (6.4) 8. Why does the integral property 3 f ( x ) dx 5 3 f ( x ) dx 1 3 f ( x ) dx make sense for a # c # b? Explain. (6.4) 9. Explain the meaning of the phrase an accumulated change over an interval. (6.5) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
a a b c b c a b a b
Definitions
Integration: The process of finding the general form of a function that has a given derivative. Also referred to as antidifferentiation. Indefinite Integral: The general form of a function that has a given derivative. Also referred to as an antiderivative. Integral Sign: The symbol 1, which indicates that integration is to be performed. Integrand: The function to the right of the integral sign. This is the function that is to be integrated. Differentials dx and dy: dx is an arbitrarily small number. We define dy 5 yr dx. Integration by Substitution: The process of rewriting a function that doesnt appear to be readily integrable in such a way that it can be integrated using basic integration rules. Riemann sum: The sum of areas of rectangles used to approximate the area of the region between the graph of a function and the horizontal axis. Definite integral of f from a to b: lim a f ( xi ) D x 5 lim a f ( xi ) D x 3 f ( x ) dx 5 n S` nS`
a
i50 i51
n21
Finding a Left-Hand Sum of f on an Interval [a, b] Using n Rectangles: b2a Calculate D x 5 . n Create a table of values starting at x 5 a and ending at x 5 b with intermediate x-values spaced D x units apart. Add up the first through the second-to-last value of f ( x ) in the table. Multiply the sum in the preceding step by D x.
Chapter 6
Study Sheet
465
Finding a Right-Hand Sum of f on an Interval [a, b] Using n Rectangles:

Calculate D x 5 n . Create a table of values starting at x 5 a and ending at x 5 b with intermediate x-values spaced D x units apart. Add up the second through the last value of f ( x ) in the table. Multiply the sum in the preceding step by D x.
b
b2a
Meaning of 3 f ( x ) d x for a Nonnegative Function f : The definite a integral is the area of the region between the graph of f and the x-axis on the interval [a, b].
b
Meaning of 3 f ( x ) d x for a Rate-of-Change Function f : The definite integral of f on the interval [a, b] is the accumulated change in F (an antiderivative of f ) over the interval [a, b]. Definite Integral Properties
3 f ( x ) dx 5 0 3 f ( x ) dx 5 2 3 f ( x ) dx 3 k ? f ( x ) dx 5 k 3 f ( x ) dx 3 3 f ( x ) 6 g ( x ) 4 dx 5 3 f ( x ) dx 6 3 g ( x ) dx 3 f ( x ) dx 5 3 f ( x ) dx 1 3 f ( x ) dx for a # c # b
a a b a c b a c a a b a b b a b b b a b a
Formulas
xn 1 1 1 C, n 2 2 1 Power Rule for Integrals: 3x n dx 5 n11 Constant Multiple Rule for Integrals: 3 3a ? g ( x ) 4 dx 5 a 3g ( x ) dx 1 C
Constant Rule for Integrals: 3k dx 5 kx 1 C Sum and Difference Rule for Integrals:
3 3 f ( x ) 6 g ( x ) 4 dx 5 3 f ( x ) dx 6 3g ( x ) dx 1 C
ax 1C Exponential Rule for Integrals: 3a x dx 5 In( a ) 1 1 Integral Rule for : 3 dx 5 In| x| 1 C x x Fundamental Theorem of Calculus: 3 f ( x ) dx 5 F( b ) 2 F( a ) , where a F is any antiderivative of f
b
466

Section 6.1 In Exercises 14, find the general antiderivative of the function.
1. g( x ) 5 2 2x 1 5 2. f ( x ) 5 3x 1 x 2 1 3 3. h( x ) 5 4x 2 x 4. g( x ) 5 x4 2 4x3 In Exercises 58, calculate the indefinite integral. 5. 3( 4x3 2 5x 2 1 ) dx u12 du 6. 3 u 8. 3( x 2 2 2x 1 3 ) dx In Exercises 910, apply the concept of integration in solving the real-world applications. 9.
Publishing Costs Golden Pillar
Section 6.2 In Exercises 1115, integrate the functions. Some integrals will require integration by substitution, while others may be integrated using basic integration rules.
11. f ( x ) 5 x( x2 2 9 ) 5 13. h( x ) 5 15. g( t ) 5 2 ln( x ) x 61t 2t 12. f ( t ) 5 2t " t 2 4 14. f ( x ) 5 2x x 11
2
2t 1 1 dt 7. 3 2t
Section 6.3 In Exercises 1620, use the left-hand sum to estimate the area between the graph of the function and the horizontal axis on the specified interval. For each exercise, calculate the sum with n 5 2, n 5 4, and n 5 10 rectangles.
16. f ( x ) 5 x2 2 1 ; 31, 54 18. h( x ) 5 ln( x ) ; 31, 24 20. g( t ) 5 "t 1 2 ; 30, 14 In Exercises 2125, use left- and right-hand sums (with n 5 4) to estimate the area between the graph of the function and the horizontal axis on the specified interval. In each exercise, calculate the left-hand sum, the right-hand sum, and the average of the two sums. The exact area, A, is given so that you can compare your estimates to the actual area. 21. f ( t ) 5 3t 2 2 2t 1 1 on 30, 14 ; A 5 1 22. g( x ) 5 x 2 "x on 31, 44 ; A 5 2 6
5
17. g( x ) 5 2 4x 1 16 ; 30, 44 19. f ( t ) 5 t 2 1 ; 33, 54 t
Publishing (www.goldenpillarpublishing. com) publishes books for aspiring authors. In 2003, the marginal printing cost for a 8.25 s 3 11 s soft-cover book was c( p ) 5 0.018 dollars per page, where p is the number of pages in the book. Including setup costs, a 100-page book had a $3.10 production cost. Find the book production cost function. 10.
Free Fall The Giant Drop is one of the most popular amusement park rides manufactured by Intamin. Riders are hoisted to a height of 227 feet and then dropped into a free fall. (Source: www.sixflags.com.) The velocity of an object that is dropped into a free fall may be modeled by v( t ) 5 2 32t feet per second, where t is the number of seconds since the object was dropped. A negative velocity indicates that the object is moving toward the ground. Given that position is the antiderivative of velocity, find the position function for the free-fall portion of the Giant Drop thrill ride.
23. h( x ) 5 2x 1
1 on 31, 24 ; A 5 3 1 ln( 2 ) < 3.693 x

2
24. g( x ) 5 "x 1 4 on 30, 54 ; A 5 12 3 25. g( x ) 5 2 3x2 1 3 on 30, 14 ; A 5 2
Chapter 6
Review Exercises
467
In Exercises 2627, estimate the area between the graph of the function and the horizontal axis. Then interpret the physical meaning of the result. 26.
Free Fall The Giant Drop is one of the
30. a 4( 2i 2 9 ) 2
i53
most popular amusement park rides manufactured by Intamin. Riders are hoisted to a height of 227 feet and then dropped into a free fall. (Source: www.sixflags.com.) The velocity of an object that is dropped into a free fall may be modeled by v( t ) 5 2 32t feet per second, where t is the number of seconds since the object was dropped. A negative velocity indicates that the object is moving toward the ground. Estimate the area of the region between the graph of v and the horizontal axis on the interval 30, 24 and interpret the meaning of the result. (Hint: The function v is entirely negative on the interval. Consequently, the left- and right-hand sums will yield a negative value. Since the entire region is below the horizontal axis, you can simply drop the negative sign to get the area estimate.) 27. Between 5:00 p.m. and 9:45 p.m. on November 22, 2002, the stream flow rate of the Columbia River below the Priest Rapids Dam increased dramatically. The flow rate during that time period may be modeled by f ( t ) 5 714.05t3 2 5282.6t2 1 21,738t 1 61,656
River Flow
In Exercises 3135, draw the graph of the integrand function. Then calculate the definite integral using geometric formulas for area.
5 2
31. 3 2x dx
2 6
32. 3 ( 2 3x 1 6 ) dx
0
33. 3 ( 2 Z x 2 3 Z 1 3 ) dx
0 2 6
34. 3 ( 2 2x 1 4 ) dx
1
35. 3 2 dx
4
In Exercises 3640, use the definite integral properties to find the numeric value of each definite
3 4
integral, given that 3 f ( x ) dx 5 2, 3 f ( x ) dx 5 5,

1 3 4 1
3 g( x ) dx 5 1, and 3 g( x ) dx 5 3.
1 3 4
36. 3 32 f ( x ) 4 dx
3
37. 3 33g( x ) 4 dx
1
38. 3 32 f ( x ) 1 4g( x ) 4 dx
3
cubic feet per second, where t is the number of hours since 5:00 p.m. (Source: waterdata.usgs.gov.) (a) Convert the units of the flow-rate function from cubic feet per second to cubic feet per hour. (b) Estimate the area of the region between the graph of f and the horizontal axis on the interval 30, 4.754 and interpret its meaning. (c) Depending upon the pool dimensions, an Olympic-size swimming pool can hold between 94,000 and 114,000 cubic feet of water. How many 100,000-cubic-foot swimming pools could be filled with the volume of water that flowed past the checkpoint below the Priest Rapids Dam on the Columbia River between 5:00 p.m. and 9:45 p.m.?
39. 3 3 f ( x ) 2 g( x ) 4 dx
1
40. 3 34 f ( x ) 2 5g( x ) 4 dx
1
In Exercises 4142, shade the regions of the graph that will be used in calculating the definite integral. x 41. 3 a 2 b dx 1 x 1 1
2
y 0.5 0.4 0.3 0.2 0.1
Section 6.4 sum.

6 i51
In Exercises 2830, calculate the 29. a ( i 2 4 ) 2

i53 7
0 0 0.5 1.0 1.5 2.0 2.5
28. a i 2 2 2i 1 1
468
2
x2 1 2 42. 3 a 2 b dx 1 x 1 1
y 2.0 1.5 1.0 0.5 0.5 0 0.5 1.0 1.5 2.0 x
declining milk consumption. (Source: www.gotmilk.com.) Considering the results of part (a) together with the graph of R, do you think that the Got milk? advertising campaign has been successful? 44.
Personal Income Based on data from 1993 to 2000, the rate of change in the annual per capita personal income in New Mexico may be modeled by R( t ) 5 8.4546t 2 1 45.944t 1 723.00
In Exercises 4344, use the properties of definite integrals to find the solution. 43. Based on data from 1985 to 1999, the rate of change in the annual per capita consumption of milk as a beverage may be modeled by R( t ) 5 0.001326t2 2 0.02192t 2 0.1541
Milk Consumption
dollars per person per year, where t is the number of years since the end of 1993. (Source: Modeled from Bureau of Economic Analysis data.) The graph of R( t ) is shown.
y 1600 1400 1200 1000 800 600 400 200 0 0 1 2 3 4 5 6 7 Years (since the end of 1993) Rate of change (in dollars per person per year)
gallons per person per year, where t is the number of years since the end of 1985. (Source: Modeled
Rate of change (Gallons per person per year) 0 0.05 0.10 0.15 0.20 0.25 0.30 y Years (since the end of 1985) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 t
The signed area of the shaded region of the graph is

7
2 3 ( 8.4546t 1 45.944t 1 723.00 ) dt
and is equal to 7153. Additionally,

7
2 3 ( 8.4546t 1 45.944t 1 723.00 ) dt 5 3713
The signed area of the shaded region of the graph is

14
2 3 ( 0.001326t 2 0.02192t 2 0.1541 ) dt
Calculate and interpret the meaning of

4
2 3 ( 8.4546t 1 45.944t 1 723.00 ) dt .
and is equal to 23.093. Additionally,

8
2 3 ( 0.001326t 2 0.02192t 2 0.1541 ) dt
5 2 1.708 (a) Calculate and interpret the meaning of

14
2 3 ( 0.001326t 2 0.02192t 2 0.1541 ) dt
Section 6.5 In Exercises 4550, use the Fundamental Theorem of Calculus to calculate the definite integral.
3 2
45. 3 ( 3x 1 1 ) dx
2 1
46. 3 ( x 2 ) dx
22 1
(b) In 1993, California milk processors launched the Got milk? advertising campaign in an effort to turn around the 20-year trend of
47. 3 ( 2 x3 ) dx 2x dx 49. 3 2 x 11 2
21 3
48. 3 ( 3e x 1 3x 2 ) dx
0 2
50. 3
1
ln( x ) dx x
Chapter 6
Review Exercises
469
In Exercises 5155, calculate the area of the region bounded by the graph of the function and the x-axis. You may find it helpful to graph the function before attempting the exercise. 51. h( x ) 5 x 2 8 between x 5 1 and x 5 5
3
dollars per person per year, where t is the number of years since the end of 1993. (Source: Modeled
from Bureau of Economic Analysis data.)
Between the end of 1995 and the end of 2000, by how much did the annual per capita personal income increase? 57.
Personal Income Based on data from 1993 to 2000, the rate of change in the annual per capita personal income in Florida may be modeled by R( t ) 5 2 18.914t2 1 209.16t 1 645.4
52. g( x ) 5 2.5 2 53. f ( x ) 5
2x between x 5 2 3 and x 5 3 x2 1 1
1 2 x between x 5 0.5 and x 5 4 x
54. f ( t ) 5 2 t 2 t 2 between t 5 0 and t 5 4
55. g( x ) 5 ( x2 2 1 )( x3 2 3x ) 3 between x 5 2 "3 and x 5 "3. In Exercises 5657, use the Fundamental Theorem of Calculus to determine the answer to the question. 56.
Personal Income Based on data from 1993 to 2000, the rate of change in the annual per capita personal income in Louisiana may be modeled by R( t ) 5 2 28.940t 1 883.18
dollars per person per year, where t is the number of years since the end of 1993. (Source: Modeled
from Bureau of Economic Analysis data.)
Between 1995 and 2000, by how much did the annual per capita personal income increase?
P R O J E C T
What to do
1. Collect at least five years worth of data that are reported annually (for example, annual income or annual tuition cost). 2. Find a model for the data. This model represents the rate of change in the cumulative sum of the data (for example, the rate of change in ones total earned income in dollars per year). 3. Find the antiderivative of the model equation. 4. Use the Fundamental Theorem of Calculus to determine the projected accumulated change over a period of at least four years, including at most one future year. 5. Interpret the meaning of the results from Step 4 and explain why the information is useful.
Example
Data Points
Johnson and Johnson Pharmaceutical Operating Profit
Meaning
From the end of 1998 through 2002, the cumulative operating profit earned by Johnson and Johnson is projected to increase by 17,350 million dollars. When making investment choices, long-term investors often look at cumulative profit over time instead of focusing on fluctuations in annual profits.
Years (Since 1997) (t) 0 1 2 3 4 Model
Operating Profit (millions of dollars per year) (P) 2,332 3,114 3,735 4,394 4,928
Fundamental Theorem A( 5 ) 2 A( 1 ) 5 20,070 2 2719.2 < 17,350
P 5 2 32.71t2 1 778.1t 1 2341 Antiderivative A 5 2 10.90t3 1 389.1t2 1 2341t
470
Chapter
Advanced Integration Techniques and Applications

T
he per capita income for a state is the total income for workers in the state divided by the states population. The rate of change in the per capita income helps us identify trends. By comparing a states rate of change in per capita income to the national average rate of change, we get a sense of the growth or decline of a states income level. By using the concept of the area between two curves, we can determine the change in the gap between the national and state average per capita income.
7.1
Integration by Parts
Use the integration by parts method to find the antiderivative of a function
7.2
Area Between Two Curves
Calculate the area of the region between two continuous curves Determine consumer and producer surpluses
7.3
Improper Integrals
Use indefinite integrals to analyze real-world situations Calculate indefinite integrals
471
472
CHAPTER 7 Advanced Integration Techniques and Applications
7.1 Integration by Parts
Use the integration by parts method to find the antiderivative of a function
As we saw in Chapter 6, many derivative rules have a corresponding integral rule. Although there is no product rule for integrals, the integration by parts method is directly related to the product rule for derivatives. In this section, we will demonstrate how to use the method of integration by parts to integrate a function. This technique is among the more challenging methods to use, so we will use it only after basic integration methods and the substitution method have failed.
Recall that the Product Rule for derivatives states that d 3u ( x ) ? v ( x ) 4 5 ur( x ) ? v ( x ) 1 vr( x ) ? u ( x ) dx where u and v are differentiable functions. If we integrate both sides of the equation with respect to x, we get d 3 e dx 3u ( x ) ? v( x ) 4 f dx 5 3 3ur( x ) ? v ( x ) 1 vr( x ) ? u( x ) 4 dx u ( x ) ? v ( x ) 5 3 3ur( x ) ? v ( x ) 4 dx 1 3 3vr( x ) ? u ( x ) 4 dx du dv 5 3c ? v ( x ) d dx 1 3 c ? u ( x ) d dx dx dx
u( x ) ? v( x ) 5 3v( x ) du 1 3u( x ) dv For notational ease, we will write u(x) as u and v(x) as v, keeping in mind that both u and v are functions of x. uv 5 3v du 1 3u dv Well solve this equation for 3u dv . 3u dv 5 uv 2 3v du This equation is the basis of the integration by parts method.
INTEGRATION BY PARTS
Let u and v be differentiable functions of x. Then 3u dv 5 uv 2 3v du
Consider the integral 3( xe x ) dx. We attempted to integrate this function by substitution in Section 6.2, but we were unable to do so. We will now attempt to integrate the function using integration by parts. In order to use this method, we
473
must first identify a function u and a function vr such that uvr 5 xe x. We typically will pick the function with the simplest derivative to be u or the most complex function that may be easily integrated to be vr. In this case, we pick u(x) 5 x and vr( x ) 5 e x Next well differentiate u with respect to x and, using the concept of differentials, solve for du. d d (u) 5 (x) dx dx du 51 dx du 5 1 dx Next well integrate both sides of the equation vr( x ) 5 e x with respect to x. vr( x ) 5 e x
x 3vr( x ) dx 5 3e dx v( x ) 5 e x
We will add in the constant C at the end of the problem
Well now return to the formula 3u dv 5 uv 2 3v du and plug in the corresponding pieces. (Recall that, by definition, dv 5 vr( x ) dx. In this case, dv 5 e x dx.) 3u dv 5 uv 2 3v du
x x x 3( x ) ( e dx ) 5 ( x ) ? ( e ) 2 3( e ) ( dx )
5 xe x 2 3e x dx 5 xe x 2 e x 1 C (In this example, as in previous examples, the constant C represents the sum of all constants generated throughout the integration process.) Lets check our work by differentiating F(x) 5 xe x 2 e x 1 C . d d 3F( x ) 4 5 ( xe x 2 e x 1 C ) dx dx 5 e x 1 xe x 2 e x
f ( x ) 5 31 ? e x 1 ( e x) ? ( x ) 4 2 e x 1 0 5 xe x
The result checks out, so we did the problem correctly. Now that weve demonstrated the basic theory that surrounds the integration by parts method, we will refine the process to increase our efficiency. Well demonstrate the streamlined process in Example 1 and then detail the steps of the process.
EXAMPLE 1
Integrating by Parts
Integrate f (x) 5 3x (2 x ) with respect to x.
474
SOLUTION We are asked to find 33x ( 2x ) dx. Since ( 3x ) is a constant, we dx
will select u 5 3x. Consequently, dv 5 2x dx. Observe that 33x ( 2x ) dx may be rewritten as 3u dv. We are now ready to do integration by parts. u 5 3x du 5 3 dx
x 33x ( 2 ) dx 5 3u dv
dv 5 2x dx v5 2x ln( 2 )
Choose u and dv Differentiate u and integrate dv with respect to x
5 uv 2 3v du 5 ( 3x ) c 5 5 2x 2x d 2 3 ( 3 dx ) ln ( 2 ) ln ( 2 )
Constant Multiple Rule Exponential Rule
3x ( 2x ) 3 2 2x dx ln ( 2 ) ln ( 2 ) 3 3x ( 2x ) 3 2x 2 c d 1C ln ( 2 ) ln ( 2 ) ln ( 2 ) 2x 3 d c 3x 2 d 1C ln ( 2 ) ln ( 2 )
c 2x d c 3x ln ( 2 )
5 c
Factor out 1 C.
2x ln( 2 )
Therefore, 33x ( 2x ) dx 5
3 d ln ( 2 )
Detailed Steps for the Integration by Parts Method Let u and v be differentiable functions of x. To integrate 3u dv, do the following:
1. Identify u and dv. The product of these two factors must equal the entire
integrand coupled with the associated dx. 2. Differentiate u and integrate dv with respect to x.
3. Write the expression uv 2 3v du in terms of x. 4. Integrate 3v du to eliminate the integral sign. 5. Simplify.
Well demonstrate these steps in the next several examples.

EXAMPLE 2
Integrate f ( x ) 5 x ln( x ) .
475
SOLUTION We must find 3x ln( x ) dx. Well pick u 5 ln( x ) and dv 5 x dx.
u 5 ln( x ) du 5 1 dx x
dv 5 x dx v5 x2 2
Choose u and dv Differentiate u and integrate dv with respect to x
x2 x2 1 uv 2 3v du 5 ln( x ) # 2 3 a # b dx 2 2 x 5 5 1 2 1 x ln( x ) 2 3x dx 2 2 1 2 1 1 x ln( x ) 2 a x2 b 1 C 2 2 2

Constant Multiple Rule Power Rule Factor out 2 x2
1
1 1 5 x2 c ln( x ) 2 d 1 C 2 2 Therefore, 3x ln( x ) dx 5 2 x2 C ln( x ) 2 2 D 1 C .

1 1
EXAMPLE
Simplify 3( 2x 1 1 )( x 2 3 ) 3 dx.
SOLUTION Well pick u 5 2x 1 1 and dv 5 (x 2 3)3 dx.
u 5 2x 1 1 du 5 2 dx uv 2 3v du 5 ( 2x 1 1 ) c 5
dv 5 (x 2 3)3 dx v5 (x 2 3)4 4
By the substitution method
(x 2 3)4 (x 2 3)4 d 2 3c d ( 2 dx ) 4 4
Constant Multiple Rule Simplify fractions Power Rule Simplify fractions Rewrite fractions Factor out
1 (x 20
2 ( 2x 1 1 )( x 2 3 ) 4 2 3( x 2 3 ) 4 dx 4 4
1 1 5 ( 2x 1 1 )( x 2 3 ) 4 2 3( x 2 3 ) 4 dx 4 2 1 1 (x 2 3)5 d 1C 5 ( 2x 1 1 )( x 2 3 ) 4 2 c 4 2 5 1 1 5 ( 2x 1 1 )( x 2 3 ) 4 2 ( x 2 3 ) 5 1 C 4 10 5 5 5 2 ( 2x 1 1 )( x 2 3 ) 4 2 ( x 2 3 ) 5 1 C 20 20 1 ( x 2 3 ) 4 35( 2x 1 1 ) 2 2 ( x 2 3 ) 4 1 C 20
2 3)4
476
5 5
1 ( x 2 3 ) 4( 10x 1 5 2 2x 1 6 ) 1 C 20 1 ( x 2 3 ) 4( 8x 1 11 ) 1 C 20
1
Simplify Group like terms
Therefore, 3( 2x 1 1 )( x 2 3 ) 3 dx 5 20 ( x 2 3 ) 4( 8x 1 11 ) 1 C.
EXAMPLE
Simplify 3( x2 )( x 1 1 ) 3 dx .
SOLUTION Let u 5 x2 and dv 5 (x 1 1)3 dx.
u 5 x2 du 5 2x dx uv 2 3v du 5 ( x2 ) c
dv 5 (x 1 1)3 dx v5 (x 1 1)4 4
By the substitution method
(x 1 1)4 (x 1 1)4 d 2 3c d ( 2x dx ) 4 4
Constant Multiple Rule
1 1 5 x2 ( x 1 1 ) 4 2 3x ( x 1 1 ) 4 dx 4 2
Observe that 3x ( x 1 1 ) 4 dx is not readily integrable. Well integrate this piece separately using integration by substitution and then substitute the result back into the equation. Let w 5 x 1 1. Then dw 5 dx and x 5 w 2 1 . Thus
4 4 3x( x 1 1 ) dx 5 3( w 2 1 )( w ) dw
5 3( w5 2 w4 ) dw w5 w6 2 1C 5 6 5 5 5 5 5 5 5w6 6w5 2 1C 30 30 1 5 w ( 5w 2 6 ) 1 C 30 1 ( x 1 1 ) 5 35( x 1 1 ) 2 64 1 C 30 1 ( x 1 1 ) 5( 5x 1 5 2 6 ) 1 C 30 1 ( x 1 1 ) 5( 5x 2 1 ) 1 C 30
Power Rule Get a common denominator

1 5 w 30
Factor out
Replace w with x 1 1 Simplify Group like terms
477
Well substitute this result back into the integration by parts equation. 1 2 1 1 1 1 x ( x 1 1 ) 4 2 3x ( x 1 1 ) 4 dx 5 x2 ( x 1 1 ) 4 2 c ( x 1 1 ) 5( 5x 2 1 ) d 1 C 4 2 4 2 30 15 2 1 5 x ( x 1 1 ) 4 2 ( x 1 1 ) 5( 5x 2 1 ) 1 C 60 60 1 5 ( x 1 1 ) 4 315x2 2 ( x 1 1 )( 5x 2 1 ) 4 1 C 60 1 5 ( x 1 1 ) 4 315x2 2 ( 5x2 1 4x 2 1 ) 4 1 C 60 1 5 ( x 1 1 ) 4 ( 10x2 2 4x 1 1 ) 1 C 60 Therefore, 3( x2 )( x 1 1 ) 3 dx 5 60 ( x 1 1 ) 4 ( 10x2 2 4x 1 1 ) 1 C . Phew!
1
Although we used the substitution method to integrate the second integral in Example 4, we could have used integration by parts to integrate the second integral. In Example 5, we will apply the integration by parts method multiple times.
EXAMPLE
Integrate f (x) 5 x2e x.
SOLUTION Were asked to find 3x 2e x dx. Well pick u 5 x2 and dv 5 e x dx.
u 5 x2 du 5 2x dx
dv 5 e x dx v 5 ex
uv 2 3v du 5 ( x 2)( e x) 2 32xe x dx 5 x 2e x 2 2 3xe x dx The remaining integral, 3xe x dx, is not readily integrable. We will integrate this function using integration by parts. We will select new values for u and dv; however, it is important to note that although the variables are the same, the functions are not equal to the u and v identified previously. u5x du 5 dx dv 5 e x dx v 5 ex
x2e x 2 2 3xe x dx 5 x2e x 2 2( xe x 2 3e x dx ) 5 x2e x 2 2 (xe x 2 e x ) 1 C 5 x2e x 2 2xe x 1 2e x 1 C 5 e x(x2 2 2x 1 2) 1 C
478
Integration by parts can also be used to integrate functions that formerly did not appear integrable. In Example 6, we will integrate the natural log function.
EXAMPLE
Integrating the Natural Logarithm Function

Simplify 3ln ( x ) dx.
SOLUTION Let u 5 ln ( x ) and dv 5 dx.
u 5 ln( x ) 1 du 5 dx x
dv 5 dx v5x
1 uv 2 3v du 5 3ln( x ) 4 ( x ) 2 3x a dx b x 5 x ln( x ) 2 31 dx 5 x ln( x ) 2 x 1 C Therefore, 3ln( x ) dx 5 x 3ln ( x ) 2 14 1 C .
5 x 3ln( x ) 2 14 1 C
EXAMPLE
Using Integration to Forecast Motor Vehicle Deaths

Based on data from 1980, 1985, 1990, and 1995, the number of motor vehicle deaths in South Carolina may be modeled by m( t ) 5 843.3 1 65.92 ln( t ) where t is the number of years since 1979. According to the model, how many motor vehicle deaths occurred in South Carolina between 1980 and 1995?
SOLUTION Since m(t) is the annual death rate, we must calculate
3 3843.3 1 65.92 ln( t ) 4 dt to determine the cumulative number of deaths

1
16
between 1980 and 1995. Deaths 5 3 3843.3 1 65.92 ln( t ) 4 dt

16
5 843.3t `
16 1 16 1
1 65.92 3 ln( t ) dt
1
16
16
Power, Constant Multiple Rules Integral of natural log function

1
5 843.3t ` 1 65.92 5t 3ln( t ) 2 146 ` 5 3843.3( 16 ) 2 843.3( 1 ) 4 5 12,649.5 1 1935.5 5 14,585
1 65.92 516 3ln( 16 ) 2 146 2 65.92 51 3ln( 1 ) 2 146
Fundamental Theorem of Calculus
479
According to the model, the cumulative number of motor vehicle deaths in South Carolina between 1980 and 1995 was 14,585 deaths.
7.1 Summary
In this section, you learned how to use the method of integration by parts to integrate a function. You discovered that this technique is among the more challenging integration methods and should be used only after basic integration methods and the substitution method have failed.

In Exercises 18, simplify the algebraic expression. 1. x 2e x 2 2( xe x 2 e x ) 2. 4x 2e x 2 ( 8xe x 2 8e x ) 3. ( 2x 2 1 ) a 4. 3x a e x 2 e 2x e 2x b 2 2 2 5. 4x a e 2x 1 1 b 2 e 2x 1 1 2 6t 1.5 t 1.5 b 2 4a b 1.5 1.5
6. 4x 2e x 2 ( 8xe x 2 8e x) 7. ln( t ) a 8. a
x2 x3 b 2 3 a ex 2 b 2 6
3t 2 3t 2 2 5t b ln( t ) 2 a 2 5t b 2 4
7.1 Exercises
In Exercises 110, integrate the function using integration by parts. 1. f ( x ) 5 xe
2x
15. h(x) 5 2x (x2 2 9) 16. f (t) 5 (t 2 1)5(2t 2 7) 17. g(t) 5 (3t2) (t 3 2 1) 19. g(x) 5 3x(x2 2 3) 21. h(x) 5 3xe3x 23. f (t) 5 e2t2 25. f ( x ) 5 2
ln(x) 2
2. g( x ) 5 5x ln( x ) 4. f ( t ) 5 6. h( t ) 5 6 1 ln( t ) t ln( t ) 23 t
18. h(x) 5 x6 (3x 2 4) 20. g( x ) 5 3ln( x ) 4 2 x
3. h( t ) 5 ( 3t 2 5 ) ln( t ) 5. h( t ) 5 t3 ln( t ) 7. f (x) 5 4xe2x+1 9. h(x) 5 e2x(2x 2 1)
22. g( t ) 5 t 2 t ln( t ) 24. h( t ) 5 ( 2t 2 3 ) ln( t )
8. f ( x ) 5 ln( x2 ) 10. f (t) 5 (3t 1 4)e2t
In Exercises 1125, integrate the function using the simplest possible method. 11. g(t) 5 3t 3 21
t
In Exercises 2630, use the integration by parts method to determine the answer to the question. 26. Textbook Sales A new college textbook edition typically generates most of its sales in the year of its publication. Sales drop off in subsequent years as a result of competition from the used book
12. h(t) 5 (2t 2 2 9) (e t )
13. f (x) 5 3x (e x 2 x)
14. g( x ) 5 ( 4x 2 1 ) 3ln( x ) 4
480
market. Suppose that the annual sales of a particular textbook may be modeled by S( t ) 5 30,000 te21.5t textbooks, where t is the number of years since the edition was published. According to the model, how many textbooks will be sold in the first three years of the edition? 27.
Real Networks Licensing Revenue
million dollars, where t is the number of years since the end of 1998. (Source: Modeled from Real
Networks, Inc., 2003 Report, p. 14.)
According to the model, what was the accumulated operating expense between the end of 1999 and the end of 2003?
Based on data from 1999 to 2003, the annual net revenue from software licensing fees for Real Networks, Inc., may be modeled by R( t ) 5 1208 ( t 2 0.88 ) e21.34t 1 56.7 million dollars, where t is the number of years since the end of 1998. (Source: Modeled from Real
31. Which rule for derivatives is most closely related to the integration by parts method for integration? Explain. 32. Picking the correct u and dv is one of the challenges of the integration by parts method. Write how you would explain to a classmate what selection strategies to use. 33. Explain how the equation
1
2 2 3 S 2x ln( x ) 1 x A x B T dx 5 x 3ln( x ) 4 is related to the integration by parts method.
According to the model, what was the accumulated net revenue from software licensing fees between the end of 2000 and the end of 2003? 28.
Real Networks Advertising Revenue
Based on data from 1999 to 2003, the annual net revenue from advertising for Real Networks, Inc., may be modeled by A( t ) 5 1430( t 2 0.95 ) e21.89t 1 6 million dollars, where t is the number of years since the end of 1998. (Source: Modeled from Real
34. To solve the integral 35x( 2x 1 1 ) dx , a classmate picks u 5 5x and dv 5 ( 2x 1 1 ) dx. Do you agree that this is a good choice for u and dv? Explain why or why not. 35. The integral 3 dx may be calculated using x integration by substitution. Can integration by parts be used to calculate 3
ln( x ) x ln( x )
According to the model, what was the accumulated net revenue from advertising between the end of 2000 and the end of 2003? 29.
Electronic Arts Profit Based on data
dx? Explain.
from 2000 to 2004, the annual gross profit for Electronic Arts, Inc., may be modeled by P( t ) 5 274( t 2 0.6 ) e20.232t 1 596 million dollars, where t is the number of years since the end of 1998. (Source: Modeled from
Electronic Arts 2004 Annual Report, p. 19.)
36. Integrate h(x) 5 4x 2e x. 37. Integrate f ( t ) 5 t 2e t . 38. Integrate g( t ) 5 6t0.5 ln( t ) . 39. Explain the relationship between the integration by parts method for integrals and the Product Rule for derivatives. 40. Explain an effective strategy for determining when to use the integration by parts method.
According to the model, what was the accumulated profit between the end of 2000 and the end of 2004? 30.
Real Networks Operating Expenses
Based on data from 1999 to 2003, the annual total operating expenses for Real Networks, Inc., may be modeled by C( t ) 5 5970( t 2 1.04 ) e21.73t 1 150
7.2 Area Between Two Curves
481
Calculate the area of the region between two continuous curves Determine consumer and producer surpluses
Electronic Arts, Inc., is arguably the most dominant force in the electronic gaming software market. Sixteen of the games the company shipped in 2002 sold more than one million copies each. Its 2002 title Harry Potter and the Sorcerers Stone generated more than $206 million in net revenues! The rate of change in the net revenues at Electronic Arts, Inc., may be modeled by r(t) 5 162.840t 3 2 823.299t 2 1 961.788t 1 65.839 million dollars per year and the rate of change in cost of goods sold may be modeled by c(t) 5 66.288t 3 2 328.464t 2 1 355.674t 1 61.435 million dollars per year where t is the number of years since the end of fiscal year 1998. (Source: Modeled from Electronic Arts, Inc., Annual Report; March 31, 2002.) By how much did the companys gross profits increase between the end of fiscal year 1998 and the end of fiscal year 2002? This problem may be interpreted graphically by calculating the area of the region between the graph of r and the graph of c. In this section, we will discuss how to calculate the area of the bounded region between two graphs. We will also explain the meaning of the definite integral of the difference between two rate of change functions. We will investigate how these concepts relate to consumer and producer surplus.
AREA BETWEEN TWO CURVES
If the graph of f lies above the graph of g on an interval [a, b], then the area of the region between the two graphs from x 5 a to x 5 b is given by 3 3 f ( x ) 2 g( x ) 4 dx
a b
y 7 6 5 4 3 2 1 0 0 1 2
y = f (x)
This result is relatively easy to verify graphically. Consider the graphs of the functions f and g shown in Figure 7.1. We see that on the interval [0, 4], f ( x ) . g( x ) . That is, the graph of f lies above the graph of g. Consider A 5 3 3 f ( x ) 2 g( x ) 4 dx. By the Sum and Difference Rule for in3
y = g (x)
tegrals, we know that this equation is equivalent to

3 3
A 5 3 f ( x ) dx 2 3 g( x ) dx
3 4 x
FIGURE 7.1
5 ( area between f and x-axis on 31, 34 )
2 ( area between g and x-axis on 31, 34 )
482
The shaded region in Figure 7.2 is the area between the graph of f and the x-axis on the interval [1, 3]. The shaded region in Figure 7.3 is the area between the graph of g and the x-axis on the interval [1, 3]. Subtracting the second area from the first area results in the area between the graphs of f and g on the interval [1, 3] (Figure 7.4).
y 7 6 5 4 3 2 1 0 0 1 2 3 4 x y = g (x) y = f (x)
y 7 6 5 4 3 2 1 0 0 1 2 3 4 x y = g (x) y = f (x)
y 7 6 5 4 3 2 1 0 0 1 2
y = f (x)
y = g (x)
FIGURE 7.2
FIGURE 7.3
FIGURE 7.4
Well do a few examples before returning to the Electronic Arts profit question.
EXAMPLE
Finding the Area of an Enclosed Region

Calculate the area of the region between the graphs of f (x) 5 x2 2 2x and g(x) 5 x 1 2 on the interval [0, 3].
y 7 6 5 4 3 2 1 0.5 1
y = f (x) y = g (x)
SOLUTION We begin by graphing the functions together and shading the appropriate region (see Figure 7.5). We see that the graph of g (x) 5 x 1 2 lies above the graph of f (x) 5 x2 2 2 x on the interval [0, 3]. We will calculate the area of the region bounded by the two graphs.
2 3 3g( x ) 2 f ( x ) 4 dx 5 3 3( x 1 2 ) 2 ( x 2 2x ) 4 dx
0 3
0.5 1.0 1.5 2.0 2.5 3.0 3.5
5 3 ( 2 x2 1 3x 1 2 ) dx
0
3 1 3 5 a 2 x3 1 x2 1 2x b ` 3 2 0
FIGURE 7.5
3 1 5 c 2 ( 3 ) 3 1 ( 3 ) 2 1 2( 3 ) d 2 ( 0 ) 3 2 5 29 1 5 21 2 27 16 2
5 10.5 The area of the region between the two graphs is 10.5 square units.
483
Just In Time
A Common Error in Calculating the Difference of Two Functions
When calculating the difference of two functions, a common error is forgetting to distribute the negative sign. For example, consider the functions f ( x ) 5 3x2 2 4x 1 7 and g( x ) 5 2x3 1 3x2 2 5. We know that f ( x ) 2 g( x ) 5 ( 3x2 2 4x 1 7 ) 2 ( 2x3 1 3x2 2 5 ) 5 3x2 2 4x 1 7 2 2x3 2 3x2 2 ( 2 5 ) 5 2 2x3 2 4x 1 12 However, a common error is to write this as f ( x ) 2 g( x ) 5 3x2 2 4x 1 7 2 2x3 1 3x2 2 5 5 2 2x 3 1 6x 2 2 4x 1 2 In this case, the negative sign was not distributed to the 3x2 and 2 5 terms. Consequently, the simplified function is incorrect.
Distribute negative sign
EXAMPLE

Calculate the area of the bounded region between the graphs of f ( x ) 5 2 x 2 1 2x 1 1 and g (x) 5 x 2 2 2x 1 1.
y 3 2 1 x y = g (x)
SOLUTION We begin by graphing the functions together and shading the bounded region (Figure 7.6). It appears that the graphs intersect at x 5 0 and x 5 2; however, we must confirm this algebraically.
f ( x ) 5 g( x ) 2 x 1 2x 1 1 5 x2 2 2x 1 1
2
0.5 1
0.5 1.0 1.5 2.0 2.5
0 5 2x2 2 4x 2x( x 2 2 ) 5 0 x 5 0, x 5 2 Our algebraic solution confirms our graphical observation. We see that the graph of f lies above the graph of g on the interval [0, 2]. We will calculate the area of the eye-shaped region between the two graphs. A 5 3 3 f ( x ) 2 g( x ) 4 dx 5 3 3( 2 x2 1 2x 1 1 ) 2 ( x2 2 2x 1 1 ) 4 dx
0 2 0 2 2
y = f (x)
FIGURE 7.6
5 3 ( 2 x2 1 2x 1 1 2 x2 1 2x 2 1 ) dx
0
484
5 3 ( 2 2x2 1 4x ) dx
0
2 2 5 a 2 x3 1 2x2 b ` 3 0
2 5 c 2 ( 2 ) 3 1 2( 2 ) 2 d 2 0 3 5 8 3
2
The area of the region between the two graphs is 2 3 square units.
EXAMPLE

Find the area of the bounded region between f (x) 5 x3 2 3x2 1 3x and g( x ) 5 2 x2 1 4x 2 2 .
SOLUTION We begin by graphing the functions together and shading the bounded regions (see Figure 7.7). The graphs intersect in three different places: x 5 2 1, x 5 1, and x 5 2. We can confirm our graphical observation by evaluating both f and g at these points and ensuring that the function values are equal at these x values.
y 2 1 1 y = f (x) 1 2 3 4 5 6 7
y = g (x)
TABLE 7.1
x 1 1 2
f (x) 5 x3 2 3x2 1 3x 7 1 2
g( x ) 5 2 x2 1 4x 2 2 7 1 2
FIGURE 7.7
Our computational results, shown in Table 7.1, confirm our graphical observation. There are two bounded regions. The first bounded region occurs on the interval [2 1, 1]. On this interval, the graph of f lies above the graph of g. The second bounded region occurs on the interval [1, 2]. On this interval, the graph of g lies above the graph of f. The combined area of the regions is given by A 5 3 3 f ( x ) 2 g( x ) 4 dx 1 3 3 g( x ) 2 f ( x ) 4 dx 5 3 3 f ( x ) 2 g( x ) 4 dx 2 3 3 f ( x ) 2 g( x ) 4 dx
21 1 21 1 1 2 1 2
Factor out a 2 1 to make the integrands equal
Since both integrands are equal, we will simplify the expression f (x) 2 g(x) and substitute the result into each integrand.
485
f ( x ) 2 g( x ) 5 x 3 2 3x 2 1 3x 2 ( 2 x 2 1 4x 2 2 ) 5 x 3 2 3x 2 1 3x 1 x 2 2 4x 1 2 5 x 3 2 2x 2 2 x 1 2 A 5 3 3 f ( x ) 2 g( x ) 4 dx 2 3 3 f ( x ) 2 g( x ) 4 dx
21 1 21
1
2 3 2
5 3 ( x3 2 2x2 2 x 1 2 ) dx 2 3 ( x 2 2x 2 x 1 2 ) dx
1
2 1 2 1 1 2 1 A 5 a x4 2 x3 2 x2 1 2x b ` 2 a x4 2 x3 2 x2 1 2x b ` 4 3 2 4 3 2 21 1
1 2 1 1 2 1 5 e c ( 1 ) 4 2 ( 1 ) 3 2 ( 1 ) 2 1 2( 1 ) d 2 c ( 2 1 ) 4 2 ( 2 1 ) 3 2 ( 2 1 ) 2 1 2( 2 1 ) d f 4 3 2 4 3 2 1 2 1 1 2 1 2 e c ( 2 ) 4 2 ( 2 ) 3 2 ( 2 ) 2 1 2( 2 ) d 2 c ( 1 ) 4 2 ( 1 ) 3 2 ( 1 ) 2 1 2( 1 ) d f 4 3 2 4 3 2 5 ca 5 13 19 8 13 b 2a 2 bd 2 ca b 2 a bd 12 12 12 12
37 1 5 3 12 12 The combined area of the bounded regions between the two graphs is 3
1 square units. 12
Difference of Accumulated Changes

In Chapter 6, we saw that the Fundamental Theorem of Calculus had special meaning when working with a rate of change function f. If the units of f were miles per hour and t was in terms of hours, then 3 f ( t ) dt indicated that the total change in distance (in miles) between t 5 a hours and t 5 b hours was F(b) 2 F(a). This notion may be applied to the difference of two rate of change functions as detailed in the following box.
a b
DIFFERENCE OF TWO ACCUMULATED CHANGES
If f and g are continuous rate of change functions defined on the interval [a, b], then the difference of the accumulated change of f from x 5 a to x 5 b and the accumulated change of g from x 5 a to x 5 b is given by 3 3 f ( x ) 2 g( x ) 4 dx
a b
486
EXAMPLE
Calculating an Accumulated Change

Based on data from 1998 to 2002, the rate of change in the annual net revenues of Electronic Arts, Inc., may be modeled by r(t) 5 162.840t 3 2 823.299t 2 1 961.788t 1 65.839 million dollars per year and the rate of change in annual cost of goods sold may be modeled by c(t) 5 66.288t 3 2 328.464t 2 1 355.674t 1 61.435 million dollars per year where t is the number of years since the end of fiscal year 1998. (Source: Modeled from Electronic Arts, Inc., Annual Report, March 31, 2002.) By how much did the companys annual net revenues, cost of goods sold, and gross profits change between the end of fiscal year 1998 and the end of fiscal year 2002?
SOLUTION The total change in the annual net revenues of Electronic Arts, Inc., between the end of 1998 and the end of 2002 may be determined by calculating the sum of the signed areas of the shaded regions between the graph of r and the horizontal axis on the interval [0, 4] (see Figure 7.8).
r (t) Rate of change (in millioms of dollars per year) 1200 1000 800 600 400 200 0 200 1 2 3 4 t
Years (since the end of 1998)
FIGURE 7.8
In other words, the change in the annual net revenues is given by

4 4
3 2 3 r ( t ) dt 5 3 ( 162.840t 2 823.299t 1 961.788t 1 65.839 ) dt
5 ( 40.710t 4 2 274.433t 3 1 480.894t 2 1 65.839t ) `
4 0
5 10,421.76 2 17,563.712 1 7694.304 1 263.356 5 815.708 million dollars
5 340.710( 4 ) 4 2 274.433( 4 ) 3 1 480.894( 4 ) 2 1 65.839( 4 ) 4 2 ( 0 )
Between the end of fiscal year 1998 and the end of fiscal year 2002, annual net revenues increased by about $815,708,000. The total change in the annual cost of goods sold between the end of fiscal year 1998 and the end of fiscal year 2002 may be determined by calculating the sum of the signed areas of the shaded regions between the graph of c and the horizontal axis on the interval [0, 4] (see Figure 7.9).
7.2 Area Between Two Curves c (t) Rate of change (in millioms of dollars per year) 1200 1000 800 600 400 200 0 200 1 2 3 4 t
487
Years (since the end of 1998)
FIGURE 7.9
In other words, the total change in the annual cost of goods sold is given by
4 4
3 2 3 c ( t ) dt 5 3 ( 66.288t 2 328.464t 1 355.674t 1 61.435 ) dt
5 4242.432 2 7007.232 1 2845.392 1 245.740 5 326.332 million dollars
5 316.572( 4 ) 2 109.488( 4 ) 1 177.837( 4 ) 1 61.435( 4 ) 4 2 ( 0 )

4 3 2
5 ( 16.572t 4 2 109.488t 3 1 177.837t 2 1 61.435t ) `
4 0
Between the end of fiscal year 1998 and the end of fiscal year 2002, the annual cost of goods sold increased by about $326,332,000. The change in annual gross profit between the end of fiscal year 1998 and the end of fiscal year 2002 is given by 3 3r ( t ) 2 c( t ) 4 dt
0 4
since the rate of change in profit is the difference of the rate of change in revenue and the rate of change in cost. However, 3 3r( t ) 2 c( t ) 4 dt 5 3 r( t ) dt 2 3 c( t ) dt
0 0 0 4 4 4
Therefore, 3 3r ( t ) 2 c( t ) 4 dt 5 815.708 2 326.332

0 4
5 489.376 million dollars Between the end of fiscal year 1998 and the end of fiscal year 2002, the annual profit increased by about $489,376,000. We have drawn the graphs of r and c together and shaded the bounded regions between the two graphs, as shown in Figure 7.10. These regions represent the accumulated change in annual profit.
488
c (t), r (t) Rate of change (in thousands of dollars per year) 1200 1000 800 600 400 200 0 200 1 2 3
r (t) c (t) Accumulated change in annual profit
Years (since the end of fiscal year 1998)
FIGURE 7.10
Note that although annual profits increased from the start of fiscal year 1998 through the end of fiscal year 2000 and from early in fiscal year 2002 through the end of fiscal year 2002, they decreased between the end of fiscal year 2000 and early in fiscal year 2002.
EXAMPLE
Calculating an Accumulated Change

Based on data from 1959 to 1989, the rate of change in the annual per capita income in Washington state may be modeled by w( t ) 5 2 5.96t 1 327.4 dollars per year and the rate of change in the per capita income in the United States may be modeled by u(t) 5 239.0 dollars per year where t is the number of years since the end of 1959. (Source: Modeled from www.census.gov.) In 1959, the per capita income in Washington state exceeded the national per capita income. From the end of 1959 until the end of 1989, by how much did the pay gap between the Washington state and the national per capita income change?
30 30
SOLUTION We can calculate 3 w ( t ) dt and 3 u ( t ) dt independently and

0 0
then calculate their difference, as was demonstrated in Example 4. However, it is often easier to calculate 3 3w( t ) 2 u( t ) 4 dt.
0 30
w( t ) 2 u( t ) 5 ( 2 5.96t 1 327.4 ) 2 ( 239.0 ) 5 2 5.96t 1 88.4 3 3w( t ) 2 u( t ) 4 dt 5 3 ( 2 5.96t 1 88.4 ) dt

0 0 30 30
5 32 2.98( 30 ) 1 88.4( 30 ) 4 2 32 2.98( 0 ) 2 1 88.4( 0 ) 4

2
5 2 2.98t2 1 88.4t `
30 0
5 2 30
489
Between 1959 and 1989, the pay gap between the annual per capita income in Washington state and the national average annual per capita income changed by $30. Since the Washington state per capita income exceeded the national per capita income in 1959, the pay gap between the two was reduced by $30 between 1959 and 1989. (If the Washington state per capita income had been less than the national per capita income in 1959, the pay gap would have increased by $30 between 1959 and 1989.) Lets look at the problem graphically. We begin by graphing both u and w (see Figure 7.11). The sum of the signed areas between the graph of u and the graph of w represents the change in the pay gap.
y Rate of change (in dollars per person per year) 300 250 200 150 100 50 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 t Years (since the end of 1959) w (t) u (t)
FIGURE 7.11
The sum of the signed areas of the shaded regions shown in the graph is 30. Another way to look at the problem graphically is to graph the function g (t) 5 w (t) 2 u (t), which represents the rate of change in the pay gap between the Washington state and the national per capita income (see Figure 7.12).
Rate of change (in dollars per person per year) y 100 50 0 50 100 Years (since the end of 1959) 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 t g (t) = w(t) u(t)
FIGURE 7.12
When g( t ) . 0 , the Washington state per capita income is growing faster than the national per capita income. When g( t ) , 0 , the Washington state per capita income is growing more slowly than the national per capita income. The sum of the signed areas of the shaded regions shown in the graph is 30. This means that, over the 30-year period from 1959 to 1989, the gap between the Washington state per capita income and the national per capita income was reduced by $30.
490
Consumer and Producer Surplus

In a free market economy, producers (businesses) seek to make as much money as possible from the products they sell while consumers (buyers) seek to spend as little money as possible for the products they want. When the selling price of an item is less than the amount a consumer is willing to spend on the item, the consumer saves money. This may be viewed as a surplus for the individual consumer. When the selling price of an item is more than the minimum amount the producer is willing to accept for the item, the producer makes money. The difference between the selling price and the price the producer would have accepted for the item represents a surplus for the producer. Lets look at a specific example. In 2003, Apple Computer Corporation sold 0.9 million iPod units at an average price of about $333 per unit. (Source: Apple Computer Corporation 2005 Annual Report, pp. 3132.) For illustration purposes, well assume that the minimum amount the company was willing to accept for each iPod was $100. We will further assume that the maximum price a consumer was willing to pay for the iPod was $350, and that the actual sales price was $333. The surplus for the buyer is Surplus 5 350 2 333 5 17 dollars The buyers surplus is 17 dollars. That is, the iPod cost the buyer 17 dollars less than the amount she was willing to pay for the iPod. The surplus for the seller is given by Surplus 5 333 2 100 5 233 dollars The sellers surplus is 233 dollars. This means that the company made 233 dollars more on the item than the minimum amount it was willing to accept. In economics, it is customary to focus on the overall relationship between an items price and the number of the items sold rather than focusing on a specific price. A consumer demand function models the item price p that a consumer is willing to pay for q items. Based on data from 2003 to 2005, a demand function model for the Apple iPod is Demand: p 5 52.1 dollars 0.00484q 1 0.152
where q is the number of iPods sold (in millions). (Source: Modeled from Apple
Computer Corporation 2005 Annual Report, pp. 3132.)
A producer supply function models the item price p that the producer is willing to accept for supplying q items to the marketplace. A hypothetical supply function for the iPod is Supply: p 5 0.1q2 1 100 dollars where q is in millions of iPods.
491
We graph the supply and demand functions simultaneously in Figure 7.13.

p iPod price (in dollars) 350 300 250 200 150 100 0 0 Demand 5 10 15 20 25 30 35 40 q iPods sold (in millions) Supply (28.3, 180)
FIGURE 7.13
As the iPod price decreases, more consumers are willing to buy the music player. As the iPod price increases, the company is more willing to supply the market with the item because of the potential for increased profits. The point of intersection of the supply and demand graphs is referred to as the equilibrium point. In this case, the equilibrium point is (28.3, 180). This means that the model predicts that 28.3 million iPods will be sold when the price is 180 dollars. At the equilibrium price, the company is unwilling to take less and the consumer is unwilling to pay more for the product. Let us now consider the area between the supply and demand curves. See Figure 7.14.
p iPod price (in dollars) 350 300 250 200 150 100 0 0 5 10 15 20 25 30 35 40 q iPods sold (in millions) Consumer surplus Producer surplus
(28.3, 180)
FIGURE 7.14
What is the meaning of the area? The units of input are millions of iPods and the units of output are dollars per iPod. So Units of area 5 ( units of input ) ( units of output ) 5 million iPods ? 5 million dollars Thus the area represents an amount of money in millions of dollars. Assuming that the item is selling at the equilibrium price of 180 dollars, the portion of the shaded region above the horizontal line p 5 180 is the consumer surplus. The consumer surplus is the total gain for all consumers buying the product. The portion of the shaded region below the horizontal line p 5 180 is the producer surplus. The producer surplus is the total gain for producers selling the product. dollars iPod
492
28.3
The definite integral 3

0
52.1 dq 0.00484q 1 0.152
calculates the area below the
demand curve on the interval from q 5 0 to q 5 28.3.

28.3
3
0
52.1 dq < 6920 0.00484q 1 0.152
Accurate to three significant digits
The area represents 6920 million dollars. However, not all of this is surplus. To determine the surplus, we must subtract the area of the rectangular region below the equilibrium price line. See Figure 7.15.
p iPod price (in dollars) 350 300 250 200 150 100 0 0 5 10 15 20 25 30 35 40 q iPods sold (in millions) Consumer surplus
(28.3, 180)
FIGURE 7.15
The area of this region is Revenue 5 ( 28.3 )( 180 ) < 5090 So the consumer surplus is Consumer surplus 5 6920 2 5090 5 1830 The consumer surplus is 1830 million dollars.
CONSUMER SURPLUS
Given the demand function p 5 f ( q ) and equilibrium point ( q0, p0 ) , the consumer surplus is given by CS 5 3 f ( q ) dq 2 p0 q0
0 q0
The process used to find the producer surplus is similar. We have already determined that the total revenue was 5090 million dollars. By subtracting the area under the supply curve, we determine what portion of the revenue is surplus. See Figure 7.16.
493
p iPod price (in dollars) 350 300 250 200 150 100 0 0 5 10 15 20 25 30 35 40 q iPods sold (in millions) Producer surplus (28.3, 180)
FIGURE 7.16
28.3
Producer surplus 5 5090 2 3

0
0.1q2 1 100 dq
< 5090 2 3590 5 1500 The producer surplus is 1500 million dollars.
PRODUCER SURPLUS
Given the supply function p 5 s( q ) and equilibrium point ( q0, p0 ), the producer surplus is given by PS 5 p0 q0 2 3 s( q ) dq
0 q0
EXAMPLE
Determining Consumer and Producer Surplus

The price per ounce for sharp cheddar cheese may be modeled by the demand function p( q ) 5 11.1 1 0.200 dollars q2 1 1
where q is the number of ounces of cheese purchased. (Source: Modeled from data
at www.safeway.com.)
Suppose the supply function for sharp cheddar cheese is s( q ) 5 0.0001q 2 1 0.100 dollars (a) Graph the demand and supply functions and identify the location of the equilibrium point, the consumer surplus, and the producer surplus. (b) Determine the coordinates of the equilibrium point. (c) Calculate the consumer and producer surplus.
494
SOLUTION
(a)
Price (in dollars per ounce)
s 0.9 0.8 0.7 0.6 0.5 Consumer 0.4 surplus 0.3 0.2 Producer surplus Equilibrium 0.1 point 0 0 5 10 15 20 25 30 35 40 q Ounces
FIGURE 7.17
(b) To determine the equilibrium point, we set the supply and demand equations equal to each other and solve. supply 5 demand 0.0001q2 1 0.100 5 0.0001q2 2 0.100 5 11.1 1 0.200 q2 1 1
11.1 q2 1 1 0.0001q 2 ( q 2 1 1 ) 2 0.100 ( q 2 1 1 ) 5 11.1 0.0001q 4 1 0.0001q 2 2 0.100q 2 2 0.100 5 11.1 0.0001q 4 2 0.0999q 2 2 11.2 5 0 This is a quadratic function in q 2, so we may solve the equation by using the Quadratic Formula. q2 5 0.0999 6 "( 2 0.0999 ) 2 2 4( 0.0001 )( 2 11.2 ) 2( 0.0001 ) 2 q < 1100 q < 33.2 s( 33.2 ) 5 0.0001( 1100 ) 1 0.100 5 0.21 The equilibrium point is (33.2, 0.21).
33.2
We substitute this value into s( q ) 5 0.0001q2 1 0.100.
(c)
CS 5 3
0
11.1 1 0.200 b dq 2 ( 0.21 )( 33.2 ) q2 1 1

Using technology
5 23.7 2 ( 6.97 ) < 16.7 The consumer surplus is $16.70.

q0
495
PS 5 p0 q0 2 3 s( q ) dq
0 33.2
5 ( 0.210 )( 33.2 ) 2 3
0
( 0.0001q2 1 0.100 ) dq
5 6.97 2 4.54 5 2.43 The producer surplus is $2.43.
Using technology
7.2 Summary
In this section, you learned how to calculate the area of the bounded region between two graphs. You also learned that the definite integral of the difference between two rate of change functions is equal to the difference between the accumulated changes in each of the individual functions.

1. Given f ( x ) 5 x 2 1 4x and g( x ) 5 7x 2 8, calculate f ( x ) 2 g( x ) . 2. Given f ( x ) 5 3x 4 2 2x 3 1 x 2 1 and g( x ) 5 2 2x 3 1 x 1 7, calculate f ( x ) 2 g( x ) . 3. Given f ( x ) 5 ln( x ) and g( x ) 5 e x, calculate f ( x ) 2 g( x ) . 4. Where do f ( x ) 5 x 2 2 4x 1 4 and g( x ) 5 2x 1 4 intersect? 5. Where do f ( x ) 5 x3 2 x and g( x ) 5 x2 2 x intersect? 6. Where do f ( x ) 5 ( x 2 3 )( x 1 2 )( x 1 4 ) and g( x ) 5 ( x 2 5 )( x 2 3 )( x 1 4 ) intersect? 7. On the interval 32 5, 54 , when is the graph of f ( x ) 5 x 3 2 x above the graph of g( x ) 5 0? 8. On the interval 32 10, 104 , when is the graph of f ( x ) 5 x 2 2 4 above the graph of g( x ) 5 2 x 2 1 4?
7.2 Exercises
In Exercises 120, calculate the combined area of the region(s) bounded by the graphs of the two functions. (On some of the exercises, you may find it helpful to use your graphing calculator to locate the intersection points of the graphs.) 1. f (x) 5 x2; g(x) 5 x 2. f (x) 5 x 2 3x 1 2; g(x) 5 2
2
4. f (x) 5 2x2 2 1; g (x) 5 1 5. f (x) 5 x2 2 4x 2 5; g ( x ) 5 2 3x 2 5 6. f (x) 5 2x2 2 1; g (x) 5 x 3 2 1 7. f ( x ) 5 2 x; g ( x ) 5 2 x2 1 3x 8. f (x) 5 x; g (x) 5 x 3 2 x 9. f (x) 5 x3 2 6x2 1 5x; g (x) 5 6x2 1 50x
3. f (x) 5 x2 2 4; g( x ) 5 2 x2 1 4
496
10. f (x) 5 2x2 1 2x 1 4; g (x) 5 x3 1 2x 1 4 11. f (x) 5 x 2 2x; g (x) 5 x 2 3x

2 3 2
12. f (x) 5 x 2 1; g (x) 5 x2 2 2x 13. f (x) 5 x3 2 x; g ( x ) 5 2 x3 1 x 14. f (x) 5 2x 1 x; g (x) 5 3x 2 x 15. f (x) 5 3x; g (x) 5 x3 16. f (x) 5 2x; g (x) 5 x3 1 1 17. f (x) 5 2x; g (x) 5 x2 18. f ( x ) 5 3 ln( x ) ; g (x) 5 2x 2 4 19. f ( x ) 5 2 1 ln( x ) ; g (x) 5 5x ln( x ) 20. f (x) 5 x2 2 4; g ( x ) 5 x In Exercises 2125, graph the functions together on the same axes. Then determine the solution by using the integral techniques demonstrated in this section. (You may integrate each function separately and then calculate the difference, or, if you prefer, you may calculate the difference between the functions and then integrate the result.) 21.
Company Cost and Profit Based on data from 1999 to 2001, the annual gross revenue of the Johnson & Johnson company and its subsidiaries may be modeled by
2
According to the models, what were the total costs of the company between 1999 and 2001? (Assume that annual net sales is the same as annual revenue.) 23.
Prison vs. College Population Based on data from 1980 to 1998, the rate of change in the number of adults in prison, in jail, on probation, or on parole may be modeled by r(t) 5 238.1
thousand people per year, where t is the number of years since 1980. (Source: Modeled from
Based on data from 1980 to 1998, the rate of change in the number of students enrolled in private colleges may be modeled by s(t) 5 1.839t 1 23.25 thousand students per year, where t is the number of years since 1980. (Source: Modeled from
Determine the accumulated change in the difference of the two populations from the end of 1980 through 1998. Interpret the meaning of the result. 24. Based on data from 1999 to 2001, the rate of change in the annual net sales of Pepsi-Cola North America may be modeled by R( t ) 5 2 131t 2 749.5
Company Operating Costs
R(t) 5 665t 2 1 1150t 1 27,357 million dollars, and the annual cost of goods sold may be modeled by C(t) 5 103t 2 1 315t 1 8539 million dollars, where t is the number of years since the end of 1999. (Source: Modeled from
Johnson & Johnson 2001 Annual Report, p. 6.)
million dollars per year, and the rate of change in the annual operating profit may be modeled by P(t) 5 12t 1 76 million dollars per year, where t is the number of years since the end of 1999. (Source: Modeled from
PepsiCo 2001 Annual Report, pp. 23 and 24.)
According to the models, what was the total profit earned between 1999 and 2001? 22.
Company Sales and Profit Based on
Determine the accumulated change in annual operating costs from the end of 1999 through 2001. 25.
Company Profit Based on data from 1999 to 2001, the rate of change in the annual net revenue of the Coca-Cola Company may be modeled by r( t ) 5 2 402t 1 806 million dollars, and the rate of change in the annual cost of goods sold may be modeled by c( t ) 5 2 355t 1 372.5
data from 1999 to 2001, the annual net sales of Gatorade/Tropicana North America may be modeled by R( t ) 5 2 107t 2 1 496t 1 3452 million dollars, and the annual operating profit may be modeled by P( t ) 5 2 18.5t 2 1 85.5t 1 433 million dollars, where t is the number of years since the end of 1999. (Source: Modeled from
PepsiCo 2001 Annual Report, p. 44.)
497
million dollars, where t is the number of years since the end of 1999. (Source: Modeled from CocaCola Company 2001 Annual Report.)
where q is the number of bottles purchased. supply function for the bottles is
(Source: Modeled from data at www.pens.com.) The
Determine the accumulated change in annual profit from the end of 1999 through 2001. In Exercises 2629, calculate the equilibrium point, the consumer surplus, and the producer surplus. 26.
Potato Supply and Demand The
s( q ) 5 0.000250q 1 0.150 dollars per bottle
demand for prepackaged Russet potatoes may be modeled by p( q ) 5 0.0027q 2 20.076q 1 0.86 dollars per pound where q is the number of pounds purchased.
(Source: Modeled from data at www.safeway.com.) The
30. Two curves intersect at three different points. How many definite integrals should be used to calculate the area of the region bounded by the curves? 31. What is the relationship between a definite integral and the area between two curves? 32. Explain what 3 3 f ( x ) 2 g ( x ) 4 dx 5 3 f ( x ) dx 2 3 g ( x ) dx
a a a b b b
supply function for the potatoes is s( q ) 5 0.001q2 1 0.15 dollars per pound 27.
Magnetic Wall Calendar Supply and Demand For sales of up to ten
means graphically. 33. If both f and g are even functions (symmetric with respect to the y-axis), what is an efficient way to calculate the area between f and g on the interval 32 a, a4 ?
2
thousand calendars, the demand for customized magnetic wall calendars may be modeled by p( q ) 5 0.1536q2 2 3.058q 1 64.24 cents per calendar where q is the number of calendars purchased (in thousands). (Source: Modeled from data at www.pens.com.) The supply function for the calendars is s( q ) 5 1.880q 1 30.22 cents per calendar 28.
Lollipop Supply and Demand The
34. The function f ( x ) 5 x1 is undefined at x 5 0. Is it possible to calculate the area of the region bounded by f and g( x ) 5 2 x 2 1 4? Explain.
In each exercise below, we are interested in the region that is bordered by all of the given functions. 35. What is the combined area of the regions bordered by the three functions f (x) 5 x2 , g( x ) 5 2 x2 1 4, and h(x) 5 2x? 36. What is the combined area of the regions bordered by the three functions f (x) 5 2x , g( x ) 5 22x , and h( x ) 5 2 x 1 4? 37. What is the combined area of the regions bordered by the three functions f (x) 5 x2 , g(x) 5 x, and h( x ) 5 2 x2 1 1? 38. What is the area of the region bordered by the three functions f (x) 5 x2 , g(x) 5 5x 2 6, and h( x ) 5 2 x2 ? 39. What is the area of the region bordered by the four functions f ( x ) 5 x, g( x ) 5 2 x 1 4, h( x ) 5 ( x 2 2 ) 2, and j( x ) 5 1.5?
demand for imprinted lollipops may be modeled by p( q ) 5 0.0144q2 2 0.504q 1 12.40 cents per lollipop where q is the number of lollipops purchased (in thousands). (Source: Modeled from data at www.pens.com.) The supply function for the lollipops is s( q ) 5 0.000131q 1 0.450 cents per lollipop 29.
Water Bottle Supply and Demand
The demand for 22-ounce sports bottles with straw tops may be modeled by p( q ) 5 1.93 dollars per bottle ( q 1 1 ) 0.116
498
7.3 Improper Integrals
Use indefinite integrals to analyze real-world situations Calculate indefinite integrals
Pandemic influenza is a situation in which a particular strain of influenza impacts a wide geographic area, affecting a large portion of the population. These super-flu strains kill millions more people than do annual flu epidemics. Pandemic influenza outbreaks have occurred in 1918, 1957, and 1968. The director of the World Health Organization warned of the high likelihood of an avian influenza pandemic in 2006. In this section, we will introduce you to the concept of improper integrals. Improper integrals may be used to analyze situations such as a pandemic flu outbreak.
We will investigate improper integrals by modeling mortality data from the 1968 Hong Kong flu (named after its place of origin). The Centers for Disease Control and Prevention recorded the following weekly totals of excess pneumoniainfluenza deathsthat is, the numbers of such deaths in excess of the average numbers to be expected. The following table shows the impact of the disease in New York City.
Excess Pneumonia-Influenza Deaths w, F, Week Flu-Related Deaths 1 2 3 4 5 6 7 8 9 10 11 12 13 14 28 50 66 156 190 156 108 68 77 33 65 24
We are interested in determining the total number of excess flu-related deaths as time (number of weeks) continues to increase. Each value of F is the number of flu-related deaths during that week. The function F(w) represents the
499
number of flu-related deaths for a given week w. We can use the definite integral
13
3 F( w ) dw
1
to determine the total of excess flu-related deaths between the first and thirteenth weeks. By creating a model for F(w), we can forecast the number of excess flu-related deaths for this particular epidemic in any future week. If we use the integral 3 F( w ) dw
1 n
we can find the total excess number of deaths after n weeks. If we imagined using this integral to determine the total excess number of deaths as n S ` , we would need to imagine an improper integral.
`
3 F( w ) dw
1
An integral having 2 ` or 1` as one of its limits of integration is called an improper integral.

`
To evaluate the improper integral 3 F( w ) dw, we will first determine a mathematical model F(w) for the given data. Through experimentation, we find that the following function best models the data. F( w ) 5 33we2(w 2 5.5) >9
2
Figure 7.18 shows a scatter plot of the data, along with the model F(w).
F 200 Excess Deaths 150 100 50 0 0 2 4 6 8 Week 10 12 w
FIGURE 7.18
EXAMPLE
Evaluating an Improper Integral

Use the data concerning excess pneumonia-influenza-related deaths to evaluate the integral
`
3 F( w ) dw
1
500
SOLUTION The area under this curve represents the total number of excess deaths between the first and thirteenth weeks. Using our calculator to evaluate the integral, we determine that
2(w 2 5.5) >9 dw < 963.9 deaths 3 33 ? w ? e
2
13
What if we want to follow the trend beyond the thirteenth week? That is, what if we want to know the total number of excess deaths for the entire flu epidemic? We could let the upper limit of integration go to infinity. Doing so yields
2(w 2 5.5) >9 dw < 964.4 deaths 3 33 ? w ? e
2
Notice that there is not a big difference between the value of the definite integral and the improper integral in Example 1. This is because this particular improper integral converges to the finite value < 964. An improper integral is said to converge if it yields a finite value. Otherwise, it is said to diverge.
CONVERGENCE OF IMPROPER INTEGRALS ON AN UNBOUNDED INTERVAL
If f ( x ) . 0 for x $ a and lim 3 f ( x ) dx 5 L

b b
bS` a
for some finite number L, then 3 f ( x ) dx is said to converge to L.

a
Because f ( x ) . 0 for most improper integrals, the above restriction of f is not unreasonable. We can make an educated guess as to whether the improper integral will converge or diverge by first analyzing its graph. If the function f (x) gets small enough, fast enough, then it will converge. This gives us a qualitative way to understand the nature of the improper integral.
EXAMPLE

`
Evaluate the improper integral 3

1
1 x2
dx.
501
SOLUTION First, examine a graph of the function on the interval 31, ` ) . Does it appear as though the function gets small enough (approaches the x-axis) fast enough that the improper integral converges?
y 8 6 4 2 0 0 2 4 6 8 x 1 x2
f (x) =
FIGURE 7.19
Note that even on the interval 31, 104 shown in Figure 7.19, it appears as 1 though the function f ( x ) 5 x 2 approaches the x-axis fast enough that the area under the curve on the interval 31, ` ) will indeed converge to a finite number. Lets analyze the improper integral symbolically to confirm our prediction.
b
We first compute the definite integral 3

1 b b
1 x2
dx .
1 22 3 x 2 dx 5 3 x dx 1 1 5 2 x 21 ` 1 52 ` x 1 1 52 11 b Since we are interested in the improper integral, we take the limit as b S ` .

bS`
b b 1
1 1 lim a 2 1 1b 5 lim a 2 b 1 lim ( 1 ) b bS` b bS` 5011 51
Because 2 b approaches zero rapidly as b S ` , the improper integral 3 1 converges to 1.
1 x2
dx
502
EXAMPLE

`
Evaluate the improper integral 3 e23x dx .

22
SOLUTION We begin by examining a graph of the function to try to determine if the function f ( x ) 5 e23x gets small enough fast enough.
y 350 300 250 200 150 100 50 3 2 1 0 f (x) = e 3 x 1 2 3 x
FIGURE 7.20
Even on the interval 32 2, 34 shown in Figure 7.20, the function seems to approach the x-axis very quickly. Imagine finding the area under this curve on the interval 32 2, ` ) . The graph is so close to the x-axis by the time we reach x 5 3 that we predict that the improper integral representing the area under the curve to the right of x 5 2 will converge to a finite number. Well confirm our prediction by evaluating the integral and finding the number to which it converges. We begin by evaluating the definite integral 3 e23x dx.
22 b
1 23x b 23x 3 e dx 5 2 3 e `
22
22
52 52 52 52
1 b ` 3e3x 22 1 1 2 3( 2 2) b 3b 2 a 3e 3e 1 1 1 26 3e3b 3e 1 e6 1 3e3b 3

`
Since we are interested in the improper integral 3 e23x dx, we take the limit of the definite integral as b S ` .
22
503
bS`
lim a 2
1 e6 e6 1 b 5 0 1 3e3b 3 3 5 e6 3
< 134.5
1 23x Because 2 3e 3b approaches zero as b S ` , the improper integral 3 e dx 22 e6 converges to 3 .
`
EXAMPLE

`

2
1 x11
dx .
SOLUTION We first look at the graph of the function in an attempt to predict whether or not the function will get small enough fast enough to cause the improper integral to converge to a finite number. See Figure 7.21.
y 0.4 0.3 0.2 0.1 0 0 2 4 6 8 x
f (x) =
1 x+1
FIGURE 7.21
From our graph, it is not clear whether the function is getting small enough fast enough. While the function does approach the x-axis, function values do not get close to zero as quickly as they did in the previous examples. We are uncertain as to whether or not this improper integral converges. Symbolic work will tell us
b
for sure. Again, we begin by evaluating the definite integral 3

2 b
b 1 3 x 1 1 dx 5 ln( x 1 1 ) ` 2 2
1 dx . x11
5 ln( b 1 1 ) 2 ln( 2 1 1 ) 5 ln( b 1 1 ) 2 ln( 3 )
504
We take the limit of the definite integral as b S ` .

bS`
lim 3ln( b 1 1 ) 2 ln( 3 ) 4 5 ` 2 ln( 3 ) 5`
Since ln( b 1 1 ) approaches infinity as b approaches infinity, the limit does not exist. Therefore, the improper integral diverges. That is, the area under the 1 function x 1 1 on 32, ` ) does not converge to a finite number but rather goes to infinity.
In Examples 1 to 4, we looked at improper integrals in which we were integrating over an infinitely wide interval. An integral is also considered improper when its integrand (the function that is being integrated) is unbounded over the interval given in the limits of integration, as demonstrated in the next example.
EXAMPLE

3

0
1 !x
dx.
SOLUTION Notice that the limits of integration are not infinite. Rather, the in-
tegrand !x approaches infinity as x approaches 0. As in the previous examples, we wish to explore the behavior of the graph near x 5 0 to determine whether the area of the region bounded by the x- and y-axes and the function will be small enough as x S 0 to make the improper integral converge. Lets look at a graph of the function over the given interval to see if it appears that the area under the curve converges to a finite number. (See Figure 7.22.)
y 8 6 4 f (x) = 2 0 0 0.5 1.0 1.5 2.0 2.5 1
x
FIGURE 7.22
Note that for this function, the upper limit of 3 provides a boundary for the area under the curve on the right side of the graph. With a lower limit of 0, the question now becomes whether the function gets close enough to the y-axis quickly enough for the improper integral to converge. It appears as though it might.
505
We evaluate the definite integral 3

3
undefined at x 5 0, we use the variable b in place of the lower limit 0. 1 21>2 dx 3 ! x dx 5 3 x b b 5 2x 1>2 `
3 b 3 b
1 !x
dx. Note that because the function is

3
5 2"x `
We take the limit as b S 01 to determine if the improper integral converges or diverges.

b S 01
5 2"3 2 2"b
lim ( 2"3 2 2"b ) 5 2"3
Notice that 2"b approaches zero as b approaches zero from the right. Thus the
3
improper integral 3
0 "x
dx converges to 2"3.
EXAMPLE

5

0
1 ( x 2 5 )2
dx.
SOLUTION The integrand
fore, the integrand becomes infinite at x 5 5. This feature is what causes this integral to be classified as improper. Again, we begin by investigating a graph of the function f ( x ) 5 ( x
1 2 5 )2 y 8 6 4 2 0 0 1 2 3 4 5 x
1 ( x 2 5 )2
has a vertical asymptote at x 5 5. There
on the interval 30, 54 .
f (x) =
1 (x 5)2
FIGURE 7.23
506
From Figure 7.23, it is difficult to tell if the function gets close enough fast enough to the asymptote x 5 5 to cause the area under the curve on the interval 30, 54 to converge to a finite number. To find out whether it converges, we evaluate the
b
definite integral 3
0
1 ( x 2 5 )2
b
dx. Note that we use the variable b in place of the

b
upper limit of integration because it is at x 5 5 that the integrand becomes infinite. 1 22 3 ( x 2 5 ) 2 dx 5 3 ( x 2 5 ) dx 0 0 5 2 ( x 2 5 ) 21 ` 5 21 b ` x250 1 1 1 b25 5
b 0
52
Because the graph of the function is approaching the vertical asymptote x 5 5 from the left, we take the limit as b S 52 to determine if the improper integral converges or diverges. (Recall that b S 52 means that we are approaching 5 from the left.)
bS5
lim a 2
2
1 1 1 1 b 5`1 b25 5 5 5`
Note that as b S 5 , the quantity b 2 5 approaches 0 from the negative side. 1 With such a small denominator, the fraction 2 b 2 5 goes to infinity as b approaches 5 from the negative side of 5. Therefore, the improper integral
2
3
0
1 ( x 2 5 )2
dx diverges.
IMPROPER INTEGRALS WITH AN INTEGRAND THAT BECOMES INFINITE
If f ( x ) is positive and continuous on the interval 3a, c ) with lim f ( x ) 5 6` xSc and lim 3 f ( x ) dx 5 L bSc
2
(where L is a finite number), then the improper integral 3 f ( x ) dx converges

a
to L. Otherwise, 3 f ( x ) dx diverges.
a
507
7.3 Summary
In this section, you learned that improper integrals may converge to a finite number, or they may diverge. You learned how to investigate the graph of a function and analyze the area under its curve over a given interval to make a prediction about the nature of the improper integral. You also discovered that symbolic work with integration and limits will tell you for sure whether an improper integral converges or diverges.

After calculating the indicated values, determine what value (if any) the function approaches as x goes to infinity. 1. Given f ( x ) 5 2. Given g( x ) 5
1 , !x 2 1 1
5. Given f ( x ) 5 6. Given g( x ) 5
f ( 0.01 ) , and f ( 0.001 ) .
2 log( x ) , 4x
calculate f ( 1 ) , f ( 0.1 ) , calculate g( 4 ) , g( 40 ) ,
calculate f ( 0 ) , f ( 99 ) ,
f ( 9999 ) , and f ( 999,999 ) . g( 1000 ) , and g( 10,000 ) . 3. Given f ( x ) 5 6x 4. Given g( x ) 5 f ( 1000 ) , and f ( 10,000 ) .
log( x ) , x 4x 2 9 , 27 2x , x2 1 4
g( 400 ) , and g( 40,000 ) . 7. Given f ( x ) 5 e2x 2 ln( x ) , calculate f ( 1 ) , f ( 100 ) , f ( 1000 ) , and f ( 10,000 ) .
1 1 x 1
!x 2 4 , !4x 2 5
calculate g( 0 ) , g( 100 ) , calculate f ( 0 ) , f ( 100 ) , calculate g( 1 ) , g( 100 ) ,
8. Given g( x ) 5 x 2 2
1 1 , calculate f ( 1 ) , f ( 10 ) ,
f ( 100 ) , and f ( 1000 ) .
g( 1000 ) , and g( 10,000 ) .
7.3 Exercises
In Exercises 12, investigate the given improper integral to determine whether it converges or diverges. Specifically, do the following: (a) Compute the integral on the intervals from 1 to 5, 1 to 10, 1 to 50, and 1 to 100. (b) Make a graph of the function that comprises the integrand of the improper integral and make a guess as to whether or not the improper integral converges. (c) Using the results of parts (a) and (b), determine whether the integral converges or diverges. 1 dx 1. 3 1 x 1 2
` `
In Exercises 315, evaluate the improper integral and state whether it converges or diverges. 1 3. 3 4 dx x 1
` ` 0
4. 3 e x dx
2` `
5. 3 x
1 `
23>2
dx
6. 3
25 `
2 dx x15
7. 3 22x dx 1 9. 3 dx ( x 2 2)2 2` 1 11. 3 dx x ln (x) 1

` 0 2
1 8. 3 dx ( x 2 2)2 2
0
10. 3 x 2e x dx 1 12. 3 2 dx x 0
2` 1
2. 3 3e2x dx
1
508
`
13. 3 e20.25x dx x 15. 3 2 dx 0 x 1 2

0 `
1 14. 3 dx 22 x 1 2
forcible rapes per 100,000 people and P is the population (in hundred-thousands), starting with 2628 (hundred-thousand) in 1995. (a) Create a graph of F versus P .
`
In Exercises 1622, evaluate the improper integral. 1 16. 3 dx x 1 1 18. 3 3 dx 1 x 1 20. 3 1.0001 dx 1 x
` ` ` `
(b) What does the integral 3

2628
0.2678e20.0008P dP
1 17. 3 2 dx 1 x 1 19. 3 6 dx 1 x
` `
represent in this situation?

`
(c) Evaluate 3
2628
0.2678e20.0008P dP and interpret
the meaning of the result in the context of the problem. Do you think the result is reasonable? Why or why not? 28. The Federal Bureau of Investigation, in its report Crime in the United States 2000, reports that the number of homicides due to arguments over money or property has been decreasing at a decreasing rate since 1991. A model for the number of homicides due to arguments over money or property is H 5 599.39e20.1094t, where H is the number of homicides and t is the number of years after 1990. (a) Create a graph of H versus t.
FBI Crime Report
`
21. 3
1
22. 3
1
"x
3
"x
dx
dx
23. Based on your work in Exercises 16 to 22, describe under what conditions an improper integral of the
`
form 3
1
1 x p dx
will converge.
24. Make a graphical argument for the conditions you described in Exercise 23. That is, use graphs of 1 functions of the form f ( x ) 5 p to explain the x conditions under which the improper integral
`
(b) What does the integral 3 599.39e20.1094t dt

1

`
3
1
1 xp
dx converges.
(c) Evaluate 3 599.39e20.1094t dt and interpret the

1
25. In general, what will be the value of the improper

`
integral 3
1
1 xp
dx for the case in which it converges?
meaning of the result in the context of the problem. Do you think the result is reasonable? Why or why not? 29. The number of cases of pediatric AIDS reported each year has been decreasing since 1992. A model for the number of new pediatric AIDS cases diagnosed annually is P 5 1276.3e20.2776t, where P is the number of cases of pediatric AIDS and t is the number of years after 1992. (Source:
Pediatric AIDS Cases
26.
Recall that the function used to model the flu pandemic situation described at the beginning of the section was
Flu Pandemic
2(w 2 5.5) >9 dw. Verify using numerical 3 33 ? w ? e
2
methods (e.g., the Trapezoidal Rule, Simpsons Rule) that the value of this improper integral is approximately 964 excess deaths. 27. The Federal Bureau of Investigation, in its report Crime in the United States 2000, reports that the number of forcible rapes has been decreasing at a decreasing rate since 1995, despite an increase in total population. A model for the rate at which the number of forcible rapes has been decreasing is F 5 0.2678e20.0008P, where F is the number of
FBI Crime Report
Modeled from www.cdc.gov.)
(a) Create a graph of P versus t.

`
(b) What does the integral 3 1276.3e20.2776t dt

1

`
(c) Evaluate 3 1276.3e20.2776t dt and interpret

0
the meaning of the result in the context of the problem. Do you think the result is reasonable? Why or why not?
509
30.
Based on data from the weekend box office, movie ticket sales for The Passion of the Christ may be modeled by s( t ) 5 200e20.4819 t million dollars per weekend, where t is the weekend number, with t 5 1 representing February 29, 2004. (Source:
Movie Ticket Sales
Explain what this bumper sticker means. Show all work. The Laplace transform is a tool for solving some differential equations. It transforms the differential equation problem into a problem involving algebra. The Laplace transform is defined as the function created from the improper integral
`
Modeled from www.rottentomatos.com data.)
(a) Create a graph of s versus t.

`
(b) What does the integral 3 200e20.4819t dt

1
F( s ) 5 3 e2st f ( t ) dt.
0

`
(c) Evaluate 3 200e20.4819t dt and interpret the

1
meaning of the result in the context of the problem. Do you think the result is reasonable? Why or why not?
That is, given a function f ( t ) , the Laplace transform operation transforms this function into the function F( s ) . For Exercises 3641, transform the given function f ( t ) into the function F( s ) using the Laplace transform operation. 36. f ( t ) 5 t 38. f ( t ) 5 t 3 37. f ( t ) 5 t 2 39. f ( t ) 5 sin t 41. f ( t ) 5 e t
31. Write three different improper integrals that are considered improper for different reasons. Explain what makes each improper and then draw a graph to support your explanation.
`
40. f ( t ) 5 cos t
32. Consider the improper integral 3 e2ax dx, where a . 0. Calculate the improper integral for different values of a. Generalize the results for positive values of a and describe why this generalization makes sense. 33. A student explains to a classmate that a particular improper integral converges because it gets small enough fast enough. Clearly explain what the word it refers to in this phrase and what is meant by small enough fast enough. 34. A calculus student who has not yet studied improper integrals reads the statement 1 1 3 x 3 dx 5 2 1 and asks you what this statement means. Explain precisely what the symbols mean and what the 1 resulting value of 2 means. 35. While driving down the road, you see a car with a bumper sticker that reads:
` ` 0
42. Review the results of Exercises 36 to 38. Create and test a conjecture about the Laplace transform of any function of the form f ( t ) 5 t n. That is, if f ( t ) 5 t n, what do you think F( s ) will be? A probability density function is defined as a function that is always greater than or equal to zero and that has the property 3 f ( x ) dx 5 1, since the probability of an event that is certain to occur is 1. For Exercises 4345, show that the given function is a probability density function. 43. Uniform distribution where f (x) 5 e 1, 0 # x # 1 0, otherwise
2
44. Normal distribution where f (x) 5 e2x >2 on ( 2 ` , ` ) 2p
45. Exponential distribution where
f ( x ) 5 2e22x on 30, ` )
Calculus is as easy as 3
2`
1 dx. x2 1 1
510

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. Which derivative rule is closely related to integration by parts? Explain. (7.1) 2. What guidelines should you apply when choosing u and dv in the integration by parts method? (7.1) 3. Under what conditions is 3 3 f ( x ) 2 g( x ) 4 dx equal to the area between f and g
a b
on 3a, b4 ? (7.2) 4. You are asked to find the area of the region bounded by the graphs of f and g. How do you determine the limits of integration for the definite integral? (7.2) 5. What is the relationship between a definite integral and the difference of two accumulated changes? (7.2) 6. What does it mean when we say that an integral is improper? (7.3)
`
7. What does it mean when we say that 3

2
1 x2 2 2
dx converges to about
0.6323? What makes the integral improper? (7.3)

5
8. What does it mean when we say that 3 integral improper? (7.3)
"2
1 x2 2 2
dx diverges? What makes the
You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
Definitions
Area Between f and g on [a, b]: The area of the region(s) between x 5 a and x 5 b that has upper and lower boundaries consisting of the graphs of f and g Improper Integral: An integral with 2 ` or 1` as one of its limits of integration, or an integral whose integrand tends to infinity at some value of the limits of integration Converging Improper Integral: An improper integral is said to converge if evaluation of the integral yields a finite value. Diverging Improper Integral: An improper integral is said to diverge if evaluation of the integral does not yield a finite value.
Integration by Parts Method: 3u dv 5 uv 2 3v du Identify u and dv. Differentiate u and integrate dv with respect to x.
Write the expression uv 2 3v du in terms of x.
Chapter 7
Review Exercises
511
Integrate 3v du to eliminate the integral sign. Simplify. Convergence of Improper Integrals on an Unbounded Interval:
If f ( x ) . 0 for x $ a and lim 3 f ( x ) dx 5 L for some finite number L, bS`

` a a
then 3 f ( x ) dx is said to converge to L.
Improper Integral with an Integrand That Becomes Infinite: If f ( x ) is positive and continuous on the interval 3a, c ) with lim f ( x ) 5 6` , lim f ( x ) 5 ` , and lim 3 f ( x ) dx 5 L (where L is a finite number), then
xSc b S c2 a b xSc
the improper integral 3 f ( x ) dx converges to L. Otherwise, 3 f ( x ) dx a a diverges.
Formulas
Integration by Parts Method: 3u dv 5 uv 2 3v du

Section 7.1 In Exercises 110, integrate the function using the simplest possible method.
1. f ( x ) 5 2 xe2x 2. g ( x ) 5 2 4x ln( x ) 3. h (x) 5 3x 2e x 4. g( t ) 5 2t ln( t ) 5. f ( x ) 5 3ln( x ) 4 3 x
x 2 2 x +1
Section 7.2 In Exercises 1115, calculate the combined area of the region(s) bounded by the graphs of the two functions. (On some of the exercises, you may find it helpful to use your graphing calculator to locate the intersection points of the graphs.)
11. f (x) 5 x 2 2 1; g ( x ) 5 2 x 2 1 17 12. f (x) 5 x 2 2 1; g(x) 5 x 3 2 3x 2 2 x 1 3 13. f (x) 5 2 x; g (x) 5 x 2 14. f (x) 5 3 x; g (x) 5 3x 3 15. f ( x ) 5 ln( x ) ; g (x) 5 0.1x
6. h (x) 5 (2x 2 1) e
7. g ( x ) 5 ( 5x 2 3 ) 3ln( x ) 4 8. f (t) 5 (t 2 1)4 (2t 2 5) 9. h (x) 5 4 xe4 x 10. g ( t ) 5 2t 2 t ln( t )
512
In Exercises 1620, graph each of the functions. Then determine the solution by using the integral techniques demonstrated in Section 7.2. 16.
Adult and Adolescent AIDS Based on data from 1981 to 1992, the rate of change in the annual number of adult and adolescent AIDS cases diagnosed in the United States may be modeled by R (t) 5 75.48t 3 2 1237t 2 1 6652t 2 4882
Table 205, p. 133.) In 2000, 58,821,000 students
were enrolled in public schools and 9,303,000 students were enrolled in private schools. Determine the accumulated change in the annual enrollment gap between private and public schools over the period 2000 to 2010. Interpret the meaning of the result. 19.
Public- and Private-School Teachers
cases per year, and the rate of change in the annual number of adult and adolescent deaths in the United States due to AIDS may be modeled by D ( t ) 5 2 107.3t 2 1 1641t 2 975.3 deaths per year, where t is the number of years since the end of 1981. (Source: Modeled from
Centers for Disease Control and Prevention data.)
Based on U.S. National Center for Educational Statistics projections for 2000 to 2010, the rate of change in the number of publicschool teachers may be modeled by g( t ) 5 2 1.4242t 1 16.11 thousand teachers per year, and the rate of change in the number of private-school teachers may be modeled by p ( t ) 5 2 0.8158t 1 1.453 thousand teachers per year, where t is the number of years since the end of 2000. (Source: Modeled
from Statistical Abstract of the United States, 2001, Table 207, p. 134.) In 2000, there were 2,850,000
Determine the accumulated change in the difference of the two populations from the end of 1981 through 1992. Interpret the meaning of the result. 17.
Company Profit Based on data from 1999 to 2001, the rate of change in the revenue of the Johnson & Johnson company and its subsidiaries may be modeled by r (t) 5 1330t 1 1150
million dollars per year, and the rate of change in the cost of goods sold may be modeled by c(t) 5 206t 1 315 million dollars, where t is the number of years since the end of 1999. (Source: Modeled from
Johnson & Johnson 2001 Annual Report.)
public-school teachers and 402,000 private-school teachers. Determine the predicted accumulated change in the annual gap between private- and publicschool teachers over the period 2000 to 2010. Interpret the meaning of the result. 20.
Wages of Clothing Producers Based on data from 1980 to 2000, the rate of change in the average hourly wage of a mens and boys furnishings production worker may be modeled by m (t) 5 0.007032t 1 0.1446
Determine the accumulated change in annual profit from the end of 1999 through 2001. 18.
Public- and Private-School Enrollment Based on U.S. National
Center for Education Statistics enrollment projections for 2000 to 2010, the rate of change in the annual number of students enrolled in public schools may be modeled by g (t) 5 0.4971t 2 2 3.022t 1 51.17 thousand students per year, and the rate of change in the number of students enrolled in private schools may be modeled by p ( t ) 5 2 19.99t 1 280.5 thousand students per year, where t is the number of years since the end of 2000. (Source: Modeled
from Statistical Abstract of the United States, 2001,
dollars per year, and the rate of change in the average hourly wage of a womens and misses outerwear production worker may be modeled by w (t) 5 0.006394t 1 0.1333 dollars per year, where t is the number of years since 1980. (Source: Modeled from Statistical
In 1980, the hourly wage of a mens and boys furnishings production worker was $4.23, and the average hourly wage of a womens and misses outerwear production worker was $4.61. Determine the accumulated change in the hourly wage gap between mens and boys furnishings and womens and misses outerwear production workers over the period 1980 to 2000.
Chapter 7
Review Exercises
513
Section 7.3 In Exercises 2128, evaluate the improper integral and state whether it converges or diverges.
1 21. 3 dx 0 x 1 4 22. 3 x 25>2 dx 1 23. 3 2 dx 3 (x 2 3)
` 0 ` ` `
representing May 5, 2002. (Source: Modeled from www.rottentomatos.com data.) (a) Create a graph of s versus t.
`
(b) What does the integral 3 200e20.5044t dt

1

`
(c) Evaluate 3 200e20.5044t dt and interpret the

1
24. 3 2e23x dx
1 `
meaning of the result in the context of the problem. Do you think the result is reasonable? Why or why not? 30. Based on data from the weekend box office, movie ticket sales for A Beautiful Mind may be modeled by B( t ) 5 15t 0.9( 0.75 ) t million dollars per weekend, where t is the weekend number, with t 5 1 representing December 25, 2001. (Source:
Movie Ticket Sales
25. 3 5 dx
2x
1 26. 3 3 dx 0 x
`
0 2
27. 3 e0.1x dx 4x dx 28. 3 2 0 2x 1 5 In Exercises 2930, apply the concept of improper integrals to solve the real-world application. 29. Based on data from the weekend box office, movie ticket sales for Spider-Man may be modeled by s( t ) 5 200e20.5044 million dollars per weekend, where t is the weekend number, with t 5 1
Movie Ticket Sales
2 `
Modeled from www.rottentomatos.com data.)
(a) Create a graph of B versus t.

`
(b) What does the integral 3 15t 0.9( 0.75 ) t dt

1

`
(c) Evaluate 3 15t 0.9( 0.75 ) t dt and interpret the

1
meaning of the result in the context of the problem. Do you think the result is reasonable? Why or why not?
P R O J E C T
What to do
1. Identify a public company that you would like to work for in the future. 2. Obtain a copy of the companys annual financial report. 3. Using five or more years of data, model the revenue of the company and the costs of the company as a function of time. 4. Graph the revenue and cost models on the same axes. 5. Calculate the accumulated change in revenue and the accumulated change in costs over a five-year period using the models. 6. Shade the region that represents the accumulated change in profit. 7. Using the results from Steps 5 and 6, calculate the accumulated change in profit over the five-year period

Publicly held companies are required to release financial data to the public through annual reports. Many of these annual reports are available online and may be accessed through the investor information page on the companys web site. Here are links to the home pages of some prominent companies.
Safeway
www.safeway.com
McDonalds
www.mcdonalds.com
Kelloggs
www.kelloggs.com
Revlon
www.revlon.com
514
Chapter
Multivariable Functions and Partial Derivatives

T
he amount of moisture in the air affects how hot it feels to a person. If the air temperature is 96F and the relative humidity is 75 percent, it feels like it is 132! The apparent temperature (what it feels like) is a multivariable function of the actual temperature and the relative humidity. Partial derivatives may be used to determine what effect a one-degree increase in the actual temperature will have on the apparent temperature if the relative humidity remains constant.
8.1
Multivariable Functions

Analyze and interpret multivariable mathematical models Evaluate multivariable functions Analyze and interpret multivariable mathematical models Calculate first- and second-order partial derivatives of multivariable functions Find the equation of a cross section of a graph of a surface Determine the meaning of partial rates of change
8.2
Partial Derivatives

8.3
Multivariable Maxima and Minima

Analyze and interpret multivariable mathematical models Find critical points of multivariable functions Locate relative extrema and saddle points of two-variable function graphs
8.4
Constrained Maxima and Minima and Applications

Analyze and interpret multivariable mathematical models Use the Lagrange Multiplier Method to find constrained maxima and minima
515
516
CHAPTER 8 Multivariable Functions and Partial Derivatives
8.1 Multivariable Functions
Analyze and interpret multivariable mathematical models Evaluate multivariable functions
Businesses seek to manage their labor and capital so as to maximize their production efficiency. Since financial resources are limited, executives have to determine how much money to invest in their employees and how much money to invest in equipment and buildings. Economists Charles Cobb and Paul Douglas formulated the multivariable Cobb-Douglas production function to model the output of small- and large-scale economies. In this section, we will discuss how to evaluate and graph multivariable functions. Additionally, well look at several applications of multivariable functions, including the Cobb-Douglas production function.
Functions requiring two or more inputs to generate a single output are called multivariable functions. For example, the heat index is a function of the temperature and the humidity. The wind chill factor is a function of temperature and wind speed. The volume of a cylindrical can is a function of its radius and its height. The market share of a business is a function of the market shares of each of its individual competitors. The value of a savings account is a function of the amount invested, the interest rate, the compounding frequency, and the amount of time the money is invested. Functions with two input variables may be represented using a three-dimensional graph.
FUNCTION OF TWO VARIABLES
A function f of two variables is a rule that associates each ordered pair of inputs ( x, y ) with a single output z 5 f ( x, y ) . The graph of a function of two variables is a three-dimensional surface.
In this context, both x and y are independent variables and z is the dependent variable. The domain of the function is the set of all ordered pairs ( x, y ) . The range of the function is the set of corresponding values of f ( x, y ) . Consider the volume of a cylindrical can. The equation for the volume is given by V 5 pr 2h, where r is the radius and h is the height of the can. We may alternatively write f ( h, r ) 5 pr 2h, where V 5 f ( h, r ) . What is the volume of a can with a five-inch height and a two-inch radius? f ( h, r ) 5 pr 2h f ( 5, 2 ) 5 p( 2 ) 2( 5 ) 5 20p < 62.83 The volume of a cylindrical can with a five-inch height and a two-inch radius is approximately 62.83 cubic inches. We may generate a table of values for V by constructing a table with a column for each of the variables (see Table 8.1).
517
TABLE 8.1
h 1 1 1 2 2 2
r 1 2 3 1 2 3
V 5 pr 2h p 4p 9p 2p 8p 18p
Notice that each value of h may be paired with multiple values of r. A more concise way to record the information is to draw a table with the rows containing the values of h and the columns containing the values of r (see Table 8.2). The value of V is contained in the body of the table.
TABLE 8.2
h T 1 2
rS 1 p 2p 2 4p 8p 3 9p 18p
To graph V, we must use a three-dimensional grid with an axis for each of the variables. One of the easiest ways to visualize this is to look at an interior corner of a classroom. The edge between one wall and the floor represents the h-axis. The edge between the other wall and the floor represents the r-axis. The edge joining the two walls represents the V-axis. The origin, ( 0, 0, 0 ) , is the point of intersection of the three axes (see Figure 8.1).
V
(0, 0, 0) h r
FIGURE 8.1
518
The hV plane is the wall formed by the h-axis and the V-axis. The rV plane is the wall formed by the r-axis and the V-axis. The hr plane is the floor formed by the h-axis and the r-axis. (Using the standard variables x, y, and z, the xz plane and the yz plane form the walls and the xy plane is the floor.) A point in this coordinate system is an ordered triple of the form ( h, r, V ) . To plot the point ( 2, 1, 2p ) , we move 2 units from the origin along the h-axis, move 1 unit from the origin in the direction of the r-axis, and finally move 2p units up in the direction of the V-axis, as shown in Figure 8.2.
V
(2, 1, 2p )
(0, 0, 0) h r
FIGURE 8.2
To plot the point ( 1, 2, 4p ) , we move 1 unit from the origin along the h-axis, move 2 units from the origin in the direction of the r-axis, and finally move 4p units up in the direction of the V-axis, as shown in Figure 8.3.
V (1, 2, 4p )
(2, 1, 2p )
(0, 0, 0) h r
FIGURE 8.3
(If h, r, or V were negative, we would move in the opposite direction. However, since the equation represents volume as a function of lengths, all variables will be nonnegative.) Visualizing what the entire surface will look like is often difficult, since were trying to represent a three-dimensional object on a two-dimensional face of a sheet of paper. However, many technologies and techniques have been developed to make it easier to do so. An Internet search on the key words 3D Graphing Freeware will reveal a variety of free graphing utilities that may be used to
519
draw the surface. Additionally, purchased graphing programs such as Maple, Mathematica, and Autograph can create beautiful 3D surfaces that can be easily rotated. To help you better visualize the surface, the graph of the cylinder volume function is shown from two different angles in Figure 8.4.
V V
FIGURE 8.4
EXAMPLE
Graphing a Multivariable Function

Generate a table of data for the multivariable function f ( x, y ) 5 x 2 1 y 2 and use a graphing utility to draw the surface.
SOLUTION To calculate the values in Table 8.3, we substitute each ordered pair ( x, y ) into the function f. For example,
f ( x, y ) 5 x 2 1 y 2 f ( 2 3, 2 2 ) 5 ( 2 3 ) 2 1 ( 2 2 ) 2 5914 5 13
TABLE 8.3
x T 3 2 1 0 1 2 3
yS 3 18 13 10 9 10 13 18 2 13 8 5 4 5 8 13 1 10 5 2 1 2 5 10 0 9 4 1 0 1 4 9 1 10 5 2 1 2 5 10 2 13 8 5 4 5 8 13 3 18 13 10 9 10 13 18
520
Notice that the low point of this surface occurs at the origin, ( 0, 0, 0 ) . As we move away from the origin in any direction, the value of z increases. Two different views of the graph of the surface are shown in Figure 8.5.
z
x x y
FIGURE 8.5a
FIGURE 8.5b
Figure 8.5a is a graph in standard position. The x-axis points to the left, the y-axis points to the right, and the z-axis points upward. In Figure 8.5b, we have rotated the axes so that we can see the surface from a different point of view.
EXAMPLE
Graphing a Multivariable Function

Generate a table of data for the multivariable function f ( x, y ) 5 y 3 2 yx 2 and use a graphing utility to draw the surface.
SOLUTION
TABLE 8.4
x T 3 2 1 0 1 2 3
yS 3 0 15 24 27 24 15 0 2 10 0 6 8 6 0 10 1 8 3 0 1 0 3 8 0 0 0 0 0 0 0 0 1 8 3 0 1 0 3 8 2 10 0 6 8 6 0 10 3 0 15 24 27 24 15 0
The graph of the function on the specified domain is shown in Figure 8.6 in standard position and with the axes rotated.

z z
521
FIGURE 8.6
EXAMPLE
Using a Multivariable Function to Forecast the Value of an Investment

A $1000 investment is to be made into an account paying 6 percent compound interest. The value of the investment compounded n times per year after t years is given by A( n, t ) 5 1000 a 1 1 0.06 nt b n
Generate a table of data for the multivariable function using n 5 1, 2, 4, 12, 365 and t 5 0, 1, 2, 3. Graph the function and interpret the meaning of ( 12, 2, 1127.16 ) .
SOLUTION
TABLE 8.5
n T
A
tS 0 1000.00 1000.00 1000.00 1000.00 1000.00 1 1060.00 1060.90 1061.36 1061.68 1061.83 2 1123.60 1125.51 1126.49 1127.16 1127.49 3 1191.02 1194.05 1195.62 1196.68 1197.20
1 2 4 12 365
t
FIGURE 8.7
When we graph the function using a window with 0 # A # 1200, the graph looks like a plane parallel to the nt plane (Figure 8.7). However, when we use the window 1000 # A # 1200, we get a better look at what is happening (Figure 8.8).
522
FIGURE 8.8
The compounding frequency, n, has a greater effect on the value of the investment as t increases. The meaning of ( 12, 2, 1127.16 ) is that a $1000 investment earning 6 percent interest compounded monthly ( t 5 12 ) at the end of two years ( n 5 2 ) will have a value of $1127.16.
EXAMPLE
Using a Multivariable Function to Forecast Business Sales

Based on data from 1998 to 2001, the net sales of Kellogg Company and subsidiaries may be modeled by S( a, e ) 5 2 0.36661a 1 0.16556e 1 4671.3 million dollars where a is the amount of money spent on advertising (in millions of dollars) and e is the number of employees working for the company. (Source: Modeled from Kellogg Company 2001 Annual Report, p. 26.) In 2001, the company had 26,424 employees, spent $519.2 million on advertising, and had $8853.3 million in net sales. If the company were to increase advertising spending to $700 million and decrease the number of employees to 25,000, what would be the predicted net sales?
SOLUTION We are asked to calculate S( 700, 25,000 ) .
S( 700, 25,000 ) 5 2 0.36661( 700 ) 1 0.16556( 25,000 ) 1 4671.3 5 8553.7 Net sales are predicted to be $8553.7 million when advertising spending is $700 million and the number of employees is 25,000. Lets look a little more closely at the coefficients of the model equation. The units of the output of the function are millions of sales dollars. Consequently, the units of each term in the sum must also be millions of sales dollars. The units of a are millions of advertising dollars. What are the units of the coefficient 0.36661? The units are sales dollars per advertising dollar, since sales dollars ? millions of advertising dollars 5 millions of sales dollars advertising dollar According to the model, for every advertising dollar, sales decrease by 0.36661 million dollars ($366,610). What are the units of the coefficient 0.16556? The units are millions of sales dollars per employee, since
523
millions of sales dollars ? employees 5 millions of sales dollars employee According to the model, adding an employee will increase sales by 0.16556 million dollars ($165,560). Although the coefficients of the model indicate that increasing advertising spending will decrease net sales, we know that if Kellogg stops advertising altogether ( a 5 0 ) , its net sales will eventually decrease, not increase. However, since the coefficient of a in the model is negative, the model does seem to suggest that the company might want to look more closely at its advertising expenditures to ensure that it is getting the desired return on its investment. The Cobb-Douglas production function, f ( L, C ) 5 kLmC n is widely used in economics. The model assumes that the output of a company, industry, or country is a function of its labor L and its capital C. (Capital may be interpreted as the dollar value of the money invested in the company, including equipment, material, and building costs.) The model requires that m, n, and k are positive constants and m 1 n 5 1.
EXAMPLE
Using a Cobb-Douglas Production Model

Based on data from 2000 and 2001, a Cobb-Douglas production model for the Coca-Cola Company is given by P( L, C ) 5 0.02506L0.0481 C 0.9519 billion unit cases of liquid beverage where L is the labor cost (including sales, payroll, and other taxes) in millions of dollars and C is the cost of capital expenditures (property, plant, and equipment) in millions of dollars. One unit case of beverage equals 24 eight-ounce servings. (Source: Modeled from Coca-Cola Company 2000 and 2001 Annual Reports.) According to the model, would it be better to spend $800 million on capital expenditures and $200 million on labor costs or $700 million on capital expenditures and $300 million on labor costs? Explain.
SOLUTION We are asked to calculate P( 200, 800 ) and P( 300, 700 ) .
P( 200, 800 ) 5 0.02506( 200 ) 0.0481( 800 ) 0.9519 5 18.8 P( 300, 700 ) 5 0.02506( 300 ) 0.0481( 700 ) 0.9519 5 16.8 Spending $200 million on labor and $800 million on capital expenditures is expected to produce 18.8 billion unit cases of beverage, while spending $300 million on labor and $700 million on capital expenditures is expected to produce 16.8 billion unit cases of beverage. The model suggests that it may be better to spend $800 million on capital expenditures and $200 million on labor; however, there may be additional constraints to take into consideration. For example, if the Coca-Cola Company were to purchase additional equipment with its capital, it might need to hire additional people to operate the new equipment. Hiring additional people would result in a corresponding increase in the labor cost.
524
EXAMPLE
Using a Multivariable Function to Find the Surface Area of a Box

A box manufacturer designed a closeable box with a square base, as shown in Figure 8.9. Find the surface area equation for the unassembled box. How many square inches of material are required to construct a box 6 inches wide, 6 inches long, and 18 inches high?
Top flap 1
2h
Side 1 w w
Top flap 1
Top flap 2
Top flap 3
Top flap 4
2h
Side 1
Side 2
Side 3
Side 4
Bottom flap 1 w
Bottom flap 2 w
Bottom flap 3 w
Bottom flap 4 w
1 4
FIGURE 8.9
SOLUTION The surface area of the unassembled box is given by
A( h, w ) 5 ( h 1 2h 1 h )( w 1 w 1 w 1 w ) 1 1 5 ( 4h )( 4w ) 1 hw 2 5 16hw 1 1 hw 2
1 w( 2h ) 4
5 16.5hw square inches The height of the box is 2h. So 2h 5 18 h59 The width and length of the box are both 6 inches, so w 5 6. A( h, w ) 5 16.5( 9 )( 6 ) 5 891 The surface area of the unassembled box is 891 square inches.
525
8.1 Summary
In this section, you learned how to evaluate multivariable functions and saw how functions of two variables may be represented graphically. You also discovered some applications of multivariable functions.

1. Given A 5 P Q1 1 n R , determine the value of A when P 5 250, r 5 0.08, n 5 12, and t 5 4. 2. Given A 5 P Q1 1 n R , determine the value of A when P 5 1200, r 5 0.0425, n 5 365, and t 5 8. 3. Given A 5 P Q1 1 n R , determine the value of P when A 5 20,000, r 5 0.0842, n 5 4, and t 5 14. 4. Given A 5 P Q1 1 n R , determine the value of P when A 5 450, r 5 0.08, n 5 2, and t 5 3. 5. Given V 5 pr 2h, determine the volume V of a cylinder with radius r 5 4 inches and height h 5 5 inches.
r
nt
nt
nt
6. Given V 5 pr 2h, determine the volume V of a cylinder with radius r 5 9 inches and height h 5 20 inches. 7. An equation for the volume of a rectangular box is V 5 lwh. What is the volume of a box that is 18 inches long, 5 inches wide, and 4 inches high? 8. Given V 5 pr 2h, determine the volume V of a cylinder with radius r 5 9 feet and height h 5 22 feet.
nt
8.1 Exercises
In Exercises 110, generate a table of data for the multivariable function using x 5 2 2, 0, 2 and y 5 2 1, 0, 1. Then use a graphing utility to graph the function. Compare the coordinates of the points in the table to the coordinates of the points on the graph of the function. 1. z 5 3x 2 2y 3. z 5 4xy 1 4 4. f ( x, y ) 5 x 2 1 2xy 1 y 2 5. f ( x, y ) 5 e2xy 6. f ( x, y ) 5 x 2 2 2xy 1 y 2 x2 1 1 7. f ( x, y ) 5 2 y 11 9. z 5 ( x 1 y ) 3 xy 8. f ( x, y ) 5 2 y 11 10. z 5 ln( x 2 y 2 1 1 ) 2. z 5 xy 2 y 2 In Exercises 1120, use the multivariable function to answer the given question. 11.
Cobb-Douglas Function Based on data from 1999 and 2000, a Cobb-Douglas model for Ford Motor Company is given by P( L, C ) 5 0.5319L0.4558 C 0.5442
thousand vehicles, where L is the total labor cost and C is the capital expenditure cost, both in millions of dollars. In 2000, Ford Motor Company spent $25,783 million on labor and produced 7424 thousand vehicles. (Source: Modeled from Ford
Motor Company 2001 Annual Report.)
According to the model, if Ford spends $20,000 million on labor and $9000 million on capital expenditures, how many vehicles will it produce?
526 12.
Housing Costs The U.S. Department of Housing and Urban Development (HUD) published a research study in January 2002 comparing the cost of building a steel-frame home to the cost of building a wood-frame home. Two homes with identical floor plans were built side by side in Beaufort, South Carolina. The material and labor costs of constructing each home are given in the following table.
closed.) What is the exterior surface area of the box with a 6-inch width, 6-inch length, and 18-inch height? 16. Surface Area Analysis In Example 6 we calculated the surface area of an unassembled box, and in Exercise 15 you calculated the exterior surface area of the assembled box. Of what practical value are each of these results? 17.
Body Mass Index The Body Mass Index (BMI) is a helpful indicator of obesity or underweight in adults. A high BMI is predictive of death from cardiovascular disease. The Centers for Disease Control and Prevention indicate that a healthy BMI for adults is between 18.5 and 24.9. They use the following guidelines.
Material Cost Steel Frame Wood Frame

Source: HUD.
Labor Cost $32,228.94 $25,236.83
$97,447.52 $91,589.38
Find an equation for the total cost of constructing w identical wood-frame homes and s steel-frame homes in Beaufort, South Carolina. How much will it cost to construct ten woodframe and five steel-frame homes? 13.
Personal Data Assistant Software SplashData is a leading
Underweight Overweight Obese
BMI less than 18.5 BMI of 25.0 to 29.9 BMI of 30.0 or more
developer of virtual wallet applications for Palm OS handhelds. As of December 2002, SplashData offered consumers six products: SplashWallet Suite ($39.95), SplashClock ($5.00), SplashID ($19.95), SplashMoney ($14.95), SplashPhoto ($19.95), and SplashShopper ($19.95). (Source:
www.splashdata.com.)
The following formula is used to calculate body mass index. 703 ? W BMI 5 H2 where W is your weight in pounds and H is your height in inches. What is your BMI? 18.
Relative Humidity The relative humidity of the air alters our perception of the temperature. As the humidity increases, the air feels hotter to us. The heat index (apparent temperature) is a function of air temperature and relative humidity, as shown in the table on p. 527. When exposed to apparent temperatures exceeding 105F for prolonged periods of time, people are at high risk of sunstroke, a potentially fatal condition. The heat index values were devised for shady, light wind conditions. Exposure to full sunshine can increase apparent temperature by up to 15F. What is the apparent temperature when the air temperature is 90F and the relative humidity is 80 percent?
Find the formula for the SplashData revenue function. How much revenue will SplashData generate if it sells 20 SplashWallet Suites, 50 SplashClocks, 10 SplashIDs, 100 SplashMoneys, 40 SplashPhotos, and 70 SplashShoppers? 14. Future Value of Investment Find the equation for the future value of an investment of P dollars into an account paying 6 percent interest compounded monthly t years after the investment is made. What is the value of a $2500 investment after 10 years? 15. Surface Area Referring to Example 6 in the section, find a formula for the exterior surface area of the assembled box with a square base. (Assume that the top and bottom of the box are
527
Relative Humidity (%) 40 110 108 106 104 136 130 124 119 114 109 105 101 97 94 91 88 85 83 81 80 137 130 124 119 114 109 104 100 96 93 89 87 84 82 80 137 131 124 118 113 108 103 99 95 91 88 85 83 81 137 130 124 117 112 106 101 97 93 89 86 84 81 137 129 123 116 110 105 100 95 91 88 84 82 136 128 121 114 108 103 98 93 89 85 82 134 126 119 112 106 100 95 90 86 83 132 124 116 109 103 97 92 88 84 129 121 113 106 100 94 89 84 135 126 117 110 102 96 90 85 131 122 113 105 98 91 86 127 117 108 100 93 86 132 121 112 103 95 87 Heat Index (Apparent Temperature) 45 50 55 60 65 70 75 80 85 90 95 100
Air Temperature ( F) 19.
102 100 98 96 94 92 90 88 86 84 82 80
Source: National Weather Service.
Wind alters our perception of the temperature. As the wind strength increases, the air feels colder to us. This phenomenon is referred to as wind chill. Wind chill temperature is a function of wind speed and air temperature and is modeled by
Wind Chill
W( T, V ) 5 35.74 1 0.6215T 2 35.75V 0.16 1 0.4275TV 0.16 degrees Fahrenheit, where T is the actual air temperature (in degrees Fahrenheit) and V is the velocity of the wind (in miles per hour). (Source: National Climatic Data Center.) When the air is 10F and the wind is blowing at 40 mph, what is the wind chill temperature? 20.
Ice Cream Cones The GE-500 SemiAutomatic is a machine that is designed
to create a variety of sizes of cones, ranging from a diameter of 20 millimeters (mm) to a diameter of 56 mm. The number of cones that can be made in an hour depends upon the diameter of the cone. The production rate of 20-mm cones is 4400 per hour, while the production rate of 42-mm cones is 2000 per hour. (Source: www.maneklalexports.com.) Let s be the number of hours spent producing 20-mm cones and l be the number of hours spent producing 42-mm cones. Find a formula for the cumulative number of cones produced as a function of s and l. How many cones are produced if 1.5 hours are spent making 20-mm cones and 3 hours are spent making 42-mm cones?
528
21. What is a multivariable function? 22. Why is it especially important to identify the meaning of every variable when working with a multivariable function? 23. How many independent variables and how many dependent variables may a multivariable function have? 24. Given f ( x, y ) 5 x 2 2 y 2, what does the equation f ( 1, y ) 5 1 2 y 2 tell us about the graph of f ? 25. Given f ( x, y ) 5 x 2 2 y 2, what does the equation f ( x, 1 ) 5 x 2 2 1 tell us about the graph of f ?
26. What is the domain of the multivariable function? "x 2 2 1 f ( x, y ) 5 y 27. What are the domain and range of the multivariable function? f ( x, y ) 5 3 1 "x 2 2 1 "y 2 2 4 28. The amount of a persons take-home pay is a multivariable function. Identify four different variables that contribute to the amount of a persons take-home pay.
8.2 Partial Derivatives
Analyze and interpret multivariable mathematical models Calculate first- and second-order partial derivatives of multivariable functions Find the equation of a cross section of a graph of a surface Determine the meaning of partial rates of change
The wind chill temperature is a function of the air temperature and the wind velocity. If the wind speed is increasing at a rate of 2 mph per hour, how quickly is the wind chill temperature changing? Questions such as these may be answered by using partial derivatives. In this section, we will demonstrate how to calculate partial derivatives. These partial rates of change allow us to determine the effect of a change in one of the input variables on the output.
Recall that the operator operator
d means take the derivative with respect to x. The dx
' means take the partial derivative with respect to x. When calculating 'x a partial derivative, all variables except the variable of differentiation are treated as constants. Consider the function z 5 x 2 2 xy. ' ' ( z ) 5 ( x 2 2 xy ) 'x 'x 'z 5 2x 2 y 'x
529
Observe that we treated the y in xy as if it were a constant. That is, when differentiating x 2 2 xy, we treated the expression as if it were x 2 2 cx (where c is a cond 2 'z stant). Since ( x 2 cx ) 5 2x 2 c, 5 2x 2 y. Lets now take the partial dx 'x derivative of the function with respect to y. ' ' ( z ) 5 ( x 2 2 xy ) 'y 'y 'z 5 2x 'y Since the variable of differentiation is y, we treated x as a constant. That is, when differentiating x 2 2 xy, we treated the expression as if it were c2 2 cy (where d 2 ' 2 c is a constant). Since ( c 2 cy ) 5 2 c, ( x 2 xy ) 5 2 x. Observe that dy 'y 'z 'z 2 . The partial derivative of a function varies based upon the variable of 'x 'y differentiation.
PARTIAL DERIVATIVES
The partial derivatives of a function f ( x, y ) are given by 'f (read the partial of f with respect to x) 'x 'f (read the partial of f with respect to y) 'y 'f To calculate , differentiate the function treating the y variable as a constant. 'x 'f To calculate , differentiate the function treating the x variable as a constant. 'y
EXAMPLE
Finding Partial Derivatives

Find the partial derivatives of f ( x, y ) 5 3xy 1 y 2 1 5.
SOLUTION
'f ' 5 ( 3xy 1 y 2 1 5 ) 'x 'x 5 d ( 3cx 1 c2 1 5 ) dx

c is a constant representing y
5 3c 5 3y
Replace c with y
530
Replacing the variable y with the constant c is an optional step used to help you apply the partial differentiation rules correctly. As you become skilled at applying the rules, the step will become unnecessary. 'f ' 5 ( 3xy 1 y 2 1 5 ) 'y 'y 5 d ( 3cy 1 y 2 1 5 ) dy
c is a constant representing x
5 3c 1 2y 5 3x 1 2y
Replace c with x
PARTIAL DERIVATIVES (ALTERNATIVE NOTATION)
The partial derivatives of a function f ( x, y ) may alternatively be written as fx (read the partial of f with respect to x) fy (read the partial of f with respect to y)
EXAMPLE
Finding Partial Derivatives

Find the partial derivatives of f ( x, y ) 5 x 2 y 2.
SOLUTION
fx 5 5
' 2 2 (x y ) 'x d 2 2 (x c ) dx
c is a constant representing y
5 2xc2 5 2 xy 2 fy 5 5 ' 2 2 (x y ) 'y d 2 2 (c y ) dy

c is a constant representing x Replace c with y
5 2c2y 5 2x 2y
Replace c with x
The units of
'f are the units of f divided by the units of y. 'y
'f are the units of f divided by the units of x. Similarly, the units of 'x
531
EXAMPLE
Interpreting the Meaning of the Partial Derivatives of the Compound Interest Function
Find A( r, t ) 5 1000 A 1 1 the partial
r 4
( 0.12, 3, 1425.76 ) and interpret the meaning of the result.

SOLUTION We will first find the partial derivatives.
derivatives
4t
of
the
compound
interest
function
. Then evaluate the partial derivatives at the point
' r 4t ' ( A ) 5 c 1000 a 1 1 b d 't 't 4 'A r r 4t 5 1000 ? ln a 1 1 b ? a 1 1 b ? 4 't 4 4 r r 4t At 5 4000 ln a 1 1 b ? a 1 1 b 4 4 'A r 4t 2 1 1 5 1000 ? 4t a 1 1 b ? a b 'r 4 4 r 4t 2 1 Ar 5 1000t a 1 1 b 4 Next, we will evaluate each of the partial derivatives at the point (0.12, 3, 1425.76). At( 0.12, 3 ) 5 4000 ln a 1 1 5 168.58 The units of the partial derivative are the units of the output (dollars) divided by the units of t (years). An investment account earning 12 percent interest (compounded quarterly) with an initial investment of $1000 is increasing in value at a rate of $168.58 per year at the end of the third year. That is, from the end of the third year to the end of the fourth year, the account value will increase by about $168.58. r 4t 2 1 Ar 5 1000t a 1 1 b 4 Ar ( 0.12, 3 ) 5 1000( 3 ) a 1 1 5 3000 ( 1.03 ) 11 5 4152.70 The units of the partial derivative are the units of the output (dollars) divided by the units of r (100 percentage points). (Observe that 0.12 ? 100 percentage points 5 12%.) 4152.70 dollars 5 41.527 dollars per percentage point 100 percentage points 0.12 4(3) 2 1 b 4 0.12 0.12 4(3) b ? a1 1 b 4 4
Exponential, Constant Multiple, and Chain Rules
Power, Constant Multiple, and Chain Rules
5 4000 ? ln( 1.03 )( 1.03 ) 12
An investment account earning 12 percent interest per year (compounded quarterly) with an initial investment of $1000 is increasing in value at a rate of $41.53
532
per percentage point increase in the interest rate at the end of the third year. That is, increasing the interest rate from 12 percent per year to 13 percent per year at the end of the third year will increase the account value at the end of the fourth year by approximately $41.53 over what it would have been without the interest rate increase.
In general, x 2n 5 x n and x m>n 5 ! x m . Consider the following examples: 1 1. ( x 2 1 7x ) 23 5 2 ( x 1 7x ) 3

1
n 3 2. ( 3x 2 1 4 ) 2>3 5 "( 3x 2 1 4 ) 2
Just In Time
Fractional and Negative Exponents
3. ( x 2 1 y 2 ) 21>2 5
"x 1 y 2
2
EXAMPLE
Interpreting the Meaning of Partial Derivatives

The distance of a cue ball from the eight ball in the game of billiards may be expressed in terms of two perpendicular components parallel to the side or end of the pool table, respectively, as shown in Figure 8.10.
8 End of table d in.
y in.
x in. Side of table
FIGURE 8.10
By the Pythagorean Theorem, d 5 "x 2 1 y 2. Find the partial derivative of the distance function. Then evaluate each partial derivative at the point ( 3, 4, 5 ) and interpret the meaning of the result.
SOLUTION We will first determine the equations of the partial derivatives.
d 5 "x 2 1 y 2
5 ( x 2 1 y 2 ) 1>2
Chain Rule treating y as a constant
1 dx 5 ( x 2 1 y 2 ) 21>2( 2x ) 2 5 x ( x 2 1 y 2 ) 21>2 1 dy 5 ( x 2 1 y 2 ) 21>2( 2y ) 2 5 y ( x 2 1 y 2 ) 21>2
Chain Rule treating x as a constant
533
Evaluating these functions at ( 3, 4, 5 ) , we determine dx( 3, 4 ) 5 ( 3 )( 32 1 42 ) 21>2 5 3 5
5 0.6 dy( 3, 4 ) 5 ( 4 )( 32 1 42 ) 21>2 5 4 5
5 0.8 When the cue ball is five inches away from the eight ball, rolling toward the eight ball one inch (along the path depicted in Figure 8.10) will change the cue balls distance from the end of the pool table by 0.6 inch and its distance from the side of the pool table by 0.8 inch.
EXAMPLE
Using Partial Derivatives to Forecast Changes in the Wind Chill Temperature

Wind chill temperature is a function of wind speed and air temperature and is modeled by W( T, V ) 5 35.74 1 0.6215T 2 35.75V 0.16 1 0.4275TV
0.16
degrees Fahrenheit, where T is the actual air temperature (in degrees Fahrenheit) and V is the velocity of the wind (in miles per hour). (Source: National Climatic Data Center.) If the wind speed is 20 mph, how much will a 1-mph change in wind speed alter the wind chill temperature if the current temperature remains at 30 degrees? If the current temperature is 30 degrees, how much will a 1-degree increase in temperature alter the wind chill temperature if the current wind speed remains at 20 mph?
SOLUTION We begin by calculating the partial derivatives of W.
W( T, V ) 5 35.74 1 0.6215T 2 35.75V 0.16 1 0.4275TV 0.16 degree of wind chill temperature 'W 5 0.6215 1 0.4275V 0.16 'T degree of actual temperature 'W 5 0.16 ( 2 35.75 ) V 20.84 1 0.16( 0.4275TV 20.84 ) 'V 5 2 5.72V 20.84 1 0.0684TV 20.84 5 0.0684V 20.84( 2 83.63 1 T ) degree of wind chill temperature miles per hour
We will now evaluate each of the partial derivatives at ( 30, 20 ) . 'W ( 30, 20 ) 5 0.6215 1 0.4275( 20 ) 0.16 'T 5 1.312 degree of wind chill temperature degree of actual temperature
534
Increasing the temperature from 30 to 31 degrees will increase the wind chill temperature by about 1.3 degrees if the wind remains at 20 mph. (Since the partial derivative does not contain the variable T, the wind chill temperature will increase by about 1.3 degrees for every 1 degree increase in temperature regardless of the initial temperature.) 'W ( 30, 20 ) 5 0.0684V 20.84 ( 2 83.63 1 T ) 'V 5 0.0684( 20 ) 20.84( 2 83.63 1 30 ) 5 2 0.2962 degree of wind chill temperature miles per hour
Increasing the wind speed from 20 to 21 miles per hour will decrease the wind chill temperature by about 0.3 degree when the temperature is 30 degrees. The colder the temperature is, the more substantial the impact a change in wind speed will have on the wind chill temperature.
Cross Sections of a Surface

When an apple is sliced in half, the exposed surface is called a cross section of the apple. If the cross section is dipped in paint and pressed on a sheet of paper, the resultant image is a two-dimensional figure (see Figure 8.11).
FIGURE 8.11
The shape of the cross section varies depending upon the point at which the apple is sliced. Figure 8.12 shows three different cross sections of the same apple.
FIGURE 8.12
535
The equation for the border of the cross section can be derived from the original multivariable equation. (The shape of an apple may be modeled by a multivariable piecewise equation. However, because of the complexity of the equation, we will focus on simpler models as we analyze cross sections.) Consider the graph of the function f ( x, y ) 5 2 x 3 1 4y 2 , shown in Figure 8.13.
z z
x x y
FIGURE 8.13
If we want to know what the cross section of the surface looks like at y 5 0 , we calculate f ( x, 0 ) . The equation of the cross-section graph is given by f ( x, 0 ) 5 2 x 3 1 4 ( 0 ) 2 5 2 x3 Slicing the surface at y 5 0 yields the graph shown in Figure 8.14.
z
z
x x y
z 25 20 15 10 5 3 2 1 5 10 15 20 25 1 2 3 x
FIGURE 8.14
Note: Looking at the graph from this angle, the x-values go from positive to negative as we read the graph from left to right.
Graphing the cross section in two dimensions yields the curve z 5 2 x 3 (Figure 8.15). dz The slope of the cross-section graph is 5 2 3x 2. What is the relationship dx between the slope of the cross-section graph and the partial derivative? Lets find fx and evaluate it at ( x, 0 ) . fx 5 2 3x 2
FIGURE 8.15
536
Observe that fx ( x, 0 ) 5 2 3x 2 . The slope equation for the cross-section graph is the same as the equation of the partial derivative of f with respect to x evaluated at ( x, 0 ) . Lets look at another cross section and see if the relationship holds true. This time well slice the surface at y 5 2 (see Figure 8.16).
z z
x x y
FIGURE 8.16
Note: Looking at the graph from this angle, the x-values go from positive to negative as we read the graph from left to right.
z 45 40 35 30 25 20 15 10 5 3 2 1 5 10 1 2 3 x
The equation of the cross-section graph when y 5 2 is given by f ( x, 2 ) . f ( x, 2 ) 5 2 x 3 1 4( 2 ) 2 5 2 x 3 1 16 Graphing the cross-section graph in two dimensions yields the graph shown in Figure 8.17. dz The slope of the cross-section graph is 5 2 3x 2. We find fx and evaluate dx it at ( x, 2 ) . fx 5 2 3x 2 Observe that fx ( x, 2 ) 5 2 3x 2. The slope equation for the cross-section graph is the same as the equation of the partial derivative of f with respect to x evaluated at ( x, 2 ) .
FIGURE 8.17
GEOMETRIC INTERPRETATION OF PARTIAL DERIVATIVES
The partial derivatives of a function f ( x, y ) represent rates of change of the graph of the surface. f ( x, y ) is the rate of change of the graph of the function f ( x, y ) when y is x held constant. It is referred to as the slope in the x-direction. f ( x, y ) is the rate of change of the graph of the function f ( x, y ) when x is y held constant. It is referred to as the slope in the y-direction.
537
Consider the graph of f ( x, y ) 5 2 x 2 1 4x 2 y 2 1 y 1 xy 1 100 (see Figure 8.18). The point ( 3, 1, 106 ) lies on the graph.
z
(3, 1, (3 1 1 10 106) )
FIGURE 8.18
Lets calculate the partial derivatives. fx( x, y ) 5 2 2x 1 4 1 y fy( x, y ) 5 2 2y 1 1 1 x Recall that fx( x, y ) is the slope in the x-direction. Thus, fx( 3, 1 ) is the slope of the graph in the x-direction at the point ( 3, 1, 106 ) . fx( x, y ) 5 2 2x 1 4 1 y fx( 3, 1 ) 5 2 2( 3 ) 1 4 1 1 5 21 There is a line tangent to the graph at ( 3, 1, 106 ) that has slope 2 1. The line is of the form z 5 2 1x 1 b, since the line is in the x-direction. Substituting in x 5 3 and z 5 106, we can find the exact equation of the line. z 5 2 1x 1 b 106 5 2 3 1 b b 5 109 So z 5 2 x 1 109 The equation of the tangent line to the graph of f in the x-direction at the point ( 3, 1, 106 ) is z 5 2 x 1 109. We draw the line on the graph of f (see Figure 8.19).
z z = x + 109
(3, (3, 3 1, 1, 106) 06 0 6) )
FIGURE 8.19
538
Recall that fy( x, y ) is the slope in the y-direction. Thus, fy ( 3, 1 ) is the slope of the graph in the y-direction at the point ( 3, 1, 106 ) . fy( x, y ) 5 2 2y 1 1 1 x fy( 3, 1 ) 5 2 2( 1 ) 1 1 1 3 52 There is a line tangent to the graph at ( 3, 1, 106 ) that has slope 2. The line is of the form z 5 2y 1 b, since the line is in the y-direction. Substituting in y 5 1 and z 5 106, we can find the exact equation of the line. z 5 2y 1 b 106 5 2( 1 ) 1 b 106 5 2 1 b b 5 104 So z 5 2y 1 104 So the equation of the tangent line to the graph of f in the y-direction at the point ( 3, 1, 106 ) is z 5 2y 1 104. We draw the line on the graph of f (see Figure 8.20).
z = 2y +104 z z = x +109
(3, (3 3, 1 1, , 106 06) 6)
FIGURE 8.20
It is important to note that in three dimensions, the graph of the equation z 5 2y 1 104 is really the plane containing the green line that we have drawn, not the line itself. However, by requiring x 5 3 , we constrain the plane to the green line. Similarly, the graph of z 5 2 x 1 109 is the graph of the plane containing the blue line we have drawn, not the line itself. However, by requiring y 5 1 , we constrain the plane to the blue line. Without these additional constraints on x and y, respectively, the graph would look like Figure 8.21.
z = 2y +104 z z = x +109
(3, 1 1, , 106 106 6) )
FIGURE 8.21
539
EXAMPLE
Finding Tangent Lines of a Multivariable Function

Given the function f ( x, y ) 5 x 2 2 xy 2, determine the equations of the tangent lines in the x-direction and in the y-direction at the point ( 3, 2, 2 3 ) .
SOLUTION The rate of change of f in the x-direction is given by
fx( x, y ) 5 2x 2 y 2 Thus the slope of the tangent line in the x-direction at ( 3, 2, 2 3 ) is fx ( 3, 2 ) 5 2( 3 ) 2 ( 2 ) 2 52 Since y is being held constant, the equation will have the form z 5 2x 1 b. We find the value of b by substituting in the point ( 3, 2, 2 3 ) . z 5 2x 1 b 2 3 5 2( 3 ) 1 b b 5 29 So z 5 2x 2 9 Thus the equation of the tangent line in the x-direction (when y is constrained to y 5 2 ) is z 5 2x 2 9. The rate of change of f in the y direction is given by fy ( x, y ) 5 2 2xy Thus the slope of the tangent line in the y-direction at ( 3, 2, 2 3 ) is fy ( 3, 2 ) 5 2 2 ( 3 )( 2 ) 5 2 12 Since x is being held constant, the equation will have the form z 5 2 12y 1 b. We find the value of b by substituting in the point ( 3, 2, 2 3 ) . z 5 2 12y 1 b 2 3 5 2 12( 2 ) 1 b b 5 21 So z 5 2 12y 1 21 Thus the equation of the tangent line in the y-direction (when x is constrained to x 5 3 ) is z 5 2 12y 1 21. We can confirm our results graphically by using a graphing utility (see Figure 8.22).
540

z z = 12y + 21
x z = 2x 9
,2
3)
FIGURE 8.22
Second-Order Partial Derivatives

Recall that the derivative of a derivative of a function is called the second derivative of the function. Likewise, a partial derivative of a partial derivative of a function is called a second-order partial derivative.
SECOND-ORDER PARTIAL DERIVATIVES
The second-order partial derivatives of a function f ( x, y ) may be written as fxx 5 fyy 5 fxy 5 ' 'f a b (read the second partial of f with respect to x) 'x 'x ' 'f a b (read the second partial of f with respect to y) 'y 'y ' 'f a b (read the mixed second partial derivative of f with respect 'y 'x to x then y) ' 'f (read the mixed second partial derivative of f with respect to a b 'x 'y y then x)
fyx 5
For all functions with continuous first and second partial derivatives, fxy 5 fyx. For all multivariable functions in this text, fxy 5 fyx.
EXAMPLE
Finding First and Second Partial Derivatives

The graph of the function f ( x, y ) 5 x 2 1 y 2 1 1 is shown in Figure 8.23. Calculate the value of the first and second partial derivatives of the function at the point ( 0, 0, 1 ) .

z
541
FIGURE 8.23
SOLUTION We have
fx 5 2x fy 5 2y fxx 5 2 fyy 5 2 fxy 5 0 fyx 5 0 Evaluating each of the partial derivatives at ( 0, 0, 1 ) , we get fx ( 0, 0 ) 5 2( 0 ) 50 fy ( 0, 0 ) 5 2( 0 ) 50 The second partial derivatives remain constant for all values of x and y. Thus fxx( 0, 0 ) 5 2 fyy( 0, 0 ) 5 2 fxy( 0, 0 ) 5 0 fyx( 0, 0 ) 5 0
EXAMPLE
Finding First and Second Partial Derivatives

The graph of the function f ( x, y ) 5 x 3 2 xy 2 is shown in Figure 8.24. Calculate the first and second partial derivatives of the function.
FIGURE 8.24
542
SOLUTION
fx 5 3x 2 2 y 2 fy 5 2 2xy fxx 5 6x fyy 5 2 2x fxy 5 2 2y fyx 5 2 2y
In the next section, we will demonstrate the key role that second-order partial derivatives play in identifying the location of the maxima and minima of a surface.
8.2 Summary
In this section, you learned how to calculate partial derivatives. You saw that these partial rates of change allow you to determine the effect on the output of a change in one of the input variables.

1. Calculate f ( 3, 5 ) given f ( x, y ) 5 4x 2 3y 1 1. 2. Calculate f ( 2 1, 4 ) given f ( x, y ) 5 xy 2 x 1 y .
2 2
6. What is the equation of the line passing through ( 2 2, 2 4 ) and ( 3, 11 ) ? 7. What is the equation of the line with slope 4 passing through ( 1, 3 ) ? 8. What is the equation of the line with slope 2 9 passing through ( 20, 19 ) ?
3. What is the equation of the line with slope 4 passing through ( 3, 4 ) ? 4. What is the equation of the line with slope 2 3 passing through ( 2 2, 5 ) ? 5. What is the equation of the line passing through ( 4, 10 ) and ( 6, 2 ) ?
8.2 Exercises
In Exercises 110, calculate the first and second partial derivatives of each of the functions. 1. f ( x, y ) 5 3xy 3. f ( x, y ) 5 4xy 2 y 2 5. W 5 t "v 7. V 5 pr 2h 2. f ( x, y ) 5 x 2y 2 xy 2 4. f ( x, y ) 5 x 3 2 y 3 6. W 5 v 2t 1.2 1 v 8. V 5 lw 2 10. z 5 xy ln( y ) In Exercises 1117, calculate and interpret the practical meaning of each of the partial derivatives. 11. Volume of a Cylinder The volume of a cylinder is given by V( h, r ) 5 pr 2h, with h and r measured in inches. What is the practical meaning of Vr ( 3, 2 ) and Vh( 3, 2 )? 12. Volume of a Box The volume of a rectangular box with a square base is given by V( l, w ) 5 lw 2, with l and w measured in inches. What is the practical meaning of Vl ( 4, 5 ) and Vw( 4, 5 ) ?
9. z 5 x ln( xy )
543
13.
Wind Chill Temperature Wind chill temperature is a function of wind speed and air temperature and is modeled by
In Exercises 1820, use partial derivatives to determine the answer to each question. 18.
Price-to-Earnings Ratio The price-to-
W( T, V ) 5 35.74 1 0.6215T 2 35.75V 0.16 1 0.4275TV 0.16 degrees Fahrenheit, where T is the actual air temperature (in degrees Fahrenheit) and V is the velocity of the wind (in miles per hour). (Source: National Climatic Data Center.) What is the practical meaning of WT ( 27, 5 ) and WV ( 27, 5 )? 14.
Wind Chill Temperature Wind chill temperature is a function of wind speed and air temperature and is modeled by
earnings ratio (P/E) is the most common measure of how expensive a stock is, and is given by R( E, P ) 5 P E
W( T, V ) 5 35.74 1 0.6215T 2 35.75V 0.16 1 0.4275TV 0.16 degrees Fahrenheit, where T is the actual air temperature (in degrees Fahrenheit) and V is the velocity of the wind (in miles per hour). (Source: National Climatic Data Center.) What is the practical meaning of WT ( 2 30, 15 ) and WV ( 2 30, 15 ) ? 15.
Body Mass Index The following
where P is the price of a share of the stock and E is the earnings per share. On December 28, 2002, Electronic Arts stock was selling for $51.50 per share and had a published earnings per share of $0.71. (Source: www.quicken.com.) If the stock price increased by $1 or the earnings per share increased by $1, what effect would this have on the P/E ratio? 19.
Company Sales Based on data from 1998 to 2001, the sales of the Kellogg Company may be modeled by
S( a, e ) 5 2 0.36661a 1 0.16556e 1 4671.3 million dollars, where a is the amount of money spent on advertising (millions of dollars) and e is the number of employees. In 2001, the company spent $519.2 million on advertising and had 26,424 employees. (Source: Modeled from Kellogg
Company 2001 Annual Report, p. 26.)
formula is used to calculate the body mass index: 703 ? W B( H, W ) 5 H2 where W is weight in pounds and H is height in inches. What is the practical meaning of BH ( 67, 155 ) and BW ( 67, 155 ) ? 16. index: 703 ? W H2 where W is weight in pounds and H is height in inches. What is the practical meaning of BH ( 67, 100 ) and BW ( 67, 100 ) ? B( H, W ) 5 17.
Price-to-Earnings Ratio The price-toBody Mass Index The following formula is used to calculate body mass
According to the model, what effect would adding one employee in 2002 have on sales? What effect would increasing advertising spending by $1 million have on sales? 20.
Cobb-Douglas Model Based on data from 1999 and 2000, a Cobb-Douglas model for Ford Motor Company is given by P( L, C ) 5 0.5319L0.4558C 0.5442
earnings ratio (P/E) is the most common measure of how expensive a stock is, and is given by P E where P is the price of a share of the stock and E is the earnings per share. What is the practical meaning of RE ( 4, 32 ) and RP ( 4, 32 ) ? R( E, P ) 5
thousand vehicles, where L is the total labor cost and C is the capital expenditure cost, both in millions of dollars. In 2000, Ford Motor Company spent $25,783 million on labor and produced 7424 thousand vehicles. (Source: Modeled from Ford
According to the model, if Ford increased labor spending by $1 million or increased capital expenditures by $1 million in 2001, what effect would each change have on the production of vehicles?
544
In Exercises 2125, a function graph with a cross section drawn at x 5 0 and y 5 2 1 is shown. Determine the equations of the tangent lines of the function in the x-direction and the y-direction at the point ( 0, 2 1, f ( 0, 2 1 )) . 21. f ( x, y ) 5 x 4 2 4y 2
z
25. f ( x, y ) 5 ex 1 y
2
22. f ( x, y ) 5 x 2 2 xy 1 y 2
z
26. If h is the heat index in points, r is the relative humidity (%), and t is the air temperature in degrees Fahrenheit, then what are the units of 'h 'r 'h , , and ?
'r 't 't
27. If h is the heat index in points, r is the relative humidity (%), and t is the air temperature in degrees Fahrenheit, then what do represent in a real-world context? 28. What is the difference in meaning between and 'x 'z ? Explain.
'x ' 'h 'r , , 'r 't
and 't
'h
23. f ( x, y ) 5 e2xy
z
29. How are partial derivatives related to derivatives? 30. What does the partial derivative operator tell 'x you to do to 3x 2 2 xy 2 1 1 in the expression ' ( 3x 2 2 xy 2 1 1 ) ? Write your answer as if you 'x were explaining the concept to someone who understands derivatives but is unfamiliar with partial derivatives.
'
24. f ( x, y ) 5
x2y 2 x 1 y2 1 1
z
31. Determine the rate of change in the x-direction and the y-direction for the function f ( x, y ) 5
ex 2 ey e xy
at the point ( 2, 2, 0 ) .
32. Give an example of a function f ( x, y ) with the property that fx ( x, y ) . 0, fy ( x, y ) . 0, and fxy ( x, y ) 5 0 for every ( x, y ) in the domain of f.
y
33. Give an example of a function f ( x, y ) with the property that f ( x, y ) . 0, fxx ( x, y ) . 0, fyy ( x, y ) . 0, and fxy ( x, y ) 5 0 for every ( x, y ) in the domain of f.
8.3 Multivariable Maxima and Minima
545
34. A function f ( x, y ) 5 ax 2 1 bxy 1 cy 2 has the property that fxx ( x, y ) 5 4, fyy ( x, y ) 5 2 8, and fxy ( x, y ) 5 3. Determine the values of a, b, and c. Then determine the values of x and y for which fx ( x, y ) and fy ( x, y ) are simultaneously equal to 1.
35. A function f ( x, y ) 5 ax 2 1 bxy 1 cy 2 has the property that fx( 1, 2 ) 5 4, fy ( 1, 2 ) 5 2 12, and fxy ( 1, 2 ) 5 0. Determine the values of a, b, and c.
Analyze and interpret multivariable mathematical models Find critical points of multivariable functions Locate relative extrema and saddle points of twovariable function graphs
An investor has $3000 to invest. She wants to get a $10,000 return on her investment without an excessive amount of risk. To achieve her goal, she has decided to invest a portion of her money in a certificate of deposit paying 6 percent interest compounded annually and the rest of her money in an aggressive growth fund that is expected to earn a maximum of 16 percent annually. How much money should she invest in each account if she wants to assume a moderate amount of risk? Questions such as this can be investigated using the techniques associated with finding multivariable extrema and saddle points. In this section, we will demonstrate the techniques used to find relative extrema and saddle points of two-variable functions. The techniques used will be similar to those covered in our discussion of single-variable optimization in Chapter 4.
We will first define relative extrema of multivariable functions.
RELATIVE EXTREMA OF MULTIVARIABLE FUNCTIONS
Consider the graph of a surface with equation z 5 f ( x, y ). The value f ( a, b ) is a relative minimum if f ( a, b ) # f ( x, y ) for all ( x, y ) in some region surrounding ( a, b ) . a relative maximum if f ( a, b ) $ f ( x, y ) for all ( x, y ) in some region surrounding ( a, b ) .
Recall that relative extrema of single-variable functions could not occur at endpoints. Similarly, relative extrema of two-variable functions may not occur on the border of the domain region. To help you visualize what relative extrema in three dimensions look like, we will label the relative extrema on the graphs of several multivariable functions (see Figure 8.25).
546
f (x, y) = x 2 6x + y 2 4y + 13 z f (x, y) = x 2 +
z
2 e y
relative maximum (0, 0, 1)
y x
(3, 2, 0) relative minimum z
z relative maximum (0, 0, 2)
relative maximum (1, 1, 6)

7 1 98 , , ( 3 3 27 )
relative minimum x y x f (x, y) = 1 + 1 x2 + 1 y2 + 1 y
f (x, y) = x 3 5x 2 7x + y 3 + y 2 y + 2
FIGURE 8.25
As was the case in two dimensions, relative extrema have horizontal tangent lines provided that the extrema dont occur at a sharp point on the graph. In addition to relative extrema, a three-dimensional surface may have a saddle point. A saddle point is a point ( a, b, c ) that is a relative maximum of one cross-section graph and a relative minimum of another cross-section graph. The portion of the graph surrounding the saddle point looks like a saddle. Consider the graph of f ( x, y ) 5 2 x 2 1 y 2 1 2, shown in Figure 8.26.
z
saddle point (0, 0, 2)
y f (x, y) = x 2 + y 2 + 2 x
FIGURE 8.26
547
If we slice the graph with the plane x 5 0, we see a cross section that looks like a concave up parabola with a vertex at ( 0, 0, 2 ) (see Figure 8.27).
f (x, y) = x 2 + y 2 + 2 z
(0, 0, 2)
FIGURE 8.27
The smallest possible value of z (in the cross section) is reached at the vertex. Similarly, if we slice the graph with the plane y 5 0, we see a cross section that looks like a concave down parabola with a vertex at ( 0, 0, 2 ) (see Figure 8.28).
z f (x, y) = x 2 + y 2 + 2
(0, ( 0, 0 0, ,2 2) )
FIGURE 8.28
The largest possible value of z (in the cross section) is reached at the vertex. Since the same point is a maximum value of one cross section and a minimum value of a different cross section, the point ( 0, 0, 2 ) is a saddle point. Later in the section, we will discuss how to find saddle points using partial derivatives. Recall from our discussion of single-variable functions that a value c in the domain of f was called a critical value of f if f r( c ) 5 0 or f r( c ) was undefined. All relative extrema occurred at critical values. A similar definition is used for two-variable functions.
CRITICAL POINT OF A TWO-VARIABLE FUNCTION
An ordered pair ( c, d ) in the domain of f is a critical point of f if fx ( c, d ) 5 0 and fy ( c, d ) 5 0 (Both statements must be true in order for ( c, d ) to be a critical point.)
548
When we refer to the critical point ( c, d ) , we mean that ( c, d ) is the pair of input values whose output is f ( c, d ) . Lets find the critical point of f ( x, y ) 5 x 2 1 y 2 1 1. fx 5 2x fy 5 2y Setting each partial derivative equal to zero and solving, we find that fx 5 2x 0 5 2x x50 fy 5 2y 0 5 2y y50
Therefore, the ordered pair ( 0, 0 ) is a critical point of f. The domain region of the function is defined by 2 ` , x , ` and 2 ` , y , ` . Since ( 0, 0 ) is in the interior of the region, it is a relative extremum candidate. To determine whether a maximum or minimum occurs at a critical point, we will apply the Second Derivative Test for Two-Variable Functions.
SECOND DERIVATIVE TEST FOR TWO-VARIABLE FUNCTIONS (D-TEST)
To determine if a relative extremum of a function f occurs at a critical point ( c, d ) , first calculate D ( c, d ) 5 fxx ( c, d ) ? fyy ( c, d ) 2 3 fxy ( c, d ) 4 2
Then f has a relative maximum at ( c, d ) if D ( c, d ) . 0 and f ( c, d ) , 0. xx f has a relative minimum at ( c, d ) if D( c, d ) . 0 and f ( c, d ) . 0. xx f has a saddle point at ( c, d ) if D( c, d ) , 0. The test is inconclusive if D( c, d ) 5 0.
To apply the D-Test, we must first find the second partial derivatives. For the function f ( x, y ) 5 x 2 1 y 2 1 1, we have the first partial derivatives fx 5 2x and fy 5 2y. Therefore, fxx 5 2 fyy 5 2 fxy 5 0 We evaluate each of the second partial derivatives at the critical point ( 0, 0 ) . In this case, the second partials are constant for all values of x and y. Thus fxx ( 0, 0 ) 5 2 fyy ( 0, 0 ) 5 2 fxy ( 0, 0 ) 5 0 We now calculate D. D( 0, 0 ) 5 2 ? 2 2 02 54
549
Since D . 0, either a maximum or a minimum occurs at ( 0, 0 ) . Since fxx( 0, 0 ) . 0, a relative minimum occurs at ( 0, 0 ) . Evaluating f at ( 0, 0 ) , we determine that f ( 0, 0 ) 5 02 1 02 1 1 51 The relative minimum occurs at ( 0, 0, 1 ) .
EXAMPLE
Locating Relative Extrema and Saddle Points

Find the location of the relative extrema and saddle points of f ( x, y ) 5 x 2 1 y 2 1 xy 2 2. The graph of the surface is shown in Figure 8.29.
z
FIGURE 8.29
SOLUTION The partial derivatives of f are
fx 5 2x 1 y fy 5 2y 1 x We set each partial derivative equal to zero and solve. 0 5 2x 1 y y 5 2 2x 0 5 2y 1 x y 5 2 0.5x
Since both equalities must be true at a critical point, we set the functions equal to each other and solve. 2 2x 5 2 0.5x 1.5x 5 0 x50 Back substituting into the equation y 5 2 2x, we determine that y 5 2 2( 0 ) 50 The critical point of the function is ( 0, 0 ) . We now find the second partial derivatives. fxx 5 2 fyy 5 2 fxy 5 1
550
Since the second partials are constant, evaluating each second partial at the critical point will result in the same values. D( 0, 0 ) 5 2 ? 2 2 1 ? 1 53 Since D( 0, 0 ) . 0 and fxx . 0, a relative minimum occurs at ( 0, 0 ) . The value of the function at ( 0, 0 ) is f ( 0, 0 ) . f ( 0, 0 ) 5 ( 0 ) 2 1 ( 0 ) 2 1 ( 0 )( 0 ) 2 2 5 22 The relative minimum occurs at ( 0, 0, 2 2 ) . There are no relative maxima or saddle points.
EXAMPLE
Locating Relative Extrema and Saddle Points

Find the location of the relative extrema and saddle points of f ( x, y ) 5 e2xy. The graph of the surface is shown from two different viewpoints in Figure 8.30.
x y
FIGURE 8.30
SOLUTION Well first find the partial derivatives of f.
fx 5 2 ye2xy fy 5 2 xe2xy To find the critical points, well set each partial equal to zero and solve. 0 5 2 ye2xy We solve the equation by setting each factor to zero. 2y 5 0 y50 Since e2xy 2 0 for all values of x and y, the only value that makes the equation zero is y 5 0. Well now set fy equal to zero. 0 5 2 xe2xy e2xy 5 0
551
We solve the equation by setting each factor to zero. x50 e2xy 5 0 Since e2xy 2 0 for all values of x and y, the only value that makes the equation zero is x 5 0. Thus, the critical point of the function is ( 0, 0 ) . Well now find the second partials and calculate the value of D at ( 0, 0 ) . fxx 5 y 2e2xy fyy 5 x 2e2xy fxy 5 2 1( e2xy ) 1 ( 2 xe2xy )( 2 y ) 5 2 e2xy( 1 2 xy ) fxx( 0, 0 ) 5 0 fyy( 0, 0 ) 5 0 fxy( 0, 0 ) 5 2 e(0)(0) 31 2 ( 0 )( 0 ) 4 5 21 D( 0, 0 ) 5 0 ? 0 2 ( 2 1 ) 2 5 21 Since D , 0, a saddle point occurs at the critical point ( 0, 0 ) . The coordinates of the saddle point are ( 0, 0, 1 ) . There are no relative extrema.
EXAMPLE
Finding Relative Extrema and Saddle Points

Find the location of the relative extrema and saddle points of f ( x, y ) 5 0.1x 2y 2. Two different views of the graph of the surface are shown in Figure 8.31.
z z
x x y
FIGURE 8.31
SOLUTION The partial derivatives of f are given by
fx 5 0.2xy 2 fy 5 0.2x 2y Setting the partial derivatives equal to zero and solving yields 0 5 0.2xy 2 0 5 xy 2 x50 or y50 0 5 0.2x 2y 0 5 x 2y x50 or y50
552
Any point of the form ( a, 0 ) or ( 0, b ) is a critical point of the function (a and b are real numbers). The function f has an infinite number of critical points. The second partial derivatives are given by fxx 5 0.2y 2 fyy 5 0.2x 2 fxy 5 0.4xy fyx 5 0.4xy D 5 ( 0.2y 2)( 0.2x 2 ) 2 ( 0.4xy ) 2 5 0.04x 2y 2 2 0.16x 2y 2 5 2 0.12 x 2y 2 Observe that D( a, 0 ) 5 0 and D( 0, b ) 5 0 for all values of a and b. The D-Test is inconclusive for all critical points. However, by evaluating the function at points near the critical points, we can determine what is happening there. Well generate a data table (Table 8.6) by increasing and decreasing each domain variable on either side of the critical points.
TABLE 8.6
x T 22 21 0 1 2
yS 22 1.6 0.4 0 0.4 1.6 21 0.4 0.1 0 0.1 0.4 0 0 0 0 0 0 1 0.4 0.1 0 0.1 0.4 2 1.6 0.4 0 0.4 1.6
The values highlighted in Table 8.6 are the function values of the critical points of f. Observe that as we move away from the highlighted values vertically or horizontally, the value of f increases or remains constant. Since f ( a, 0 ) # f ( x, y ) and f ( 0, b ) # f ( x, y ) for all values of a, b, x, and y, relative minima occur at all of the critical points. There are no relative maxima or saddle points.
EXAMPLE
Using Multivariable Functions to Forecast Investment Growth

An investor has $3000 to invest in two separate accounts. The first account is a certificate of deposit guaranteed to earn 6 percent annually. The second account is an aggressive growth fund expected to earn between 4 percent and 16 percent annually. The combined dollar value of the return on the investment accounts after 20 years is given by V 5 x( 1.06 ) 20 1 ( 3000 2 x )( 1 1 y ) 20 2 3000
553
where x is the amount of money invested in the certificate of deposit and y is the growth rate (as a decimal) of the aggressive growth fund. (If the growth rate is 8 percent, y 5 0.08.) A graph of the function is shown in Figure 8.32. The green plane is a graph of the function z 5 0 and is drawn to provide a point of reference. When the graph of V is above the green plane, the investor makes money. When the graph of V is below the green plane, the investor loses money.
z z
FIGURE 8.32
Find all critical points of V and determine the location of relative extrema and saddle points.
SOLUTION
Vx 5 ( 1.06 ) 20 2 ( 1 1 y ) 20 Vy 5 20( 3000 2 x )( 1 1 y ) 19 We set each partial derivative equal to zero. 0 5 ( 1.06 ) 20 2 ( 1 1 y ) 20 ( 1 1 y ) 20 5 ( 1.06 ) 20 1 1 y 5 1.06 y 5 0.06 0 5 20( 3000 2 x )( 1 1 y ) 19 3000 2 x 5 0 or ( 1 1 y ) 19 5 0 x 5 3000 y 5 21 Recall that in order to be a critical point, the ordered pair must satisfy both equations. Since 2 1 2 0.06, we throw out the value y 5 2 1. Thus the only critical point is ( 3000, 0.06 ) . Well now find the second partials. Vxx 5 0 Vyy 5 380( 3000 2 x )( 1 1 y ) 18 D 5 ( 0 ) 3380( 3000 2 x )( 1 1 y ) 184 2 32 20( 1 1 y ) 194 2 5 2 400( 1 1 y ) 38 D( 3000, 0.06 ) 5 2 400( 1 1 0.06 ) 38 D( 3000, 0.06 ) , 0
We only need to know the sign of D
Vxy 5 2 20( 1 1 y ) 19
554
Since D , 0, a saddle point occurs at ( 3000, 0.06, 6621.41 ) . However, from the context of the problem, we know that 0 # x # 3000. Therefore, this point sits on the border of the domain region. As with relative extrema, a saddle point may not sit on the border of the domain region. We conclude that the function does not have any relative extrema or saddle points on the domain region defined by 0 # x # 3000 and 2 0.04 # y # 0.16.
Lets look at the table of data for the function in the region surrounding the critical point given in Example 4 (Table 8.7).
TABLE 8.7
x T 2,600 2,700 2,800 2,900 3,000
yS 20.04 5,515.35 5,791.87 6,068.38 6,344.89 6,621.41 0.00 5,738.55 5,959.27 6,179.98 6,400.69 6,621.41 0.04 6,215.00 6,316.60 6,418.20 6,519.81 6,621.41 0.06 6,621.41 6,621.41 6,621.41 6,621.41 6,621.41 0.08 7,202.94 7,057.55 6,912.17 6,766.79 6,621.41 0.12 9,197.07 8,553.15 7,909.24 7,265.32 6,621.41 0.16 13,122.86 11,497.49 9,872.13 8,246.77 6,621.41
From the table, we see that investing all of the money in the certificate of deposit or earning exactly 6 percent on the amount of money invested in the aggressive growth account will yield a combined return of $6621.41 on the $3000 investment. As the amount of money invested in the certificate of deposit (x) decreases, the amount of the return becomes more volatile. If $2600 is invested in the certificate of deposit and $400 is invested in the aggressive growth account, the combined return could be as low as $5515.35 or as high as $13,122.86. Many investors are willing to risk losing $1106.06 ($6621.41 $5515.35) for the potential to earn an additional $6501.45 ($13,122.86 $6621.41). Investors who are willing to risk all $3000 on the aggressive growth account could lose as much as $1673.99 of their initial investment or earn as much as $55,382.28 on top of their initial investment. Most financial advisers encourage investors to diversify their investments among a variety of investment accounts with varying levels of risk in order to achieve a return that meets the investors goals.
EXAMPLE
Using Multivariable Functions to Forecast Product Sales

A software company produces two versions of its lead product: one for the Windows operating system and the other for the Macintosh operating system. Through market research, the company determines that at a price of p dollars, approximately q copies of the Windows version will be sold, and at a price of s dollars, approximately v copies of the Macintosh version will be sold. The relationship between the price of each item and sales may be modeled by q5 100,000 1 100 2 0.2 p p and v5 20,000 1 50 2 0.1s s
Determine the price the company should charge for each version in order to maximize the combined revenue from sales.
555
SOLUTION The revenue from the sale of q copies of the Windows version is
given by pq 5 p a 100,000 1 100 2 0.2p b p
5 100,000 1 100p 2 0.2p 2 The revenue from the sale of v copies of the Macintosh version is given by sv 5 s a 20,000 1 50 2 0.1s b s
5 20,000 1 50s 2 0.1s2 The combined revenue from the sales of the two products is given by R( p, s ) 5 100,000 1 100p 2 0.2p2 1 20,000 1 50s 2 0.1s2 5 100p 2 0.2p2 1 50s 2 0.1s2 1 120,000 From the graph of R on the region defined by 0 # p # 500 and 0 # s # 500 (see Figure 8.33), we can see that the revenue function has a relative maximum.
R R
s s p
FIGURE 8.33
To find the location of the relative maximum, we first find the partial derivatives. Rp 5 100 2 0.4p Rs 5 50 2 0.2s We then set each partial derivative equal to zero and solve. 0 5 100 2 0.4p 0.4p 5 100 p 5 250 Rpp 5 2 0.4 Rss 5 2 0.2 Rps 5 0 D( 250, 250 ) 5 ( 2 0.4 )( 2 0.2 ) 2 0 5 0.08 0 5 50 2 0.2s 0.2s 5 50 s 5 250
The critical point is ( 250, 250 ) . The second partial derivatives are
556
Since D . 0 and Rpp , 0, a relative maximum occurs at ( 250, 250 ) . To maximize revenue, the company should charge $250 for each version of the software. The maximum combined revenue is $138,750, with $112,500 coming from the Windows version and $26,250 coming from the Macintosh version.
EXAMPLE
Finding Relative Extrema and Saddle Points

Find the relative extrema and saddle points of f ( x, y ) 5 x 4 2 4x 3 1 4x 2 2 y 2. Two different views of the graph are shown in Figure 8.34.
z z
FIGURE 8.34
SOLUTION We will first find the critical points of f. We begin by finding the partial derivatives of f.
fx 5 4x 3 2 12x 2 1 8x fy 5 2 2y We then set each partial derivative equal to 0 and solve. 4x 3 2 12x 2 1 8x 5 0 4x( x 2 3x 1 2 ) 5 0
2
2 2y 5 0 y50
4x( x 2 1 )( x 2 2 ) 5 0 x 5 0, 1, 2 We must pair up each x-value with each y-value to form the critical points. The critical points are ( 0, 0 ) , ( 1, 0 ) , and ( 2, 0 ) . Next we will find the second partial derivatives and perform the D-Test on each critical point. fxx 5 12x 2 2 24x 1 8 5 4( 3x 2 2 6x 1 2 ) fyy 5 2 2 fxy 5 0 D 5 ( 12x 2 2 24x 1 8 )( 2 2 ) 2 ( 0 )( 0 ) 5 2 8( 3x 2 2 6x 1 2 ) When there are many critical points, it is often helpful to construct a table of data (see Table 8.8).
557
TABLE 8.8
Critical Point ( 0, 0 ) ( 1, 0 ) ( 2, 0 )
z
fxx 5 4( 3x 2 2 6x 1 2 ) 8 4 8
D 5 2 8( 3x 2 2 6x 1 2 ) 2 16 8 2 16
Graphical Interpretation Saddle point Relative maximum Saddle point
In order to understand better what a saddle point looks like, well zoom in on the region surrounding the critical point at ( 0, 0 ) (see Figure 8.35). Looking at the graph from the x z plane (Figure 8.36), it looks as if a relative minimum occurs at the critical point.
z
FIGURE 8.35
FIGURE 8.36
z
However, looking at the graph from the y z plane (Figure 8.37), it looks as if a relative maximum occurs at the critical point. When a graph has a saddle point, the point will look like a minimum from one plane and a maximum from the other plane. (The planes will not necessarily be the xz plane and y z plane.) To determine the actual coordinates of each of the extrema and saddle points, we evaluate the function at the critical points (see Table 8.9).
TABLE 8.9
FIGURE 8.37
x T 0 1 2
y 0 0 1 0
From the table, we see the relative maximum occurs at ( 1, 0, 1 ) . Saddle points occur at ( 0, 0, 0 ) and ( 2, 0, 0 ) . These results are consistent with our earlier conclusions.
558
8.3 Summary
In this section, you learned the techniques used to find relative extrema and saddle points of two-variable functions. In the next section, we will look at finding absolute extrema subject to function constraints.

In Exercises 18, solve for x. 1. x 2 2 5x 2 6 5 0 2. x 2 2 6x 1 8 5 0 3. 2x 2 2 2 5 0 4. 4x 2 2 24x 2 28 5 0 5. x 3 2 14x 2 1 24x 5 0 6. x 3 2 x 5 0 7. x 4 2 4x 2 5 0 8. 2xe x 5 0
8.3 Exercises
In Exercises 125, find the relative extrema and saddle points of the function. 1. f ( x, y ) 5 x 2 1 4y 2 2 10
z x y
3. f ( x, y ) 5 2 x 2 2 y 3 1 3y 2 x
2. f ( x, y ) 5 x 2 2 4xy
z
4. f ( x, y ) 5 x 4 2 4x 2 1 y 4 2 9y 2 1 1
x y x y
559
5. f ( x, y ) 5 x 2y 2 y 2 1 x
z
10. f ( x, y ) 5 2
x2 2 4 y2 1 4
z
y
4 2
6. f ( x, y ) 5 x 2 4x 1 y 2 9y 1 1
4
11. f ( x, y ) 5
z
x2 1 1 y2 1 1
z
y
3 2
2
7. f ( x, y ) 5 x 2 4x 1 y 2 y
4
12. f ( x, y ) 5 e2x 2 y
8. f ( x, y ) 5 x 3 2 2x 1 y 3 2 2y
z
13. f ( x, y ) 5
21 1 y 4 2 4y 2 x 11
2
9. f ( x, y ) 5 x 3 2 6x 1 xy 2 y 2
z
14. f ( x, y ) 5 e xy 2 4
x y z
560
15. f ( x, y ) 5 x 3 2 3x 1 y 3 2 12y
z
20. f ( x, y ) 5
(x 2 1)2 1 (y 2 2)2 e(x 2 1) 1 (y 2 2)

2 2
y x
16. f ( x, y ) 5 x 2 2 4x 1 y 2 2 y 2 xy
z
21. f ( x, y ) 5
x
x 22 y 2 e x 1y
2 2
z y
17. f ( x, y ) 5 x 2 x 2 3y 1 3y
2 2
22. f ( x, y ) 5
x2 2 4 21 y21 1
z
y x y
2
18. f ( x, y ) 5 ln( x 1 1 ) 1 ln( y 1 1 )

2 2
23. f ( x, y ) 5 e2x 1 2x 2 y 1 y 1 1
2
19. f ( x, y ) 5
x2 1 y2 1 1 e x 1y
2 2
24. f ( x, y ) 5 "x 2 2 2x 1 y 2 2 4y 1 5
z
x y
8.4 Constrained Maxima and Minima and Applications
561
25. f ( x, y ) 5 x 3y 1 xy 3 2 4xy
27. In order for an ordered pair ( a, b ) to be a critical point of a function f, what criteria must be met? 28. Explain how to determine if a critical point is a relative minimum of a two-variable function. 29. If the D-Test is inconclusive in determining whether a critical point is a relative extremum or a saddle point, how else may you determine the behavior of the graph at the critical point? 30. A classmate claims that if fxx ( a, b ) 5 fxy ( a, b ) 5 fyy ( a, b ) 5 0, then ( a, b ) is a critical point of f. Is your classmate correct? Explain.
26. Graphically speaking, what is the difference between a relative extremum and a saddle point?
Analyze and interpret multivariable mathematical models Use the Lagrange Multiplier Method to find constrained maxima and minima
Based on data from 1999 and 2000, a Cobb-Douglas model for Ford Motor Company is given by P( L, C ) 5 0.5319L0.4558C 0.5442 thousand vehicles, where L is the total labor cost and C is the capital expenditure cost, both in millions of dollars. In 2000, Ford Motor Company spent $25,783 million on labor and produced 7424 thousand vehicles. (Source: Modeled from Ford Motor Company 2001 Annual Report.) Assuming that Ford had $34,131 million to spend on labor and capital in 2000, how much should it have spent on labor and how much should it have spent on capital expenditures in order to maximize vehicle production? Questions such as this may be answered by solving a constrained optimization problem. (See Example 3.) In this section, we will demonstrate how to use the Lagrange Multiplier Method to solve constrained multivariable optimization problems. This technique may be used in two-variable and multivariable optimization problems. We will first demonstrate the technique with an example and then formally define the process.
562
EXAMPLE
Minimizing a Multivariable Function with the Lagrange Multiplier Method

Minimize f ( x, y ) 5 x 2 2 2y subject to the constraint 2 2x 1 y 5 4 (see Figure 8.38).
z f (x, y) = x 2 2y 2x + y = 4
x y
FIGURE 8.38
SOLUTION We are to minimize f ( x, y ) 5 x 2 2 2y subject to the constraint
2 2x 1 y 5 4. As shown in Figure 8.38, the graph of the constraint is a plane that cuts through the graph of f ( x, y ) 5 x 2 2 2y . We are to minimize f along the intersection of the graph of f and the graph of the constraint. We have 2 2x 1 y 5 4 2 2x 1 y 2 4 5 0 We will call the constraint function g ( x, y ) . That is, g ( x, y ) 5 2 2x 1 y 2 4 Observe that g ( x, y ) 5 0. We will now introduce a new function F( x, y, l ) . The third input variable of the function is called the Lagrange multiplier and is represented by the Greek letter l (read lambda). The function F( x, y, l ) is called the Lagrange function and is equal to the function to be maximized plus l times the constraint function. That is, F ( x, y, l ) 5 f ( x, y ) 1 lg ( x, y ) 5 x 2 2 2y 1 l ( 2 2x 1 y 2 4 ) 5 x 2 2 2y 2 2lx 1 ly 2 4l We will next take the partial derivative of F with respect to each of the input variables. Fx 5 2x 2 2l Fy 5 2 2 1 l Fl 5 2 2x 1 y 2 4
563
By design, Fl 5 g ( x, y ) 5 0. We will now set the other two partial derivatives equal to zero and solve for l. Fx 5 2x 2 2l 0 5 2x 2 2l 2l 5 2x l5x Now we set the lambdas equal to each other and solve. x52 So x 5 2 . We will substitute this result back into the constraint equation. g ( x, y ) 5 0 2 2x 1 y 2 4 5 0 2 2x 1 y 5 4 2 2( 2 ) 1 y 5 4 24 1 y 5 4 y58 The critical point of the objective function subject to the constraint is ( 2, 8 ). (Recall that the objective function is the function that is to be maximized or minimized.) It is important to note that the D-Test works only for unconstrained optimization and cannot be used to solve a constrained problem. Although the Lagrange Multiplier Method finds the critical points of the constrained objective function, it does not tell us whether maxima or minima (or neither) occur at these points. Lets construct a table of values for f at points near the critical point that also satisfy the constraint equation 2 2x 1 y 5 4 (see Table 8.10). Points satisfying the constraint equation nearby (2, 8) include (0, 4), (1, 6), (3, 10), and (4, 12).
TABLE 8.10
Fy 5 2 2 1 l 0 5 22 1 l l52
x 0 1 2 3 4
y 4 6 8 10 12
f ( x, y ) 28 211 212 211 28
The constrained minimum of f is ( 2, 8, 2 12 ) . The approach used in Example 1 is referred to as the Lagrange Multiplier Method. The steps of the method are detailed in the following box.
564
Lagrange Multiplier Method for Solving Constrained Optimization Problems To optimize an objective function f ( x, y ) subject to the constraint g( x, y ) 5 0, do the following: 1. Define F( x, y, l ) 5 f ( x, y ) 1 lg( x, y ) . 2. Set the partial derivatives of F equal to zero Fx 5 0 Fy 5 0 Fl 5 0
and solve to find the critical points. 3. The solution to the constrained optimization problem (if it exists) will occur at one of these critical points.
EXAMPLE
Maximizing a Multivariable Function with the Lagrange Multiplier Method

The United States Postal Service classifies a package as Parcel PostTM if its length plus its girth is less than or equal to 130 inches. (Source: U.S. Postal Service.) (The girth is the distance around the package at its thickest point.) What are the dimensions of the rectangular packing box with a square base that has the greatest volume and meets the requirements for Parcel Post?
SOLUTION We are looking for the dimensions of the rectangular box with maximum volume (see Figure 8.39).
w l
FIGURE 8.39
The volume of the box is a multivariable function given by V( l, w ) 5 lw 2 We have the constraint that the length plus the girth must not exceed 130 inches. A package with maximum volume will use all of the 130 inches allowed. Since the length of the package is l and the girth of the package is 4w, we have 130 5 l 1 4w 0 5 l 1 4w 2 130 We will call the constraint function g( l, w ) . That is, g( l, w ) 5 l 1 4w 2 130 Observe that g( l, w ) 5 0.
565
We will now find the Lagrange function F( l, w, l ) . Recall that it is equal to the function to be maximized plus l times the constraint function. That is, F( l, w, l ) 5 V( l, w ) 1 lg( l, w ) 5 lw 2 1 l ( l 1 4w 2 130 ) 5 lw 2 1 ll 1 4lw 2 130l We will next take the partial derivative of F with respect to each of the input variables. Fl 5 w 2 1 l Fw 5 2lw 1 4l Fl 5 l 1 4w 2 130 By design, Fl 5 g( l, w ) 5 0 . We will now set the other two partial derivatives equal to zero and solve for l. 0 5 w2 1 l l 5 2w
2
0 5 2lw 1 4l 4l 5 2 2lw l 5 2 0.5lw
Now we set the lambdas equal to each other and solve. 2 w 2 5 2 0.5lw w 2 2 0.5lw 5 0 w( w 2 0.5l ) 5 0 So w 5 0 or w 5 0.5l. We will substitute each of these results back into the constraint equation. g ( l, w ) 5 0 l 1 4w 2 130 5 0 l 1 4( 0 ) 2 130 5 0 l 5 130 We now back substitute these results to solve for w. g ( l, w ) 5 0 l 1 4w 2 130 5 0 130 1 4w 2 130 5 0 4w 5 0 w50 or g ( l, w ) 5 0 l 1 4w 2 130 5 0 43.33 1 4w 2 130 5 0 4w 5 86.67 w < 21.67 or g ( l, w ) 5 0 l 1 4w 2 130 5 0 l 1 4( 0.5l ) 2 130 5 0 3l 5 130 l < 43.33
Since it doesnt make sense to have a box with width 0 inches, we throw out the extraneous solution ( 130, 0 ) . The box with maximum volume is 43.33 inches long, 21.67 inches high, and 21.67 inches wide.
As stated before, the Lagrange Multiplier Method finds the critical points of the constrained objective function, but it does not tell us whether maxima or minima (or neither) occur at these points. However, in most applied problems, we can figure out the status of the points from the context of the problem. In Example 2,
566
we threw out the critical point (130, 0) because it didnt make sense to talk about the volume of a box with width 0. Since conceptually we knew that there had to be a box with maximum volume, we concluded that the other critical point would yield the optimal solution. Locating critical points is sometimes a challenge. However, approaching the task in a methodical way makes it easier. Consider the following technique for finding critical points.
Hints for Finding the Critical Points

1. Solve Fx 5 0 and Fy 5 0 for l. 2. Eliminate l by setting the resultant equations equal to each other. 3. Substitute the result from Step 2 into the equation g ( x, y ) 5 0 and solve
for x and y.
EXAMPLE
Using the Lagrange Multiplier Method in a Real-World Setting

Based on data from 1999 and 2000, a Cobb-Douglas model for Ford Motor Company is given by P( L, C ) 5 0.5319L0.4558C 0.5442 thousand vehicles, where L is the total labor cost and C is the capital expenditure cost, both in millions of dollars. In 2000, Ford Motor Company spent $25,783 million on labor and produced 7424 thousand vehicles. (Source: Modeled from Ford Motor Company 2001 Annual Report.) According to the model, how much should Ford have spent on labor and how much should it have spent on capital if its goal was to maximize production? According to its 2001 annual report, Ford had $34,131 million to spend on labor and capital in 2000. constraint L 1 C 5 34,131. We define g ( L, C ) 5 L 1 C 2 34,131. Then F ( L, C ) 5 P ( L, C ) 1 lg ( L, C ) 5 0.5319L0.4558C 0.5442 1 l( L 1 C 2 34,131 ) 5 0.5319L0.4558C 0.5442 1 lL 1 lC 2 34,131l Next we find the partial derivatives of F. FL 5 0.24244L20.5442C 0.5442 1 l FC 5 0.28946L0.4558C 20.4558 1 l Fl 5 L 1 C 2 34,131 We set the first two partials equal to zero and solve for l. 0 5 0.24244L 20.5442C 0.5442 1 l l 5 2 0.24244L 20.5442C 0.5442 0 5 0.28946L0.4558C 20.4558 1 l l 5 2 0.28946L0.4558C 20.4558
SOLUTION We want to maximize P ( L, C ) 5 0.5319L0.4558C 0.5442 subject to the
567
We then set the equations equal to each other and solve. 2 0.24244L20.5442C 0.5442 5 2 0.28946L0.4558C 20.4558 L0.4558 2 0.24244 C 0.5442 5 L20.5442 2 0.28946 C 20.4558 L 5 0.83756C We substitute this result into the third equation, Fl 5 L 1 C 2 34,131. Since Fl 5 g( L, C ) , Fl 5 0. Thus L 1 C 2 34,131 5 0 and L 1 C 2 34,131 5 0 0.83756C 1 C 2 34,131 5 0 1.83756C 5 34,131 C 5 18,574 We know that L 1 C 5 34,131; thus L 1 18,574 5 34,131 L 5 15,557 According to our model, when $15,557 million is invested in labor and $18,574 is invested in capital expenditures, vehicle production will be maximized. Since P( 15,557, 18,574 ) 5 9113, we conclude that a maximum of 9113 thousand vehicles could have been produced in 2000.
Since L0.83756C
EXAMPLE
Maximizing Volume with the Lagrange Multiplier Method

A box designer wants to construct a closeable box with a square base, as shown in Figure 8.40. The box will be used to store products in a warehouse.
Top flap 1
2h
Side 1 w w
Top flap 1
Top flap 2
Top flap 3
Top flap 4
2h
Side 1
Side 2
Side 3
Side 4
Bottom flap 1 w
Bottom flap 2 w
Bottom flap 3 w
Bottom flap 4 w
1 4
FIGURE 8.40
The designer requires that the sum of the height and width of the piece of cardboard used to make the box be at most 120 inches. What are the dimensions of the box with the maximum volume?
568
SOLUTION We are asked to maximize the volume of the box subject to the constraint. The volume of the box is
V( h, w ) 5 ( 2h )( w )( w ) 5 2hw 2 The constraint is given by ( h 1 2h 1 h ) 1 ( w 1 w 1 w 1 w 1 0.25w ) 5 120 4h 1 4.25w 5 120 4h 1 4.25w 2 120 5 0 Well call this function g( h, w ) . g( h, w ) 5 4h 1 4.25w 2 120 The Lagrange function F( h, w, l ) is given by F( h, w, l ) 5 V( h, w ) 1 lg ( h, w ) 5 2hw 2 1 l( 4h 1 4.25w 2 120 ) 5 2hw 2 1 4hl 1 4.25wl 2 120l Well now find the partial derivatives of F. Fh 5 2w 2 1 4l Fw 5 4hw 1 4.25l Fl 5 4h 1 4.25w 2 120 Observe that Fl 5 g ( h, w ) 5 0 . We will now set the other two partial derivatives equal to zero and solve for l. 0 5 2w 2 1 4l 4l 5 2 2w
2
0 5 4hw 1 4.25l 4.25l 5 2 4hw l52 52 4 hw 4.25 16 hw 17
l 5 2 0.5w 2 1 5 2 w2 2
We now set the equations equal to each other and solve. 1 16 2 w 2 5 2 hw 2 17 w2 5 w2 2 w Aw 2

32
32 hw 17
32 hw 5 0 17 32 hB 5 0 17
So w 5 0 or w 5 h. In the context of the problem, we know that w 5 0 will not 17 maximize the volume. So w 5 of g( h, w ) .
32 h. We 17
will substitute this result into the equation
569
4h 1 4.25 A
4h 1 4.25w 2 120 5 0 32 h B 2 120 5 0 17
4h 1 8h 5 120 12h 5 120 h 5 10
w5 5
32 h 17 32 ( 10 ) 17
< 18.82 Note that the box has height 2h and width w. The dimensions of the box with maximum volume are 18.82 inches 3 18.82 inches 3 20.00 inches. The volume of this box is 7084 cubic inches.
Although the problem in Example 4 could have been solved using optimization methods from Chapter 4, there is value in the Lagrange Multiplier Method, especially when working with functions of more than two variables. In Example 5, we will repeat the problem given in Example 4; however, this time we will allow the box to have a rectangular base instead of a square base.
EXAMPLE
Maximizing Volume with the Lagrange Multiplier Method

A box designer wants to construct a closeable box as shown in Figure 8.41. The designer requires that the sum of the height and width of the piece of cardboard used to make the box be 120 inches. What are the dimensions of the box with the maximum volume?
Top flap 1
2h
Side 1 l w
Top flap 1
Top flap 2
Top flap 3
Top flap 4
2h
Side 1
Side 2
Side 3
Side 4
Bottom flap 1 w
Bottom flap 2 l
Bottom flap 3 w
Bottom flap 4 l
1 4
FIGURE 8.41
570
SOLUTION We are to maximize V( h, l, w ) 5 2hlw subject to the constraint 4h 1 2l 1 2.25w 5 120. We have g( h, l, w ) 5 4h 1 2l 1 2.25w 2 120. Note that g( h, l, w ) 5 0. The Lagrange function is given by
F( h, l, w, l ) 5 V( h, l, w ) 1 lg( h, l, w ) 5 2hlw 1 l( 4h 1 2l 1 2.25w 2 120 ) 5 2hlw 1 4hl 1 2ll 1 2.25wl 2 120l The partial derivatives are Fh 5 2lw 1 4l Fl 5 2hw 1 2l Fw 5 2hl 1 2.25l Fl 5 4h 1 2l 1 2.25w 2 120 We will set the partial derivatives Fh, Fl , and Fw equal to 0 and solve each one for l. 0 5 2lw 1 4l 4l 5 2 2lw 1 l 5 2 lw 2 0 5 2hw 1 2l 2l 5 2 2hw l 5 2 hw 0 5 2hl 1 2.25l 2.25l 5 2 2hl 8 l 5 2 hl 9 We set the first two equations equal to each other to get h in terms of l. 1 2 lw 5 2 hw 2 1 2 lw 1 hw 5 0 2
1 w A2 l 1 hB 5 0 2
1
So w 5 0 or h 5 l. Since we are looking for the maximum volume, we throw 2 out w 5 0. Next we set the second two equations equal to each other to get w in terms of l. 8 2 hw 5 2 hl 9 8 hw 2 hl 5 0 9 8 h aw 2 lb 5 0 9 So h 5 0 or w 5 l. Since we are looking for the maximum volume, we throw 9 out h 5 0.
8
571
We substitute the values w 5 9 l and h 5 2 l into g( l, w, h ) 5 0 to solve for l. 1 8 4 A l B 1 2l 1 2.25 A l B 2 120 5 0 2 9 2l 1 2l 1 4h 1 2l 1 2.25w 2 120 5 0
18 l 2 120 5 0 9 6l 5 120 l 5 20
Then we back substitute to solve for w and h. w5 5 5 8 l 9 8 ( 20 ) 9 160 9
< 17.78 1 h 5 ( 20 ) 2 5 10 Recall that the height of the box is 2h. The dimensions of the box with maximum volume are 20 inches 3 17.78 inches 3 10 inches. The maximum volume is 7112 cubic inches.
Graphical Interpretation of Constrained Optimization Problems

Graphically speaking, a constraint slices the graph of the surface, creating a cross section. A constrained maximum is the largest value of the objective function when the domain is restricted to the cross section. Similarly, a constrained minimum is the smallest value of the objective function when the domain is restricted to the cross section. Using a graphing utility to graph the objective function and its constraint helps us to identify the location of constrained extrema visually. However, we will check our work using the Lagrange Multiplier Method. It is easy to overlook extrema when relying on graphical methods alone, as will be illustrated in Example 6.
EXAMPLE
Maximizing a Multivariable Function with the Lagrange Multiplier Method

Maximize f ( x, y ) 5 2 x 3 1 6x 2 2 y 2 1 10 10x 2 y 5 50. subject to the constraint
572
SOLUTION The graph of the function f is shown in Figure 8.42, together with the graph of the constraint equation.
f (x, y) = x 3 + 6x 2 y 2 + 10 z 10x y = 50
x y
FIGURE 8.42
The constrained maximum is the highest point along the curve of intersection between the two graphs. We let g( x, y ) 5 10x 2 y 2 50. The Lagrange function is F( x, y, l ) 5 f ( x, y ) 1 lg( x, y ) 5 2 x 3 1 6x 2 2 y 2 1 10 1 l( 10x 2 y 2 50 ) 5 2 x 3 1 6x 2 2 y 2 1 10 1 10xl 2 yl 2 50l The partial derivatives are Fx 5 2 3x 2 1 12x 1 10l Fy 5 2 2y 2 l Fl 5 10x 2 y 2 50 We set Fx and Fy equal to zero and solve for l. 0 5 2 3x 2 1 12x 1 10l 2 10l 5 2 3x 1 12x
2
0 5 2 2y 2 l l 5 2 2y
l 5 0.3x 2 1.2x
2
We set the equations equal to each other and solve for y in terms of x. 0.3x 2 2 1.2x 5 2 2y y 5 2 0.15x 2 1 0.6x Since g( x, y ) 5 0, 10x 2 y 2 50 5 0 10x 2 ( 2 0.15x 1 0.6x ) 2 50 5 0
2
0.15x 2 1 9.4x 2 50 5 0 15x 2 1 940x 2 5000 5 0 3x 2 1 188x 2 1000 5 0
573
We solve for x by using the Quadratic Formula. x5 5 5 5 2 b 6 "b2 2 4ac 2a
2 188 6 "47,344 6 2 188 6 217.6 6 or
2 ( 188 ) 6 "( 188 ) 2 2 4( 3 )( 2 1000 ) 2( 3 )
5 4.931 y 5 2 0.15x 2 1 0.6x
2 67.60 y 5 2 0.15x 2 1 0.6x 5 2 0.15( 2 67.60 ) 2 1 0.6( 2 67.60 ) 5 2 726.0
5 2 0.15( 4.931 ) 2 1 0.6( 4.931 ) 5 2 0.6886
The critical points are ( 4.931, 2 0.6886 ) and ( 2 67.60, 2 726.0 ) . Evaluating f at each of the critical points, we get f ( 4.931, 2 0.6886 ) 5 35.52 and f ( 2 67.60, 2 726.0 ) 5 2 190,732
The constrained maximum occurs at ( 4.931, 2 0.6886, 35.52 ) . We also discovered a constrained minimum. (The minimum is not shown in the original graph of the function.)
8.4 Summary
In this section, you learned how to use the Lagrange Multiplier Method to solve constrained multivariable optimization problems. You discovered that graphing utilities are helpful but not foolproof in identifying the location of constrained extrema.

1. Solve 2lw 1 4l 5 0 for l. 2. Solve 2hw 1 2l 5 0 for l. 3. Solve 2hl 1 2.25l 5 0 for l. 4. Solve
1 2 lw 4
5. Solve 2 hw 5 2 5hl for h. 6. Solve 2 w2 5 2 0.5lw for w. 7. Solve 2 3xy 5 4yz for y. 8. Solve x2 1 7x 1 19 5 0 for x.
5 2 hw for w.
574
8.4 Exercises
In Exercises 110, solve the constrained optimization problem using the Lagrange Multiplier Method. 1. Minimize f ( x, y ) 5 x 2 2 2y subject to the constraint 2 2x 1 y 5 5.
z
5. Maximize and minimize f ( x, y ) 5 2 x 3 1 4x 2 1 2y 2 subject to the constraint y 5 x .

z
y x x y
2. Maximize f ( x, y ) 5 2 x 2 2 y 2 1 16 subject to the constraint x 2 2y 5 4 .
6. Minimize f ( x, y ) 5 x 2 1 3y 2 subject to the constraint 2x 1 y 5 2 4.
7. Minimize f ( x, y ) 5 4xy subject to the constraint 5x 2 y 5 2. 3. Maximize f ( x, y ) 5 x 2 2 y 2 subject to the constraint y 5 3x 2 1 .

z
8. Maximize f ( x, y ) 5 2 4xy subject to the constraint 2 2x 1 y 5 2 4.
4. Minimize f ( x, y ) 5 xy subject to the constraint 2x 2 y 5 4.

z x
575
9. Maximize and minimize f ( x, y ) 5 x 3 2 20x 1 2y 3 1 y 2 subject to the constraint y 5 2x 1 3 .
The designer requires that the sum of the height and width of the piece of cardboard used to make the box be at most 180 inches. What are the dimensions of the box with the maximum volume? 12. Cylindrical Can Volume The volume of a cylindrical can is given by V( h, r ) 5 pr 2h, where r is the radius and h is the height of the can, both measured in inches. Because of manufacturing equipment limitations, the sum of the diameter and the height of the can cannot exceed 12 inches. Find the dimensions of the can with maximum volume. 13.
Cobb-Douglas Model Based on data from 1999 and 2000, a Cobb-Douglas model for Ford Motor Company is given by
10. Maximize f ( x, y ) 5 x 2y 1 2y 2 subject to the constraint x 2 4y 5 24 .

z
In Exercises 1115, use the Lagrange Multiplier Method to find the answer to the question. 11. Box Design A box designer wants to construct a closeable box with a square base as shown in the figure.
P( L, C ) 5 0.5319L0.4558C 0.5442 thousand vehicles, where L is the total labor cost and C is the capital expenditure cost, both in millions of dollars. In 2000, Ford Motor Company spent $25,783 million on labor and produced 7424 thousand vehicles. (Source: Modeled from Ford
Assuming that the model is valid for future years, how much should Ford spend on labor and how much should it spend on capital in order to maximize vehicle production if its annual budget for labor and capital expenditures is $40,000 million? 14. Landscape Design A landscape architect is designing a garden for a client as shown in the following figure.
Top flap 1
2h
Side 1 w w
2y h Top flap 1 Top flap 2 Top flap 3 Top flap 4 x 2h Side 1 Side 2 Side 3 Side 4 arc length= y
Bottom flap 1 w
Bottom flap 2 w
Bottom flap 3 w
Bottom flap 4 w
1 4
She intends to line the flat edges of the garden with shrubs and the curved edge of the garden with flowers. The shrubs cost $25 a linear foot, and the flowers cost $20 a linear foot, including labor. The client is willing to pay up to
576
$1000 and wants as large a garden as possible. What are the dimensions of the garden? (Hint: The area of the garden is given by 1 A( x, y ) 5 2xy 1 py 2.) 2 15.
Cobb-Douglas Model Based on data from 2000 to 2001, a Cobb-Douglas model for the Coca-Cola Company is given by P( L, C ) 5 0.02506L0.0481C 0.9519
is the maximum number of unit cases that can be produced with a labor and capital budget of $917 million?
16. What is constrained optimization? 17. What geometric shape is defined by a constraint of the form ax 1 by 5 c? 18. Once youve identified a critical point of a constrained optimization problem, what process should you use to determine if the point is a maximum or a minimum? 19. Is it possible for a critical point of a constrained optimization function not to be a maximum or a minimum? 20. Do all constrained optimization functions have one or more relative extrema?
billion unit cases of beverage, where L is the amount of money spent on labor (including sales, payroll, and other taxes) and C is the amount of money spent on capital expenditures, both in millions of dollars. (One unit case of beverage equals 24 eight-ounce servings.) In 2001, CocaCola spent a combined total of $917 million on labor and capital expenditures and produced 17.8 billion unit cases of beverage. (Source: Coca-Cola 2001 Annual Report.) According to the model, what
After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. How many independent and how many dependent variables does a function of two variables have? (8.1) 2. What is an effective way to construct a table of values for a function of two variables? (8.1) 3. What is the difference between a derivative and a partial derivative? (8.2) 4. What do the partial derivatives of a function f ( x, y ) represent graphically? (8.2) 5. Explain the meaning of the term critical point with regard to single-variable functions versus two-variable functions. (5.1, 8.3) 6. What does it mean when we say that the D-Test was inconclusive? (8.4) 7. Does the D-Test work for both constrained and unconstrained optimization problems? Explain. (8.3, 8.4) 8. What is the relationship between the Lagrange function and the function to be optimized? (8.4)
Chapter 8
Study Sheet
577
9. Are the critical points of a Lagrange function the same as the critical points of the function to be optimized? (8.4) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
Definitions
Multivariable Function: A function with two or more inputs (independent variables) Function of Two Variables: A function that associates each ordered pair of inputs (x, y) with a single output z 5 f ( x, y ) Surface: The graph of a function of two variables Partial Derivative: The derivative of a function of two variables when one of the independent variables is treated as a constant Partial Derivative of f with Respect to x: Partial Derivative of f with Respect to y: 'f or fx 'x 'f or fy 'y ' 'f a b or fxx 'x 'x ' 'f a b or fyy 'y 'y
Second Partial Derivative of f with Respect to x: Second Partial Derivative of f with Respect to y:
Mixed Second Partial Derivative of f with Respect to x then y: ' 'f a b or fxy 'y 'x Mixed Second Partial Derivative of f with Respect to y then x: ' 'f a b or fyx 'x 'y Relative Minimum of f( x, y ) : The value f ( a, b ) with the property that f ( a, b ) # f ( x, y ) for all (x, y) in some region surrounding (a, b) Relative Maximum of f( x, y ) : The value f ( a, b ) with the property that f ( a, b ) $ f ( x, y ) for all (x, y) in some region surrounding (a, b) Saddle Point of f( x, y ) : A point (a, b, c) that is a relative maximum of one cross section of a graph and a relative minimum of another cross section of the graph Critical Point of a Two-Variable Function: An ordered pair (c, d) such that fx ( c, d ) 5 0 and fy ( c, d ) 5 0
578
Second Derivative Test for Two-Variable Functions (D-Test) Let (c, d) be a critical point of f, where D ( c, d ) 5 fxx ( c, d ) ? fyy ( c, d ) 2 3 fxy ( c, d ) 4 2. Then f has a relative maximum at (c, d) if D ( c, d ) . 0 and f ( c, d ) , 0. xx f has a relative minimum at (c, d) if D ( c, d ) . 0 and f ( c, d ) . 0. xx f has a saddle point at (c, d) if D ( c, d ) , 0. the test is inconclusive if D ( c, d ) 5 0. Lagrange Multiplier Method for Solving Constrained Optimization Problems To optimize the objective function f ( x, y ) subject to the constraint g ( x, y ) 5 0, do the following. 1. Define F ( x, y, l ) 5 f ( x, y ) 1 lg ( x, y ) . 2. Set the partial derivatives of F equal to zero and solve to find the critical points. 3. The solution (if it exists) of the optimization problem will occur at one of these critical points. Method to Find Critical Points of a Lagrange Function 1. Solve Fx 5 0 and Fy 5 0 for l. 2. Eliminate l by setting the resultant equations equal to each other. 3. Substitute the result from Step 2 into the equation g ( x, y ) 5 0 and solve for x and y.
Formulas
Lagrange Function: F ( x, y, l ) 5 f ( x, y ) 1 lg ( x, y )

Section 8.1 In Exercises 13, generate a table of data for the multivariable function using x 5 2 2, 0, 2 and y 5 2 1, 0, 1 . Then use a graphing utility to graph the function. Compare the coordinates of the points in the table to the coordinates of the points on the graph of the function.
1. z 5 4x 2 5y 2. z 5 2xy 2 3y 2 3. z 5 xy 1 4y In Exercises 45, use the multivariable function to answer the given question. 4.
Price-to-Earnings Ratio The price-to-
earnings ratio (P/E) is the most common measure of how expensive a stock is and is given by P R5 E where P is the price of a share of the stock and E is the earnings per share. Growth stocks typically have high P/Es compared to the overall market.
Chapter 8
Review Exercises
579
Investors are willing to pay more for these stocks because they expect the companys earnings and stock price to rise. Stocks with low P/Es are often considered to be overlooked value stocks. As of December 28, 2002, the share prices of four major companies were Apple Computer: $14.06 General Electric: $24.70 Microsoft: $52.97 Electronic Arts: $51.50 Their earnings per share were Apple Computer: $0.18 General Electric: $1.41 Microsoft: $1.41 Electronic Arts: $0.71
(Source: www.quicken.com.)
Section 8.2 In Exercises 610, calculate the first and second partial derivatives of each of the functions.
6. f ( x, y ) 5 6xy 7. f ( x, y ) 5 2x 2y 2 4xy 2 9. z 5 xy ln( y ) 8. f ( x, y ) 5 5xy 2 5y 2 10. z 5 e x ln( y )
In Exercises 1112, interpret the practical meaning of each of the partial derivatives. 11. Cylindrical Can Volume The volume of a cylindrical can is given by V( h, r ) 5 pr 2h, where r is the radius and h is the height of the can, both measured in inches. Find Vh and Vr . Then evaluate the partial derivatives at ( 4, 5 ) and interpret the meaning of the results. 12. Landscape Design The area of the garden shown in the figure is given by 1 A( x, y ) 5 2xy 1 py 2 . 2
Calculate the P/E ratio for each of the companies. According to your calculations, which of the companies would you classify as growth stocks? 5.
Debt/Asset Ratio The debt/asset ratio
shows the proportion of a companys assets that are financed through debt and is calculated by L R5 A where L is the total value of the companys liabilities and A is the total value of its assets. If most of the companys assets are financed through equity (stock ownership interest in a company), the ratio is less than 1. If most of the companys assets are financed through debt, the ratio is greater than 1. Companies, and individuals, with high debt/asset ratios may be in danger if creditors start to demand repayment of debt. As of December 28, 2002, the total debt of each of three major companies was Apple Computer: $316.0 million General Electric: $254,522.0 million Microsoft: $0.0 million Their reported debt/equity ratios were Apple Computer: 0.08 General Electric: 4.25 Microsoft: 0.00
(Source: www.quicken.com.)
2y
arc length= y
Find Ax and Ay . Then evaluate the partial derivatives at ( 6, 4 ) and interpret the meaning of the results. In Exercises 1314, a function graph with a cross section drawn at x 5 0 and y 5 2 1 is shown. Determine the equations of the tangent lines of the function in the x-direction and y-direction at the point ( 0, 2 1, f ( 0, 2 1 ) ) . 13. f ( x, y ) 5 x 2 y 1 2
z
x y
Determine the value of the assets of Apple Computer and General Electric. Then interpret the meaning of the debt/equity ratio for Microsoft.
580
14. f ( x, y ) 5 x 2 2 4y 1 2
z
18. Maximize f ( x, y ) 5 2 x 3 1 3x 1 4y 3 subject to 2 2x 1 y 5 2 6.

z
y x y
Section 8.3 In Exercises 1516, find the relative extrema and saddle points of the function.
15. f ( x, y ) 5 x 2 2 xy 1 4y 2 1 10
z
In Exercises 1920, use the Lagrange Multiplier Method to find the answer to the question. 19. Landscape Design A landscape architect is designing a garden for a client as shown in the following figure.
2y
x y
16. f ( x, y ) 5 2 x 3 1 3x 1 4y 2 y 2
z
arc length= y
She intends to line the flat edges of the garden with shrubs and the curved edge of the garden with flowers. The shrubs cost $30 a linear foot, and the flowers cost $40 a linear foot, including labor. The client is willing to pay up to $1200 and wants as large a garden as possible. What are the dimensions of the garden? 20.
Cobb-Douglas Model Based on data from 2000 to 2001, a Cobb-Douglas model for the Coca-Cola Company is given by P( L, C ) 5 0.02506L0.0481C 0.9519 billion unit cases of beverage, where L is the amount of money spent on labor (including sales, payroll, and other taxes) and C is the amount of money spent on capital expenditures, both in millions of dollars. (One unit case of beverage equals 24 eight-ounce servings.) In 2001, CocaCola spent a combined total of $917 million on labor and capital expenditures and produced 17.8 billion unit cases of beverage. (Source: Coca-Cola
Section 8.4 In Exercises 1718, solve the constrained optimization problem using the Lagrange Multiplier Method.
17. Maximize f ( x, y ) 5 2 x 3 1 3x 1 4y 2 y 2 subject to the constraint 2 2x 1 y 5 0.
z
2001 Annual Report.)
Assuming that the model works for future years, what is the maximum number of unit cases that can be produced with a labor and capital budget of $998 million?
P R O J E C T
What to do
1. Go to www.weather.com and enter your zip code to access your local forecast. 2. Use the wind chill formula to calculate the current wind chill temperature. The wind chill formula is given by W( T, V ) 5 35.74 1 0.6215T 2 35.75V 0.16 1 0.4275TV 0.16 where T is the temperature (in degrees Fahrenheit) and V is the wind speed (in miles per hour). 3. Compare the calculated wind chill temperature to the Feels Like temperature displayed on the forecast. 4. Estimate how much the Feels Like temperature will change if the wind speed increases by 1 mph.
Local weather forecast

50F
Fair
UV Index: Dew Point: Humidity: Visibility: Pressure: Wind:
Feels Like 48F

1 Minimal 35F 57% Unlimited 30.25 inches and falling From the West Southwest at 6 mph
As reported at Seattle-Tacoma, WA Last Updated Tuesday, January 7, 2003, at 11:56 AM Pacific Standard Time (Tuesday, 2:56 PM EST).
Source: www.weather.com. Reprinted by permission.
581
Chapter
Trigonometric Functions
O
n August 13, 2003, the largest electrical blackout in North American history struck Connecticut, New Jersey, New York, Massachusetts, Michigan, Ohio, Pennsylvania, Vermont, and the Canadian province of Ontario. The outage impacted the lives of an estimated 50 million people, leaving some without power for days. Financial losses due to the outage were estimated to be between $4 billion and $10 billion.
(Source: U.S.-Canada Power System Outage Task Force, reports.energy.gov.) The photo shows
the New York skyline during the outage. One factor that contributes to power outages is consumer energy consumption. When consumer demand nears or exceeds available power supplies, the power grid is stressed and power outages become more likely. To reduce the frequency of such outages, energy organizations monitor and attempt to predict power usage each day throughout the year. Thankfully, their predictions are often remarkably accurate due to the cyclical nature of daily power consumption.
9.1
Trigonometric Functions, Equations, and Graphs
Create and graph a trigonometric function given data Determine the amplitude, period, horizontal shift, and vertical shift of a trigonometric function Solve an equation involving trigonometric functions
9.2
Derivatives of Trigonometric Functions
Find and interpret the derivative of a trigonometric function Use technology to find and interpret the derivative of a trigonometric function
9.3
Integrals of Trigonometric Functions
Find and interpret the integral of a trigonometric function Use technology to find and interpret the definite integral of a trigonometric function
582
9.1 Trigonometric Functions, Equations, and Graphs
583
Create and graph a trigonometric function given data Determine the amplitude, period, horizontal shift, and vertical shift of a trigonometric function Solve an equation involving trigonometric functions
When people decide to move to or visit a new city, one of the first things they are interested in knowing is what the weather is like. For the past several years, the Phoenix, Arizona, metropolitan area has been one of the fastestgrowing areas in the United States. Many people are moving to the Phoenix area because of the warm temperatures and frequent sunshine. In fact, the Phoenix area is known as the Valley of the Sun because of its weather. The data in Table 9.1 show the average monthly high temperatures in Phoenix. (Source: phoenix.about.com.) From the table, we can see that the temperatures are relatively low in the winter months of January and February, increase in the spring season, and are highest in the summer months of June, July, and August. The temperatures decrease in the fall and winter seasons. We predict that this trend will continue from year to year. Figure 9.1 shows the data from the table repeated for two years.
TABLE 9.1
Temperature (in degrees F)
Average High Temperature (in F) Month January February March April May June July August September October November December t 1 2 3 4 5 6 7 8 9 10 11 12 Temperature 66 70 75 84 93 103 105 103 99 88 75 66
100 90 80 70 60 0 0 5 10 15 Month 20
FIGURE 9.1
This scatter plot is unique compared to other data we have seen because the data repeat in yearly cycles. Data that repeat in cycles are called periodic. Other natural phenomena that can be described as periodic include the height of the ocean tide, the height of an object moving in a circular path, electrical voltage, and many animal populations. We will explore these periodic phenomena later in this section.
584
CHAPTER 9 Trigonometric Functions
The Sine and Cosine Functions

Periodic data often may be modeled by trigonometric functions of the form where Z A Z is the amplitude of the function, B is the frequency of the function (number of periods that fit in the interval 30, 2p4 ), C is the horizontal shift, and D is the vertical shift. Such functions are commonly referred to as sinusoidal functions. To construct a model, we need to determine four key components of the model: the amplitude, the period, the vertical shift, and the horizontal shift. We first consider the simplest periodic functions, y 5 sin t and y 5 cos t For these functions, A 5 B 5 1 and C 5 D 5 0. The graphs of the functions are given in Figure 9.2.
y 1 0.75 0.50 0.25 2 3
2 2 2
y 5 A sin 3B( t 2 C ) 4 1 D
or
y 5 A cos 3B( t 2 C ) 4 1 D
y = cos t
3 2 t
2
0.50 0.75 1 y = sin t
FIGURE 9.2
We see that these two functions have the same amplitude, period, and vertical shift. However, their horizontal placement on the x-axis differs. Observe that the sine graph is equivalent to the cosine graph shifted to the right. We can imagine lifting the function y 5 cos t and shifting it horizontally to the right by ing the graph of y 5 sin t and shifting it to the left by
p 2 p 2
radians.
It would then line up with the function y 5 sin t. Similarly, we can imagine liftradians to get the graph of y 5 cos t. We will transform these simple sinusoidal graphs as we construct models of periodic data. To prepare us to construct periodic models from scatter plots of data sets, we will look at the four key characteristics of sinusoidal functions in greater detail. The amplitude of a sinusoidal function is the vertical distance from an imaginary horizontal line (midline) that runs through the center of the scatter plot. The amplitude can be found by subtracting the minimum data value from the maximum data value (thus determining the distance between them) and dividing by 2.
585
By subtracting the amplitude from the maximum data value or adding it to the minimum data value, we can find the midline for the data set.
Finding the Amplitude of a Sinusoidal Model The amplitude of a sinusoidal model is given by Amplitude 5 maximum function value 2 minimum function value 2
A similar strategy may be used to find the equation of the midline.
Finding the Midline of a Sinusoidal Model The midline of a sinusoidal model is given by y5 maximum function value 1 minimum function value 2 y 5 maximum function value 2 amplitude or y 5 minimum function value 1 amplitude
Alternatively, the midline equation may be calculated as
EXAMPLE
Determining the Midline and Amplitude of a Sinusoidal Model

Find the amplitude and midline of the function that models the average monthly high temperature in Phoenix (Table 9.1).
SOLUTION First, we find the amplitude by subtracting the minimum temperature (66) from the maximum temperature (105) and dividing the result by 2.
100 90 80 70 60 0 0 5 10 15 Month 20
Amplitude 5 A 5 5
105 2 66 2 39 2
5 19.5 Adding this value to the minimum value of 66 or subtracting it from the maximum value of 105 yields 85.5. Therefore, the midline through the data set is located at y 5 85.5, as shown in Figure 9.3. Note that the distance from the midline to the maximum temperature of 105 or from the midline to the minimum temperature of 66 is equal to the amplitude, 19.5.
FIGURE 9.3
586
The period of a function is the length of time it takes for the function to complete one cycle. Recall that the period of y 5 sin t and y 5 cos t is 2p. However, the period of the Phoenix temperature data is 12 months because the data values start to repeat after 1 year. Therefore, we need to adjust the function y 5 sin t so that it has a period of 12 rather than 2p. Recall that the frequency, B, tells us how many periods of the function can fit into an interval 2p units long. We can determine B by dividing 2p by the length of the period of the function being modeled. B5 5 5 So
p 6
2p period 2p 12 p 6
< 0.52 (just over half) of the period of the temperature function will fit into
p 2p
an interval 2p units long. Will one entire period of the temperature function fit into a single period of y 5 sin Q t R ? Observe that since B 5 , it follows that 6 period period 5
2p . Thus B
the period of y 5 sin Q t R is given by 6 Period 5 2p p 6
5 12 So the period of y 5 sin Q t R is long enough to contain exactly one period of the 6 temperature function. Both functions have a period of 12.
p
Finding the Frequency and Period of a Sinusoidal Model The frequency B of a sinusoidal model y 5 A sin 3B( t 2 C ) 4 1 D or y 5 A cos 3B( t 2 C ) 4 1 D is given by B5 2p period
The period of a sinusoidal model y 5 A sin 3B( t 2 C ) 4 1 D or y 5 A cos 3B( t 2 C ) 4 1 D is given by Period 5 2p B
The vertical shift is the distance that the midline of the basic sinusoidal function is shifted, and is equal to D in the above equations. If D . 0, the shift is upward. If D , 0, the shift is downward.
587
EXAMPLE
Determining the Vertical Shift of a Sinusoidal Model

Determine the amount of vertical shift of the sinusoidal function that models the average monthly high temperature in Phoenix (Table 9.1).
SOLUTION The maximum data value is 105 and the minimum data value is 66. The midline will lie halfway between these two values. We find the average of these values by adding them and dividing by 2.
105 1 66 5 85.5 2 Notice that this is the same value for the midline that we found in Example 1. The model contains a vertical shift of 85.5. We now have three of the four key components of the trigonometric model for the temperature function. Filling in the values for A, B, and D, we get the following model. p y 5 19.5 sin c ( t 2 C ) d 1 85.5 6 What about the horizontal shift C ? We will first consider the case in which C 5 0 in order to get a sense of how well our model is representing the data. The graph in Figure 9.4 shows the original temperature data along with the model. Notice that the model represents the data well, except that it is shifted to the left. To shift the model to the right, we must find a value of C that will shift the function the appropriate distance to the right.
100 90 80 70 60 0 0 5 10 15 Month 20
FIGURE 9.4
EXAMPLE
Determining the Horizontal Shift of a Sinusoidal Model

Determine the horizontal shift C of the function y 5 19.5 sin S ( t 2 C ) T 1 85.5 6 that will accurately model the average monthly high temperature in Phoenix.
SOLUTION By looking at the scatter plot in Figure 9.4, we see that the maxp
imum temperature of 105 occurs when t 5 7. Our model p y 5 19.5 sin c ( t 2 0 ) d 1 85.5 6 first reaches its maximum temperature at t 5 3. Therefore, we need to shift the model 4 units to the right to align it with the temperature scatter plot. Letting C 5 4 yields a new model: p y 5 19.5 sin c ( t 2 4 ) d 1 85.5 6
588
How well does the new model fit the temperature data? Lets graph the new model together with the scatter plot (Figure 9.5).
100 90 80 70 60 0 0 2 4 6 8 Month 10
FIGURE 9.5
The new model fits the data well. In practice, we may need to adjust the value of C through a process of guess and check in order to get the best possible model for our data.
Modeling Data with Trigonometric Functions To create a sinusoidal function of the form y 5 A sin 3B( t 2 C ) 4 1 D for the purpose of modeling data that are periodic, we do the following.
1. Find the value of A, the amplitude:
A5
maximum data value 2 minimum data value 2

2p . period
2. Find the value of B, the frequency, by first determining the period and
then using B 5
3. Find the value of D, the vertical shift:
D5
4. Find the value of C, the horizontal shift, through a process of trial and error.
We can also use the general form y 5 A cos 3B( t 2 C ) 4 1 D to model periodic data. Recall that the relationship between y 5 sin t and y 5 cos t is that each is a horizontal shift of
p 2
radians from the other. In the case of the model for the
average monthly high temperature in Phoenix, we can change to a cosine model by changing the value of C. This can be accomplished by shifting the sine model
p 2
< 1.57 radians.
589
EXAMPLE
Using a Cosine Function to Model Data

Find a model of the form y 5 A cos 3B( t 2 C ) 4 1 D for the Phoenix average monthly high temperature data (Table 9.1).
SOLUTION We first consider the model found in Example 3:
p y 5 19.5 sin c ( t 2 4 ) d 1 85.5 6 In this model, if we were to distribute the 6 , we would have p 4p y 5 19.5 sin a t 2 1 85.5 6 6 b In changing the sine model to a cosine model, we need to add the above version of the model. 4p p 4p 3p 1 5 1 6 2 6 6 5 The cosine model can be written as p 7p y 5 19.5 cos a t 2 1 85.5 6 6 b If we factor the common factor of
p 6 p 2 p
to the
4p 6
term in
7p 6
from inside the parentheses, we get
p y 5 19.5 cos c ( t 2 7 ) d 1 85.5 6 We can also model real-world phenomena that are not given as data in table form. In the next series of examples, we will investigate the height above the ground of a Ferris wheel rider. The Ferris wheel is named after George Ferris, a civil engineer who created the first Ferris wheel for the Chicago Worlds Fair in 1893. The original Ferris wheel had a diameter of 250 feet, rose 266 feet above the ground, and carried 36 cars, each able to accommodate 60 people. Traveling at an angular speed of 72 degrees per minute, it took 5 minutes to make a complete revolution. It cost $250,000 to complete the project. For our first example, we will consider a more conventional Ferris wheel. Consider a Ferris wheel that is 40 feet in diameter and 10 feet above the ground at its lowest point. We will assume that the Ferris wheel completes one turn in 20 seconds. We can plot the height above the ground of a Ferris wheel rider as a
590
function of time. For this graph, we will assume that the rider begins in position 1 (Figure 9.6) and moves counterclockwise.
h 2 Height (in feet) 40 30 20 10 0 0 5 10 15 Time (in seconds) 1 5 3 2 4
3 4
6 7
6 7
FIGURE 9.6
As Figure 9.6 suggests, we can model a Ferris wheel riders height above the ground as a function of time by using a sine function. This is another example of a periodic function.
EXAMPLE
Modeling the Height of a Ferris Wheel Rider Above the Ground

Construct a periodic model of a Ferris wheel riders height above the ground as a function of time given that the wheel has a 40-foot diameter, is 10 feet above the ground at its lowest point, and completes one revolution in 20 seconds.
SOLUTION We want to create a model of the form y 5 A sin 3B( t 2 C ) 4 1 D. We will begin by finding the amplitude A for this situation. The Ferris wheel has a diameter of 40 feet and is positioned 10 feet above the ground at its lowest point. Thus the maximum height is 50 feet and the minimum height is 10 feet. The amplitude is
A5
50 2 10 5 20 2
Next, we will determine the value of B in our general model. Since the period (the length of time for the Ferris wheel to complete one revolution) is 20 seconds, B5 p 2p 5 20 10
Third, we will determine the value of D, the vertical shift. The vertical shift, or midline, will represent the center of the Ferris wheel, which is 30 feet above the ground. We can also compute the value of D by finding the average of the maximum and minimum heights. D5 50 1 10 5 30 2
591
50 Height (in feet) 40 30 20 10 0 0 5 10 15 Time (in seconds)
Finally, we will determine the value of C, the horizontal shift. In this example, the function is not shifted horizontally, so the value of C is 0. This leads to the final model for this situation: y 5 20 sin a p t b 1 30 10
A graph of this model (Figure 9.7) confirms that we have an accurate model of the Ferris wheels height as a function of time.
FIGURE 9.7
Once we have constructed the mathematical model for a specific situation, we can use the model to analyze different types of situations and do additional analysis. For example, imagine that in the Ferris wheel situation we wish to determine the height of the Ferris wheel rider at a specific point in time. Or, we may wish to determine how high above the ground the Ferris wheel rider will be after a certain amount of time has elapsed. Examples 6 and 7 reveal how we could solve such problems.
EXAMPLE 6
Calculating the Position of a Ferris Wheel Rider

Use the Ferris wheel model constructed in Example 5 to determine the height above the ground of the Ferris wheel rider after 7 seconds. Assume the Ferris wheel begins at position 1 when we start the time clock and travels counterclockwise.
SOLUTION We will express our model in function notation, where y 5 h( t ) . Here, h represents the height of the rider above the ground at time t. The height above the ground of the Ferris wheel rider is given by
50 Height (in feet) 40 30 20 10 0 0
(7, 46.18)
y 5 h( t ) 5 20 sin a
p t b 1 30 10
We need to find the height after 7 seconds have elapsed. Therefore, we need to compute h( 7 ) . h( 7 ) 5 20 sin a < 46.18
5 10 15 Time (in seconds)
p ? 7 b 1 30 10
FIGURE 9.8
After 7 seconds have elapsed, the Ferris wheel rider will be about 46.18 feet above the ground. We can confirm this graphically using technology (Figure 9.8). See the Technology Tip: Solving an Equation Graphically at the end of this section.
592
Just In Time
Solving Trigonometric Equations
In much the same way that we solve a linear equation, such as 4x 1 3 5 5, by performing inverse operations to isolate the variable x, we solve a trigonometric equation by first isolating the trigonometric term. 4 cos( 2x ) 1 3 5 5 4 cos( 2x ) 5 2 cos( 2x ) 5 1 2
Given Subtract 3 Divide by 4
We now solve for 2x by performing the inverse operation of cosine, arccosine. Note that arccos 3cos ( 2x ) 4 5 2x. Because of the periodic nature of the cosine function, there are infinitely many values of x that satisfy the equation. We will limit our answers to the interval 0 # x # 2p, or one period. 1 arccos 3cos( 2x ) 4 5 arccos a b 2 2x 5 x5 p 5p 7p 11p , , , 3 3 3 3 p 5p 7p 11p , , , 6 6 6 6
Use p as a reference angle 3
Divide by 2
EXAMPLE
Finding the Time(s) When a Ferris Wheel Rider Is at a Given Height

Use the Ferris wheel model constructed in Example 5 to determine the time(s) at which the Ferris wheel rider is 35 feet above the ground.
SOLUTION We need to determine the time it takes to reach a height of 35 feet. Thus, we need to solve the equation
y 5 h( t ) 5 35 5 20 sin a
p t b 1 30 10
We first consider a graphical solution to this equation. We use the graphing calculator to graph the function y 5 h( t ) along with the function y 5 35. We find the solution to the equation by finding the intersection of the two graphs (Figure 9.9).
FIGURE 9.9
Notice that the two graphs intersect in more than one place. Therefore, there is more than one time at which the Ferris wheel rider is 35 feet above the ground.
593
However, this should not surprise us, considering the periodic nature of the function. We first find the times at which the Ferris wheel rider is 35 feet above the ground during the first revolution of the Ferris wheel. We use the graphing calculator to determine these points of intersection (Figure 9.10).
FIGURE 9.10
The calculator screen shots in Figure 9.10 show that the Ferris wheel rider is 35 feet above the ground about 0.8 second and about 9.2 seconds into the wheels first revolution. p Alternatively, we can solve the equation 35 5 20 sin Q 10 t R 1 30 symbolically. Our first goal in solving this trigonometric equation is to isolate sin Q 10 tR . 35 5 20 sin a 5 5 20 sin a p t 1 30 10 b p t 10 b
Subtract 30 from both sides Divide both sides by 20
p
5 p 5 sin a t b 20 10 1 p 5 sin a t b 4 10
Next, we take the inverse sine of each side of the equation. Recall that the inverse sine function, written as sin21( x ) or arcsin ( x ) , is the inverse operation for the sine function. 1 p 5 sin a t b 4 10 1 p sin21 a b 5 sin21 c sin a t b d 4 10 1 p sin21 a b 5 t 4 10 10 21 1 sin a b 5 t p 4
1
Multiply both sides by

10 . p
10 p
We will use the calculator to find sin21 Q 4 R and then multiply by t5 10 21 1 sin a b < 0.8 p 4
Consistent with the graphical method of solution, we find that after about 0.8 second, the Ferris wheel rider is 35 feet above the ground. However, this method does not
594
provide us with the second time that the Ferris wheel rider is 35 feet above the ground during the first complete rotation. To determine the second time that the Ferris wheel rider is 35 feet above the ground, we must recall concepts from trigonometry. 1 1 We know that the angle whose sine is 4 that is, sin21 A 4 B is about 0.25 radian. We also know that sine is positive in Quadrants I and II of the Cartesian plane. Our task is to determine the Quadrant II angle whose sine is 4.
y 1
Notice that there are two angles whose sine is 4, one in Quadrant I and one in Quadrant II. We already know that the Quadrant I angle is about 0.25 radian. The Quadrant II angle is found by subtracting 0.25 radian from p radians.
1
1 4
1 4
FIGURE 9.11
We see from Figure 9.11 that there are two angles whose sine ratio is 4: 1 arcsin a b < 0.25 4 and 1 arcsin a b < p 2 0.25 < 2.89 4
We find the second time that the Ferris wheel rider is 35 feet above the ground by multiplying the second angle by
10 . p
10 ( 2.89 ) < 9.2 p The Ferris wheel rider is 35 feet above the ground at about t 5 0.8 second and about t 5 9.2 seconds during the first revolution.
EXAMPLE
Finding the Time(s) When a Ferris Wheel Rider Is at a Given Height

Use the Ferris wheel model constructed in Example 5 to determine the time(s) at which the Ferris wheel rider is 12 feet above the ground during the first revolution of the Ferris wheel. Assume that the Ferris wheel is at position 1 when we start the time clock, and that it travels counterclockwise. Solve the problem twice, once graphically and once symbolically.
SOLUTION We need to determine the time it takes for the Ferris wheel to reach a height of 12 feet. Thus, we need to solve the equation
12 5 20 sin a
p t 1 30 10 b
595
We first consider a graphical solution to this equation. Using the graphing calculator, we graph the function h( t ) 5 20 sin Q 10 t R 1 30 and the function y 5 12. We use the calculator to find the points of intersection (Figure 9.12).
p
FIGURE 9.12
Figure 9.12 shows that the Ferris wheel rider reaches a height of 12 feet after about 13.6 seconds and 16.4 seconds. To solve this equation symbolically, we follow the same steps shown in Example 7. 12 5 20 sin a 2 18 5 20 sin a p t 1 30 10 b p t 10 b
2 18 p 5 sin a t b 20 10 29 p 5 sin a t b 10 10 sin21 a sin21 a

y
29 p 5 sin21 c sin a t b d b 10 10 29 p t 5 10 b 10
29 . 10
We must determine the two angles whose sine is
Again, we will use a
drawing along with the graphing calculator to help us determine these values. Since the sine ratio is negative in Quadrants III and IV , we will draw reference angles in these quadrants (Figure 9.13).
x 9
10
1 1
10
FIGURE 9.13
596
Using the graphing calculator, we find that the Quadrant IV angle is

29
angle can be found by p 1 1.12 < 4.26 radians. We now multiply these two answers by
10 p
sin21 A 10 B < 2 1.12 radians or 2p 2 1.12 < 5.16 radians. The Quadrant III to get the final results. t5 t5 t< t< 10 21 2 9 sin a p 10 b 29 10 ? c p 2 sin21 a bd p 10 10 ? ( p 1 1.12 ) p 10 ? 4.26 < 13.6 p
Also, t5 t5 t< t< 10 21 2 9 sin a b p 10 10 29 c 2p 2 sin21 a bd p 10 10 ( 2p 2 1.12 ) p 10 ? 5.16 < 16.4 p
Consistent with our graphical results, we find that the Ferris wheel is 12 feet above the ground after about 13.6 seconds and after about 16.4 seconds during the first revolution.
The Tangent Function

Recall that the tangent function is also a periodic function.
THE TANGENT FUNCTION
tan t 5
sin t cos t
Because the definition of the tangent function is a quotient with cos t in the denominator, the function is undefined whenever cos t 5 0. cos t 5 0 at t 5 c ,2
3p p p 3p ,2 2 , 2 , 2 , 2
c . In fact, cos t 5 0 if t is of the form t 5
p 2
1 kp,
597
where k is any integer. Thus, the graph of y 5 tan t has a vertical asymptote whenever t is of the form t 5
p 2
1 kp (Figure 9.14).
y 2 1.5 1 0.5 x
2
3
2 2
3 2
1 1.5 2
FIGURE 9.14
Most models that we will use involve the sine and cosine functions.
Using Technology to Solve an Equation Graphically

Graphing calculators, such as the TI-83 Plus, can be used to solve equations graphically. They are especially useful when we want to determine the domain value that yields a given range value. In the following Technology Tip, we detail how to solve the equation 1.5 5 2 sin( x 2 1 ) .
Solving an Equation Graphically 1. Bring up the Y5 Editor by pressing the Y button. Type the left side of the equation in Y1 and the right side of the equation in Y2. Make sure you use parentheses as needed.
2. Edit the viewing window by pressing the WINDOW button and editing the parameters. Then press the GRAPH key to draw the graph.
(Continued)
598
3. To determine the point(s) where the two graphs intersect, determine the solution of the equation. Press 2ND then TRACE and choose 5: intersect.
4. Identify the first curve by pressing ENTER . Notice that the function Y1 is indicated as the first curve.
5. Identify the second curve by pressing ENTER . Notice that the function Y2 is indicated as the second curve.
6. The last on-screen prompt asks for a guess. Use the left and right arrow keys to move the cursor near the point of intersection that you are interested in finding. Once there, press ENTER .
7. The calculator displays the point of intersection. This is the value of x that satisfies the original equation. In this case, one of the solutions to the equation 1.5 5 2 sin( x 2 1 ) is x 5 2 4.44.
Solving an Equation Graphically To solve the equation f ( x ) 5 b graphically, graph y 5 f ( x ) and y 5 b. Then estimate the coordinates of the point(s) of intersection visually from the graph.
6 4 2
y = f(x) y=b
3 2 1 0
1 2
599
9.1 Summary
In this section, you learned how to construct a model, both symbolically and graphically, for data or phenomena that are periodic in nature. By finding the amplitude, the period, the horizontal shift, and the vertical shift, you were able to construct a model of the form y 5 A sin 3B( t 2 C ) 4 1 D. You discovered that you can also choose to use the form y 5 A cos 3B( t 2 C ) 4 1 D by recognizing the relationship between the sine and cosine functions. Finally, you were able to use a model to answer questions pertaining to a real-life situation by solving equations involving trigonometric functions.

1. A classmate is solving a trigonometric equation and finds that cos( x ) 5 1. The classmate indicates that the next step is to divide both sides by cos to isolate the variable x. Explain to your classmate why this is not a viable option and what the expression cos( x ) means. 2. You are helping a classmate solve the trigonometric equation 44 1 2 cos( x ) 5 26. Your classmate begins by writing 46 cos( x ) 5 26. Explain to your classmate why this is not a viable option and what the expression 44 1 2 cos( x ) means. 3. A classmate is solving the trigonometric equation cos( x 1 2 ) 5 2 1 and begins by writing cos( x ) 1 cos( 2 ) 5 2 1. Explain to your classmate why this is not a viable option and what the expression cos( x 1 2 ) means. In Exercises 414, solve the equation for 0 # u , 2p. If possible, find exact answers. If not possible, use your calculator to find decimal approximations. 4. cos( 2u ) 5 2 1 2 5. sin ( 2u ) 1 4 5 8 6. sin2 u 2 1 5 0 7. cos2 u 2 1 5 0 8. tan2 u 2 1 5 0 10. 2 cos( 3u ) 5 "3 9. 2 cos u sin u 2 sin u 5 0
11. 5 sin( u 1 1 ) 2 2 5 1 12. 2 2 cos( 2u 2 p ) 1 1 5 2 13. 3 cos a u 2 p 2152 2b
14. "2 tan( 2u ) 5 1
600
9.1 Exercises
In Exercises 15, determine if a periodic function would be appropriate to model the situation described. Explain why you think your conclusion is correct. 1. The population of an animal species living in the wild with respect to time in months 2. The distance above the ground of a bungee jumper with respect to time in seconds 3. The population of the United States with respect to time in years 4. The depth of water in the ocean as measured from low tide to high tide, and back to low tide 5. The distance from Earth of an orbiting satellite In Exercises 610, determine the value of the amplitude, the period, the horizontal shift, and the vertical shift. 6. y 5 10 sin 32( x 2 1 ) 4 1 5 7. y 5 cos( x 1 p ) 8.
y 5 4 3 2 1 0 0 1 2 3 4 5 x
10. 0
x 0.63 1.26 1.89 2.51 3.14 3.76 4.40 5.03 5.65 6.28
y 1 1.85 1.93 1.20 20.03 21.25 21.94 21.83 20.95 0.32 1.46
In Exercises 1114, you will model the hours of daylight in various cities and then evaluate the significance of the results. 11. The number of hours of daylight at the end of each month in Miami, Florida, is given in the table.
Hours of Daylight in Miami, FL
Hours of Daylight in Miami Month January February

1 2 3 4 5 x
9.
t 31 59 90 120 151 181 212 243 273 304 334 365
Hours of Daylight 11.00 11.50 12.50 13.00 13.67 13.75 13.33 12.67 12.00 11.17 10.67 10.50
y 2 0 2 4 6
March April May June July August September October November December
601
(a) Find a trigonometric model representing the number of daylight hours as a function of the day of the year t. (b) Graph the data and the model using your graphing calculator. (c) Interpret the meaning of the amplitude in the context of the data. (d) Interpret the meaning of the vertical shift in the context of the data. (e) Interpret the meaning of the horizontal shift in the context of the data. 12.
Hours of Daylight in Phoenix, AZ The
13.
Hours of Daylight in Juneau, AK The
number of hours of daylight at the end of each month in Juneau, Alaska, is given in the table. Hours of Daylight in Juneau Month January February March April May June July August September October November December t 31 59 90 120 151 181 212 243 273 304 334 365 Hours of Daylight 8.25 10.50 13.00 15.67 17.75 18.17 16.50 14.00 11.50 9.00 7.00 6.50
number of hours of daylight at the end of each month in Phoenix, Arizona, is given in the table. Hours of Daylight in Phoenix Month January February March April May June July August September October November December t 31 59 90 120 151 181 212 243 273 304 334 365 Hours of Daylight 10.50 11.33 12.50 13.50 14.25 14.33 13.83 13.00 11.83 10.83 10.00 10.00 14.
(a) Find a trigonometric model representing the number of daylight hours as a function of the day of the year t. (b) Graph the data and the model using your graphing calculator. (c) Interpret the meaning of the amplitude in the context of the data. (d) Interpret the meaning of the vertical shift in the context of the data. (e) Interpret the meaning of the horizontal shift in the context of the data.
Hours of Daylight in Sydney, AU The
(a) Find a trigonometric model representing the number of daylight hours as a function of the day of the year t. (b) Graph the data and the model using your graphing calculator. (c) Interpret the meaning of the amplitude in the context of the data. (d) Interpret the meaning of the vertical shift in the context of the data. (e) Interpret the meaning of the horizontal shift in the context of the data.
number of hours of daylight at the end of each month in Sydney, Australia, is given in the table.
602
Hours of Daylight in Sydney Month January February March April May June July August September October November December t 31 59 90 120 151 181 212 243 273 304 334 365 Hours of Daylight 13.75 12.83 11.75 10.75 10.00 10.00 10.33 11.33 12.33 13.50 14.25 14.50
18. Review the models you created for Exercises 1114. Compare the value of D in each case. When compared with the other cities, does the value of D for a particular city make sense? Why or why not? Exercises 1921 focus on additional real-world phenomena that may be modeled by periodic functions. 19.
Alternating Current In the United States, electricity is provided to homes and businesses using alternating current. Alternating current is electric current that varies between 165 volts and 2 165 volts. This voltage is supplied at a rate of 60 cycles per second. (a) Create a model of electrical voltage as a function of time. (b) Identify and interpret the amplitude, period, vertical shift, and horizontal shift in this situation. Moon OrbitPerigee The moons
20.
(a) Find a trigonometric model representing the number of daylight hours as a function of the day of the year t. (b) Graph the data and the model using your graphing calculator. (c) Interpret the meaning of the amplitude in the context of the data. (d) Interpret the meaning of the vertical shift in the context of the data. (e) Interpret the meaning of the horizontal shift in the context of the data. 15. Review the models you created for Exercises 1114. Compare the value of A in each case. When compared with the other cities, does the value of A for a particular city make sense? Why or why not? 16. Review the models you created for Exercises 1114. Compare the value of B in each case. Why is the value of B the same? When compared with the other cities, does the value of B for a particular city make sense? Why or why not? 17. Review the models you created for Exercises 1114. Compare the value of C in each case. Why is the value of C the same? When compared with the other cities, does the value of C for a particular city make sense? Why or why not?
distance from the earth changes as the moon orbits the earth in an elliptical path. The point at which the moon is closest to the earth is known as the perigee. The perigee changes from month to month in a periodic fashion. For the year 2006, the greatest perigee distance was 368,614 kilometers, on May 22. (Source: www.fourmilab.ch.) The least perigee distance was 356,884 kilometers, on February 27. (a) Create a sinusoidal model of the perigee distance as a function of time in days (b) Identify and interpret the amplitude, period, vertical shift, and horizontal shift in this situation. (c) What was the perigee distance on August 10, 2006? (d) On what date or dates in 2006 was the perigee distance approximately 363,735 kilometers? 21.
Moon OrbitApogee The moons distance from the earth changes as the moon orbits the earth in an elliptical path. The point at which the moon is farthest from the earth is known as the apogee. The apogee changes from month to month in a periodic fashion. For the year 2006, the greatest apogee distance was 406,498 kilometers, on September 22. (Source: www.fourmilab.ch.) The least apogee distance was 404,079 kilometers, on June 4.
603
(a) Create a sinusoidal model of the apogee distance as a function of time in days (b) Identify and interpret the amplitude, period, vertical shift, and horizontal shift in this situation. (c) What was the apogee distance on August 26, 2006? (d) On what date or dates in 2006 was the apogee distance approximately 404,571 kilometers? In Exercises 2226, create a sinusoidal model for the given graph. 22.
y 5 4 3 2 1 0 0 2 4 6 8 10 x
26.
y 2 0 2 4 6 0.5 1 1.5 x
27. Sketch several examples of graphs that are periodic and several that are not periodic. Describe what makes a graph periodic. 28. The figure shows the graphs of P( t ) 5 2 cos t and Q( t ) 5 4 cos t. Identify which graph is P( t ) and which is Q( t ) . Provide a justification for your choice.
y 4 2 6 4 2 2 4 2 4 6 x
23.
y 2.0 1.6 1.2 0.8 0.4 0 0 1 2 3 4 5 6 x
24.
y 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1 y 5 4 3 2 1 0 0 2
29. The figure shows the graphs of R( t ) 5 cos( 2t ) and S( t ) 5 cos( 4t ) . Identify which graph is R( t ) and which is S( t ) . Provide a justification for your choice.
y
0.5
1.5
1 0.5 2 1 0.5 1 1 2 x
25.
30. Without using technology, describe the appearance of the graph of y 5 ( cos x ) 2 as compared with y 5 cos x. Explain how you know your description is correct. Then check and correct your work by graphing y 5 ( cos x ) 2 using technology.
604
31. Without using technology, describe the appearance of the graph of y 5 sin x 1 x as compared with y 5 sin x. Explain how you know your description is correct. Then check and correct your work by graphing y 5 sin x 1 x using technology.
33. For each of the parameters A, B, C, and D, describe what the parameter represents in the context of the first Ferris wheel. 34. Assuming that a rider begins his continuous ride at the lowest point, determine where the rider will be after 2 minutes. 35. Assume that the rider will complete 2 revolutions. Determine each time that the rider will be 100 feet above the ground. 36. Describe both the graphical impact and the symbolic impact on the model if the speed of the Ferris wheel is increased or decreased.
For these exercises, refer to the description of the first Ferris wheel on page 589. It had a diameter of 250 feet, was 16 feet above the ground at its lowest point, and completed one revolution in 5 minutes. 32. Construct a periodic model of the Ferris wheel riders height above the ground as a function of time.
9.2 Derivatives of Trigonometric Functions
Find and interpret the derivative of a trigonometric function Use technology to find and interpret the derivative of a trigonometric function
Meteorologists and others may be interested in knowing how fast the temperature is increasing or decreasing per month in a particular city. In this section, we will revisit the data concerning the average high temperatures in Phoenix, Arizona. This time, however, we will focus on the rate at which the temperatures are changing in degrees per month. In this section, we will show how to calculate the instantaneous rate of change of a trigonometric function. We will show that when the rate of change is positive, the function is increasing; when the rate of change is negative, the function is decreasing.
Recall that in Section 9.1 we modeled the average monthly high temperatures for Phoenix, Arizona, with the function T( t ) 5 19.5 sin S ( t 2 4 ) T 1 85.5. Using 6 this model, we construct a table of values for the temperature function. As we read down the table, we estimate the instantaneous rate of change in temperature in a given month by calculating the average rate of change between the temperature in the month prior and the month after the given month. We are interested in understanding these monthly rates of change and how they relate to the original model of the temperature.
p
605
TABLE 9.2
Average High Temperatures (F) Estimate of Instantaneous Rate of Change 68.6 2 68.6 50 2 75.8 2 66.0 5 4.9 2 85.5 2 68.6 < 8.4 2 95.3 2 75.8 2 102.4 2 85.5 2 105.0 2 95.3 2 102.4 2 102.4 2 95.3 2 105.0 2 85.5 2 102.4 2 75.8 2 95.3 2 68.6 2 85.5 2 < 9.8 < 8.4 < 4.9 50 < 2 4.9 < 2 8.4 < 2 9.8 < 2 8.4
Month December January February March April May June July August September October November December January
t 0 1 2 3 4 5 6 7 8 9 10 11 12 13
Temperature 68.6 66.0 68.6 75.8 85.5 95.3 102.4 105.0 102.4 95.3 85.5 75.8 68.6 66.0
66.0 2 75.8 5 2 4.9 2
Change in temperature (in degrees F)
10 5 0 3 5 6 9 12 15 18 21
We will now look at a scatter plot showing the estimated rate of change in monthly temperatures as a function of the month. We repeat the rates of change for a second year to show the periodic nature of the situation (Figure 9.15). We see that the rate of change in temperature appears periodic and may be modeled using a periodic function. How is this periodic function, which models the rate of change in temperature, related to the model of the temperature T( t ) 5 19.5 sin S ( t 2 4 ) T 1 85.5? We will answer this question over the next 6 several examples.
p
10 Month
FIGURE 9.15
606
EXAMPLE
Constructing a Sinusoidal Model from a Data Set

Construct a sinusoidal model for the rates of change in temperature presented in Table 9.2 and Figure 9.15.
SOLUTION First, we observe that the midline seems to be the x-axisthat is, the line y 5 0. Next, we compute the amplitude.
A5
9.8 2 ( 2 9.8 ) 2 5 9.8
Change in temperature (in degrees F)
10 5 0
Model with Pre-shift model shift
The amplitude is 9.8. Third, we determine the period and use it to find the value of B. Since the data repeat every 12 months, we can compute the value of B for our model. B5 2p p 5 12 6
3 5
9 12 15 18 21
10 Month
Next, we determine the value of D, the vertical shift. Since the x-axis appears to be the midline, D 5 0. Finally, we use the graph of the model along with the scatter plot to help determine the value of C, the horizontal shift. We shift the graph to the right 4 units in an attempt to find a good model of the data. A model for the data is d( t ) 5 9.8 cos S ( t 2 4 ) T . See Figure 9.16. (We will explain later why we chose 6 a cosine function to model these data.) The intent of Example 1 was to demonstrate the relationship between a periodic function (in this case, T( t ) 5 19.5 sin S 6 ( t 2 4 ) T 1 85.5) and a model for its derivative (in this case, d( t ) 5 9.8 cos S 6 ( t 2 4 ) T ). The derivative function was determined by finding the estimated instantaneous rate of change in the temperature in a given month. The approximated derivative of the sine function p T( t ) 5 19.5 sin c ( t 2 4 ) d 1 85.5 6 is the cosine function d( t ) 5 9.8 cos S 6 ( t 2 4 ) T . Lets examine the pieces of the function T( t ) to see where the derivative function d( t ) comes from. First, if we take the derivative of the constant 85.5, we get 0. Since the approximated derivative function has a vertical shift of 0, it makes sense that the constant 85.5 would become 0 in the derivative. Second, observe that the sine function became the cosine function in the approximated derivative. p Third, because of the Chain Rule, the argument 6 ( t 2 4 ) remains the same in the approximated derivative. Finally, the leading coefficient of 19.5 becomes 9.8 in the approximated derivative model. This is roughly equal to the product of 19.5 and
Q p 6 p p p p
FIGURE 9.16
? 19.5 < 10.2 R . Taken together, these observations make a strong argument
p 6
that the derivative of a sine function is a cosine function.
9.2 Derivatives of Trigonometric Functions y 1 0.75 0.50 0.25 0 0.25 0.50 0.75 1 slope = 0 slope = dx = cos x
dy
607
slope = 0 y = sin x slope = 1

2
THE DERIVATIVE OF A SINE FUNCTION
A sine function y 5 a sin( u ) 1 k has derivative function yr 5 a cos u ? ur This can be further verified by looking at the simplest sine function, y 5 sin x, and sketching the graph of its derivative function (Figure 9.17). Reading the slopes from y 5 sin x, we create the derivative function 5 cos x. dx
dy
slope = 1 slope = 1
3 2
y 1 (0, 1) 0.75 0.50 0.25 0 0.25 0.50 0.75 1
(2, 1)
Just In Time
Graphically Interpreting Instantaneous Rate of Change
( ) (
2
,0 2
3 ,0 2
)
3 2
y = cos x (, 1)
We can understand the behavior of derivative (rate of change) functions by estimating their graphs. We think of the derivative as the slope of the tangent line to a curve at a point, and use this idea to estimate the graph of the derivative. Plotting these estimates versus the independent variable on another axis gives us information about the derivative function. For example, we know that the derivative of a quadratic function is a linear function. This is shown graphically below.
FIGURE 9.17
slope y is more 9 negative 8 7 6 5 4 3 2 1 1 y 8 6 4 2 1 1 2 3 4 5 6 x 1 2 3 4 5 6 x slope is zero slope is less negative slope is less positive slope is more positive
2 4 6
608
We can find further graphical evidence for this relationship if we notice that on the interval S 0, itive, but gets
S 0, p R , the slope of any tangent line to the graph of y 5 sin x is pos2 p less positive as x approaches 2 . We see this in the graph of the
derivative y 5 cos x, since the function is positive but decreases on the interval
p R . Similar graphical 2 d that dx( sin x ) 5 cos x.
arguments for other intervals can be made as we verify
EXAMPLE
Finding the Derivative of a Sine Function

Calculate the derivative of the function y 5 4 sin( 2x ) 1 5.
SOLUTION
y 5 4 sin( 2x ) 1 5 yr 5 4 cos( 2x ) ? 2 1 0 yr 5 8 cos( 2x ) The extra 2 in the product comes from the Chain Rule. Observe that dx( 2x ) 5 2.
d
Constant Multiple, Chain, Constant, and Sum Rules
EXAMPLE
Finding the Derivative of a Sine Function

Calculate the derivative of the function y 5 8.5 sin S 12( x 2 6 ) T 1 75.
SOLUTION
p
y 5 8.5 sin B
p ( x 2 6 ) R 1 75 12 p p (x 2 6) R ? 10 12 12
Constant Multiple, Chain, Constant, and Sum Rules
yr 5 8.5 cos B yr 5
8.5p p cos B ( x 2 6 ) R 12 12 p (x 2 6) R 12
yr < 2.225 cos B
609
Derivative of the Cosine Function

We can use a graphical analysis of the slopes of tangent lines to y 5 cos x to determine that dx( cos x ) 5 2 sin x (Figure 9.18).
y slope = 0 1 0.75 0.50 0.25 0 0.25 0.50 0.75 1 slope = 0 slope = dx = sin x 3 ,1 2
dy
slope = 0
y = cos x slope = 1 slope = 1

2
3 2
y 1 0.75 0.50 0.25 0 0.25 0.50 0.75 1 (0, 0)
(, 0)
2
(2, 0) x 3
2
y = sin x , 1) ( 2
FIGURE 9.18
If we measure the slope of the tangent line to the function y 5 cos x at a variety of x values and then plot these slope values, we find that the result is the graph of y 5 2 sin x. We can find further graphical evidence for this relationship if we notice that on the interval ( 0, p ) , the slope of any tangent line to the graph of y 5 cos x is negative, and becomes more and more negative as x approaches the derivative y 5 2 sin x, since the function decreases on the interval S 0, then increases on the interval
p Q , 2 d p R 2 p 2
and then less and less negative as x approaches p. We can see this in the graph of and p T . Similar graphical arguments for other intervals
can be made as we verify that dx ( cos x ) 5 2 sin x.
610
THE DERIVATIVE OF A COSINE FUNCTION
A cosine function y 5 a cos( u ) 1 k has derivative function yr 5 2 a sin u ? ur
EXAMPLE
Finding the Derivative of a Cosine Function

Calculate the derivative of the function y 5 2 4 cos( 3x 1 2 ) .
SOLUTION
y 5 2 4 cos( 3x 1 2 ) yr 5 2 4 ? 2 sin( 3x 1 2 ) ? 3 yr 5 12 sin( 3x 1 2 ) The extra 3 in the product comes from the Chain Rule. Observe that
d ( 3x dx
Constant Multiple, Chain, and Sum Rules
1 2 ) 5 3.
EXAMPLE
Finding the Derivative of a Cosine Function

Calculate the derivative of the function y 5 25.25 cos S ( x 2 8 ) T 2 16. 6
SOLUTION
p
p y 5 25.25 cos c ( x 2 8 ) d 2 16 6 p p yr 5 25.25 ? 2 sin c ( x 2 8 ) d ? 6 6 yr 5 2 p 25.25p sin c ( x 2 8 ) d 6 6

p yr < 2 13.221 sin c ( x 2 8 ) d 6

Now that we have seen how to compute the derivative, lets turn our attention to interpreting the meaning of the derivative of a sinusoidal function in the context of a modeling situation.
611
EXAMPLE
Interpreting the Derivative of a Sine Model

A model for the average monthly high temperature in Phoenix, Arizona, is h( t ) 5 19.5 sin S 6 ( t 2 4 ) T 1 85.5. Evaluate hr( 5 ) and interpret its meaning.
SOLUTION We find the derivative function hr( t ) and evaluate it at t 5 5.
p
p p degrees hr( t ) 5 19.5 cos c ( t 2 4 ) d ? 6 6 month p hr( 5 ) < 10.21 cos c ( 5 2 4 ) d 6 p hr( 5 ) < 10.21 cos a b 6 hr( 5 ) < 8.842 degrees per month What does this derivative mean? The derivative may be interpreted as an instantaneous rate of changein this case, the rate at which the temperature is changing. Therefore, hr( 5 ) < 8.842 means that the temperature in month 5 (May) is increasing at a rate of 8.842 degrees per month. That is, during May, we expect the temperature to increase by approximately 8.8 degrees.
EXAMPLE 7

The model given in Section 9.1 for a Ferris wheel riders vertical position was p h( t ) 5 20 sin Q 10t R 1 30 feet above the ground at time t seconds. Find the derivative of this function at time t 5 10 and interpret its meaning.
SOLUTION
h( t ) 5 20 sin a
p t b 1 30 10 p p feet tb ? 10 10 second
hr( t ) 5 20 ? cos a hr( t ) 5
20p p cos a t b 10 10 p t b feet per second 10 p ? 10 b 10
hr( t ) < 6.28 cos a
We now determine the value of this derivative function at time t 5 10. hr( 10 ) < 6.28 cos a
hr( 10 ) < 6.28 cos( p ) hr( 10 ) < 2 6.28 feet per second Because the original function measured height above the ground at any time, the derivative measures how the height is changing over time. In this case, the negative derivative indicates that the height is decreasing at a rate of 6.28 feet per second at time t 5 10 seconds. That is, the Ferris wheel rider is on the way down.
612
EXAMPLE

The number of hours of daylight in Juneau, Alaska, can be modeled by H( t ) 5 5.75 sin c 2p ( t 2 80 ) d 1 12.25 365
where t is the number of days from January 1 of the given year. Find and explain the meaning of the derivative at t 5 80, t 5 173, t 5 263, and t 5 356.
SOLUTION We will graph the function H( t ) 5 5.75 sin S 2p ( t 2 80 ) T 1 12.25
365
FIGURE 9.19
and then use the calculator to numerically determine the derivative at each value of t. (Refer to the Technology Tip Evaluating the Derivative at a Point in Section 5.1 for keystrokes.) At t 5 80 (spring equinox), the derivative is Hr( 80 ) < 0.099 hour of daylight per day. That is, 80 days after January 1, the number of hours of daylight is increasing at a rate of 0.099 hour per day (about 5.94 minutes per day). That is, between day 80 and day 81, the number of hours of sunlight will increase by about 5.94 minutes (Figure 9.19). At t 5 173 (summer solstice), the derivative is Hr( 173 ) < 2 0.003 hour of daylight per day. That is, the number of hours of daylight is decreasing at a rate of 0.003 hour per day (about 10.8 seconds per day). In other words, between day 173 and day 174, the number of hours of sunlight will decrease by about 10.8 seconds (Figure 9.20).
FIGURE 9.20
At t 5 263 (fall equinox), the derivative is Hr( 263 ) < 2 0.099 hour of daylight per day. That is, the number of hours of daylight is decreasing at a rate of 0.099 hour per day (about 5.94 minutes per day). In other words, between day 263 and day 264, the number of hours of sunlight will decrease by about 5.94 minutes (Figure 9.21).
FIGURE 9.21
At t 5 356 (winter solstice), the derivative is Hr( 356 ) < 0.004 hour of daylight per day. That is, the number of hours of daylight is increasing at a rate of 0.004 hour per day (about 14.4 seconds per day). In other words, between day 356 and day 357, the number of hours of sunlight will increase by about 14.4 seconds (Figure 9.22). Notice that on the first day of summer (day 173) and the first day of winter (day 356), the number of hours of sunlight is not changing very much. On the first day of spring (day 80) and the first day of fall (day 263), the number of hours of daylight is changing at the greatest rate.
FIGURE 9.22
613
The Derivative of the Tangent Function

To determine the derivative of a tangent function, we begin with the definition tan t 5 cos t and apply the Quotient Rule. d d sin t ( tan t ) 5 a b dt dt cos t 5 5 5 ( cos t ? cos t ) 2 ( sin t ? 2 sin t ) ( cos t ) 2 cos2 t 1 sin2 t cos2 t 1 cos2 t
Since cos2 t 1 sin2 t 5 1 Since 1 5 sec t cos t
sin t
5 sec2 t
If we are computing a derivative using a function u 5 g( t ) in place of the variable t, we must use the Chain Rule in conjunction with the tangent rule.
THE DERIVATIVE OF A TANGENT FUNCTION
d ( a tan u ) 5 a sec2 u ? ur dt
EXAMPLE
Finding the Derivative of a Tangent Function

Compute the derivative of the function y 5 2 2 tan( e3t 1 2t )
SOLUTION
y 5 2 2 tan( e3t 1 2t ) yr 5 2 2 sec2( e3t 1 2t ) ? ( e3t ? 3 1 2 ) yr 5 ( 2 6e 2 4 ) sec ( e 1 2t )

3t 2 3t
Just In Time
Finding the Equation of a Line Given Two Points
To find the equation of a line given two points, we first find the slope of the line. Next we find the y-intercept of the line and write the equation in slopeintercept form, y 5 mx 1 b. For example, to find the equation of the line between the points ( 2, 5 ) and ( 2 1, 3 ) , we first find the slope. m5 523 2 5 2 2 (21) 3
(Continued)
614
Next, we find the value of b by substituting one of the known points into y 5 mx 1 b. 2 y5 x1b 3 2 5 5 (2) 1 b 3 55 4 1b 3
11 5b 3 The equation of the line passing through the points ( 2, 5 ) and ( 2 1, 3 ) is y 5 3x 1

2 11 . 3
EXAMPLE 10
Finding the Equation of the Tangent Line of a Tangent Function at a Point

Find the equation of the tangent line of y 5 4 tan( 3t 1 2 ) at t 5 0.
SOLUTION The problem is a bit confusing because the word tangent is used to
sin x
refer to the identity tan x 5 as well as to the line whose slope is equal to the cos x slope of the curve at the given point. Reading closely, we determine that we are to find the equation of the line that intersects y 5 4 tan( 3t 1 2 ) at t 5 0 and has slope equal to the slope of the curve at t 5 0. We first find the slope equation. y 5 4 tan( 3t 1 2 ) yr 5 4 ? sec2( 3t 1 2 ) ? 3 yr 5 12 ? sec2( 3t 1 2 )
Were interested in the slope at t 5 0.
yr 5 12 ? sec2 33( 0 ) 1 24 5 12 sec2( 2 ) 1 5 12 2 cos ( 2 ) < 69.29
So the slope is 69.29. To find a point on the line, we evaluate the original function at t 5 0. y 5 4 tan 33( 0 ) 1 24 5 4 tan( 2 ) < 2 8.740
So the point ( 0,2 8.740 ) is on the line and is the y-intercept. The equation of the tangent line is given by y 5 mx 1 b y 5 69.29x 2 8.740
615
9.2 Summary
In this section, you learned how to find and interpret derivatives of the basic trigonometric functions. You learned that
d d 3 sin dt
and dt 3tan t4 5 sec2 t. You also learned that the other rules of differentiation, such as the Chain Rule and the Constant Multiple Rule, must be used in conjunction with these trigonometric derivatives.
t4 5 cos t,
d 3cos dt
t4 5 2 sin t,

In Exercises 15, match the graph of the function with the graph of its derivative given in AE. 1.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
2.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
3.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
4.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
5.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
A.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
B.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
C.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
D.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
E.
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
(Continued)
616
In Exercises 611, find the equation of the line that passes through the given points. 6. ( 2 2, 2 ) and ( 4, 2 3 ) 7. ( 2 4, 2 5 ) and ( 2, 2 ) 8. ( 8, 2 3 ) and ( 2 5, 2 ) 9. ( 3, 5 ) and ( 5, 7 ) 10. ( 3, 5 ) and ( 3, 2 2 ) 11. ( 2 3, 7 ) and ( 5, 7 )
9.2 Exercises
1. Imagine that you are trying to convince someone that
d 3sin x4 dx
23. y 5 cos( 2 x ) ; x 5
5 cos x. Write a convincing
p 2
argument, including graphs and symbols as appropriate, to make your case. 2. Create a graphical argument to support the fact that y 5 tan x and then sketch a graph of its derivative. Explain why it makes sense that the derivative function is the function y 5 sec2 x. In Exercises 320, find the derivative of the function. 3. f ( x ) 5 3 sin x 4. g ( x ) 5 2 2 cos x 5. d( t ) 5 2 2 sin( 3t 1 6 ) 6. n( t ) 5 2 cos( pt 2 3 ) 1 5 7. y 5 5 tan x p 8. T( t ) 5 3.2 sin a t 1 2 b 2 12.5 6 9. B( t ) 5 2 8.6 cos a 11. f ( x ) 5 2 sin2 x 13. r( u ) 5 2 tan2 u 15. y 5 32 cos( x 3 2 1 ) 4 3 17. z( u ) 5 sin u ? cos u 19. y 5 1 tan x p t 2 6 b 1 56.7 12 12. g( x ) 5 "cos x 16. y 5 x ? cos x 18. g( t ) 5 20. y 5 cos t t 3 2 sin t
d 3tan x4 dx
p 24. y 5 2 sin c ( x 2 2 ) d 1 8; x 5 4 6 25. y 5 tan x; x 5 0 Exercises 2629 deal with hours-of-daylight models at various locations around the world. 26.
Hours of Daylight in Phoenix, AZ
5 sec2 x. Begin with the graph of
A function that models the number of hours of daylight in Phoenix, Arizona, as a function of the number of days t since January 1 is P( t ) 5 2.15 sin c 2p ( t 2 80 ) d 1 12.15 365
where t 5 0 represents January 1. (Source: Modeled from data at aa.usno.navy.mil.) (a) Compute Pr( t ) . (b) Calculate and interpret Pr( 90 ) . (c) Explain why the sign (1 or 2) of your result in part (b) makes sense. 27.
10. y 5 3 tan( 2x 2 5 ) 1 9
14. y 5 sin2( x 2 1 4 )
A function that models the number of hours of daylight in Miami, Florida, as a function of the number of days t since January 1 is M( t ) 5 1.625 sin c 2p ( t 2 80 ) d 1 12.125 365
cos x sin x
where t 5 0 represents January 1. (Source: Modeled from data at aa.usno.navy.mil.) (a) Compute Mr( t ) . (b) Calculate and interpret Mr( 90 ) . (c) Explain why the sign (1 or 2) of the result in part (b) makes sense. 28.
Hours of Daylight in Sydney, AU
21. Explain how to find the 50th derivative of y 5 sin x quickly and easily. In Exercises 2225, find the equation of the tangent line at the indicated domain value. 22. y 5 sin( 2x ) ; x 5 p
A function that models the number of hours of daylight in Sydney, Australia, as a
9.2 Derivatives of Trigonometric Functions p
617
function of the number of days t since January 1 is S( t ) 5 2.15 cos c 2p ( t 1 10 ) d 1 12.15 365
per second and at an angle of 20 Q radians R , a 9 baseball player can hit a base ball 417 feet for a home run. A model for the horizontal distance d of the baseball as a function of the time t in seconds is p d( t ) 5 c 242 cos a b 2 147.4 d t feet 9 (a) Compute the derivative of the function and interpret its meaning in the context of the problem. (b) Compute and interpret dr( 5.3 ) . 32.
Projectile MotionBaseball Five seconds after being hit with an initial velocity of 242 feet per second, a model for the vertical height h of the baseball as a function of the angle of impact u is
where t 5 0 represents January 1. (Source:

Modeled from data at aa.usno.navy.mil.)
(a) Compute Sr( t ) . (b) Compute and interpret Sr( 90 ) . (c) Explain why the sign (1 or 2) of your result in part (b) makes sense. 29.
Hours of Daylight in Juneau, AK
A function that models the number of hours of daylight in Juneau, Alaska, as a function of the number of days t since January 1 is J( t ) 5 5.835 sin c 2p ( t 2 80 ) d 1 12.335 365
where t 5 0 represents January 1. (Source: Modeled from data at aa.usno.navy.mil.) (a) Compute Jr( t ) . (b) Compute and interpret Jr( 90 ) . (c) Explain why the sign (1 or 2) of your result in part (b) makes sense. Exercises 3033 deal with the motion of a projectile. Projectile motion refers to the motion of an object, such as a baseball hit by a bat, that is propelled into the air, only to return to the ground due to the effects of gravity. 30. Using the website http://www.exploratorium.edu/ baseball/scientificslugger.html, the authors determined that with an initial velocity of 242 feet p per second and at an angle of 20 Q 9 radians R , a baseball player can hit a baseball 417 feet for a home run. A model for the vertical height h of the baseball as a function of the time t in seconds is
Projectile MotionBaseball
h( u ) 5 3 1 1210 sin( u ) 2 400 feet (a) Compute the derivative of the function and interpret its meaning in the context of the problem. (b) Compute and interpret hr( 10 ) . 33.
Projectile MotionBaseball Five seconds after being hit, a model for the horizontal distance d traveled by a baseball as a function of the angle of impact u is
d( u ) 5 1210 cos( u ) 2 737 feet (a) Compute the derivative of the function and interpret its meaning in the context of the problem. p (b) Compute and interpret dr A 18 B .
34. Consider the function f ( x ) 5 sin x 1 x. Using graphs and formulas, explain why f r( x ) . 0 for all x. 35. Consider the function f ( x ) 5 cos x 2 x. Using graphs and formulas, explain why f r( x ) # 0 for all x. 36. Graph the function y 5 cos( 2x ) and its derivative. Write a few sentences explaining why the graph of the derivative visually makes sense when compared with the graph of y 5 cos( 2x ) . 37. Determine a function F( x ) such that Fr( x ) 5 2 sin x ? cos x. 38. Determine a function F( x ) such that Fr( x ) 5 sin x ? sec2 x 1 sin x.
p h( t ) 5 3 1 242 sin a b t 2 16t 2 feet 9 (a) Compute the derivative of the function and interpret its meaning in the context of the problem. (b) Use the derivative to determine the height of the ball when it reaches its maximum height. 31. Using the website http://www.exploratorium.edu/ baseball/scientificslugger.html, the authors determined that with an initial velocity of 242 feet
Projectile MotionBaseball
618
CHAPTER 9 Trigonometric Functions sin x . xS0 x cos x 2 1 . x
41. When x is measured in degrees, find lim 39. Use the limit definition of the derivative,
f (x 1 h) 2 f (x) lim , h hS0 d 3sin x4 dx
to show that
5 cos x.
42. When x is measured in degrees, find lim
xS0
40. Use the limit definition of the derivative,

hS0
lim
d 3cos x4 dx
f (x 1 h) 2 f (x) , h
to show that
43. Use the results of Exercises 39 42 to show that, d p when x is measured in degrees, dx 3sin x4 5 180 cos x and dx 3cos x4 5 2 180 sin x.
d p
5 2 sin x.
9.3 Integrals of Trigonometric Functions
Find and interpret the integral of a trigonometric function Use technology to find and interpret the definite integral of a trigonometric function
We saw earlier that a model for the rate of change in monthly temperatures in Phoenix, Arizona, could be given by p d( t ) 5 9.8 cos c ( t 2 4 ) d 6 where t represents time in months (t 5 1 represents January) and d (t ) represents the rate of change in monthly temperature measured in degrees per month. In this section, we will discuss how to find an accumulation function given a rate-of-change function such as the difference-in-temperatures model.
11 10 9 8 7 6 5 4 3 2 1 0 0 1 2 3 4 5 Month 6 7
Rate of change in temperature
In the Phoenix difference-in-temperatures model, d( 6 ) < 4.9 means that in month 6 (June), the average monthly high temperature in Phoenix is increasing at
7
a rate of about 4.9 degrees Fahrenheit per month. What does 3 d( t ) dt represent
7 1
in this context? Recall that 3 d( t ) dt represents the area under the curve d(t) on
1
the interval [1, 7], as shown in Figure 9.23. If we analyze the units used in this situation, the area under the curve can be interpreted as
degrees month
? month 5 degrees. That is, with vertical units of
degrees month
and
FIGURE 9.23
horizontal units of month, the area of a representative rectangle under the curve would have units
degrees month
? month 5 degrees. The area underneath the curve represents the

p
total change in degrees from month 1 (January) to month 7 (July). To compute this area, we will need to find the antiderivative of d( t ) 5 9.8 cos S ( t 2 4 ) T . Recall 6 that
d 3sin x4 dx
5 cos x . Therefore, the antiderivative of a function involving cosine
will be a function involving sine.
619
THE INTEGRAL OF A COSINE FUNCTION
3cos x dx 5 sin x 1 C
We will need to apply our rules of integration to find the integrals of more complicated sinusoidal functions such as the one we are currently analyzing.
EXAMPLE 1
Finding an Accumulated Change in Temperature

Find the total change in the average monthly high temperature from January to July in Phoenix, Arizona. A model for the rate of change in Phoenix monthly high temperatures is given by p d( t ) 5 9.8 cos c ( t 2 4 ) d degrees per month 6 where t is the month, with t 5 1 corresponding to January.
SOLUTION We need to compute 3 9.8 cos S ( t 2 4 )T dt. 6
1 7
p p 3 9.8 cos c 6 ( t 2 4 ) d dt 5 9.8 3 cos c 6 ( t 2 4 ) d dt 1 1 Let u 5

7
p p 6 ( t 2 4 ) . Then du 5 dt and du 5 dt. p 6 6

b
p 6 9.8 3 cos c ( t 2 4 ) d dt 5 9.8 3 cos( u ) du p 6 1 a 5 9.8 sin( u ) ? 6 b ` pa
p 6 7 5 9.8 sin c ( t 2 4 ) d ? ` p1 6 p 6 p 6 5 a 9.8 sin c ( 7 2 4 ) d ? b 2 a 9.8 sin c ( 1 2 4 ) d ? b p p 6 6 p 6 2p 6 5 a 9.8 sin a b ? b 2 a 9.8 sin a b ? b p p 2 2 5 a 9.8 ? 1 ? 6 6 b 2 a 9.8 ? 2 1 ? b p p
5 18.717 2 ( 2 18.717 ) < 37.4 degrees According to our model, the average monthly high temperature in Phoenix, Arizona, increases by nearly about 37 degrees Fahrenheit from January to July.
620
Using Technology to Compute Definite Integrals of Trigonometric Functions

We can use technology to numerically compute definite integrals by using the Technology Tip introduced in Section 6.4.
EXAMPLE
Using Technology to Find a Definite Integral Graphically

Use a graphing calculator to confirm the results of Example 1.
SOLUTION Applying the aforementioned Technology Tip, we use the graphing
calculator to confirm that 3 9.8 cos S ( t 2 4 ) T dt < 37.4 (Figure 9.24). 6

1
The calculator confirms our earlier conclusion that the average monthly high temperature in Phoenix increases by about 44 degrees from January to July.
FIGURE 9.24
EXAMPLE
Finding the Antiderivative of a Cosine Function

Evaluate 33 cos( 2t 1 4 ) dt .
SOLUTION
33 cos( 2t 1 4 ) dt 5 3 3cos( 2t 1 4 ) dt 5 3 sin( 2t 1 4 ) ? 1 1C 2

1 Using substitution with u 5 2t 1 4 and dt 5 du 2
3 5 sin( 2t 1 4 ) 1 C 2 It is important to remember that we need to apply the Chain Rule in reverse as we evaluate the antiderivative. If we check our answer by taking the derivative, we see that, by the Chain Rule, the derivative of 2t 1 4 is 2. Therefore, we must multiply our antiderivative by the reciprocal of 2, or .
1 2
EXAMPLE
Interpreting a Definite Integral

A model for the velocity of a piston moving up and down is given by v ( t ) 5 4 cos( pt ) where t is time measured in seconds and v(t) is velocity measured in inches per second. When the velocity is positive, the piston is moving upward. When the

2.5
621
velocity is negative, the piston is moving downward. Find 3 v ( t ) dt and inter1.5
pret the result in the context of this situation.

SOLUTION We first examine a graph of the velocity function (Figure 9.25). The function is positive (above the x-axis) over the interval [1.5, 2.5], meaning that the piston is moving in the upward direction during the entire interval. We compute the definite integral as follows.
2.5 2.5
3 v( t ) dt 5 3 4 cos( pt ) dt
1.5 1.5 2.5
FIGURE 9.25
5 4 3 cos( pt ) dt
1.5
5 4 sin( pt ) ? 5
1 2.5 ` p 1.5
4 4 sin( 2.5p ) 2 sin( 1.5p ) p p
5 1.273 2 ( 2 1.273 ) 5 2.546

2.5
The definite integral 3 v( t ) dt represents the area under the velocity curve. If we
1.5
analyze the units, we find that
inches second
? second 5 inches. This area is interpreted
as the total distance, measured in inches, traveled by the piston during the one second between t 5 1.5 and t 5 2.5. In this case, the piston traveled about 2.5 inches in the upward direction during that one-second interval.
Just In Time
Absolute Value Functions
When thinking about the absolute value of a function, we consider only the magnitude of the vertical portion of the function value and not the direction. That is, if the function value is 2 7, we record it as 7 when graphing the absolute value. Compare the graphs of y 5 x 2 1 4x and y 5 Z x 2 1 4x Z . The graph of y 5 x 2 1 4x is:
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
(Continued)
622
Notice that f ( x ) , 0 when 0 , x , 4. To graph y 5 Z x 2 1 4x Z , imagine that f ( x ) . 0 when 0 , x , 4.

y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
EXAMPLE
Interpreting a Definite Integral

1
Using the model for the velocity of a piston given in Example 4, find 3 v ( t ) dt
0
and interpret the result in the context of the situation.

SOLUTION We first examine a graph of the velocity function (Figure 9.26). The piston is moving in the upward direction on the interval [0, 0.5) and in the downward direction on the interval (0.5, 1]. We compute the definite integral.
1 1
3 v ( t ) dt 5 3 4 cos( pt ) dt
0 0 1
FIGURE 9.26
5 4 3 cos( pt ) dt
0
5 4 sin( pt ) ? 5
1 1 ` p0
4 4 sin( p ) 2 sin( 0 ) p p
5 0 2 (0) 50
1
The definite integral 3 v( t ) dt represents the sum of the signed areas between the
0
velocity curve and the horizontal axis. Since the piston travels the same distance in the upward direction as it does in the downward direction, the definite integral shows us that the net distance traveled by the piston is zero (after 1 second, it is in the same position that it was when it started).
623
If we wanted to find the total distance traveled by the piston in Example 5 without regard to direction, we could determine how far it traveled while traveling upward (on the interval [0, 0.5)) and how far it traveled while traveling downward (on the interval (0.5, 1]), and then add the distances together. If we take the absolute value of these definite integrals, we will be adding total distance traveled without regard to direction traveled. We need to compute ` 3 v( t ) dt ` 1 ` 3 v( t ) dt ` . This time, we will perform
0 0.5 0.5 1
the computation using the graphing calculator (Figures 9.27 and 9.28).
0.5 1
3 v ( t ) dt < 1.273
0
3 v( t ) dt < 2 1.273
0.5
FIGURE 9.27
0.5 1
FIGURE 9.28
2 3 v( t ) dt 2 1 2 3 v( t ) dt 2 < Z 1.273 Z 1 Z 2 1.273 Z 5 2.546

0 0.5
The piston traveled a total distance of 2.546 inches during the one-second interval [0, 1].
The Integral of the Sine Function

We can also find the antiderivative of a sine function. Recall that d ( cos x ) 5 2 sin x dx 2 d ( cos x ) 5 sin x dx
d ( 2 cos x ) 5 sin x dx Therefore, if we are looking for the antiderivative of a sine function, we need a negative cosine function.
THE INTEGRAL OF A SINE FUNCTION
3sin x dx 5 2 cos x 1 C
We can use this definition to find the average temperature in Phoenix, Arizona, over a given period of time.
624
EXAMPLE
Determining the Average Value of a Function

A model for the average monthly high temperature in Phoenix, Arizona, is given by p T( t ) 5 19.5 sin c ( t 2 4 ) d 1 85.5 degrees Fahrenheit 6 where t is time in months (t = 1 represents January). Use this model to find the average monthly high temperature in Phoenix between January and March.
3
SOLUTION We will first compute the definite integral 3 T( t ) dt.

1
p 3 T ( t ) dt 5 3 c 19.5 sin c 6 ( t 2 4 ) d 1 85.5 d dt

1 1
3 p 6 5 2 19.5 cos c ( t 2 4 ) d ? 1 85.5t ` p 6 1
p 6 p 6 5 a 2 19.5 cos c ( 3 2 4 ) d ? 1 85.5( 3 ) b 2 a 2 19.5 cos c ( 1 2 4 ) d ? 1 85.5( 1 ) b p p 6 6 5 a 2 19.5 cos a 5 138.747 We can confirm our work using the graphing calculator (Figure 9.29). 2p 6 2p 6 b ? 1 256.5 b 2 a 2 19.5 cos a b ? 1 85.5 b p p 6 2
5 224.247 2 85.5
FIGURE 9.29
What does our result represent? When we analyze the units, we find that
3
the definite integral 3 T( t ) dt has units degree-months. These units are not
1
meaningful to us and do not provide an answer to the question posed. We want to determine the average monthly high temperature in Phoenix between January and March. To determine this value, we divide the value of the definite integral (138.747 degree-months) by 2 (the change in time). Analyzing the units shows that this operation will produce a meaningful result.
3
3 T( t ) dt
1
138.747 degree ? months 5 69.374 degrees 2 months
The average high monthly temperature in Phoenix, Arizona, during the months January, February, and March is about 69 degrees Fahrenheit.
625
EXAMPLE
Determining the Average Value of a Function

The number of hours of daylight per day in Juneau, Alaska, can be modeled by H( t ) 5 5.75 sin c 2p ( t 2 80 ) d 1 12.25 365
where t is the number of days from January 1 of the given year. Find the average number of hours of daylight per day in Juneau during the month of December.
SOLUTION November 30 is the 334th day of the year and December 31 is the
365
365th day of the year. We first compute the definite integral 3 H( t ) dt. This will
334
give us the number of hour-days from the end of November 30 to the end of December 31.
365 365
3 H( t ) dt 5 3
334 334
c 5.75 sin c
2p ( t 2 80 ) d 1 12.25 d dt 365
5 2 5.75 cos c
365 2p 365 ( t 2 80 ) d ? 1 12.25t ` 365 2p 334
5 a 2 5.75 cos c
2p 365 ( 365 2 80 ) d ? 1 12.25( 365 ) b 365 2p 2p 365 ( 334 2 80 ) d ? 1 12.25( 334 ) b 365 2p
2 a 2 5.75 cos c 5 a 2 5.75 cos a 5 204.079
570p 365 508p 365 b ? 1 4471.25 b 2 a 2 5.75 cos a b ? 1 4091.5 b 365 2p 365 2p
5 4406.966 2 4202.887 We can confirm our work using the graphing calculator (Figure 9.30).
365
When we analyze the units, we find that the definite integral 3 H( t ) dt has
334
FIGURE 9.30
units of hours-days. As we saw before, these units are not meaningful to us and do not provide an answer to the question posed. We want to determine the average number of hours of daylight per day in Juneau during the month of December. To determine this value, we divide the result of the definite integral (204.079 hours-days) by 31 (the number of days in December). Analyzing the units shows that this operation will produce a meaningful result.
365
3 H( t ) dt
334
31
204.079 hours ? days 5 6.583 hours 31 days
During the month of December, Juneau enjoys an average of about 6.6 hours of daylight each day.
626
The Integral of the Tangent Function

To determine the integral of the tangent function, we will again take advantage of the definition tan t 5 cos t and integrate using the substitution method. sin x 3tan x dx 5 3 a cos x b dx Let u 5 cos x. Then du 5 2 sin x dx. sin x 3tan x dx 5 3 a cos x b dx 1 5 32 du u 5 2 ln Z u Z 1 C 5 2 ln Z cos x Z 1 C
Since u 5 cos x and du 5 2 sin x dx
sin t
THE INTEGRAL OF A TANGENT FUNCTION
3tan x dx 5 2 ln Z cos x Z 1 C
EXAMPLE
Finding the Antiderivative of a Tangent Function

Evaluate the integral 32 tan( 3x ) dx.
SOLUTION
32 tan( 3x ) dx 5 2 3tan( 3x ) dx 5 2( 2 ln Z cos( 3x ) Z ) ? 1 1C 3
2 5 2 ln Z cos( 3x ) Z 1 C 3
9.3 Summary
In this section, you learned how to find and interpret antiderivatives of trigonometric functions. You learned that 3sin t dt 5 2 cos t 1 C, 3cos t dt 5 sin t 1 C, and 3tan t dt 5 2 ln Z cos t Z 1 C. You also learned that the other rules of integration, such as substitution and the Constant Multiple Rule, must be used in conjunction with these trigonometric integrals.
627

In Exercises 15, you are given the graph of the function y 5 f ( x ) . Use this graph to sketch a graph of the function y 5 Z f ( x ) Z . 1.
y 2 1 3 2 1 1 2 3 4 1 2 3 x
4.
y 2 1 3 2 1 1 2 3 4 1 2 3 x
5.
y 0.5 0.25
2.
y 8 6 4 2 4 3 2 1 2 1 2 x 4 2 0.25 0.5 2 4
3.
y 1 0.5 3
2 2 2
3 x
2
9.3 Exercises
In Exercises 120, find the general antiderivative of the function. 1. f ( x ) 5 2 sin x 2. g( x ) 5 9 2 cos x 1 3. r( t ) 5 cos t 3 4. f ( t ) 5 2 5 sin( t 2 4 ) 5. v( t ) 5 3 tan t 7. s( t ) 5 2 sin c 6. a( t ) 5 5 cos 32( t 2 5 ) 4 2 6 p (t 2 4) d 1 5 12
628
8. s( x ) 5 sin( 2x ) 2 1 2 9. h( x ) 5 3 cos a x b 1 x 3 10. f ( x ) 5 2 cos( 4x 2 3 ) 1 6 11. p( x ) 5 x ? sin x 2 12. r( x ) 5 2x 3 ? cos( x 4 ) 13. f ( t ) 5 sin t ? cos t 14. g( x ) 5 ( 1 1 sin x ) 5 ? cos x sin( 2x ) 15. h( x ) 5 1 1 cos( 2x ) 16. r( u ) 5 u ? sin u 17. y 5 x 2 ? cos x 18. f ( x ) 5 x ? tan( x 2 ) cos t 19. y 5 sin t 20. h( x ) 5 sin2 x ? cos x In Exercises 2135, calculate the indefinite integral. 21. 32 sin( 2x ) dx x 22. 32 3 cos a b dx 3 23. 35 tan( 4x ) dx 2 4t 24. 3sin a b dt 9 25. 3cos( 2 2 x ) dx 26. 3"cos t ? sin t dt 27. 3esin t cos t dt cos t 28. 3 dt 2 2 sin t 2 sin u 29. 3 du ! 1 1 cos u 30. 3sin( 2p 2 t ) dt p 31. 33 sin c ( x 1 2 ) d dx 5
2p 32. 32 2 cos c ( t 1 2 ) d dt 365 33. 3tan x ? sec2 x dx 34. 3x ? tan( x 2 ) dx 35. 3( t 2 cos 2t ) dt In Exercises 3639, apply the concept of integration to solve the real-world application. 36.
Hours of Daylight in Phoenix, AZ
A function that models the number of hours of daylight per day in Phoenix, Arizona, as a function of time t in days is P( t ) 5 2.15 sin c 2p ( t 2 80 ) d 1 12.15 365

Modeled from aa.usno.navy.mil.)
Find the average number of hours of daylight per day over an entire year in Phoenix. 37.
A function that models the number of hours of daylight per day in Miami, Florida, as a function of time t in days is M( t ) 5 1.625 sin c 2p ( t 2 80 ) d 1 12.125 365

Find the average number of hours of daylight per day over an entire year in Miami. 38.
Hours of Daylight in Sydney, AU
A function that models the number of hours of daylight per day in Sydney, Australia, as a function of time t in days is S( t ) 5 2.15 cos c 2p ( t 1 10 ) d 1 12.15 365

Find the average number of hours of daylight per day over an entire year in Sydney.
629
39.
Hours of Daylight in Juneau, AK
42.
Average Temperature in Barrow, AK
A function that models the number of hours of daylight per day in Juneau, Alaska, as a function of time t in days is J( t ) 5 5.835 sin c 2p ( t 2 80 ) d 1 12.335 365
A function that models the average daily temperature for a particular month m in Barrow, Alaska, is given by T( m ) 5 29.44 sin( 0.5700m 2 2.656 ) 1 12.47 where m 5 1 represents January. (Source: Modeled from lwf.ncdc.noaa.gov.) Find the average temperature in Barrow, Alaska, during the period from November to March. 43.
Moon Orbit The moons distance from the earth changes as the moon orbits the earth in an elliptical path. The point at which the moon is farthest from the earth is known as the apogee. The point at which the moon is closest to the earth is known as the perigee. The perigee and apogee change from month to month in a periodic fashion. For the year 2006, the greatest apogee distance was 406,498 kilometers, on September 22. The least apogee distance was 404,079 kilometers, on June 4. (Source:

Find the average number of hours of daylight per day over an entire year in Juneau. 40.
Electricity Use For most homeowners
in the Phoenix, Arizona, area, the majority of electrical usage is for cooling the home during the hot summer months. A model for the amount of electricity used each month by one of the authors in 2002 is E( t ) 5 756 sin( 0.652t 1 2.947 ) 1 2033.258 where E is measured in kilowatt-hours and t is measured in months (t 5 1 corresponds to the end of January). (Source: Modeled from authors
electric bill.)
www.fourmilab.chrthview.)
(a) Graph E as a function of t.

12
(b) Evaluate and interpret 3 E( t ) dt.

0
(a) Create a sinusoidal model of the apogee distance as a function of time in days. (b) Find the average apogee distance for the year. 44.
Average Public K8 Enrollment
(c) Find the average amount of energy used per month (in kilowatt-hours) during the year. 41.
Electricity Use For most homeowners in the Phoenix, Arizona, area, the majority of electrical usage is for cooling the home during the hot summer months. A model for the cost of the electricity used each month by one of the authors in 2002 is
A function that models the number of students, in thousands, enrolled annually in grades K8 in the United States between 1970 and 1999 is y 5 3091 sin( 0.2041t 1 2.052 ) 1 30,135 where t 5 0 corresponds to 1970. (Source:
Modeled from Statistical Abstract of the United States, 2001, Table 232, p. 149.) Find the average number of
C( t ) 5 106.36 sin( 0.61t 1 3.11 ) 1 165.16 where C is the billed amount in dollars and t is measured in months (t 5 1 corresponds to January). (Source: Modeled from authors electric
bill.)
students enrolled annually between 1990 and 1999. 45.

Average Temperature in Colorado Springs, CO A function that models
(a) Graph C as a function of t.

12
the average daily temperature in Colorado Springs, Colorado, during a particular month m is given by T( m ) 5 20.28 sin( 0.5498m 2 2.342 ) 1 48.67 where m 5 1 represents January. (Source: Modeled from lwf.ncdc.noaa.gov.) Find the average daily temperature in Colorado Springs during the period from November to March.
(b) Evaluate and interpret 3 C( t ) dt.

0
(c) Find the average amount billed per month during the year.
630
50. Provide a geometric and symbolic explanation for 46. Find the first positive value of a that makes the statement 3 ( sin x 1 cos x ) dx 5 0 true. Create a
0 a
why 3
p>4
2p>4
Z tan x Z dx 5 ln 2.
graph showing why your result makes sense. 47. When explaining integration of trigonometric functions to a classmate, you say that 3cos x dx 5 sin x 1 C . The classmate does not understand what 1C means in the context of integration. What would you tell your classmate to help her understand? 48. Consider the function g( x ) 5 3 cos t dt. Write a
0 x
51. Evaluate 3ex sin x dx using repeated integration by parts. 52. Evaluate 3ex cos x dx using repeated integration by parts. 53. Evaluate 3cot x dx. 54. Evaluate the improper integral 3 tan x dx. What
0
p> 2
description telling what the output of the function g( x ) will be. Then determine the values of x such that g( x ) 5 0. 49. Provide a geometric and symbolic explanation for
p
makes the integral improper? 55. Evaluate 3sec x dx. Try as many methods as you can think of. If you get stuck, research the actual method used and then write an explanation of the process and why it makes sense.
why 3 Z cos x Z dx 5 2.
0
After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What do we mean when we refer to a periodic function? (9.1) 2. When working with a periodic function, we often refer to its amplitude, period, horizontal shift, and vertical shift. Explain what these terms mean in the context of a periodic function. (9.1) 3. How do you use equations, graphs, and technology to solve a trigonometric equation? (9.1) 4. Why does it make sense to say that the derivative of y 5 sin x is yr 5 cos x? Why does it make sense to say that the derivative of y 5 cos x is yr 5 2 sin x? (9.2)
Chapter 9
Study Sheet
631
5. When analyzing a periodic function representing an animal population, what would the derivative function represent? (9.2) 6. In what types of situations would it be appropriate to use the Chain Rule when taking the derivative of a trigonometric function? (9.2) 7. Why does it make sense to say that 3sin x dx 5 2 cos x 1 C ? Why does it make sense to say that 3cos x dx 5 sin x 1 C ? (9.3) 8. When analyzing a periodic function representing change in temperature in degrees per month, what would the area under the curve over a specific interval represent? (9.3) 9. In what types of situations would it make sense to find the average value of a function? How would we find this average, and what would it mean, generally speaking? (9.3) You should also be familiar with the following definitions, procedures and tests, and formulas that were emphasized in this chapter.
Definitions
Midline: An imaginary horizontal line that runs through the center of a sinusoidal function Amplitude: The vertical distance from an imaginary midline to the maximum or minimum value of a sinusoidal function Period: The domain interval required for a periodic function to complete one cycle Vertical Shift: The distance that the midline of the basic sinusoidal function is shifted
Procedures and Tests
Modeling Data with Trigonometric Functions Find the value of A, the amplitude, by using the equation A5
Find the value of B, the frequency, by first determining the period and 2p then using B 5 period. Find the value of D, the vertical shift, by using the equation D5 maximum data value 1 minimum data value 2
Find the value of C, the horizontal shift, through a process of trial and error.
Formulas
The Derivative of a Sine Function:
d 3a sin( u ) 1 k4 5 a cos u ? ur dx
The Derivative of a Cosine Function: d 3a cos( u ) 1 k4 5 2 a sin u ? ur dx
632
The Derivative of a Tangent Function:
d ( a tan u ) 5 a sec 2 u ? ur dt
The Integral of a Sine Function: 3sin x dx 5 2 cos x 1 C The Integral of a Cosine Function: 3cos x dx 5 sin x 1 C The Integral of a Tangent Function: 3tan x dx 5 2 ln Z cos x Z 1 C

Section 9.1 In Exercises 15, determine the value of the amplitude, period, phase shift, and vertical shift.
1. y 5 2 2 sin 3p( x 1 2 ) 4 1 1 p 2. y 5 cos a 2x 2 b 2 3 2 3.
y 4 3 2 1 0 0 2 4 6 8 10 12 x
2
5.
y 3 2 1
2
In Exercises 69, solve the trigonometric equation for 0 # x , 2p. 6. 2 3 5 cos 2x 2

y 3 2 1 0 2 4 6 8 10 12 x
p 23 2
4.
7. 2 2 sin 3p( x 1 2 ) 4 1 1 5 1 8. 2 cos x 5 2 1 9. 2 sin x 5 "3
Chapter 9
Review Exercises
633
In Exercises 1011, apply sinusoidal modeling to solve the real-world application. 10. According to data found at www.neurolab.jsc.nasa .gov, NASA space shuttle missions fly at many different inclinations. Most commonly, the space shuttle orbits the earth at an inclination of 28.4. This means that the path of the space shuttle is a sinusoidal path between 28.4 north latitude and 28.4 south latitude. It takes 90 minutes for the space shuttle to orbit the earth. Using positive values to represent measurements north of the equator and negative values to represent measurements south of the equator, answer the questions that follow. (a) Create a model describing the distance D north or south of the equator (as a measure of latitude) as a function of time t in minutes. Assume that t 5 0 represents the beginning of the orbit. Assume that D( 0 ) 5 0. (b) Relative to the equator, where is the space shuttle 30 minutes after beginning its orbit? (c) Determine the first two times after beginning its orbit that the space shuttle is right above the equator. (d) Determine the first time after beginning its orbit that the space shuttle is at 10o south latitude.
Space Shuttle Orbit
p 14. P( t ) 5 2 2 cos a t 1 1 b 2 6 3 15. c( x ) 5 "sin( 2x ) 16. y 5 2 tan( 6x 1 p ) 2 2 In Exercises 1718, find the equation of the tangent line at the indicated domain value. p 17. y 5 cos( 2x ) ; x 5 2 p 18. y 5 5 sin c ( x 2 1 ) d ; x 5 2 3 In Exercises 1921, apply the differentiation rules of trigonometric functions to solve the real-world problem. 19.
Natural Gas Storage Data from the Quantitative Environmental Learning Project (www.seattlecentral.edu/qelp) show the variation in natural gas storage in the U.S. (almost all underground) with time (on a weekly basis) from April 1997 to March 2000, covering three gas years. These data may be modeled by the function
G( t ) 5 877 sin( 0.121739t 2 2.03472 ) 1 2061.39 where t is the number of weeks since April 1997 and G( t ) is the amount of natural gas in billions of cubic feet. (a) Compute Gr( t ) . (b) Compute and interpret Gr( 84 ) . (c) Explain why the sign (1 or 2) of your result in part (b) makes sense. 20. Sardine Population Data from the Centre for Population Biology indicate that the population of sardines in the South Pacific Ocean near Chile, South America, varies sinusoidally with time. (Source: www.sw.ic.ac.uk.) A model for the population S as a function of time t in years since 1974 is given by S( t ) 5 4774 sin( 0.271431t 2 0.567782 ) 1 4650 (a) Compute Sr( t ) . (b) Compute and interpret Sr( 10 ) . (c) Explain why the sign (1 or 2) of your result in part (b) makes sense. 21. Data from the Quantitative Environmental Learning Project (www.seattlecentral.edu/qelp) show that the number of pounds of Dungeness crab harvested
Dungeness Crab Harvest
11.
Data from the Centre for Population Biology indicate that the population of the mallard duck in the Woodworth, North Dakota, area varied sinusoidally between 1965 and 1981. The minimum population size of 33 occurred in 1974 and the maximum population size of 78 occurred in 1976. (Source: www.
Population Data
sw.ic.ac.uk.)
(a) Create a model describing the population P of the mallard duck as a function of time t in years since 1965. (b) What was the mallard duck population in 1980? (c) Find the second time after 1965 that the mallard duck population was 40 ducks.
Section 9.2 In Exercises 1216, find the derivative of the function.

12. f ( x ) 5 2 4 sin( 2x ) 13. d( t ) 5 5 sin( 4t 2 8 )
634
each year varies sinusoidally. This is due to the southward-directed winds in the late spring that drive crab larvae onshore. These winds come in 5-to-8-year cycles, bringing the harvest down in similar cycles. A model for the number of pounds (in millions) of Dungeness crab harvested per year is D( t ) 5 6.762 sin( 1.05723t 2 0.342318 ) 1 12.4328 where t is the number of years since 1986. (a) Compute Dr( t ) . (b) Compute and interpret Dr( 2 ) . (c) Explain why the sign (1 or 2) of your result in part (b) makes sense.
In Exercises 3031, apply integration of trigonometric functions to solve the real-world problem. 30. Data from the Quantitative Environmental Learning Project (www.seattlecentral.edu/qelp) show that the number of pounds of Dungeness crab harvested each year varies sinusoidally. This is due to the southward-directed winds in the late spring that drive crab larvae onshore. These winds come in 5- to 8-year cycles, bringing the harvest down in similar cycles. A model for the number of pounds (in millions) of Dungeness crab harvested per year is
Dungeness Crab Harvest
D( t ) 5 6.762 sin( 1.05723t 2 0.342318 ) 1 12.4328 where t is the number of years since 1986. (a) Graph D as a function of t.
13
Section 9.3 In Exercises 2225, find the general antiderivative of the function.
22. f ( x ) 5 4 cos( 2x ) p 23. s( t ) 5 sin c ( t 1 2 ) d 2 1 6 24. p( x ) 5 x 2 ? sin x 3 25. f ( t ) 5 sin( 2t ) ? cos( 2t ) In Exercises 2629, calculate the indefinite integral. 26. 3 23 cos( 2 x ) dx 27. 3sin( cos x ) ? sin x dx 2 sin t dt 28. 3 1 1 cos t p 29. 3 c 3 sin a 2 t b 1 5 d dt 2 31.
(b) Evaluate and interpret 3 D( t ) dt.

0
(c) Find the average size of a Dungeness crab harvest between 1986 and 1999. Data from the Centre for Population Biology indicate that the population of the Canadian lynx in Minnesota varied sinusoidally with time between 1949 and 1981. (Source: www.sw.ic.ac.uk.) A model for the population L as a function of time t in years since 1949 is given by 2p L( t ) 5 200 sin c ( x 2 1 ) d 1 200 11
Canadian Lynx Population
(a) Graph L as a function of t.

32
(b) Evaluate and interpret 3 L( t ) dt.

0
(c) Find the average annual population of the Canadian lynx in Minnesota between 1949 and 1981.
P R O J E C T
What to do
1. Collect data showing the average temperature each month and the number of hours of daylight for selected days during a year for a city of your choice. 2. Model each data set with a sinusoidal function. 3. Find the derivative of each model function. 4. Evaluate the derivative of each function at three different points. 5. Interpret the real-life meaning of the results of Exercise 4. 6. Use integration to find the average temperature and the average number of hours of daylight over a period of time of your choice. 7. Write an article that could be used by your citys chamber of commerce to promote the city. Be sure to refer to the results of your mathematical computations.

The following web sites may have data that you will find helpful for this project. http://aa.usno.navy.mil/ http://www.weatherbase.com/ You may wish to visit your citys chamber of commerce web page to see the types of promotional publications it creates. For this project, pretend that your article will be included in such a publication. Some examples can be found at the following sites. http://www.duluthchamber.com/other/communityguide.aspx http://www.gilbertaz.com/go/lifestyle.php http://www.grandmn.com/relocation/index.html
635
10
Chapter
Differential Equations
T
he more things change, the more they stay the same is a common clich. It is used to make the point that change may be underfoot, but human nature really hasnt changed much. In mathematics, we study the idea of change to understand quantities that really are changing. How has a population changed over time? What about a financial investment, a learning curve, or the amount of pollution in a lake? The population of the town of Gilbert, Arizona, has been increasing at an increasing rate for the past several years. (Source: www.usbeacon.com.) Town managers seek to understand the rate at which new residents are moving into the town. This information is crucial as town officials make decisions regarding the infrastructure of the town. Future school and road construction, water availability, and emergency services are examples of important issues that need to be discussed as a town grows in population. A study of differential equations can help in creating accurate mathematical models that can aid in making wise decisions regarding town growth.
10.1 Slope Fields

Construct a slope field for a given differential equation Interpret a slope field representing a real-world situation Use a slope field to describe solution curves to differential equations Use Eulers Method to approximate solutions to differential equations
10.2 Eulers Method
10.3 Separable Differential Equations and Applications
Determine general and particular solutions of differential equations using the method of separation of variables Use separation of variables to solve differential equations that model realworld situations
10.4 Differential Equations: Limited Growth and Logistic Models

Find the equation of a limited growth model Find the equation of a logistic growth model Analyze and interpret the real-life meaning of differential equations
10.5 First-Order Linear Differential Equations
636
Solve first-order linear differential equations using the Reverse Product Rule, also known as the method of integrating factors Model real-world situations using firstorder linear differential equations
10.1 Slope Fields
637
10.1 Slope Fields
Construct a slope field for a given differential equation Interpret a slope field representing a real-world situation Use a slope field to describe solution curves to differential equations
We can describe the rate at which a population (of animals, bacteria, or even money) grows using the differential equation dP 5 kP dt This equation states that the rate of change of the population (derivative) is directly proportional to the amount of population present. If the population P is small, then the rate of change will be relatively small. If the population P is large, then the rate of change will be relatively large. If you think about it, it makes sense that the more of a particular animal there is, the faster the population will grow! This differential equation assumes that there are unlimited resources to sustain growth, that growth is continuous, and that no other populations in the ecosystem are growing. Different models will be needed for populations that dont meet these criteria. In this section, we will show how to solve equations involving a function and one or more of its derivatives by using a graphical approach.
There are many situations in which we know more about the rate at which a function is changing than we do about the value of the function itselffor example, the rate at which the temperature of a cooling object changes, the rate at which a rumor spreads through a population, or the rate at which a person learns a new task. Each of these situations can be modeled using a differential equation. A differential equation is an equation that contains an unknown function and some of its derivatives. The differential equation introduced previously contains an unknown function P and its first derivative
dP . dt
Our challenge is to understand and
sometimes compute the unknown function contained in the differential equation. We call this unknown function the solution to the differential equation. We can find the solution to a differential equation graphically, numerically, and symbolically. We will first focus on the graphical representation of the unknown function contained in the differential equation. Looking back at the differential equation for the rate of population growth,
dP dt
5 kP, we see that the derivative of a function P is equal to a constant times the
dependent variable P. We use the fact that the derivative represents the slope of the tangent line to the function P to create a slope field. This slope field allows us to understand graphically the nature of the population-versus-time function. Suppose that a population is growing at a continuous rate equal to 25 percent of its current population. The particular differential equation representing this special case is dP 5 0.25P dt
638
CHAPTER 10
This differential equation states that the rate of growth (the derivative) is equal to 0.25 times the value of P at a particular time. Lets choose a number of points ( t, P ) and compute the slope at each point using the differential equation (Table 10.1). Note that this differential equation depends only on P.
TABLE 10.1
(t, P)
P 9 8 7 6 5 4 3 2 1 0 0 1 2 3 4 5 6 7
dP 5 0.25P dt 0.25 0.25 1.00 1.25 1.25 2.00 2.25
(1, 1) (9, 1) (5, 4) (2, 5) (8, 5) (3, 8) (6, 9)

8 9 t
FIGURE 10.1
At each of these points, we draw a short line segment having the calculated slope (Figure 10.1). We can do this for any number of points. When we draw short line segments at many points showing slopes given by the differential equation, the resulting graph is called a slope field. A portion of the slope field for
dP dt
5 0.25P is shown
P 9 8 7 6 5 4 3 2 1 0
in Figure 10.2. By analyzing the slope field, we are able to understand the nature of the unknown function. For example, given the initial condition that P 5 2 when t 5 0, we can sketch the particular solution curve by following the slope at each point as we move from the point ( 0, 2 ) . See Figure 10.3.
P 9 8 7 6 5 4 3 2 1 0
FIGURE 10.2
2 3
8 9 t
FIGURE 10.3
9 t
Notice that the slope of the solution curve at any given point is approximately the same as the slopes of the short line segments adjacent to the point.
10.1 Slope Fields
639
By studying this slope field, we can better understand the nature of the population growth that was described earlier. If we hold t constant and allow P to change, we see that the rate of growth is greater as we move vertically through the slope field. That is, if we had started with a population larger than 2, the initial rate of growth would also have been larger. We can also see from the solution curve that as time moves on, the population increases, and it does so at an increasing rate.
Using Technology to Create Slope Fields

Although the TI-83 Plus doesnt come preloaded with a slope field program, several free slope field programs are available at www.ticalc.org. One such program is slpfld.8xp, written by Mike Miller of Corban College. We transferred the downloaded program from our computer to our calculator using the TI-CONNECT software. At the end of this section is a Technology Tip designed to help you use this program.
EXAMPLE
Using a Slope Field to Model the Spread of a Rumor

A differential equation that can be used to model the rate at which a rumor is spreading through a population is dR 5 0.25R( 1 2 R ) dt where R represents the percentage of people who have heard the rumor, and ( 1 2 R ) represents the percentage of people who have not heard the rumor, at a given time t. Use a slope field to analyze the nature of the solutions to this differential equation.
SOLUTION First, we observe that if R 5 0 (no one knows the rumor),
dR dt
5 0.
R 1.00 0.75 0.50 0.25 0 0 1 2 3 4 5 6 7 8 9 t
This means that if no one knows the rumor, the rumor cannot spread. On the other dR hand, if R 5 1, dt 5 0. This means that if everyone knows the rumor, the rumor also cannot spread. We should expect to see this characteristic of the differential equation in the slope field. We will create the slope field (Figure 10.4). We see that when R is relatively small (look toward the bottom of the slope field), the rate of spread of the rumor is relatively small. In fact, as we said earlier, if R 5 0,
dR dt
5 0. We see that as R increases (look in the middle of the slope field),
the rate of spread of the rumor becomes greater. The rumor is spreading at the greatest rate when some but not all people have heard it. There are plenty of people to tell the rumor to and plenty of people to hear it. As R approaches 1, the rate of spread of the rumor is small once again. As more and more people have heard the rumor, there are fewer people who have yet to hear it, thus slowing its rate of spread. As mentioned earlier, when R 5 1,
dR dt
FIGURE 10.4
5 0.
640
CHAPTER 10
EXAMPLE
Using a Slope Field to Analyze Heating and Cooling

A differential equation that expresses the rate at which the temperature of a cooling object, such as a hot cup of coffee, changes is given by dT 5 k( T 2 R ) dt where T represents the temperature of the object, R represents the temperature of the surrounding air (for example, room temperature), t represents time, and k is the constant of proportionality. Use a slope field and a table of slope values to analyze the nature of the solutions to this differential equation if k 5 2 0.125 and R 5 75F.
SOLUTION We will analyze the differential equation
dT dt
5 2 0.125( T 2 75 ) .
First, we notice that when the temperature of the cooling object is much greater than the temperature of the surrounding air, the rate of change of the temperature is relatively great. As the temperature of the cooling object approaches the temperature of the surrounding air, the rate of change becomes less. In this case, if dT the temperature of the surrounding air reaches room temperature, then dt 5 0. If the temperature of the object is below room temperature, the rate of change of temperature will be positive as the object warms to room temperature (Table 10.2).
TABLE 10.2
T 150 125 100 85 75 65 50 25 0
dT 5 2 0.125( T 2 75 ) dt dT 5 2 0.125( 150 2 75 ) 5 2 9.375 dt dT 5 2 0.125( 125 2 75 ) 5 2 6.25 dt dT 5 2 0.125( 100 2 75 ) 5 2 3.125 dt dT 5 2 0.125( 85 2 75 ) 5 2 1.25 dt dT 5 2 0.125( 75 2 75 ) 5 0 dt dT 5 2 0.125( 65 2 75 ) 5 1.25 dt dT 5 2 0.125( 50 2 75 ) 5 3.125 dt dT 5 2 0.125( 25 2 75 ) 5 6.25 dt dT 5 2 0.125( 0 2 75 ) 5 9.375 dt
10.1 Slope Fields T 150 135 120 105 90 75 60 45 30 15 0
641
The slope field will confirm these characteristics (Figure 10.5). We see that the short segments are very steep and negative when the temperature is high. As the temperature gets closer to room temperature, the slopes become less steep. We see horizontal line segments at room temperature, supporting our earlier statement that when T 5 75,
dT dt
5 0. We see positive slopes when the temperature is below
room temperature. Cooling objects cool quickly initially, and then cool at a slower rate as the temperature of the object approaches room temperature. Objects that are below room temperature warm up to room temperature.
0 2 4 6 8 10 12 14 16 18 20 t
FIGURE 10.5
EXAMPLE
Using a Slope Field to Analyze Ice Formation

In many states in the eastern and midwestern United States, lakes freeze during the winter months. As ice is formed, it acts as insulation, and the rate at which the ice forms decreases as the thickness of the ice increases. This occurs because heat from the water below travels up through the ice into the air. As it does, more ice is formed, but at a slower and slower rate as the ice becomes thicker. A model for the rate at which ice forms is dI k 5 dt I where I represents the thickness of the ice in inches, t represents time in days, and k is a constant of proportionality. Use a slope field to analyze the nature of the solutions to this differential equation, given k 5 1.
SOLUTION We see from the differential equation
dI dt dI dt
small,
is relatively large. As I increases,

I 10 9 8 7 6 5 4 3 2 1 0 0 3 6
1 I
5 I that when I is relatively
decreases. The slope field should
confirm these observations (Figure 10.6).
9 12 15 18 21 24 27 t
FIGURE 10.6
If we look near the bottom of the slope field, we see that the slope segments are steep. This means that
dI dt
is relatively large. As I increases (we move vertically up
the slope field), the slope segments get less and less steep. As described, the rate at which the ice forms is decreasing as the ice gets thicker. The rate at which the thickness of the ice changes is near 0 when I gets relatively large (look at the top of the slope field).
642
CHAPTER 10
EXAMPLE
Using a Slope Field to Analyze Carrying Capacity

The population of a particular species will often have a carrying capacity. This is the number of animals at which the population ceases to change in size. Unlike the differential equation for unlimited population growth given in Example 1, a more realistic model for population growth is given by dP P 5 kP a 1 2 b dt K where P represents the population of the species, k is a constant of proportionality, and K is the carrying capacity. Given a slope field for the case in which k 5 0.15 and K 5 10,000, sketch two particular solution curves, one where P , 10,000 and one where P . 10,000. Describe the growth of the species over time.
SOLUTION We will examine the differential equation
dP dt
5 0.15P Q 1 2 10,000 R .
The slope field for this differential equation is shown in Figure 10.7.
P 14,000 12,000 10,000 8000 6000 4000 2000 0 0 5 10 15 20 25 30 35 t
FIGURE 10.7
We will sketch the particular solution curve for the initial condition t 5 0, P 5 1000 (Figure 10.8).
P 14,000 12,000 10,000 8000 6000 4000 2000 0 0 5 10 15 20 25 30 35 t
FIGURE 10.8
This particular solution curve shows that when the original population contains 1000 members, it grows slowly initially. As the population increases, the rate at which it grows increases. As the population gets close to 10,000, the rate at which the population is growing decreases until the rate gets close to zero.
10.1 Slope Fields
643
Next, we will sketch the particular solution curve for the initial condition t 5 0, P 5 15,000 (Figure 10.9).
P 14,000 12,000 10,000 8000 6000 4000 2000 0 0 5 10 15 20 25 30 35 t
FIGURE 10.9
This solution curve shows that when the population initially contains 15,000 members, it decreases toward the carrying capacity of 10,000 members. This decrease is relatively quick initially, but slows as the population nears 10,000.
10.1 Summary
In this section, you learned how to create and interpret a slope field for a given differential equation. By analyzing the slope field, you were able to gain an understanding of the nature of the solution to the differential equation. The solution to the differential equation is the unknown function that satisfies the differential equation. Even though we do not necessarily know this solution function, we can understand its nature by examining the slope field created from the differential equation.
Creating a Slope Field 1. Download slpfld.8xp from www.ticalc.org. Then press PRGM to bring up the program list and then select
SLPFLD.
(Continued)
644
CHAPTER 10
2. Press ENTER twice to bring up the welcome screen.
3. Present ENTER again. Several options are displayed. For this example, we will use dx 5 0.25y.
dy
4. Select 1:input dy/dx. Use the key sequence ALPHA 1 to bring up the y variable. Then press ENTER to return to the options screen.
5. If you know the solution to the differential equation, select 2:Known solution and type the solution. Otherwise, go to step 6. (For our example, we dont know the solution.)
6. Select 3:Change Window to specify the dimensions of the viewing window. We will use the standard window and high resolution.
7. Select 4:View slp field to view the slope field. If you want to adjust the window and redraw the slope field, press ENTER and adjust the window settings using the appropriate option in step 6 above.
10.1 Slope Fields
645
10.1 Exercises
In Exercises 16, match each slope field with the corresponding differential equation. Note that there are six slope fields but eight differential equations (labeled AH), so two equations will not have matches. A. dy 5x dx E. dy 5 sin x ? sin y dx 5.
y 4 2 4 2 2 4 x
2 4
dy B. 5x2y dx dy 2x C. 5 2 dx y D. 1. dy x 5 y dx
y 4 2 4 2 2 4
dy F. 5 cos y dx dy G. 5y dx H. dy 5y2x dx
6.
3 2 2
3 x
2
3
2 2
3
2
2 4
2.
y 4 2 4 2 2 4 x
In Exercises 710, do the following: (a) Sketch a solution curve through the given point on the slope field. (b) Write a description of the behavior of this solution curve and what it represents in the context of the problem. 7. Animal Population Growth Suppose a population of animals is modeled by the differential equation P dP 5 0.3P a 1 2 b dt 1200 where P is the population in thousands and t is time in years. In this case, when t 5 0, P 5 2. The corresponding slope field follows.
P 1000
2 4
3.
3 2 2
800 600 400 200 0 0 10 20 30 t
4 2 3
2 2
4.
y 4 2 4 2 2 4 x
2 4
8. Learning Curve Suppose that the rate at which an employee learns a new task during a period of training time is modeled by the differential equation dL 5 0.5( 1 2 L ) dt
646
CHAPTER 10
where L is the level of performance and t is time in days. In this case, when t 5 0, L 5 0. The corresponding slope field follows.
L 0.8 0.6 0.4 0.2 0 0 2 4 6 8 t
11. Changing Volume Suppose that a salt-water mixture is being poured into a tank of salt water at a rate of 3 gallons per minute. Furthermore, salt water is draining from the tank at a rate of 2 gallons per minute. A differential equation that models the rate at which the amount of salt in the tank is changing is dS 2S 562 dt 51t where S is measured in pounds of salt and t is time in minutes. Choose the slope field that represents this situation, and explain your choice. A. S
25 20 15 10 5 0 0 2 4 6 8 t
9. Growth of an Investment Suppose that the rate at which an investment grows, if $2000 is added to the account each year and the account earns 6% interest compounded continuously, is modeled by the differential equation dM 5 0.06M 1 2000 dt Initially, when t 5 0, M 5 0, where t is time in years and M is the amount of money in the account. The corresponding slope field follows.
M 30,000 20,000 10,000 0 0 2 4 6 8 t
B.
S 25 20 15 10 5 0 0 2 4 6 8 t
10. Newtons Law of Cooling Suppose that the rate at which a hot cup of coffee cools to room temperature (76F) is modeled by the differential equation dT 5 2 0.04095( T 2 76 ) dt where T is the temperature (F) of the coffee after t minutes. When t 5 0, T 5 186F. The corresponding slope field follows.
T 180 160 140 120 100 80 0 0
12. Drug Elimination The rate at which a particular drug leaves a patients body after administration of the drug has stopped is given by dA 5 2 0.0201A dt where A is the amount of the drug in the body after t hours. Choose the slope field that represents this situation, and explain your choice. A. A
80 60 40 20
5 10 15 20 25 t
0 0
10 20 30 40 50 t
10.1 Slope Fields
647
B.
A 80 60 40 20 0 0 10 20 30 40 50 t
where A is measured in dollars and t is measured in years. Use a slope field to analyze the nature of the solutions to this differential equation. 15. Time of Death When the forensic investigator arrives at the scene of a homicide, one of the most important things to be determined is the time of death of the victim. Suppose that a differential equation modeling the rate of change in the temperature of the cooling deceased body is given by dT 5 2 0.1282( T 2 70 ) dt where T is measured in degrees Fahrenheit and t is measured in hours. Use a slope field to analyze the nature of the solutions to this differential equation. 16.
Pediatric AIDS Based on data from
13.
Population of the United States
A differential model that represents the rate of change of the U.S. population from 1790 to 2000 is dP 5 160"P dt where P is measured in people and t is measured in years. (Source: Modeled from www.census.gov.) (a) Complete the table of values. P 5,000,000 50,000,000 100,000,000 150,000,000 200,000,000 250,000,000 (b) Describe the population growth of the United States as indicated by the differential equation and the corresponding table. (c) Sketch a slope field corresponding to this differential equation. (d) Explain what information the slope field provides about the population of the United States. (e) In 1790, the population of the United States was about 5,300,000 people. Sketch the solution curve through this point on your slope field. 14. Investment Value The rate of change of an investment account earning continuous compound interest is given by dA 5 0.061A dt dP dt
1992 to 2001, the estimated rate at which the number of new pediatric AIDS cases in the United States is changing can be modeled by dN 2 3.804e0.07758t 5 dt ( 1 1 0.0533e0.07758t ) 2 where N is the number of cases and t is the number of years since the end of 1992. (Source:
Modeled from Centers for Disease Control and Prevention data.) Use a slope field to analyze the
nature of the solutions to this differential equation. 17.

School Internet Access Based on data from 1994 to 2000, the rate of change in the percentage of public-school classrooms with Internet access may be modeled by
dy 5 0.01077y( 85.88 2 y ) dt where y is percentage points per year and t is the number of years since the end of 1994. (Source:
Modeled from Statistical Abstract of the United States, 2001, Table 243, p. 155.) Use a slope field to
analyze the nature of the solutions to this differential equation. 18.

National Internet Usage Based on data from 1995 to 1999 and U.S. Census Bureau projections for 2000 to 2004, the rate of change in annual per capita Internet usage may be modeled by
dy 5 0.002882y( 232.2 2 y ) dt
648
CHAPTER 10
where y is hours per year and t is the number of years since 1995. (Source: Modeled from Statistical
Use a slope field to analyze the nature of the solutions to this differential equation. 19. Spread of Information Suppose that in a city of 350,000 people, 40,000 people were watching television or listening to the radio when the news of a bioterrorist attack was first broadcast. For the purpose of the model, assume that all 40,000 people heard the news simultaneously, 5 minutes after the attack. A differential equation that represents the rate at which the news spread through the population is given by dS 5 0.0243( 350,000 2 S ) dt where S is the number of people who have heard the news and t is time in minutes. Use a slope field to analyze the nature of the solutions to this differential equation. 20. Aphids and Ladybugs To control the number of aphids in a garden, people may introduce some ladybugs into the garden, since ladybugs eat aphids. Suppose that the rate at which the ladybug population changes with respect to the aphid population is given by the differential equation dL 1.5A 2 0.03AL 5 dA 2 0.65L 1 0.0002AL where L represents the number of ladybugs and A represents the number of aphids. Use a slope field to analyze the nature of the solutions to this differential equation.
23. A classmate sketches a solution curve through a point on a slope field and then erases the slope field. He hands you the following graph of a solution curve without the slope field and tells you that the differential equation, of y only (see graph). Sketch the solution curve through (0, 4) and explain how you know your solution is correct.
y 8 6 4 2 0 0 2 4 6 8 10 t dy , dt
is in terms
24. A classmate sketches a solution curve through a point on a slope field and then erases the slope field. She hands you the following graph of a solution curve without the slope field and tells you that the differential equation,
dy , dt
is in terms of t
only (see graph). Sketch the solution curve through (0, 2) and explain how you know your solution is correct.
y 8 6 4 2 3 2 1 0 1 2 3 t
21. A classmate who missed class asks you to explain the concept of a slope field. Describe what a slope field is, how it is created, and what purpose it serves. 22. Consider the differential equation Find all values of y such that
dy dt dy dt
25. For the differential equation

dy dt
5 y( y 1 1 )( y 2 3 ) , describe the intervals

dy dt dy dt
5 2y( 1 2 y ) .
(with respect to y) on which

dy dt
. 0, on which
5 0. Describe the
, 0, and on which
5 0.
characteristics of the slope field for values of y that are greater than, in between, and less than the values of y such that
dy dt
5 0.
10.2 Eulers Method
649
10.2 Eulers Method

Getting Started
Use Eulers Method to approximate solutions to differential equations
In February of 1992, a 79-year-old woman sustained third-degree burns on over 6 percent of her body when she spilled a cup of McDonalds coffee on herself. She requested that McDonalds pay the associated medical expenses, and when the company denied her request, she filed suit. During the trial, a McDonalds quality assurance manager testified that the company required that coffee in the pot be held at 180 to 190 degrees Fahrenheit. Liquids at 180 degrees will cause third-degree burns in 2 to 7 seconds. If the temperature of the coffee had been 25 degrees cooler, the woman might have avoided serious injury. The jury initially awarded the woman $2.7 million in damages; however, a judge later reduced this amount to $480,000. Ultimately, the woman and McDonalds settled the case for an undisclosed amount. (Sources: www.lectlaw.com; www.consumerrights.net.) How long does it take for a 186-degree cup of coffee to cool to 155 degrees? In this section, we will show how Eulers Method may be used to answer questions such as this.
A differential equation model for the rate of change of the temperature of the coffee in question is dT 5 2 0.041( T 2 75 ) dt where T is measured in degrees Fahrenheit and t is measured in minutes. While we may be able to get a sense of how the coffee might cool by examining a slope field associated with a differential equation model for the cooling coffee, we are not able to determine the temperature of the coffee at a specific time. Or, more importantly, in the case of the woman and the hot coffee, we cant determine from a slope field alone how long it will take for a cup of coffee to cool to a particular temperature. Our differential equation tells us how fast the temperature is changing when the temperature of the coffee is T. For example, the initial temperature of the coffee is 186 degrees. Then dT 5 2 0.041( 186 2 75 ) dt 5 2 4.551 Initially, the coffee is cooling (since
dT dt
is negative) at a rate of 4.551 degrees per
minute. If this rate were to continue for 1 minute, the coffee would cool to 181.449 degrees. 186 degrees 2 4.551 degrees ? 1 minute 5 181.449 degrees minute
Since the rate at which the coffee cools depends on the temperature of the coffee, we could recompute the rate of cooling after 1 minute. dT 5 2 0.041( 181.449 2 75 ) dt 5 2 4.364
650
CHAPTER 10
Now the coffee is cooling at a rate of 4.364 degrees per minute. Lets assume that this rate continues for 1 more minute. After 2 minutes, the temperature of the coffee would be 181.449 degrees 2 4.364 degrees ? 1 minute 5 177.085 degrees minute
We can continue this process of computing and recomputing the rate of change of temperature every minute, and keep track of our work in a table (Table 10.3).
TABLE 10.3
Time Elapsed (minutes) 0 1 2 3 4 5 6 7 8
Temperature (degrees Fahrenheit) 186 181.449 177.085 177.085 2 4.185 5 172.9 172.9 2 4.014 5 168.886 168.886 2 3.849 5 165.037 165.037 2 3.692 5 161.345 161.345 2 3.40 5 157.805 157.805 2 3.395 5 154.410
dT 5 2 0.041( T 2 75 ) dt (degrees per minute) 2 4.551 2 4.364 dT 5 2 0.041( 177.085 2 75 ) 5 2 4.185 dt dT 5 2 0.041( 172.900 2 75 ) 5 2 4.014 dt dT 5 2 0.041( 168.886 2 75 ) 5 2 3.849 dt dT 5 2 0.041( 165.037 2 75 ) 5 2 3.692 dt dT 5 2 0.041( 161.345 2 75 ) 5 2 3.540 dt dT 5 2 0.041( 157.805 2 75 ) 5 2 3.395 dt
T 185 180 175 170 165 160 155 0 0 1 2 3 4 5 6 7 t
We see that by computing the slope and following that trajectory for 1 minute, and then recomputing the slope to get a new trajectory, we eventually find that the temperature of the coffee reaches 155 degrees sometime between 7 and 8 minutes. We are making a big assumption in this process. Note that we are assuming that the rate of change remains the same over each 1-minute interval. Since the rate of change depends on the temperature, we know that the rate of change changes the moment the temperature changes. However, this method does provide a numerical approximation to the temperature of the coffee over time. If we look at the slope field for this differential equation along with a scatter plot of the data in Table 10.3, we can see how the slopes direct the path of the temperatures from one point to the next (Figure 10.10).
FIGURE 10.10
10.2 Eulers Method
651
We could get more accurate results by recomputing slopes more frequently than every minute. The more frequently we recompute, the more accurate our results. However, the more frequently we recompute, the more work we will need to do! The frequency at which we recompute slopes for use in computing function values is called the step size. In Table 10.3, the step size is 1. With a step size other than 1, it is important that we multiply the slope at a particular temperature by the step size. This numerical process of computing function values over small intervals is known as Eulers Method.
EULERS METHOD
Given a differential equation dt , subsequent function values can be computed using the formula ynext 5 ycurrent 1 where D t is the step size. dy ? Dt dt
dy
EXAMPLE
Using Eulers Method to Determine the Temperature of Cooling Coffee

Use Eulers Method with a step size of 0.5 to construct a table of approximate function values for the solution to the differential equation that models the rate of change of the temperature of a hot cup of coffee with respect to time, in minutes: dT 5 2 0.041( T 2 75 ) dt If the initial temperature of the coffee is 186 degrees Fahrenheit, determine the time needed for the coffee to cool to 155 degrees.
SOLUTION From our previous work, we know that the temperature of the coffee reached 155 degrees sometime between 7 and 8 minutes. With a step size of 0.5, we should get a more accurate result. (Note: Although we are rounding when we write the numbers in the table, we are keeping the more accurate values in our calculator as we do the computations. Consequently, if you do the calculations using
the rounded values for
dT dt
recorded in Table 10.4 instead of the calculator values,
you will get slightly different results.)
652
CHAPTER 10
TABLE 10.4
Time Elapsed (minutes) 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0
Temperature (degrees Fahrenheit) 186 186 2 4.551 ? 0.5 5 183.725 183.725 2 4.458 ? 0.5 5 181.496 181.496 2 4.366 ? 0.5 5 179.312 179.312 2 4.277 ? 0.5 5 177.174 177.174 2 4.189 ? 0.5 5 175.080 175.080 2 4.103 ? 0.5 5 173.028 173.028 2 4.019 ? 0.5 5 171.018 171.018 2 3.937 ? 0.5 5 169.050 169.050 2 3.861 ? 0.5 5 167.122 167.122 2 3.777 ? 0.5 5 165.233 165.233 2 3.700 ? 0.5 5 163.384 163.384 2 3.624 ? 0.5 5 161.572 161.572 2 3.549 ? 0.5 5 159.797 159.797 2 3.477 ? 0.5 5 158.059 158.059 2 3.405 ? 0.5 5 156.356 156.356 2 3.336 ? 0.5 5 154.688
dT 5 2 0.041( T 2 75 ) dt (degrees per minute) 2 4.551 dT 5 2 0.041( 183.725 2 75 ) 5 2 4.458 dt dT 5 2 0.041( 181.496 2 75 ) 5 2 4.366 dt dT 5 2 0.041( 179.312 2 75 ) 5 2 4.277 dt dT 5 2 0.041( 177.174 2 75 ) 5 2 4.189 dt dT 5 2 0.041( 175.080 2 75 ) 5 2 4.103 dt dT 5 2 0.041( 173.028 2 75 ) 5 2 4.019 dt dT 5 2 0.041( 171.018 2 75 ) 5 2 3.937 dt dT 5 2 0.041( 169.050 2 75 ) 5 2 3.856 dt dT 5 2 0.041( 167.122 2 75 ) 5 2 3.777 dt dT 5 2 0.041( 165.233 2 75 ) 5 2 3.700 dt dT 5 2 0.041( 163.384 2 75 ) 5 2 3.624 dt dT 5 2 0.041( 161.572 2 75 ) 5 2 3.549 dt dT 5 2 0.041( 159.797 2 75 ) 5 2 3.477 dt dT 5 2 0.041( 158.059 2 75 ) 5 2 3.405 dt dT 5 2 0.041( 156.356 2 75 ) 5 2 3.336 dt
We see that the coffee reached a temperature of 155 degrees sometime between 7.5 and 8 minutes.
10.2 Eulers Method
653
Performing Eulers Method with paper-and-pencil techniques can be a tedious process. Fortunately, there are free downloadable programs for the TI-83 Plus at www.ticalc.org that perform Eulers Method. One such program, euler.8xp, is written by Mike Miller of Corban College. We downloaded the program to our calculator using TI-CONNECT. A Technology Tip detailing how to use this program can be found at the end of this section.
EXAMPLE
Using Eulers Method to Model the Spread of a Computer Virus

The Cooperative Association for Internet Data Analysis reports that the Code Red computer virus infected over 359,000 computers in 14 hours on July 19, 2001. (Source: www.caida.org.) Data found at this web site show that the rate at which the virus spread was initially slow, then increased quickly, and then slowed again toward the end of the epidemic. The rate-of-change data, showing the infection rate in computers per minute, can be modeled by the differential equation dN 5 2 8.727 3 1028 N 2 1 0.0251N 1 200 dt where t 5 0 corresponds to 2 p.m. on July 19. The model is accurate for t 5 0 to t 5 240 minutes. That is, the model represents the time interval beginning at 2 p.m. on July 19 and ending at 6 p.m. on July 19. Use Eulers Method and a step size of 1 to determine how many computers were affected after 2 hours (120 minutes). Initially, 12,500 computers were affected by the virus.
SOLUTION We will use the calculator program euler.8xp to answer the question (Figure 10.11).
FIGURE 10.11
According to the Eulers Method approximation, after 2 hours, 182,392 computers were affected by the Code Red virus.
654
CHAPTER 10
EXAMPLE
Using Eulers Method to Model the Spread of a Computer Virus

A differential equation representing the rate, in computers per minute, at which the Code Red virus affected computers around the world in July 2001 is given by dN 5 2 8.727 3 1028 N 2 1 0.0251N 1 200 dt Use Eulers Method and a step size of 1 to determine when the number of affected computers reached 20,000.
SOLUTION We will use the calculator program euler.8xp to answer the question. The computer program does not let us solve for time (x in the program), so we will have to guess and check. From our previous work, we guess that the number of affected computers will be 20,000 after less than 30 minutes. We will try 30 minutes and see if the number of affected computers is 20,000 during this 30-minute time interval (Figure 10.12).
FIGURE 10.12
We see that after 30 minutes, the number of affected computers is over 33,000 computers. If we watch the progress of Eulers Method, we can see that we surpass the desired value of 20,000 affected computers after between 13 and 14 minutes.
Eulers Method 1. Download euler.8xp from www.ticalc.org. Then press the PRGM button to bring up the list of programs loaded to the calculator.
2. Select 2:Euler from the list and press ENTER . For this example, we will use
dy dx
5 2 0.041( y 2 75 ) .
(Continued)
10.2 Eulers Method
655
3. Type in the differential equation and press ENTER .
4. Type in the initial value for x and press ENTER . For this example, we will use x 5 0.
5. Type in the initial value for y and press ENTER . For this example, we will use y 5 168.
6. Type in the number of steps and press ENTER . For this example, we will use steps 5 16.
7. Type in the final value of x and press ENTER . For this example, we will use x 5 8.
8. Press ENTER repeatedly to show the various iterations.
(Continued)
656
CHAPTER 10
9. After the last point is shown, the calculator indicates that the values have been pasted in lists L1 and L2 of the Statistics menu.
10. To access the lists, press ENTER .
STAT
Eulers Method 1. Create a worksheet similar to the one shown here. The D x 5
final x 2 initial x . steps
(Continued)
10.2 Eulers Method dy
657
2. Eulers Method states that yn 1 1 5 yn 1 dx ? D x. We use the given values and the formula Set cell B5 equal to B2 (the initial y) and cell C5 equal to 2 0.041*(B52 75).
dy dx
5 0.041( y 2 75 ) in this example.
3. In cell B6, type B5+C5*$E$2. Then copy this formula to cells B7 through B21. Similarly, copy the formula in C5 to cells C6 through C21. For this example, the x value that is paired with yn is xn 5 0 1 n ? D x. For example, x3 5 0 1 3( 0.5 ) 5 1.5. We estimate that the point (1.5, 162.40) is on the graph of the function.
10.2 Summary
In this section, you learned a numerical technique for solving a differential equation. Recall that a solution to a differential equation is the unknown function that satisfies the differential equation. This numerical technique, called Eulers Method, enables you to find the y values of the solution function without actually finding the function. In the next section, you will learn how to use symbolic techniques to find the solution to the differential equation.
658
CHAPTER 10
10.2 Exercises
In Exercises 14, use Eulers Method to approximate the value of y(4) given the differential equation and initial condition. Use a step size of 0.5 and solve the problem with paper and pencil. 1. 2. 3. 4. dy 5 2x 2 4y, y( 0 ) 5 2 dx dy 5 2 3x, y( 0 ) 5 1 dx dy 5 "y, y( 0 ) 5 1 dx dy 5 sin y, y( 0 ) 5 1 dx Exercises 1116 deal with hypothetical applications in which Eulers Method may be used. 11. Investment Growth The rate at which an investment grows, if $2000 is added to the account each year and the account earns 6 percent interest compounded continuously, is modeled by the differential equation dM 5 0.06M 1 2000 dt Initially, at time t 5 0, M 5 0. Use Eulers Method and a step size of 0.5 to determine when the amount of money in the account reaches $100,000. 12. Changing Volume A salt-water mixture is being poured into a tank of salt water at a rate of 3 gallons per minute. Furthermore, water is draining from the tank at a rate of 2 gallons per minute. A differential equation that models the rate at which the amount of salt in the tank is changing is 2S dS 562 dt 51t where S is measured in pounds of salt and t is time in minutes. When t 5 0, S 5 10. Use Eulers Method and a step size of 0.5 to determine when the amount of salt in the tank will be 25 pounds. 13. Drug Elimination The rate at which a particular drug leaves a patients body after administration of the drug has stopped is given by dA 5 2 0.0201A dt where A is the amount of the drug in the body after t hours. When t 5 0, A 5 200. Use Eulers Method and a step size of 0.5 to determine when the amount of medicine in the body is A 5 170. 14. Investment Growth The rate of change of an investment account earning continuous compound interest is given by dA 5 0.061A dt where A is measured in dollars and t is measured in years. Use Eulers Method and a step size of 0.5 to estimate the amount of money in the account after 20 years. The account initially contains $1000.
In Exercises 58, use Eulers Method to approximate the value of y(4) given the differential equation and initial condition. Use a step size of 0.1 and use technology. Compare your answers to those found in Exercises 14. 5. 6. 7. 8. dy 5 2x 2 4y, y( 0 ) 5 2 dx dy 5 2 3x, y( 0 ) 5 1 dx dy 5 "y, y( 0 ) 5 1 dx dy 5 sin y, y( 0 ) 5 1 dx
Exercises 910 require you to analyze student claims related to Eulers Method. 9. A student claims that if the solution to a differential equation is a function whose graph is concave down, using Eulers Method to approximate the solution will always result in an overestimate of the exact solution. Do you agree? Justify your conclusion. 10. A student claims that if the solution to a differential equation is a function whose graph is concave down, using Eulers Method to approximate the solution may yield an overestimate or underestimate of the exact solution, depending on the step size. Do you agree? Justify your conclusion.
10.2 Eulers Method
659
15. Homicide Investigation When the forensic investigator arrives at the scene of a homicide, one of the most important things to be determined is the time of death of the victim. Suppose that a differential equation modeling the rate of change in the temperature of the cooling deceased body is given by dT 5 2 0.1282( T 2 70 ) dt where T is measured in degrees Fahrenheit and t is measured in minutes. Assume that the initial temperature of the body was 95F. Use Eulers Method to determine how long it took for the body to cool to 72F. 16. News Spread Suppose that in a city of 350,000 people, 40,000 people were watching television or listening to the radio when the news of a bioterrorist attack was first broadcast. For the purpose of the model, assume that all 40,000 people heard the news simultaneously, 5 minutes after the attack. A differential equation that represents the rate at which the news spread through the population is given by dS 5 0.0243( 350,000 2 S ) dt where S is the number of people who have heard the news and t is time in minutes. Use Eulers Method to determine when half of the people in the city had heard the news. Exercises 1718 give you an opportunity to investigate the details of Eulers Method. 17. Explain why using smaller and smaller step sizes with Eulers Method produces more accurate results. What would need to be done to ensure no error in the results obtained using Eulers Method? 18. Suppose that you need to explain Eulers Method to a student who missed class. Create an explanation that describes what Eulers Method is used for and how it works. In Exercises 1925, use Eulers Method and technology to answer the question. Use a step size of 0.1. 19.
Population of the United States
United States from 1790 to 2000 is
where P is measured in people and t is measured in years since 1790. If the population of the United States was 5,300,000 in 1790, use Eulers Method to estimate when the population of the United States was 300,000,000. (Source: Statistical Abstract
dP dt
5 160"P,
20.
1992 to 2001, the estimated rate at which the number of new pediatric AIDS cases in the United States is changing can be modeled by dN 2 3.804e0.07758t 5 dt ( 1 1 0.0533e0.07758t )
2
where N is the number of AIDS cases and t is the number of years since 1992. (Source: Modeled from Assume that when t 5 0, N 5 963. Use Eulers Method to determine when the number of new pediatric AIDS cases will be 900. 21.
School Internet Access Based on data from 1994 to 2000, the rate of change in the percentage of public-school classrooms with Internet access may be modeled by
Centers for Disease Control and Prevention data.)
dy 5 0.01077y( 85.88 2 y ) dt where y is percentage points per year and t is the number of years since 1994. (Source: Statistical
Assume that when t 5 0, y 5 50 percent. Use Eulers Method to determine when the percentage of public-school classrooms with Internet access will be 80 percent.
22.
National Internet Usage Based on data from 1995 to 1999 and U.S. Census Bureau projections for 2000 to 2004, the rate of change in annual per capita Internet usage may be modeled by
dy 5 0.002882y( 232.2 2 y ) dt where y is hours of usage per year and t is the number of years since 1995. (Source: Statistical
A differential equation that models the rate of change of the population of the
Assume that when t 5 0, y 5 10. Use Eulers Method to determine when the number of hours of usage per year will be 200 hours.
660 23.
CHAPTER 10
Number of Students Attending Maricopa Community Colleges The
rate at which the number of students served by the Maricopa Community College District (Arizona) is increasing may be modeled by dN N 5 2.63( N 2 165 ) a 1 2 b dt 279 where N is the number of students served in thousands and t is the time in years since 1995. (Source: www.maricopa.edu.) In this case, when t 5 1, N 5 165.5. Use Eulers Method and a step size of 0.5 to determine when the number of students will be N 5 200 (thousand). 24.
Learning Curve in Construction
26. Use words to explain what is happening when we use Eulers Method, ynext 5 ycurrent 1
dy dt
? D t, to
approximate solutions to differential equations. 27. Explain the role that step size plays in the accuracy of the solutions given by Eulers Method. 28. Imagine using Eulers Method to approximate solutions to a differential equation. When compared with the graph of the actual solution, in which of the following cases will Eulers Method overestimate the solution, underestimate the solution, or do neither? Explain how you know.
y 4 3 2 1 0 0 y 4 3 2 1 0 0 y 4 3 2 1 0 0 1 2 3 4 x 1 2 3 4 x 1 2 3 4 x
A study showed that workers constructing a 20-story prison facility were able to work faster in one aspect of construction as they completed each floor. In constructing floors 9 through 18, a particular aspect of construction involving spandrel beams proved to be quite repetitive, and so workers were able to become more efficient as they finished each floor. The rate at which the number of man-hours per square foot decreased may be modeled by dM 5 2 3.19M 2 1 0.84M 2 0.06 dF where M is the number of man-hours per square foot and F is the floor sequence. That is, F 5 1 refers to the ninth floor. (Source: www.asceditor.usm .edu.) In this case, when F 5 1, M 5 0.272. Use Eulers Method and a step size of 0.5 to determine at which floor the number of man-hours per square foot will be 0.2. 25. The rate at which a hot cup of coffee cools to room temperature (76F) can be modeled by the differential equation
Newtons Law of Cooling
dT 5 2 0.04095( T 2 76 ) dt where T is the temperature of the coffee after t minutes. (Source: Modeled from authors data.) When t 5 0, T 5 186 F. Use Eulers Method and a step size of 0.5 to determine when the temperature of the coffee will be 100F.
29. The following table was created using Eulers Method to approximate the solutions to the differential equation
dy dt
5 2 y( y 2 4 ) . Notice
particularly the behavior of the approximations late in the table.
10.2 Eulers Method
661
Provide a possible explanation for these data. Does the differential equation support these data? Why or why not? t 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 y 1.000000 2.125000 3.619141 4.136034 3.925043 4.035371 3.981845 4.008954 3.995493 4.002246 3.998875 4.000562 3.999719 4.000141 3.999930 4.000035 3.999982 4.000009 3.999996 4.000002 3.999999 New y Value 2.125000 3.619141 4.136034 3.925043 4.035371 3.981845 4.008954 3.995493 4.002246 3.998875 4.000562 3.999719 4.000141 3.999930 4.000035 3.999982 4.000009 3.999996 4.000002 3.999999 4.000001
dy dt
31. Use Eulers Method to complete the table, given the differential equation condition y( 0 ) 5 3.
dy dx
23x 2y
with initial
5 x
0 0.5 1.0 1.5 2.0
32. Use Eulers Method to create a table like the one in Exercise 31, given the differential equation
dy dx
23x 2y
with initial condition y( 0 ) 5 3. This
time, however, use a step size of 0.25. 33. Use Eulers Method to create a table like the one in Exercise 31, given the differential equation
dy dx
23x 2y
with initial condition y( 0 ) 5 3. This
time, however, use a step size of 0.10. 34. Compare your results from Exercises 3133. What do you notice? How similar or different are the results? Specifically, compare the values of y( 2 ) in each case. 35. The formula y2 1 2 x2 5 9 represents the exact solution curve to the differential equation presented in Exercises 3133. Determine the positive value of y( 2 ) from this formula and compare it to your previous results using Eulers Method. Describe the difference between the approximate values of y( 2 ) and the actual value of y( 2 ) as the step size gets smaller.
3
30. Consider the differential equation
5 5. Would
Eulers Method produce more accurate results when D t 5 0.5 or when D t 5 0.1? Explain your reasoning.
662
CHAPTER 10
Determine general and particular solutions of differential equations using the method of separation of variables Use separation of variables to solve differential equations that model real-world situations
In the previous section, we saw how to qualitatively analyze a differential equation. While this provides important graphical information about the nature of the solution to the differential equation, we also can do symbolic work to find a specific solution to a differential equation given an initial condition. Recall that a slope field represents many solutions of a differential equation, and that we can visualize a particular solution curve passing through a particular point in the slope field. In this section, we will learn how to find the equation for the solution curve. We will also discuss Newtons Law of Heating and Cooling and other differential equation applications.
Just In Time
Rules of Exponents
Recall that x a ? x b 5 x a 1 b for x 2 0. We can also use this rule in reverse: e x 1 C 5 e x ? eC
Just In Time
Solving Exponential Equations
Logarithms are used to solve exponential equations such as e2x 5 3. We take the logarithm of both sides of the equation and use the properties of logarithms to solve for x. ln( e2x ) 5 ln 3 2x ? ln e 5 ln 3 2x 5 ln 3 x5 ln 3 2
Property of logarithms ln e 5 1 Divide by 2
EXAMPLE
Solving a Differential Equation

Find a function F( t ) such that
dF dt
5 F.
663
SOLUTION
dF 5F dt dF 5 dt F 1 dF 5 dt F 1 3 F dF 5 3dt ln Z F Z 5 t 1 C e
t1C
Multiply by dt and divide by F Rewrite dF 1 as dF F F
Take the integral of both sides Integrate Recall that ln Z y Z 5 x is equivalent to ex 5 Z y Z Rules of exponents
5 ZF Z
F 5 6et ? eC 5 6eCet
Letting k 5 eC, we get the general solution of the differential equation: F( t ) 5 6ket Any function of this form is a solution of the differential equation
t t t
example, F( t ) 5 2e , F( t ) 5 2 3e , and F( t ) 5 e are all particular solutions of the differential equation.
dF dt
5 F. For
EXAMPLE
Finding the Particular Solution of a Differential Equation

Find the particular solution of the differential equation
SOLUTION
dF dt
5 F 2 given F( 0 ) 5 4.
dF 5 F2 dt dF 5 dt F2 1 dF 5 dt F2 F 22dF 5 dt
22 3F dF 5 3dt
Divide by F 2 and multiply by dt Rewrite Rewrite dF 1 as 2 dF F2 F 1 dF as F 22dF F2
Take the integral of both sides
F 21 5t1C 21 F 21 5 2 t 1 C 1 5 2t 1 C F F5 1 2t 1 C
Integrate using the Power Rule Multiply both sides by 21 Rewrite F 21 as 1 F
Divide by 2 t 1 C and multiply by F
664
CHAPTER 10
This is the general solution of the differential equation. However, since we know that F( 0 ) 5 4, we are able to find the particular solution. F5 45 45 C5 1 2t 1 C 1 20 1 C 1 C 1 4 1 . We can verify the accuracy 2t 1 1 4
Substitute F 5 4 and t 5 0
Therefore, the particular solution is F( t ) 5 of our work by differentiating this function. d d 3F( t ) 4 5 dt dt 1
2t 1
1 4
dF d 1 21 5 a2t 1 b dt dt 4 1 22 5 21 a 2t 1 b (21) 4 1 22 5 a2t 1 b 4 5 1 1 2 a2 t 1 b 4 1 2t 1 1 4 2
5 F2 Since the result is equivalent to the differential equation we were given at the start of the problem, we believe that our calculated particular solution, 1 F( t ) 5 1 , is correct. To verify, we check to see that F( 0 ) 5 4. 2t 1
4
F( 0 ) 5 54
1 2(0) 1 1 4
We used the separation of variables process to solve the differential equations in Examples 1 and 2. We will now formally describe the process.
665
SOLVING SEPARABLE FIRST-ORDER DIFFERENTIAL EQUATIONS
A first-order differential equation is said to be separable if it may be written in the form dy f (x) 5 dx g( y ) for some functions f ( x ) and g( y ) . The solution of the differential equation is obtained by moving the x and y variables to opposite sides of the equal sign and integrating. That is, 3g( y ) dy 5 3f ( x ) dx It is essential that all x terms be grouped with the dx and all y terms be grouped with the dy.
EXAMPLE
Solving a Differential Equation to Determine Investment Value

The rate of change of an investment account earning compound interest may be given by the first-order differential equation constant. (If k 5 0, the value of the investment remains constant. If k is negative, the investment is losing value.) Find the general solution of the differential equation.
SOLUTION We will solve the differential equation using the separation of varidA dt
5 kA, where k is a positive
ables process. dA 5 kA dt dA 5 k dt A 1 3 A dA 5 3k dt
Separate the variables Take the integral of both sides
Integrate ln Z A Z 5 kt 1 C A, the account value, is assumed to be positive ln A 5 kt 1 C kt 1 C A5e In general, In A 5 b is equivalent to A 5 eb 5 e kt ? eC Rules of exponents kt 5 Pe , where P 5 eC
The general solution of the differential equation is A 5 Pe kt. We immediately recognize this result as the continuous compound interest formula. The standard r nt compound interest formula A 5 P Q 1 1 n R may be converted to the form A 5 Pe kt by letting r n a 1 1 b 5 ek n r n k 5 ln c a 1 1 b d n
666
CHAPTER 10
A DIFFERENTIAL EQUATION FOR CONTINUOUS COMPOUND INTEREST
The value of an investment account earning continuous compound interest is A 5 Pe kt dollars where A is the value of the investment after t years, P is the initial value of the investment (in dollars), and k is the continuous interest rate. The rate of change in the value of the investment is given by dA dollars 5 kA dt year
EXAMPLE
Finding the Particular Solution of a Differential Equation

The rate of change of an investment account earning compound interest is given by
dA dt
5 kA, where k is a positive constant. The initial account value was $1000.
At the end of the third year, the account value was $1120. Find the particular solution of the differential equation.
SOLUTION The general solution of the differential equation
dA dt
5 kA is
A 5 Pe . We know that A( 0 ) 5 1000 and A( 3 ) 5 1120.

kt
A 5 Pe kt 1000 5 Pe k(0) 1000 5 P ? 1 P 5 1000 Thus A 5 1000e . We will find the value of k by substituting the second point, (3, 1120), into this equation.
kt
A 5 1000ekt 1120 5 1000ek(3) 1.12 5 e3k ln( 1.12 ) 5 ln( e3k ) 0.1133 5 3k k 5 0.03778 Therefore, A 5 1000e is the particular solution of the differential equation. This equation may also be written as A 5 1000( 1.0385 ) t, since e0.03778 5 1.0385. The account is earning interest at a rate of 3.778 percent compounded continuously, or at a rate of 3.85 percent compounded annually.
0.03778t
Newtons Law of Heating and Cooling

The world-renowned scientist and scholar Sir Isaac Newton determined that the rate of change in an objects temperature is proportional to the difference between the constant temperature of the environment surrounding the object and
667
the objects temperature. This observation is summarized in Newtons Law of Heating and Cooling.
NEWTONS LAW OF HEATING AND COOLING
Let T be the temperature of an object at time t and A be the temperature of the environment surrounding the object (ambient temperature). Then dT 5 k( T 2 A ) dt where k is a constant that varies depending on the physical properties of the object. According to Newtons Law of Heating and Cooling, dT 5 k( T 2 A ) . dt We will use substitution to rewrite the equation. We define the function y 5 T 2 A and differentiate this function with respect to t. d d (y) 5 (T 2 A) dt dt dy d d 5 (T) 2 (A) dt dt dt 5 5 5 So
dy dt
dT dA 2 dt dt dT 20 dt dT dt
dT dt
dA 5 0 because A is a constant. dt
dT . dt
Rewriting the cooling equation
5 k( T 2 A ) in terms of y, we get
dy 5 ky dt since
dT dt
dy dt
and T 2 A 5 y. Separating the variables, we rewrite the equation as 1 dy 5 k dt y
Integrating both sides of the equation, we get 1 3 y dy 5 3( k dt ) ln Z y Z 5 kt 1 C Recall that if x 5 ln Z y Z , then ex 5 Z y Z . Consequently, ln Z y Z 5 kt 1 C Z y Z 5 ekt 1 C y 5 6( ekt ? eC )
668
CHAPTER 10
Now well back substitute y 5 T 2 A. Note that because the object is cooling, T . A and, consequently, T 2 A . 0. T 2 A 5 ( ekt ? eC ) T 5 A 1 ( ekt ? eC ) T 5 A 1 Sekt, where S 5 eC Thus the general solution of the differential equation T 5 A 1 Sekt.
EXAMPLE 5
dT dt
5 k( T 2 A ) is
Using Newtons Law of Heating and Cooling

To test Newtons Law of Heating and Cooling, the author heated a cup of water to 186F and placed it in a room at 76F. The temperature of the water was recorded at 2-minute time intervals, as shown in Table 10.5.
TABLE 10.5
Hot Water Cooling in a 76F Room t, Minutes 0 2 4 6 8 10 12 14 T, Temperature (degrees Fahrenheit) 186 175 167 159 153 147 142 138
Use Newtons Law of Heating and Cooling to find a model for the water temperature.
SOLUTION We know that the general solution of the differential equation given by Newtons Law of Heating and Cooling is T 5 A 1 Sekt. Well substitute values from the table into the equation to determine the numeric values of k and S. We know that A 5 76, since the ambient temperature is 76F. Additionally, we know that at time t 5 0, T 5 186F.
T 5 76 1 Sekt 186 5 76 1 Sek(0) 110 5 Se0 110 5 S
669
Our equation T 5 A 1 Sekt may now be rewritten as T( t ) 5 76 1 110ekt Well use the table values t 5 14 and T 5 138 to find the value of k. (Note that we can select any pair of table values. Since were developing a model, selecting a different pair of table values would slightly alter the model equation.) 138 5 76 1 110ek(14) 62 5 110e14k 0.5636 5 e14k ln( 0.5636 ) 5 14k k 5 2 0.04095 Since k is negative, the temperature of the water is dropping at a continuous rate of 0.04095F per minute. The temperature of the water may be modeled by T( t ) 5 76 1 110e20.04095t degrees Fahrenheit Lets see how well this function fits the table of data (Table 10.6).
TABLE 10.6
Take the natural log of both sides
Hot Water Cooling in a 76F Room t, Minutes 0 2 4 6 8 10 12 14 T, Temperature (degrees Fahrenheit) 186 175 167 159 153 147 142 138 M, Model Temperature (degrees Fahrenheit) 186 177 169 162 155 149 143 138
Newtons Law of Heating and Cooling gave a remarkably good estimate for the measured temperature. The discrepancies between the two temperatures may be due to measurement device error or some other physical phenomenon. As with any mathematical model, some error is to be expected.
EXAMPLE
Using a Differential Equation to Model Cooling Rates

According to the model developed in Example 5, at what rate was the water cooling 4, 8, and 12 minutes into the cooling period?
670
CHAPTER 10
SOLUTION The model for the water temperature was given by
T( t ) 5 76 1 110e20.04095t degrees t minutes after the temperature of the water was measured to be 186 degrees. The rate of change in the temperature is given by the differential equation dT 5 k( T 2 A ) dt 5 2 0.04095( T 2 76 ) Were asked to evaluate this function at t 5 4, t 5 8, and t 5 12. dT ` 5 2 0.04095 3T( 4 ) 2 764 dt t 5 4 5 2 0.04095( 169.4 2 76 ) 5 2 3.82 Four minutes into the cooling period, the water was cooling at a rate of 3.82 degrees per minute. dT ` 5 2 0.04095 3T( 8 ) 2 764 dt t 5 8 5 2 0.04095( 155.3 2 76 ) 5 2 3.25 Eight minutes into the cooling period, the water was cooling at a rate of 3.25 degrees per minute. dT ` 5 2 0.04095 3T( 12 ) 2 764 dt t 5 12 5 2 0.04095( 143.3 2 76 ) 5 2 2.76 Twelve minutes into the cooling period, the water was cooling at a rate of 2.76 degrees per minute. Assuming coffee cools at the same rate as water, we estimate that it would take 8 minutes for a cup of McDonalds coffee to cool from 186 degrees to 155 degrees Fahrenheit. However, the size and shape of the coffee cup, the addition of cream and sugar, and other environmental variables could dramatically impact the accuracy of our estimate.
10.3 Summary
In this section, you learned how to use integral calculus to solve first-order differential equations. You also learned how to apply differential equations using Newtons Law of Heating and Cooling and the continuous compound interest formula.
671

In Exercises 15, write an equivalent expression using the addition property of exponents. 1. eM 1 N 2. x C 1 5 3. e2 2 K 4. eC ? e6 5. x 25 ? x 5 In Exercises 610, solve the exponential equation by using logarithms. 6. e3x 5 5 7. 5eC 5 25 8. 3e x 1 2 5 9 9. 2 2e x 2 5 1 3 5 2 10. 10 x 2 5 5 100
10.3 Exercises
In Exercises 110, find the general solution of the separable differential equation. 1. 3. 5. 7. 9. dy x 5 y dx dy 5 "y dx dy y2 5 2 dx x dy 5 0.25( y 2 4 ) dx dy 5 4y 2 y 2 dx a Hint: 3 y( M
dy 2 y)
16. 17. 18. 19. 20.
dA 5 4 2 2A; A( 0 ) 5 0 dt dy 1 5 y( 4 2 y ) ; y( 0 ) 5 dx 2 dA 5 A2 2 4A; A( 0 ) 5 8 dt dy 5 2y 2 y 2; y( 0 ) 5 100 dx dA 5 A( 10 2 A ) ; A( 0 ) 5 5 dt
2. 4. 6. 8.
dA 5 3A2 dt dA 3A 5 dt t dA 5 4( 20 2 A ) dt dy 5 4y dx
5 M ln ` M
1
y `b 2y
In Exercises 2128, use Newtons Law of Heating and Cooling to find the equation of the model that best fits the data. Then answer the given questions. 21. The author placed a cup of ice water in a 76F room and recorded the temperature of the water at 10-minute intervals, as shown in the table.
Water Temperature
10.
dA 5 6 2 3A dt
dy 2 y)
In Exercises 1120, find the particular solution of the differential equation. a Hint: 3y( M dy 2x 11. 5 ; y( 1 ) 5 4 y dx 12. 13. 14. 15. dA 5 4A2; A( 0 ) 5 2 1 dt dy 5 3y; y( 0 ) 5 2 1 dx 2A dA 5 ; A( 0 ) 5 0 dt t dy 5 25( y 2 5 ) ; y( 0 ) 5 0 dx 5 M ln ` M
1 y `b 2y
Cold Water Warming in a 76F Room t, Minutes 0 10 20 30 40 50 T, Temperature (degrees Fahrenheit) 34.3 40.3 43.8 48.2 51.8 54.7
672
CHAPTER 10
(a) Find a model for the temperature of the water at time t minutes. (b) Determine the rate of change in the waters temperature at 10 minutes and 50 minutes into the warming period. 22. How long will it take for a 60F cup of water to freeze in a 10F freezer? (At 32F, water freezes.) Because the result will vary depending on the quantity of water and the shape of the container, assume that for the container in question the water temperature will be 46F at the end of the eighth minute.
Water Temperature
dropped to 56.9F. If the victims internal body temperature was 98F at the time of death, how long had the victim been dead when the forensic investigator arrived on the scene? 28. Time of Death According to the model developed in Exercise 27, at what rate was the corpse cooling 4 hours, 6 hours, and 12 hours into the cooling period? Exercises 2932 deal with the notion of continuous growth. 29. Investment Value A $1000 investment has a continuous interest rate of 10 percent. Write an equation for the value of the investment after t years. 30. Investment Value The value of an investment is modeled by A 5 2000e0.08t dollars, where t is the number of years since 2006. What will be the annual rate of change in the value of the investment when the account value is $3000? $4000? $5000? 31. Investment Value The rate of change of an investment account earning continuous compound interest is given by
dA dt
23.
Cakes are typically baked at 350F. If a pan filled with cake batter is moved from a 75F kitchen into a 350F oven, how long will it take for the cake batter to heat up to 200F? Because the result will vary depending on the quantity of the batter and the shape of the pan, assume that for the container in question the batter temperature will be 100F at the end of the third minute.
Cake Temperature
24.
A cake is removed from a 350F oven and placed on a cooling rack in a 75F room. If the temperature of the cake was 350F when it was removed from the oven, and its temperature is 250F 2 minutes later, how long will it take for the cake to cool to 100F?
Cake Temperature
constant. The initial account value was $2500. At the end of the third year, the account value was $4200. Find the particular solution of the differential equation. 32. Investment Value The rate of change of an investment account earning continuous compound interest is given by
dA dt
25. Time of Death When the forensic investigator arrives at the scene of a homicide, one of the most important things to be determined is the time of death of the victim. Suppose that the victims internal body temperature was 84.5F when the body was discovered in a 70F room. Thirty minutes later, the victims body temperature had dropped to 83.6F. If the victims internal body temperature was 98F at the time of death, how long had the victim been dead when the forensic investigator arrived on the scene? 26. Time of Death According to the model developed in Exercise 25, at what rate was the corpses temperature dropping 1 hour, 2 hours, and 4 hours into the cooling period? 27. Time of Death Suppose that a homicide victims internal body temperature was 59.1F when the body was discovered in a 30F room. Thirty minutes later, the victims body temperature had
constant. The initial account value was $420. At the end of the fifth year, the account value was $630. Find the particular solution of the differential equation.
33. Explain how y 5 t, y 5 3t 1 2, and y 5 5t 1 4 all can be solutions of the differential equation
dy dt
y11 . t11
34. Explain why the separation of variables method cannot be used to solve the differential equation
dy dt
5 2y 1 t.
673
35. A classmate solved a differential equation and found y 5 et to be a solution. Determine what separable differential equation would have this solution. You will also need to determine initial conditions.
2
37. Consider the differential equation dx 5 xy 3. Explain why the general solution of this differential equation would not satisfy the initial condition y( 0 ) 5 0.
dy
36. Of the following three differential equations, which can be solved using separation of variables? Explain how you know. A. B. C. dy 5 t3 dt dy t3 5 y dt dy 5 t3 1 y dt 38. Show that the derivative of A 5 Pe kt is equivalent to kA. 39. Find the general solution of the differential equation 2y 1 yrx2 5 5 40. Find the general solution of the differential equation 3x 2y 2 1 2yyrx 3 5 0
Find the equation of a limited growth model Find the equation of a logistic growth model Analyze and interpret the real-life meaning of differential equations
On September 11, 2001, many of us woke up to the horrific news of the terrorist attack on the World Trade Center. News bulletins announcing the attack flooded the radio and television channels. Initially, the news of the attack spread very rapidly; however, as the day progressed, the news spread more slowly, since most people had already received word of the attack. The spread of information, such as the news of the September 11 attack, may often be modeled mathematically. In this section, we will continue our discussion of differential equations and show how they can be used to find limited growth and logistic models.
Limited Growth Model

Sociologists often assume that the rate at which news spreads via mass media is proportional to the number of people who have not yet heard the news. This type of growth is called limited growth. For a limited growth function y with a maximum value of M, the rate of change,
dy , dt
is proportional to the difference 5 k( M 2 y ) .
between the present value of y and M. That is,
dy dt
EXAMPLE
Using Differential Equations to Forecast the Spread of Information

At 8:46 a.m. on September 11, 2001, the hijacked American Airlines Flight 11 slammed into the North Tower of the World Trade Center in New York City. Eighteen minutes later, United Airlines Flight 175 crashed into the South Tower of the
674
CHAPTER 10
World Trade Center. By the end of the day, thousands of people were dead as a result of the deadliest terrorist attack in American history. Suppose that in a community of 10,000 people, 3000 people were watching television or listening to the radio when the news broke on September 11, 2001. For the purpose of the model, well assume that all 3000 people heard the news simultaneously five minutes after the first attack. (It takes time for news crews to reach the site of any newsworthy event.) Find the equation for the limited growth model that models the spread of the news of the attack. Then determine how long it took for 90 percent of the community to hear the news and at what rate people were hearing the news at that time.
SOLUTION The maximum number of people that can be informed is 10,000, so M 5 10,000. Additionally, we know that when t 5 5, y 5 3000. We have
dy 5 k (M 2 y) dt dy 5 k ( 10,000 2 y ) dt We can use the separation of variables process to find the equation of the general solution and then use the point (5, 3000) to find the particular solution. dy 5 k ( 10,000 2 y ) dt dy 5 k dt ( 10,000 2 y ) Let u 5 10,000 2 y; then du 5 2 dy. Rewriting the left-hand side of the equation in terms of u, we get 2 du 5 k dt u 1 2 3 du 5 3k dt u 2 ln Z u Z 5 kt 1 C 2 ln u 5 kt 1 C ln u 5 2 kt 2 C u 5 e2kt 2 C 10,000 2 y 5 e2kte2C y 5 10,000 2 Se
2kt
Since u . 0
Since u 5 10,000 2 y Where S 5 e 2 C
At t 5 0, nobody had heard the news of the attack because the information hadnt yet been broadcasted to the community. (It wasnt until five minutes after the crash that the first 3000 people in the community heard the news.) We can use this information to find the value of S. y ( 0 ) 5 10,000 2 Se2k(0) 0 5 10,000 2 Se0 S 5 10,000
675
A general solution to the differential equation is y 5 10,000 2 10,000e2kt 5 10,000( 1 2 e2kt ) Well now use the point (5, 3000) to find the particular solution. 3000 5 10,000( 1 2 e2k(5) ) 0.3 5 1 2 e25k 2 0.7 5 2 e25k 0.7 5 e25k ln ( 0.7 ) 5 2 5k k5 ln ( 0.7 ) 25
5 0.07133 The limited growth model that fits the data is given by y 5 10,000( 1 2 e20.07133t ) Since 90 percent of 10,000 is 9000, we need to determine when 9000 people had heard the news. 9000 5 10,000( 1 2 e20.07133t ) 0.9 5 1 2 e20.07133t 2 0.1 5 2 e20.07133t 0.1 5 e20.07133t ln ( 0.1 ) 5 2 0.07133t t5 ln ( 0.1 ) 2 0.07133
5 32.28 We estimate that 32 minutes after the attack, 90 percent of the people in the community had heard the news. Well now determine at what rate the word was spreading. dy 5 k ( 10,000 2 y ) dt 5 0.07133 ( 10,000 2 y ) dy ` 5 0.07133 ( 10,000 2 9000 ) dt t 5 32.28 5 71.33 Approximately 32 minutes after the attack, the news was spreading at a rate of roughly 71 people per minute.
[Note: y( 32.28 ) 5 9000]
In Example 1, we solved the given differential equation by using the separation of variables method. Using this same approach, we can derive the general solution to the differential equation
dy dt
5 k( M 2 y ) .
676
CHAPTER 10
LIMITED GROWTH
Assume that the rate of growth of a function y with maximum value M is proportional to the difference between the present value of y and M. That is, dy 5 k( M 2 y ) dt The solution to the differential equation with initial condition y (0) 5 0 is given by y 5 M ( 1 2 e2kt ) and has the graph shown in Figure 10.13.
y y=M
FIGURE 10.13
A limited growth model may be used to model the cumulative number of people infected by a disease during a bioterrorist attack once quarantine efforts are underway. In the journal article Modeling Potential Responses to Smallpox as a Bioterrorist Weapon, scientists at the Centers for Disease Control and Prevention and at Don Millar and Associates demonstrated the catastrophic effect of a smallpox attack. Table 10.7 shows how rapidly the cumulative number of smallpox cases would increase should there be no intervention.
TABLE 10.7 Estimates of Cumulative Total Smallpox Cases After 365 Days with No Intervention
Number Initially Infecteda 10 10 1,000 1,000

aNumber
Number Infected per Infectious Personb 1.5 3.0 1.5 3.0
Cumulative Total Number of Smallpox Cases, Days Postreleasec 30 Days 31 64 3,094 6,397 90 Days 214 4,478 21,372 447,794 180 Days 2,190 2.2 million 219,006 222 million 365 Days 224 thousand 774 billion 22 million 77 trillion
initially infected refers to those who are exposed during a release so that they subsequently become infectious to others. This scenario excludes those who are exposed but either do not become ill (i.e., are immune or are not exposed to an infectious dose) or do not become infectious (residual immunity from prior vaccination may be sufficient to prevent onward transmission). bThe number of persons infected per infectious person is the transmission rate. cAssumes an unlimited supply of smallpox-susceptible persons. Source: Emerging Infectious Diseases, vol. 7, no. 6, NovemberDecember 2001, p. 959.
677
Figure 10.14 shows the effectiveness of adequate intervention once smallpox is detected. If 25 percent of the people infected are quarantined daily, the cumulative number of people infected will continue to increase at an increasing rate; however, if 50 percent of the people infected are quarantined daily, the cumulative number of people infected will increase at a decreasing rate. The graphs in the bottom right-hand corner of the figure show that the cumulative number of people infected may be modeled by a limited growth function when 50 percent of the infected individuals are quarantined daily.
Quarantine only: Remove 25% per day Quarantine only: Remove 50% per day
Daily cases
120 Daily cases (in thousands) 100 100 80 60 40 20 0 Start day 30 80 60 Start day 25 40 20 0 Daily cases 140 Start day 45 120
Total cases
5 4 3 2 1 0 1 50 99 148 197 246 295 344 1 50 99 148 197 246 295 344 Days post-release Days post-release Start day 45 Start day 30 Total cases (in thousands) Total cases (in millions) 6 8 7 6 5 4 3 2 1 0
Result: No Control
Start day 25
Result: Control
Figure 4. Daily and total cases of smallpox after quarantining infectious persons at two daily rates and three post-release start dates. The graphs demonstrate that if quarantine is the only intervention used, a daily removal rate of 50% is needed to stop transmission within 365 days post-release. At a 25% daily removal rate of infectious persons by quarantine, a cohort of all those entering the first day of overt symptoms (i.e., rash) is entirely removed within 17 days (18 to 20 days post-incubation) after the first day of overt symptoms, with 90% removed within 9 days. At a 50% daily removal of infectious persons by quarantine, a cohort of all those entering their first day of overt symptoms (i.e., rash) is entirely removed within 7 days (8 to 10 days post-incubation) after the first day of overt symptoms, with 90% removed within 4 days. The daily rate of removal (quarantine) relates only to the removal of those who are infectious (i.e., overtly symptomatic). The rate does not include any persons who may be quarantined along with overtly symptomatic patients, such as unvaccinated household contacts. Data generated by assuming 100 persons initially infected and a transmission rate of 3 persons infected per infectious person. For clarity, the graphs of daily cases do not include the assumed 100 initially infected persons. The graphs of total cases include the 100 initially infected.
Source: Emerging Infectious Diseases, vol. 7, no. 6, NovemberDecember 2001, p. 959.
FIGURE 10.14
EXAMPLE
Using Differential Equations to Forecast the Effect of a Bioterrorist Attack

Reading from the graph in the bottom right corner of Figure 10.14, we can construct a limited growth model. It appears that a maximum of 7000 people will be infected with smallpox as a result of a bioterrorist attack. It also appears that 99 days after the attack, 4750 people are infected. Do the following:
678
CHAPTER 10
(a) Determine the equation of the limited growth model. (This models the cumulative number of people infected if 50 percent of the infected people are quarantined daily beginning on the 45th day after the attack.) (b) Determine the rate at which the disease is spreading on the 50th and 100th days after the attack.
SOLUTION
(a) The equation of the limited growth model is given by y 5 M( 1 2 e2kt ) Since a maximum of 7000 people will be infected, M 5 7000. y 5 7000( 1 2 e2kt ) Additionally, we know that (99, 4750) lies on the graph of the function. 4750 5 7000( 1 2 e2k(99) ) 0.6786 5 1 2 e2k(99) 2 0.3214 5 2 e299k 0.3214 5 e299k ln( 0.3214 ) 5 2 99k k5 ln( 0.3214 ) 2 99
5 0.01147 Thus y 5 7000( 1 2 e20.01147t ) models the number of people infected t days after an attack. (Note: Since we are modeling the equation from the graph, our model lacks precision. In addition, the smallpox graph doesnt appear to pass through (0, 0), and our model does. However, despite the limitations of our model, it does give us a rough idea of what would happen in a bioterrorist smallpox attack.) (b) We can use the differential equation to estimate the rate at which the disease is spreading. dy 5 k (M 2 y) dt 5 0.01147 ( 7000 2 y ) dy ` 5 0.01147 37000 2 y ( 50 ) 4 dt t 5 50 5 0.01147 ( 7000 2 3055 ) 5 45 On the 50th day, we estimate that the disease is spreading at a rate of 45 people per day. That is, approximately 45 additional people are infected between the 50th and 51st days. dy ` 5 0.01147 37000 2 y ( 100 ) 4 dt t 5 100 5 0.01147 ( 7000 2 4777 ) 5 25
679
On the 100th day, we estimate that the disease is spreading at a rate of 25 people per day. That is, approximately 25 additional people are infected between the 100th and 101st days.
Commercial fish growers seek to create a fish habitat that will maximize the size of each fish in the least amount of time. The Von Bertalanffy Limited Growth model is widely used in the fish industry. The model assumes that there is a maximal length, L`, that a fish will attain under optimal conditions and that the rate of change in the length of the fish,
dL , dt
is proportional to the difference between the

dL dt
maximal length and the current length. That is,
5 k ( L ` 2 L ) . Solving this
differential equation, we can find the general form of the Von Bertalanffy Limited Growth model. dL 5 k ( L` 2 L ) dt dL 5 k dt L` 2 L 1 3 L 2 L dL 5 3 k dt ` 2 ln Z L` 2 L Z 5 kt 1 C ln ( L` 2 L ) 5 2 kt 2 C L` 2 L 5 e
2k(t 1 C)
Separation of variables
Since dL 32 ln( L` 2 L ) 4 5 L 2 L `
d 1
Since L` 2 L . 0
L 5 L` 2 e2k(t 1 C) The solution to the differential equation is commonly written as L 5 L`( 1 2 e 2 k(t 2 t ) ) , where t0 is a negative constant that varies from species to species. The constant t0 has the property that L(t0) 5 0. Our solution may be converted from L 5 L` 2 e2k(t 1 C) to L 5 L`( 1 2 e 2 k(t 2 t ) ) by letting
0 0
` C5 2 t0. (No matter what the value of C is, we can always find a t0 that k makes this equality true.)
2 ln( L )
EXAMPLE
Forecasting Fish Size with a Von Bertalanffy Limited Growth Model

Turkish scientists conducted a study on the Eastern Black Sea between 1991 and 1996 and determined the values of k, L`, and t0 for economically important fish species in the region. (Source: www.tagem.gov.tr.) The whiting fish species has the Von Bertalanffy growing constants given in Table 10.8.
680
CHAPTER 10
TABLE 10.8
Whiting Female Male
k 0.11 0.136
L` 43.3 cm 34.2 cm
t0 1.91 years 2.02 years
Source: www.fishbase.org.
(a) According to the model, how long is a two-year-old female whiting? (b) According to the model, how long is a two-year-old male whiting? (c) Does the two-year-old female whiting grow faster than the two-year-old male?
SOLUTION
(a) We know that L( t ) 5 L`( 1 2 e2k(t 2 t ) ) . We have L` 5 43.3, k 5 0.11, and t0 5 2 1.91. Substituting these values into the Von Bertalanffy Limited Growth model yields L ( t ) 5 43.3 ( 1 2 e20.113t 2 (21.91)4 ) . We determine the length of a two-year-old female whiting by evaluating this function at t 5 2.
0
L( 2 ) 5 43.3 ( 1 2 e20.1132 2 (21.91)4 ) 5 43.3 ( 1 2 e20.11(3.91) ) 5 43.3 ( 0.3496 ) < 15.1 We estimate the length of a two-year-old female whiting to be about 15.1 cm. (b) We know that L ( t ) 5 L`( 1 2 e2k(t 2 t ) ) . We have L` 5 34.2, k 5 0.136, and t0 5 2 2.02. Substituting these values into the Von Bertalanffy Limited Growth model yields L( t ) 5 34.2 ( 1 2 e20.136 3t 2 (22.02)4 ) . We determine the length of a two-year-old male whiting by evaluating this function at t 5 2.
0
L( 2 ) 5 34.2 ( 1 2 e20.13632 2 (22.02)4 ) 5 34.2 ( 1 2 e20.136(4.02) ) 5 34.2 ( 0.4212 ) < 14.4 We estimate the length of a two-year-old male whiting to be about 14.4 cm. (c) Finally, we need to evaluate each of the differential equations at t 5 2. Recall that the general form of the differential equation is
Female:
dL ` 5 0.11 343.3 2 L( 2 ) 4 dt t 5 2 5 0.11 ( 43.3 2 15.1 ) 5 3.1
dL dt
5 k ( L` 2 L ) .
A two-year-old female whiting grows at a rate of 3.1 cm per year.
Male:
dL ` 5 0.136 334.2 2 L ( 2 ) 4 dt t 5 2 5 0.136 ( 34.2 2 14.4 ) 5 2.7 A two-year-old male whiting grows at a rate of 2.7 cm per year.
681
From our calculations we conclude that the two-year-old female whiting grows faster than the two-year-old male whiting.
Logistic Growth
When a successful product is introduced to the market, cumulative product sales often increase very slowly initially; however, as the popularity of the product increases, sales increase rapidly. Then, as the market becomes saturated with the product, cumulative sales growth again slows. This type of growth is called logistic growth.
LOGISTIC GROWTH
Assume that the rate of growth of a function y with maximum value M is proportional to the product of the present value of y and the difference between the present value of y and M. That is, dy 5 ky ( M 2 y ) dt The solution to this differential equation with initial condition y (0) 5 is given by y5 M 1 1 Se2kMt M 11S
The graph of the solution is an s-shaped (sigmoidal) graph (see Figure 10.15).
y y=M
FIGURE 10.15
For example, consider the digital video disc (DVD) player. The DVD player was introduced into the market in the first quarter of 1997. By the end of 2002, nearly 50 percent of U.S. homes had DVD capability. (Source: DVD Entertainment Group.) Looking at a scatter plot of the cumulative number of DVD players sold (Figure 10.16), it initially looks as if the cumulative number of DVD players sold is increasing exponentially.
682
CHAPTER 10
50,000 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5000 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Quarters (since first quarter 1997)
FIGURE 10.16
However, when we attempt to model the data with an exponential function (Figure 10.17), we see that the exponential function ends up growing much more rapidly than the data.
50,000 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5000 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Quarters (since first quarter 1997) DVD players (in thousands) DVD players (in thousands)
FIGURE 10.17
However, if we model the data with a logistic function (Figure 10.18), we see that DVD player sales appear to exhibit logistic growth behavior.
70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 0 5 10 15 20 25 Quarters (since the end of first quarter 1997) 30 35
FIGURE 10.18
Using logistic regression on the TI-83 Plus, we determine that the logistic model for the given DVD player data is given by D( t ) 5 68,571 thousand DVD players 1 1 288.8e20.28708t
where t is the number of quarters since the end of the first quarter of 1997.
DVD players (in thousands)
683
Plotting the graph of D together with the scatter plot of the data (Figure 10.19), we see that the solution fits the data remarkably well.
DVD players (in thousands) 70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 0 5 10 15 20 25 Quarters (since the end of first quarter 1997) 30 35
FIGURE 10.19
How rapidly were DVD player sales increasing at the end of the first quarter of 2000? We could differentiate D ( t ) 5 1 at t 5 12. However, it may be easier to calculate Dr( 12 ) using the differential equation. dD 5 kD (M 2 D) dt Since 2 0.28708t 5 2 Mkt, k 5 M . Since M 5 68,571, k 5 differential equation may now be written as follows: dD 0.28708 5 a b D ( 68,571 2 D ) dt 68,571 5 0.28708D 2 0.000004187D 2 dD ` 5 0.28708D ( 12 ) 2 0.000004187 3D ( 12 ) 4 2 dt t 5 12 Dr( 12 ) 5 0.28708 ( 6713.1 ) 2 0.000004187 ( 6713.1 ) 2 5 1927.20 2 188.69 < 1739 At the end of the first quarter of 2000, the cumulative number of DVD players sold was increasing at a rate of 1739 thousand DVD players per quarter. That is, from the end of the first quarter of 2000 to the end of the second quarter of 2000, we estimate that 1739 thousand DVD players were sold. How rapidly will DVD player sales be increasing at the end of the first quarter of 2006? That is, what is Dr( 36 ) ? dD ` 5 0.28708D ( 36 ) 2 0.000004187 3D ( 36 ) 4 2 dt t 5 36 Dr( 36 ) 5 0.28708 ( 67,934 ) 2 0.000004187 ( 67,934 ) 2 5 19,502 2 19,323 5 179
0.28708 0.28708 . 68,571 68,571 1 288.8e20.28708t
and evaluate the derivative
The
684
CHAPTER 10
At the end of the first quarter of 2006, we forecast that the cumulative number of DVD players sold will be increasing at a rate of 179 thousand DVD players per quarter. That is, from the end of the first quarter of 2006 to the end of the second quarter of 2006, about 179 thousand DVD players will be sold. We are more skeptical of this estimate than of the 2000 estimate, since the first quarter of 2006 (t 5 36) is so far outside the raw data set ( 1 # t # 22 ) . In practice, it is easiest to create a logistic model by using logistic regression. The following Technology Tip demonstrates how to find a logistic model for a set of data.
EXAMPLE 4
Using a Differential Equation to Forecast the Price of a VCR

Based on data from 1997 to 2002, the average price of a video cassette recorder (VCR) may be modeled by P(t) 5 110.0 1 65 dollars 1 1 0.1955e1.150t
where t is the number of years since the end of 1997. The dramatic drop in price may be attributed in part to the introduction of the DVD player to the market in the first quarter of 1997. (Source: Modeled from Consumer Electronics Association data.) How quickly was the average price of a VCR changing at the end of 1999 and the end of 2001?
SOLUTION The function P (t) is a logistic function shifted vertically 65 units. Well define a new function N(t) 5 P(t) 2 65. We have
N(t) 5
110.0 1 1 0.1955e1.150 t
This function is a solution to the differential equation dN 5 kN ( 110.0 2 N ) dt 5 a Since

dN dt
2 1.150 b N ( 110.0 2 N ) 110.0
5 2 1.15N 1 0.01045N 2 5
dP , dt
dP 5 2 1.15N 1 0.01045N 2 dt dP ` 5 2 1.15N ( 2 ) 1 0.01045 3N ( 2 ) 4 2 dt t 5 2 5 2 1.15 ( 37.29 ) 1 0.01045 ( 37.29 ) 2 5 2 28.35 At the end of 1999, the price of a VCR was dropping at a rate of about $28.35 per year. dP ` 5 2 1.15N( 4 ) 1 0.01045 3N( 4 ) 4 2 dt t 5 4 5 2 1.15( 5.379 ) 1 0.01045 ( 5.379 ) 2 5 2 5.88
685
At the end of 2001, the price of a VCR was dropping at a rate of about $5.88 per year. The introduction of the DVD player had a dramatic impact on the price of a VCR. In the past several examples, we have demonstrated the use of both limited and logistic growth models. Both types of models approach a constant value as the domain values grow large; however, near the origin the models behave quite differently. A limited growth model grows very rapidly at first, while a logistic model initially grows very slowly. Consequently, when choosing a model to use, it is important to consider the expected growth behavior of the function at the start of the time period in addition to looking at the shape of the scatter plot.
Logistic Regression 1. Enter the data using the Statistics Menu List Editor. (Refer to Section 1.3 if youve forgotten how to do this.)
2. Bring up the Statistics CALC menu, select item B:Logistic, and press ENTER .
3. If you want to automatically paste the regression equation into the Y editor, press the key sequence VARS Y-VARS; 1:Function; 1:Y1 and press ENTER . Otherwise press ENTER .
10.4 Summary
In this section, you learned how to use differential equations to find limited growth and logistic growth models. You also practiced interpreting the results of differential equations.
686
CHAPTER 10
10.4 Exercises
In Exercises 16, determine the equation of the limited growth model for each species of fish and then answer the associated questions. You may find it helpful to refer to Example 3 in this section. 1.
Red Mullet
3.
Flounder
Flounder Female Red Mullet Female Male k 0.23 0.21 L` 25.6 cm 22.2 cm t0 1.38 years 2.08 years Male
k 0.197 0.22
L` 44.98 cm 33.0 cm
(a) How long is a male, three-year-old red mullet? (b) How long is a female, three-year-old red mullet? (c) Is the three-year-old male or the three-year-old female red mullet growing faster? 2.
Turbot Female Male
Turbot
(a) At what rate does a 20.0-cm-long male flounder grow? (b) At what rate does a 20.0-cm-long female flounder grow? (c) Which is older, a 20-cm-long male flounder or a 20-cm-long female flounder? Explain. 4.
Pickerel Female
Pickerel Fish
k 0.232 0.793
L` 17.6 cm 20.6 cm
k 0.115 0.143
L` 103.4 cm 77.3 cm
t0 0.93 year 1.22 years
Male
(a) At what rate does a 50.0-cm-long male turbot grow? (b) At what rate does a 50.0-cm-long female turbot grow? (c) Which is older, a 50.0-cm-long male turbot or a 50.0-cm-long female turbot? Explain.
(a) How long is a male, four-year-old pickerel? (b) How long is a female, four-year-old pickerel? (c) Is the four-year-old male or the four-year-old female pickerel growing faster? 5.
Copper Shark
687
Copper Shark
k 0.038
L` 385.0 cm
t0 3.477 years
deaths, where t is the number of years since the end of 1981. (Source: Modeled from Centers for
Disease Control and Prevention data.)
(a) How long is a three-year-old copper shark? (b) At what rate is the length of a three-year-old copper shark changing? 6.
Cowcod
Cowcod
k 0.052
L` 86.9 cm
t0 1.94 years
(a) How long is a two-year-old cowcod? (b) At what rate is the length of a two-year-old cowcod changing? In Exercises 712, use the logistic growth model and associated differential equation to answer the questions. 7.
(a) According to the model, how many adult and adolescent AIDS deaths occurred in the United States in 1995? (b) According to the model, at what rate were adult and adolescent AIDS deaths increasing at the end of 1995? (c) The number of adult and adolescent AIDS deaths in the United States has decreased every year since 1995, in part because of the increased availability of drug treatments and AIDS prevention efforts. In 2001, there were 8063 deaths. If the number of AIDS deaths had followed the logistic model, how many deaths would have occurred in 2002? 9.
School Internet Access Based on data from 1994 to 2000, the rate of change in the percentage of public-school classrooms with Internet access may be modeled by dy 5 0.01077y (85.88 2 y) dt
1992 to 2001, the estimated number of new pediatric AIDS cases in the United States in year t may be modeled by 919.9 1 90 P(t) 5 1 1 0.0533e0.07758t cases, where t is the number of years since the end of 1992. (Source: Modeled from Centers for Disease
Control and Prevention data.)
percentage points per year, where t is the number of years since the end of 1994. In 2000, 77 percent of public-school classrooms had Internet access. (Source: Modeled from Statistical Abstract of
The rate of change of the pediatric AIDS function P is the same as the rate of change of the function N(t) 5
919.9 . 1 1 0.0533e0.07758t
(a) Find the particular solution to the differential equation and interpret its real-world meaning. (b) At what rate was the percentage of public schools with Internet access changing at the end of 2000? 10.
National Internet Usage Based on data from 1995 to 1999 and U.S. Census Bureau projections for 2000 to 2004, the rate of change in annual per capita Internet usage may be modeled by dy 5 0.002882y (232.2 2 y) dt
(a) How many new pediatric AIDS cases are estimated to have occurred in 2000? (b) What is the differential equation whose solution is N(t)? (c) At what rate was the estimated number of new pediatric AIDS cases changing at the end of 2000? (d) Based on the results of part (c) and the graph of P(t), does it look like efforts to reduce pediatric AIDS in the United States are generating positive results? Explain. 8.
Adult and Adolescent AIDS Based on data from 1981 to 1995, the number of adult and adolescent deaths due to AIDS in a given year in the United States may be modeled by 53,955 P5 1 1 38.834e20.45127t
hours per year, where t is the number of years since the end of 1995. (Source: Statistical Abstract of the United States, 2001, Table 1125, p. 704.) In 1999, the annual per capita Internet usage was 99 hours. (a) Find the particular solution to the differential equation and interpret its real-world meaning. (b) At what rate was the annual per capita Internet usage changing at the end of 2000?
688 11.
CHAPTER 10
Hotel/Motel Occupancy Based on data from 1990 to 1999, the average hotel/motel room rate may be modeled by
news of the attack. How long did it take for 75 percent of the city to hear the news, and at what rate was the news spreading at that time? 14.
Video Game Sales Blizzard
27.80 1 57 ( 1 1 47.23e20.6425t ) dollars per day, where t is the number of years since the end of 1990. (Source: Statistical Abstract of R( t ) 5
The rate of change of the room rate function R is the same as the rate of change of the function 27.80 ( 1 1 47.23e20.6425t ) (a) Find the differential equation whose solution is N. (b) Determine at what rate the hotel/motel room rate was changing at the end of 1997 and at the end of 1999. N( t ) 5
Entertainment released the engaging realtime strategy game Warcraft III: Reign of Chaos on July 3, 2002. By July 22, 2002, more than a million copies of the game had been sold. (Source:
www.pcgameworld.com.)
Model the Warcraft III game sales, assuming that at the end of the twentieth day after the game was released, 1 million copies of the game had been sold. Furthermore, assume that a total of 8 million copies of the game will be sold over the life of the game. According to the model, how long did it take for 4 million copies to be sold, and at what rate were sales increasing at that time? In Exercises 1520, use logistic regression to find the logistic model for the data. Then answer the given questions. 15.
Deadly Fights over Money
12.
Information Technology Based on data from 1990 to 2000, the percentage of the economy attributed to the information technology sector may be modeled by
2.956 P( t ) 5 1 7.9 1 1 30.15e20.5219t percent, where t is the number of years since the end of 1990. (Source: Modeled from Statistical
Homicides Due to Arguments over Money or Property
The rate of change in the information technology percentage of the economy function P is the same as the rate of change of the function 2.956 N( t ) 5 1 1 30.15e20.5219t (a) Find the differential equation whose solution is N. (b) Determine at what rate the information technology percentage of the economy was changing at the end of 1995 and at the end of 2000. In Exercises 1314, use a limited growth function to model the data. Then answer the given questions. 13. Spread of Information Suppose that in a city of 350,000 people, 40,000 people were watching television or listening to the radio when the news of a bioterrorist attack was first broadcast. For the purpose of the model, assume that all 40,000 people heard the news simultaneously five minutes after the attack. Model the spread of the
Years since 1990 (t) 1 2 3 4 5 6 7 8 9 10
Homicides (H ) 520 483 445 387 338 328 287 241 213 206
Source: Crime in the United States 1995, 2000, Uniform Crime Report, FBI.
According to the model, at what rate were money-related homicides decreasing in 2001?
689
16.
Cassette Tape Market Share
Years Since 1993 (t) 0 1 2 3 4 5 6 7 8 9
Percent of Music Market (percentage points) (P) 38.0 32.1 25.1 19.3 18.2 14.8 8.0 4.9 3.4 2.4
(b) IFC Films was the distributor for My Big Fat Greek Wedding. (Source: www.boxofficeguru .com.) If you were a marketing consultant to IFC Films, what would you tell the company about forecasted box office sales beyond Week 51? 18.
Resource Value
Value of Fabricated Metals Shipments
Years Since 1992 (t) 0 1 2 3 4 5 6 7 8
Shipment Value (millions of dollars) (V ) 170,403 177,967 194,113 212,444 222,995 242,812 253,720 256,900 258,960
Source: Recording Industry of America.
According to the model, in what year will cassette tape market share drop below 1 percent? At what rate will the cassette tape market share be decreasing at that time? 17.
Movie Box Office Sales
My Big Fat Greek Wedding

Cumulative Gross Box Office Sales (dollars) 597,362 19,340,988 95,824,732 199,574,370 236,448,697 241,437,427
Weekend Apr. 1921 (2002) June 28 30 Sept. 6 8 Nov. 1517 Jan. 2426 (2003) Apr. 4 6
Week # 1 11 21 31 41 51
(Hint: Before creating the model, align the data by subtracting 170,000 from each value of V. After doing logistic regression, add back the 170,000 to the resultant model equation.) According to the model, was the value of fabricated metals shipments increasing more rapidly in 1995 or in 1997?
Source: www.boxofficeguru.com.
(a) According to the model, were cumulative box office sales for My Big Fat Greek Wedding increasing at a higher rate in Week 1 or Week 51?
690 19.
CHAPTER 10
TV Homes with Cable
20.
U.S. Army Personnel
Years Since 1970 (t) 0 5 10 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
TV Homes with Cable (percent) (C ) 6.7 12.6 19.9 42.8 45.6 47.7 49.4 52.8 56.4 58.9 60.2 61.4 62.4 63.4 65.3 66.5 67.2 67.5
Years Personnel Since 1980 (thousands) (t) (P ) 0 2 4 6 8 10 12 14 16 18 20 777 780 780 781 772 732 610 541 491 484 482
(Hint: Before creating the model, align the data by subtracting 480 from each value of P. After doing logistic regression, add back the 480 to the resultant model equation.) According to the model, how many people were in the U.S. Army in 2005, and at what rate was that number changing?
According to the model, what percentage of TV homes will have cable in 2006, and at what rate will that percentage be increasing?
21. Consider the limited growth differential equation model

dy dt
5 0.05 ( M 2 y ) for y , M. Describe
what happens to the graphs of solution curves for different values of M. Be as specific as you can. 22. Consider the logistic growth differential equation model
dy dt
5 0.03y ( M 2 y ) for y , M. Describe
what happens to the graphs of solution curves for different values of M. Be as specific as you can. 23. Explain why, for the limited growth differential equation model
dy dt
5 k ( M 2 y ) , the solution
curve will always be an increasing function for t $ 0 and y $ 0.
691
24. Explain why, for the logistic growth differential equation model
dy dt
5 ky( M 2 y ) , the solution
curve will always be an increasing function for t $ 0 and y $ 0. 25. Using the idea of rate of change, describe the shape of the limited growth solution curve as compared to the shape of the logistic growth solution curve. Specifically, compare the concavity of each solution curve and explain what it means.
26. Prove that if
y 5 M( 1 2 e2kt ) . 27. Prove that if y 5 1

M , 1 Se2kMt
dy dt
5 k( M 2 y ) and y (0) 5 0, then

dy dt
then
5 ky ( M 2 y ) .
28. Show that if D( t ) 5 1
Dr( t ) 5 0.001 3D( t ) 4 5100 2 3D( t ) 46 .
100 , 1 200e20.1t
then
Solve first-order linear differential equations using the Reverse Product Rule, also known as the method of integrating factors Model real-world situations using first-order linear differential equations
Pollution can cause problems in lakes and streams. Differential equations can be used to model the flow of pollution into and out of a lake or stream. In one particular case, Trout Lake in north central Minnesota, there were two main sources of pollution. First, from about 1907 to 1940, local iron ore mining operations dumped ore tailings into Trout Lake. (Iron ore tailings are the waste product generated in the mining process.) Second, two local cities, Bovey and Coleraine, discharged waste water into Trout Lake from the early 1900s until 1987. Trout Lake serves as the source of Trout Creek, which flows into the Swan River and then into the Mississippi River. To determine the amount of pollution in the lake, scientists need to measure the rate at which pollution enters into the lake and the rate at which water leaves the lake. We will use this scenario of pollution entering a lake from multiple sources and leaving a lake into a river as a context in which to study first-order linear differential equations. In this section, we will discuss how to solve nonseparable differential equations using the integrating factor method, also known as the Reverse Product Rule method.
All of the differential equations we have encountered so far have been first-order differential equations. A first-order differential equation is one that contains only first derivatives of the dependent variable. Most of the differential equations that we have seen so far have also been linear. A first-order differential equation is linear if it can be written in the form dy 5 g( t ) ? y 1 r( t ) dt Note that it is the dependent variable, y, that determines whether or not a differential equation is linear. Study Table 10.9.
692
CHAPTER 10
TABLE 10.9
Linear Differential Equations dy 5 t 2 ? y 1 sin t dt dP 54?P17 dt dw 5 (t 1 7) ? w 2 8 dt dQ 5Q?t28?t dt
Nonlinear Differential Equations dy 5 2 ? y2 dt dy 5 2t 1 sin y dt dP 2 e P 5 2P 1 6 dt dv 5 ln t 1 sin v dt
Solving Linear Differential Equations

To solve a linear differential equation, we can apply a method known as the Reverse Product Rule or the integrating factor method. For the moment, we will focus on the Reverse Product Rule. Recall that the Product Rule for derivatives can be expressed as d dv du 3u ? v4 5 u ? 1v ? dt dt dt where u and v are assumed to be functions of t. To demonstrate the Reverse Product Rule method for solving linear differential equations, we will begin with an example.
EXAMPLE
Solving a Differential Equation Using the Reverse Product Rule Method

Solve
dy dt
5 2 ? y 1 (t 1 3).
4 t
SOLUTION First, we notice that we do indeed have a linear first-order differ-
ential equation with g( t ) 5 2 t and r( t ) 5 ( t 1 3 ) . We can rewrite the differential equation as dy 4 5 2 ? y 1 (t 1 3) dt t dy 4 1 ? y 5 (t 1 3) dt t

Add 4 ? y to both sides. t
Notice that the left side of the equation nearly looks like the result of the Product Rule. We can see that if we multiply both sides of the equation by t 4, the left side of the equation will represent the derivative of the product t 4 ? y, since dy d 4 3t ? y4 5 t 4 ? 1 4t 3 ? y dt dt
693
To reiterate, after multiplying both sides of the equation by t 4, we obtain t4 ? dy 4 1 t 4 ? ? y 5 t 4( t 1 3 ) dt t dy 1 4t 3 ? y 5 t 5 1 3t 4 dt
t4 ?
The left side of the equation, which now represents the results of the Product Rule, can be rewritten as follows. d 4 3t ? y4 5 t 5 1 3t 4 dt We now integrate both sides of this equation with respect to t and solve for y. 1 3 3t 4 ? y4 5 t 6 1 t 5 1 C 6 5 1 6 3 5 t 1 t 1C 6 5 y5 t4 1 3 C y 5 t2 1 t 1 4 6 5 t We were able to solve the differential equation in Example 1 by multiplying both sides of the equation by t 4, which allowed us to apply the Reverse Product Rule. But how did we know that multiplying by t 4 would allow us to apply the Reverse Product Rule? This special factor that allows us to apply the Reverse Product Rule is known as the integrating factor. To compute the integrating factor, we must first rewrite the equation as follows. dy 5 g( t ) ? y 1 r( t ) dt dy 2 g( t ) ? y 5 r( t ) dt We then compute the integrating factor, a( t ) .
2g(t)dt a( t ) 5 e #
At this point you might wonder, Why is the integrating factor found in this way? We will explore the origins of the integrating factor later. For now, lets focus our attention on the idea that the integrating factor enables us to employ the Reverse Product Rule in solving the differential equation.
EXAMPLE 2
Using an Integrating Factor to Solve a Differential Equation

Solve
dy dt
5 2 ? y 2 cos t.
SOLUTION First, we notice that we do indeed have a linear first-order differential equation with g( t ) 5 2 and r( t ) 5 2 cos t. We can rewrite the differential equation as
dy 2 2 ? y 5 2 cos t dt
694
CHAPTER 10
Next, we compute the integrating factor. a( t ) 5 e # 5 e22t

22 dt
We can see that if we multiply both sides of the equation by e22t, the left side of the equation will represent the derivative of the product e22t ? y. d 22t dy 3e ? y4 5 e22t ? 2 2e22t ? y dt dt To reiterate, after multiplying both sides of the equation by e22t, we obtain e22t ? dy 2 2e22t ? y 5 2 cos t ? e22t dt
The left side of the equation, which now represents the results of the Product Rule, can be rewritten as follows. d 22t 3e ? y4 5 2 cos t ? e22t dt We now integrate both sides of this equation with respect to t and solve for y. e22t ? y 5 2 1 ? cos t ? e22t 2 ? sin t ? e22t 1 C 5 5 2 1 ? cos t ? e22t 2 ? sin t ? e22t 1 C 5 5 y5 e22t 2 1 y 5 ? cos t 2 ? sin t 1 Ce 2t 5 5
Using integration by parts
Solving a First-Order Linear Differential Equation To solve a first-order linear differential equation of the form dy 5 g( t ) ? y 1 r( t ) dt
1. Subtract the g( t ) ? y term from both sides. 2. Compute the integrating factor
2g(t)dt a( t ) 5 e #
3. Multiply both sides of the equation by a( t ) . 4. Rewrite the left side of the equation by applying the Reverse Product Rule. 5. Integrate both sides of the equation with respect to t. 6. Solve for y.
Before returning to the Trout Lake scenario, we consider a similar situation in Example 3.
695
Just In Time
For example,
Rewriting Rational Expressions
In this section, it might be helpful to algebraically rewrite rational expressions.

2 ? (5 1 t)3 1 C (5 1 t)2
might be rewritten as follows.
2 ? (5 1 t)3 1 C 2 ? (5 1 t)3 C 5 1 2 (5 1 t) (5 1 t)2 (5 1 t)2 5 2 ? ( 5 1 t ) 1 C( 5 1 t ) 22 5 2 ? (5 1 t) 1 C (5 1 t)2
EXAMPLE
Using Differential Equations to Model Flow Rates

Imagine that we have a 25-gallon tank that initially contains 5 gallons of salt water containing 2 pounds of salt. Salt water with a concentration of 2 pounds of salt per gallon is pumped into the tank at a rate of 3 gallons per minute. A wellmixed solution is draining from the tank at a rate of 2 gallons per minute. How much salt is in the tank after 10 minutes?
SOLUTION One way to model the rate at which salt is entering and leaving the
tank is Rate of change 5 rate in 2 rate out We want to predict how much salt is in the tank after 10 minutes. We can see that the rate of change,
dS , dt pounds of salt . Therefore, minute pounds of salt as well. In the minute
is measured in
we want the rate in problem statement,
and the rate out to be measured in
we see that salt water with a concentration of 2 pounds of salt per gallon is being pumped into the tank at a rate of 3 gallons per minute. Therefore, Rate in 5 3 56 gallons pounds ?2 minute gallon pounds minute
gallons
The salt-water solution is draining from the tank at a rate of 2 minute . Since the volume of salt water in the tank is changing over time as well as the amount of salt in the tank, we can compute the rate as follows. Rate out 5 2 5 gallons S pounds ? minute 5 1 t gallon
To compute these rates, we multiply the rate of flow A measured in concentration of salt in the water A measured in
pounds gallon
2S pounds 5 1 t minute
B. In the case of water draining
gallons minute
B by the
696
CHAPTER 10
out, the concentration changes over time. Therefore, we use the variable S to represent the amount of salt in the tank. The expression 5 1 t represents the initial amount of salt water in the tank (5 gallons) plus the 1 gallon per minute net rate of change in volume (3 gallons per minute in minus 2 gallons per minute out) times t minutes. More specifically, 5 gallons 1 1 Finally, we have Rate of change 5 rate in 2 rate out dS 2S 562 dt 51t This differential equation can be solved using the method of integrating factors (Reverse Product Rule), since the equation is linear and can be rewritten as follows. dS 2S 1 56 dt 51t dS 2 1 ?S56 dt 51t We compute the integrating factor.
2 a( t ) 5 e #5 1 t dt
gallon ? t minutes 5 5 1 t minute
5 e 2 ln(5 1 t) 5 e ln(5 1 t)
2
5 (5 1 t)2 We multiply both sides of the differential equation by the integrating factor and rewrite the equation so that we can use the Reverse Product Rule. (5 1 t)2 dS 2 1 (5 1 t)2 ? ? S 5 6 ? (5 1 t)2 dt 51t dS 1 (5 1 t) ? 2 ? S 5 6 ? (5 1 t)2 dt d C( 5 1 t ) 2 ? SD 5 6 ? ( 5 1 t ) 2 dt We now integrate on both sides with respect to t and solve for S. d C( 5 1 t ) 2 ? SD 5 6 ? ( 5 1 t ) 2 dt (5 1 t)2 ? S 5 2 ? (5 1 t)3 1 C S5 2 ? (5 1 t)3 1 C (5 1 t)2 C (5 1 t)2
(5 1 t)2
S 5 2 ? (5 1 t) 1
697
Initially, there were 2 pounds of salt in the tank. Using this information, we can determine the value of C. S 5 2 ? (5 1 t) 1 2 5 2 ? (5 1 0) 1 2 5 10 1 28 5 C 25 200 (5 1 t)2 C 25 C (5 1 t)2 C (5 1 0)2
C 5 2 200 S 5 2 ? (5 1 t) 2
We were asked to determine the amount of salt in the tank after 10 minutes. S 5 2 ? (5 1 t) 2 200 (5 1 t)2 200 ( 5 1 10 ) 2
S 5 2 ? ( 5 1 10 ) 2 S 5 30 2 200 225
S < 29.1 pounds of salt After 10 minutes, there are about 29.1 pounds of salt in the tank. We will now consider the situation involving pollution entering and leaving a lake. It is very similar to the salty tank problem, but we will replace salt with pollution and tank with lake.
EXAMPLE
Using Differential Equations to Model Lake Pollution

A small lake contains 150,000 cubic meters of water. Initially, the lake contains only clean water provided by one main source, which causes water to flow into the lake at a rate of 1470 cubic meters per day. There is a single outlet that allows water to leave the lake at a rate of 1470 cubic meters per day. Suppose that pollution in the form of waste water is pumped into the lake source at a concentration of 3 kilograms per 1000 cubic meters, at a rate of 900 cubic meters per day. Pollution in the form of ore tailings (a sand-like substance) is pumped into the lake at a rate of 30 cubic meters per day. The tailings settle to the bottom of the lake, causing the volume of the lake to be reduced by 30 cubic meters per day. We will assume that the pollution and tailings are instantly and evenly dispersed throughout the lake. The reduction in volume and increase in the rate of flow into the lake is compensated by the outflow, which is increasing to 2400 cubic meters per day. Describe the amount of pollution in the lake over time. What happens in the long run?
698
CHAPTER 10
SOLUTION We will again begin with the model
Rate of change 5 rate in 2 rate out We can see that the rate of change, Rate in 5 900 5 2.7
dP , dt
would be measured in
kilograms of pollution . day
cubic meters kilograms ?3 day 1000 cubic meters kilograms day

cubic meters . Since the volume day
Lake water is draining from the lake at a rate of 2400
of water in the lake is changing over time, we can compute the rate out as follows. The concentration changes over time. Therefore, we use the variable P to represent the amount of pollution in the lake. The expression 150,000 2 30t represents the initial amount of water in the lake (150,000 cubic meters) minus 30 cubic meters per day (rate of change of volume caused by the tailings settling at the bottom of the lake) times t days. More specifically, 150,000 cubic meters 2 30 Rate out 5 2400 5 5 cubic meters ? t days 5 150,000 2 30t day
cubic meters P kilograms ? day 150,000 2 30t cubic meters
2400P kilograms 150,000 2 30t day 80P 5000 2 t
cubic meters day
We see that to compute these rates, we multiply the rate of flow A measured in Finally, we have
kilograms . B by the concentration of pollution in the water Ameasured in cubic meter B
Rate of change 5 rate in 2 rate out dP 80P 5 2.7 2 dt 5000 2 t We can solve this differential equation by using the integrating factor method. dP 80P 1 5 2.7 dt 5000 2 t dP 80 1 ? P 5 2.7 dt 5000 2 t We compute the integrating factor. a( t ) 5 e (50002t) dt
3
80
5 e280 ? ln(5000 2 t) 5 e ln (5000 2 t)

2 80
5 ( 5000 2 t ) 280
699
Next, we multiply both sides of the differential equation by the integrating factor and rewrite the equation so that we can apply the Reverse Product Rule. ( 5000 2 t ) 280 dP 80 1 ( 5000 2 t ) 280 ? ? P 5 2.7 ? ( 5000 2 t ) 280 dt ( 5000 2 t ) ( 5000 2 t ) 280 dP 80 1 ? P 5 2.7 ? ( 5000 2 t ) 280 dt ( 5000 2 t ) 81 d 3( 5000 2 t ) 280 ? P4 5 2.7 ? ( 5000 2 t ) 280 dt We now integrate both sides with respect to t and solve for P. d C ( 5000 2 t ) 280 ? P D 5 2.7 ? ( 5000 2 t ) 280 dt ( 5000 2 t ) 280 ? P 5 2.7 ? ( 5000 2 t ) 279 1 C 79
2.7 ? ( 5000 2 t ) 279 1 C 79 P5 ( 5000 2 t ) 280 P5 2.7 ? ( 5000 2 t ) 1 C ? ( 5000 2 t ) 80 79
Using the fact that the pond was free of pollution initially, we can find the value of C. P5 2.7 ? ( 5000 2 t ) 1 C ? ( 5000 2 t ) 80 79 2.7 ? ( 5000 2 0 ) 05 1 C ? ( 5000 2 0 ) 80 79 13,500 1 C ? 500080 79 13,500 79
05 2C ? 500080 5
C52
13,500 79 ? 500080 2.7 ? ( 5000 2 t ) 13,500 P5 2 ? ( 5000 2 t ) 80 79 79 ? 500080 P5 13,500 5000 2 t 80 2.7 ? ( 5000 2 t ) 2 ? a b 79 79 5000
We can now analyze this solution by describing how the amount of pollution in the lake changes over time. First, we will look at a slope field (Figure 10.20) of the original differential equation and a solution curve with initial condition P( 0 ) 5 0.
700
CHAPTER 10
Differential Equations P 150
100
50
0 0 1000 2000
FIGURE 10.20
We see that the level of pollution in the lake increases very quickly initially and then decreases slowly over time. Eventually, the pollution from the waste water will not be present in the lake. However, over time, the lake will be completely filled with the tailings.
The idea that the rate of change is equal to the rate going in minus the rate going out can be applied to other situations as well. Consider a bank account that is accumulating interest (rate of money into the account) but is decreasing in value due to periodic withdrawals (rate of money out of the account). The overall change in the amount of money in the account can be modeled using the rate in minus rate out model. Consider a bank account into which periodic contributions are being made and from which no money is being withdrawn. The account value is increasing due to the periodic contributions and the interest earned. Both the contribution rate and the rate of interest earned are rates in. We have two different rates in, so the overall rate of change of money in the account can be modeled using the model rate in plus rate in. We will use this approach in Example 5.
EXAMPLE
Using Differential Equations to Plan for Retirement

To prepare for retirement, a 25-year-old woman opens a savings account that earns 6 percent interest per year, compounded continuously. She contributes $2000 per year to the account. She wants to retire at age 55. At that time, she plans to stop contributing to the account and begin withdrawing from the account at a rate of $60,000 per year. How long will the money last?
SOLUTION First, we will consider the growth of the account prior to retirement. Then we will develop a different model to analyze the account after retirement. Before retirement, a differential equation that models the growth of the account is
dM 5 0.06 ? M 1 2000 dt
701
where M( t ) represents the amount of money in the account after t years. We can solve this linear differential equation by using integrating factors. dM 5 0.06 ? M 1 2000 dt dM 2 0.06 ? M 5 2000 dt We compute the integrating factor.
2 0.06dt a( t ) 5 e#
5 e20.06t We multiply both sides of the differential equation by the integrating factor. e20.06t ? dM 2 0.06 ? e20.06t ? M 5 2000 ? e20.06t dt
We apply the Reverse Product Rule, integrate both sides with respect to t, and solve for M. e20.06t ? dM 2 0.06 ? e20.06t ? M 5 2000 ? e20.06t dt d 20.06t 3e ? M4 5 2000 ? e20.06t dt e20.06t ? M 5 2 2 M5 2000 20.06t ?e 1C 0.06 2000 20.06t ?e 1C 0.06 e20.06t
M 5 Ce0.06t 2 33,333.3 With the initial condition M( 0 ) 5 0 (there was no money in the account initially), we can determine the value of C. M 5 Ce0.06t 2 33,333.3 0 5 Ce0.06(0) 2 33,333.3 33,333.3 5 C M 5 33,333.3e0.06t 2 33,333.3 Since it will be 30 years before the woman retires, we can determine how much money will be in the account at that time. M 5 33,333.3e0.06 ? 30 2 33,333.3 5 $ 168,321.58 Now we can compute how long this money will last after retirement if $60,000 is withdrawn each year. dM 5 0.06 ? M 2 60,000 dt dM 2 0.06 ? M 5 2 60,000 dt
702
CHAPTER 10
Again, the integrating factor is e20.06t. e20.06t ? dM 2 0.06 ? e20.06t ? M 5 2 60,000 ? e20.06t dt d 20.06t 3e ? M4 5 2 60,000 ? e20.06t dt e20.06t ? M 5 60,000 20.06t ?e 1C 0.06
60,000 20.06t ?e 1C 0.06 M5 e20.06t M 5 1,000,000 1 Ce0.06t Using the initial condition M( 0 ) 5 $ 168,321.58 (the amount of money in the account prior to retirement), we can determine the value of C. 168,321.58 5 1,000,000 1 Ce0.06 ? 0 C 5 2 831,678.42 M 5 1,000,000 2 831,678.42e0.06 ? t We now can determine how long it will be before M 5 0. 0 5 1,000,000 2 831,678.42e0.06 ? t 831,678.42e0.06 ? t 5 1,000,000 e0.06 ? t 5 1,000,000 831,678.42 1,000,000 b 831,678.42 1,000,000 b 831,678.42 0.06
0.06t 5 ln a t5 ln a
< 3.07 The money will last a little over 3 years.
The Integrating Factor

We have been solving first-order linear differential equations using the Reverse Product Rule, or integrating factor method. Why do we compute the integrating 2 g(t)dt factor using e # ? The origins of this integrating factor follow. We begin with the general linear differential equation and rewrite it as we have done in this section. dy 5 g( t ) ? y 1 r( t ) dt dy 2 g( t ) ? y 5 r( t ) dt
703
We multiply both sides of the equation by a( t ) . a( t ) ? dy 2 a( t ) ? g( t ) ? y 5 a( t ) ? r( t ) dt
Suppose that the left side of the equation represents the derivative d dy 3a( t ) ? y4 5 a( t ) ? 2 a( t ) ? g( t ) ? y dt dt where dt 3a( t ) 4 5 2 a( t ) ? g( t ) . We also know that
d
d dy da 3a( t ) ? y4 5 a( t ) ? 1y? dt dt dt Equating these two expressions gives a( t ) ? dy dy da 2 a( t ) ? g( t ) ? y 5 a( t ) ? 1y? dt dt dt
We now solve for a( t ) , the integrating factor. To do this, we first subtract the term dy a( t ) ? dt from both sides of the equation. 2 a( t ) ? g( t ) ? y 5 y ? da dt
Next, we assume that the solution function y is not zero and divide both sides by y. 2 a( t ) ? g( t ) 5 da dt
Next, we divide both sides by a( t ) and multiply both sides by dt. 2 g( t ) dt 5 1 da a( t )
We can integrate both sides with respect to t. 32 g( t ) dt 5 ln Z a( t ) Z Finally, we solve for a( t ) . 32 g( t ) dt 5 ln Z a( t ) Z ln Z a( t ) Z 5 32 g( t ) dt

2g(t) dt eln Za(t)Z 5 e #
a( t ) 5 e #
2g(t) dt
As long as we multiply both sides of our linear differential equation by the inte2g(t) dt grating factor a( t ) 5 e # , we are able to use the Reverse Product Rule method.
704
CHAPTER 10
10.5 Summary
In this section, you learned how to solve first-order linear differential equations. You learned that one method for solving certain types of linear differential equations is the Reverse Product Rule method, also known as the integrating factor method. By multiplying both sides of the equation by the appropriate factor, you can express one side of the equation as a derivative of a product. Then, after integrating both sides with respect to t, you can solve for the dependent variable. This technique is helpful in solving problems that can be modeled using the rate in minus rate out template.

In Exercises 15, rewrite the rational expression, if possible. 1. 2. 3. (x 1 3)3 1 5 (x 1 3)2 5 ? (n 2 6)3 1 5 ? (n 2 6) 1 1 n26 x5 2 x3 1 x 2 5 x2 4. 5. 2 ? (S 1 5)2 1 C (S 1 5)3 (w 1 2)2 1 3 ? (w 1 5) 1 1 w16
10.5 Exercises
In Exercises 15, state whether the differential equation can be solved using separation of variables or the Reverse Product Rule (integrating factor method). You do not need to solve the equation. dy 3y 1. 5 10x 2 1 x dx dy 2x 2 2. 5 3 dx y dy 5 xy 1 3 3. dx dy 4. y 2 2x 5 2 x 3e x dx dy 5 x 1 xy 5. dx
2
In Exercises 620, solve the differential equation using separation of variables or the Reverse Product Rule, as appropriate. 6. 3 7. 8. 9. 10. dy 1 6y 5 50 cos ( 2x ) dx
dy 1 2y 5 e 2x dx dy 2 2yt 5 t dt dN 2N 1 5 6t dt t dP 2 t 2e2P 5 0 dt
705
11. ( x 2 1 ) 12.
dy 1 y 5 2 4x dx
dG 5 t2 dt
flows out of the tank at a rate of 1 gallon per minute. How much salt is in the tank after 5 minutes? How much salt is in the tank when the tank is full? 27.
Swimming Pool In Arizona, many people have backyard pools. One residents 15,000-gallon backyard swimming pool contains 0.5 part per million of chlorine. When the cleaning system is running, it allows 0.00015 part per million of chlorine per gallon to enter the pool, at a rate of 100 gallons per minute. The pool water is pumped out at the same rate. The desired amount of chlorine in the pool is 1.5 parts per million. (Source: www.paddockpoolsandspas.com.) (a) Write a differential equation to model this situation. (b) Create a slope field corresponding to the differential equation you created in part (a). Provide an analysis of this slope field in the context of the situation. (c) How long will the cleaning system need to run for the desired level of chlorine to be reached?
dv 13. t 1 v 5 t ? ln t dt 14. dS 3S 542 dt 10 1 t dt 1 4t 5 "2x dx
15. 4x 16.
dy 5 0.2y dt dC 1 C 5 tet dt
17. t 18. 19. 20.
dy 5 x3 1 x3y dx dy 5 cos x 2 y cos x dx dy 5 sin x 1 y sin x dx 28.
In Exercises 2125, solve the differential equation with the given initial condition. 21. 22. dy 1 y 5 4e x; y( 0 ) 5 4 dx dy 1 y 5 x; y( 0 ) 5 1 dx dy 1 4y 5 6x 2; y( 1 ) 5 4 dx
23. 4x 24. x 25.
dy 1 y 5 x 4 cos x; y( p ) 5 1 dx
dS S 582 ; S( 0 ) 5 5 dt 15 1 t
In Exercises 2630, use differential equation concepts to model the real-world scenarios. 26. Salt Concentration in a Tank A 150-gallon tank initially contains 50 pounds of salt dissolved in 75 gallons of water. Brine containing 5 pounds of salt per gallon flows into the tank at a rate of 2 gallons per minute, and the well-stirred mixture
The clean room in which computer chips are manufactured is often kept at Class 1 status. This means that there is not more than one particle of dirt (of size 0.5 micron) per cubic foot of air. The air in a large clean room is exchanged 10 times per minute to maintain Class 1 status. Intel Corporation is building a new clean room called Fab 32 in Chandler, Arizona. It will have 184,000 square feet of clean room space. (Source: www.intel.com.) Assuming 10-foot ceilings, this clean room will contain 1,840,000 cubic feet of space. Assume this clean room contains 900,000,000 particles of dirt. Before manufacturing can begin, this clean room must be classified as a Class 1 clean room. A ventilation system brings in nearly perfectly clean air (1 3 1026 particles per cubic foot) at a rate of 18,400,000 cubic feet per minute while an equal volume of air leaves the room. Assume that the air mixes completely during the process. (a) Write a differential equation to model this situation. (b) Use Eulers Method with a step size of D t 5 0.01 to determine how long it will
Semiconductor Manufacturing
706
CHAPTER 10
take for the clean room to be ready for manufacturing (Class 1). (c) Determine exactly how long will it take for the air in the clean room to be clean enough for manufacturing (Class 1). Compare your exact answer with your answer using Eulers method (part b). 29. Saving for College A childs parents begin a college fund when the child is 2 years old. They initially place $500 into the account, which pays interest at a rate of 2.5 percent per year, compounded continuously. In addition, they arrange for an additional $2600 per year to be automatically deposited into the account. (This is equal to $50 per week.) (a) Write a differential equation for the amount of money in the account after t years. (b) Find the amount of money in the account when the student is ready to attend college at age 18. 30. College Savings Plan Value Suppose that the child in Exercise 29 attends college beginning at age 18 and withdraws $13,000 per year from the account for educational expenses. At this time, the parents discontinue their contributions to the account. (a) Write a differential equation for the amount of money in the account t years after the first withdrawal. (b) Will the student have enough money in the account to complete a 4-year degree?
34. Find the value of k that allows you to solve the differential equation
dy dt
5 kty 1 2e2t . (Hint: Use

2
the integrating factor method and then choose a value of k that will enable you to integrate successfully.) 35. Describe the conditions under which a differential equation used to model a salt tank scenario would not be expressed in terms of the independent variable t.
36. Write an application problem similar to the scenarios presented in this section (like the saltwater tank, polluted lake, or investment account problem) that matches the differential equation dy y 562 dt 20 2 t 37. Write an application problem similar to the scenarios presented in this section (like the saltwater tank, polluted lake, or investment account problem) that matches the differential equation dy 5 0.12y 2 500 dt 38. Write an application problem that matches the differential equation dy y 5 10 2 dt 100 Use one of the scenarios presented in this section as a guideline.
31. The integrating factor method can also be called the Reverse Product Rule method. Explain why Reverse Product Rule is an appropriate name for this method of solving certain linear differential equations. 32. The differential equation
dy dt
39. Write an application problem that matches the differential equation dy 5 0.085y dt Use one of the scenarios presented in this section as a guideline. 40. Write an application problem that matches the differential equation dy y 5 0.02y a 1 2 b dt 2 Use one of the scenarios presented in this section as a guideline.
5 y 2 1 3t cannot be
solved using the integrating factor method. Explain. 33. Imagine you are explaining to a classmate who missed class the rate in 2 rate out model used in this section. Write an explanation that would make sense to your classmate.
707

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. How is a slope field created? What does each of the short line segments in a slope field represent? (10.1) dy 2. When creating a slope field for a particular differential equation dx, how should the axes be labeled? Why? (10.1) 3. How is a particular differential equation used to approximate solutions using Eulers Method? (10.2) 4. What role does step size play in the accuracy of the solution when using Eulers Method to approximate the solutions of a differential equation? (10.2) 5. How is solving a differential equation using separation of variables different from solving a differential equation using Eulers Method? (10.3) 6. Why is it helpful to separate the variables when solving particular differential equations? (10.3) 7. How does a limited growth differential equation model compare with a logistic growth differential equation model? (10.4) 8. What does the parameter M represent in the differential equations 9. What types of differential equations can be solved using the integrating factor method? (10.5) 10. How is the integrating factor computed, and how is it helpful when solving certain differential equations? (10.5) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
dy dt
5 k( M 2 y ) and
dy dt
5 ky( M 2 y ) ? (10.4)
Definitions
Differential Equation: An equation that contains an unknown function and one or more of its derivatives First-Order Differential Equation: A differential equation that contains only first derivatives of the dependent variable Solution of a Differential Equation: A function in numerical, graphical, or symbolic form that satisfies a differential equation Slope Field: An array of slope marks in the xy plane representing solutions to a differential equation Step Size: The increment used for the independent variable when numerically approximating solutions to a differential equation using Eulers Method Eulers Method: A numerical process of computing function values over small intervals Separable Differential Equation: A first-order differential equation is said to be separable if it may be written as dx 5 g( y ) for some functions f ( x ) and g( y ) .
dy f (x)
708
CHAPTER 10
Limited Growth: Assumes that the rate of growth of a function y with maximum value M is proportional to the difference between the present value of y and M Logistic Growth: Assumes that the rate of growth of a function y with maximum value M is proportional to the product of the present value of y and the difference between the present value of y and M Linear Differential Equation: A differential equation that can be written in the form
dy dt
5 g( t ) ? y 1 r( t )
Solving Separable First-Order Differential Equations The solution to a differential equation is obtained by moving the x and y variables to opposite sides of the equal sign and integrating. That is, 3g( y ) dy 5 3f ( x ) dx. Solving a First-Order Linear Differential Equation To solve a first-order linear differential equation of the form
dy dt
5 g( t ) ? y 1 r( t ) :
1. Subtract the g( t ) ? y term from both sides. 2. Compute the integrating factor a( t ) 5 e 2g(t) dt
3
3. Multiply both sides of the equation by a( t ) . 4. Rewrite the left side of the equation by applying the Reverse Product Rule. 5. Integrate both sides of the equation with respect to t. 6. Solve for y.
Formulas
Eulers Method: ynext 5 ycurrent 1
dy dt
? Dt
Continuous Compound Interest: If dA 5 kA dollars, then A 5 Pekt dt year dollars. Newtons Law of Heating and Cooling: Let T be the temperature of an object at time t and let A be the temperature of the environment surrounding the object (ambient temperature). Then Limited Growth: If y 5 M( 1 2 e ) .
2kt
dT dt
5 k( T 2 A ) , where k is a
constant that varies depending on the physical properties of the object.
dy dt dy dt
5 k( M 2 y ) and y( 0 ) 5 0, then 5 ky( M 2 y ) and y( 0 ) 5 1

M , 1S
Logistic Growth: If y51

M . 1 Se2kMt
then
Integrating Factor: a( t ) 5 e 2g(t) dt

3
709

Section 10.1 In Exercises 16, match the given slope fields with the differential equations A through H. Note that there are six slope fields but eight differential equations, so two equations will not have matches.
A. C. E. G. dy 5y dx dy y 5 x dx dy 5 x2 dx dy 1 5 y dx B. D. F. H. dy 5 y2 dx dy 5 sin y dx dy 5 cos y dx dy 5 2y2 dx 4.
y 6 4 2 0 2 2 4 6 8 x
5. y
8 6 4 2 0 0 2 4 6 8 x
1. y
8 6 4 2
6. y
8 6
0 y
4 2 0 0 2 4 6 8 x
2.
3 2
3. y
8 6 4 2 0 0 2 4 6 8 x
In Exercises 710, do the following: (a) Sketch a solution curve on the slope field through the given value. (b) Write a description of the behavior of this solution curve and what it represents in the context of the problem. 7.
Internet Use The rate at which the number of hours of Internet use per person per year in the United States has changed can be modeled by the differential equation
H dH 5 0.82H a 1 2 b dt 232
710
CHAPTER 10
where H is the projected number of hours of Internet usage and t is time in years since 1995. H 5 5. The corresponding slope field follows.
H 200 150 100 50 0 0 2 4 6 8 x
10.
(Source: Statistical Abstract of the United States, 2001, Table 1125, p. 704.) In this case, when t 5 0,
Newtons Law of Cooling One of the authors placed a fresh pitcher of lemonade into a refrigerator, that maintains a constant temperature of 42.4F. The rate at which the lemonade cools may be modeled by the
differential equation
where T is the temperature of the lemonade t minutes after being placed in the refrigerator. When t 5 0, T 5 79F. The corresponding slope field follows.
T 80 70 60 50 40 0 0 20 40 60 80 t
dT dt
5 2 0.0294( T 2 42.4 ) ,
8. Torricellis Law Water is draining from a cylindrical tank with a 6-foot radius through a hole at the bottom of the tank with a radius of 1 foot. The rate at which the height of the water in the tank changes with respect to time obeys Torricellis Law, which may be modeled by
dh dt
the tank in feet and t is time in seconds. In this case, when t 5 0, h 5 10. The corresponding slope field follows.
y 10 8 6 4 2 0 0 10 20 30 x
5 2 9"h, where h is the height of the water in

2
Section 10.2 In Exercises 1113, use Eulers Method to approximate the value of y( 4 ) given the differential equation and initial condition. Use a step size of 0.5 and make your calculations using paper and pencil.
11. 12. 13. dy 5 3x 2 y, y( 0 ) 5 1 dx dy 5 2 "x, y( 0 ) 5 4 dx dy 5 cos y, y( 0 ) 5 1 dx
9. Population Model Consider a hypothetical situation in which the rate of change of the population of a community is given by
dP dt
5 ( 0.25 cos t ) ? P, where P is the population
In Exercises 1416, use Eulers Method to approximate the value of y( 4 ) given the differential equation and initial condition. Use a step size of 0.1 and use technology. Compare your answers to those found in Exercises 1113. 14. 15. 16. dy 5 3x 2 y, y( 0 ) 5 1 dx dy 5 2 "x, y( 0 ) 5 4 dx dy 5 cos y, y( 0 ) 5 1 dx
in thousands and t is time in years. In this case, when t 5 0, P 5 2. The corresponding slope field follows.
P 10 8 6 4 2 0 0
2
3 2
2 5
2
711
In Exercises 1720, use Eulers Method to approximate the indicated solution of the differential equation. 17.
Internet Use The rate at which the
Section 10.3 In Exercises 2124, find the general solution of the separable differential equation.
21. 22. 23. 24. dy x 5 dx y11 dB 5 2B3 dt dy 5 y 3>2 dx dA 5A 5 2 dt t
number of hours of Internet use per person per year in the United States has changed can be modeled by the differential equation dH H 5 0.82H a 1 2 b dt 232 where H is the projected number of hours of Internet usage and t is time in years since 1995.
H 5 5. Use Eulers Method and a step size of 0.5 to determine when the number of hours of Internet use per person per year will be 200 hours. 18. Torricellis Law Water is draining from a cylindrical tank with a 6-foot radius through a hole with a radius of 1 foot at the bottom of the tank. The rate at which the height of the water in the tank changes with respect to time obeys Torricellis Law, which may be modeled by
dh dt
(Source: Statistical Abstract of the United States, 2001, Table 1125, p. 704.) In this case, when t 5 0,
In Exercises 2528, find the particular solution of the differential equation. a Hint: 3y( M 2 y ) 5 M ln ` M dy 3x 5 ; y( 1 ) 5 1 25. y dx
dy 1 y `b 2y
26. 27. 28.
dA 5 2 2A2; A( 0 ) 5 2 dt dy 5 y( 2 2 y ) ; y( 0 ) 5 1 dx dy 5y 5 ; y( 1 ) 5 2 dt 2t
the tank in feet and t is time in seconds. In this case, when t 5 0, h 5 10. Use Eulers Method and a step size of 0.5 to determine when the height of the water will be 5 feet.
2 2 9"h,
where h is the height of the water in
19. Population Model Consider a hypothetical situation in which the rate of change of the population of a community is given by
dP dt
In Exercises 2931, use separation of variables to solve the problem. 29. Torricellis Law Water is draining from a cylindrical tank with a 6-foot radius through a hole with a radius of 1 foot at the bottom of the tank. The rate at which the height of the water in the tank changes with respect to time obeys Torricellis Law, which may be modeled by dh 2 5 2 "h dt 9 where h is the height of the water in the tank in feet and t is time in seconds. In this case, when t 5 0, h 5 10. Determine when the height of the water will be h 5 5 feet. 30. Population Model Consider a hypothetical situation in which the rate of change of the population of a community is given by dP 5 ( 0.25 cos t ) ? P dt
5 ( 0.25 cos t ) ? P, where P is the population
in thousands and t is time in years. In this case, when t 5 0, P 5 2. Use Eulers Method and a step size of 0.5 to determine when the population of the community will be 3000. 20.
Newtons Law of Cooling One of the authors placed a fresh pitcher of lemonade into a refrigerator that maintains a constant temperature of 42.4F. The rate at which the lemonade cools may be modeled by the
differential equation
where T is the temperature of the lemonade t minutes after being placed in the refrigerator. When t 5 0, T 5 79F. Use Eulers Method and a step size of 0.5 to determine when the temperature of the lemonade will be 50F.
dT dt
5 2 0.0294( T 2 42.4 ) ,
712
CHAPTER 10
where P is the population in thousands and t is time in years. In this case, when t 5 0, P 5 2. Determine when the population of the community will be P 5 1.75 thousand for the first time. 31. One of the authors placed a fresh pitcher of lemonade into a refrigerator that maintains a constant temperature of 42.4F. The rate at which the lemonade cools may be modeled by the differential equation
Newtons Law of Cooling
The rate of change in the number of game units sold, S, is the same as the rate of change of the function N( t ) 5 121.8 1 1 17.23e21.794t
dT 5 2 0.0294( T 2 42.4 ) dt where T is the temperature of the lemonade t minutes after being placed in the refrigerator. When t 5 0, T 5 79F. Determine when the temperature of the lemonade will be 50F.
(a) Find the differential equation whose solution is N. (b) Determine at what rate computer and video game sales were changing in 1997 and in 1999. In Exercises 3435, determine the equation of the limited growth model for each species of fish and then answer the associated questions. 34. Rainbow Trout
Rainbow Trout in Australia
Section 10.4 In Exercises 3233, find the model indicated and answer the associated question.
32. Cumulative textbook sales may often be modeled by a limited growth model. A publishers marketing analyst predicted that a new edition of a textbook would sell 12,900 copies in its first year and 20,500 copies over the life of the edition. (a) Find the limited growth model for the cumulative number of sales of the edition. (b) Calculate the cumulative number of books sold by the end of the second year. (c) Determine the rate of change in the cumulative number of textbook sales at the end of the second year.
Textbook Sales Electronic Game Sales Based on data from 1996 to 2000, the number of copies of computer and video games sold worldwide may be modeled by
k 0.340
L` 52.0 cm
t0 2 0.08 year
(a) How long is a three-year-old rainbow trout? (b) At what rate is the length of a three-year-old rainbow trout changing? 35.
Albacore Tuna off California
Removed due to copyright permissions restrictions.
33.
S( t ) 5
121.8 1 100 1 1 17.23e21.794t
Albacore Tuna
k 0.233
L` 109.0 cm
t0 2 2.31 years
million units, where t is the number of years since the end of 1996. (Source: Modeled from Interactive
Digital Software Association State of the Industry Report, 20002001.)
(a) How long is a two-year-old albacore tuna? (b) At what rate is the length of a two-year-old albacore tuna changing?
713
In Exercises 3637, use logistic regression to find the logistic model for the data. Then answer the given questions. 36.
Number of Different Banks
37.
U.S. Homicide Rate
Years Since 1990 (t) 0 1 2 3 4 5 6 7 8 9 10
Homicides per 100,000 People (H) 9.4 9.8 9.3 9.5 9 8.2 7.4 6.8 6.3 5.7 5.5
Years Since 1984 (t) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Banks (B) 17,900 18,033 17,876 17,325 16,562 15,829 15,192 14,517 13,891 13,261 12,641 12,002 11,478 10,923 10,463 10,221 9,908
Source: FBI Uniform Crime Reports, 2000.
(Hint: Align the data by subtracting 5.4 from H. After finding the model using logistic regression, add back the 5.4.) According to the model, at what rate was the number of homicides per hundred thousand people declining in 2000?
Section 10.5 In Exercises 3840, state whether the differential equation can be solved using separation of variables or the Reverse Product Rule (integrating factor). You do not need to solve the equation.
38. y 1 x 39. 40. dy 5 x2 dx
(Hint: Align the data by subtracting 9900 from B. After finding the model using logistic regression, add back the 9900.) According to the model, at what rate was the number of banks declining in 2000?
dy 6y 5 dx x( y 1 1 ) dy 1 4xy 5 3x dx
In Exercises 4145, solve the differential equation using separation of variables or the Reverse Product Rule, as appropriate. 41. 6y dy 5 dx x( y 1 1 )
714 42.
CHAPTER 10
dy 2 2y 5 8 dx dy 5 x2 dx dy 50 dx
43. y 1 x
44. x 1 y 2 45.
dy 1 4xy 5 3x dx
interest at a rate of 3.0 percent per year, compounded continuously. In addition, they arrange for an additional $5200 per year to be automatically deposited into the account. (a) Write a differential equation for the amount of money in the account after t years. (b) Find the amount of money in the account when the student is ready to attend college at age 18. 50. College Fund Suppose that the child in Exercise 49 attends college beginning at age 18 and withdraws $12,000 from the account for educational expenses. At this time, the parents discontinue the $5200 annual deposits. (a) Write a differential equation for the amount of money in the account after t years. (b) Will the student have enough money in the account to complete a 4-year degree? 51. Salt-Water Concentration A 50-gallon tank initially contains 12 pounds of salt dissolved in 25 gallons of water. Brine containing 1 pound per gallon of salt flows into the tank at the rate of 4 gallons per minute, and the well-stirred mixture flows out of the tank at the rate of 2 gallons per minute. How long does it take for the tank to fill? How much salt is in the tank when the tank is full?
In Exercises 4648, solve the differential equation with the given initial condition. 46. x2 1 2 1 47. 48. 1 dy 5 0; y( 2 1 ) 5 1 y dx
dS 3 1 ? S 5 6; S( 0 ) 5 2 dt 12 1 3t dI 1 30I 5 4 sin( 2t ) ; I( 0 ) 5 1 dt
In Exercises 4951, set up a differential equation and use it to find the answer to the question. 49. Saving for College A childs parents begin a college fund when the child is 10 years old. They initially place $750 into the account, which pays
P R O J E C T
1 0
The Minnesota Department of Natural Resources (DNR) has the following mission: Our mission is to work with citizens to conserve and manage the states natural resources, to provide outdoor recreation opportunities, and to provide for commercial uses of natural resources in a way that creates a sustainable quality of life. (Source: www.dnr.state.mn.us.) One of the controversies faced by Minnesota DNR officials each year regards their antlerless-only permits policy. The goal of this policy is to help control the deer population. Officials do not want the deer population to be overhunted, nor do they want the deer population to explode. In this Make It Real project, you will investigate this situation. The differential equation D dD 5 1.05( D 2 600 ) a 1 2 b dt 1300 may be used to predict the number of deer in the state of Minnesota. (Source: Modeled from data found at www.deercrash.com.) In this equation, D is the number of deer in thousands and t is time in years, where t 5 6 corresponds to 1996.
What to do
1. Use slope fields and Eulers Method to demonstrate the growth of the deer population in Minnesota over time if, in 1996, there were D 5 797 deer (that is, 797,000 deer). 2. Suppose the Minnesota DNR officials are considering allowing the public to harvest more deer. They need to decide how many more deer per year they should allow to be harvested. The differential equation that represents this new situation is dD D 5 1.05( D 2 600 ) a 1 2 b 2K dt 1300 The parameter K represents the rate at which deer are harvested each year. Choose different values of K and use slope fields and Eulers Method to describe the effects on the deer population of allowing the public to harvest additional deer. 3. Write a report illustrating the effects that various choices of K will have on future deer populations. Again, use the fact that, in 1996, there were D 5 797 (thousand) deer. (Continued)
715
What to do (Continued)
Address the following questions: Are there values of K for which the deer population will continue to grow? Are there values of K for which the deer population will eventually decrease to zero? Conclude your report by providing a recommendation to the Minnesota DNR concerning a range of values for K that will best control the deer population.
Useful web sites

The following web sites contain information regarding the deer population in Minnesota. These web sites may provide you with useful background information for your report. http://www.dnr.state.mn.us/fwt/back_issues/november00/article9.html http://www.deercrash.com/10year/minnesota.htm#data http://www.mnforsustain.org/wolf_mech_and_nelson_wolves_deer_harvest.htm http://www.citypages.com/letters/detail.asp?TID51561 http://www.dnr.state.mn.us/hunting/deer/index.html
716
11
Chapter
Sequences and Series

n February of 2008, the Super Bowl came to the city of Glendale, Arizona. Many large cities around the country work very hard to compete for the right to host large events such as a Super Bowl. Such events can have a huge economic impact not only on the host city, but on many of the surrounding cities. While the economic impact of the Super Bowl on a city is difficult to determine, it is common to consider both direct and indirect impact. The direct impact involves money spent by people that might not normally have been spent. Visitors, as well as local folk, spend money on restaurants, lodging, souvenirs, and other items. The indirect impact involves the re-spending of the money that was spent as part of the direct impact. Local store and restaurant owners, for example, use the influx of money created by the Super Bowl to pay employees, buy the goods needed to operate their business, and so on. This money helps to strengthen the local economy. Sequences and series can be used to understand and describe real-world scenarios such as the economic impact of the Super Bowl on a city.
11.1 Sequences
Determine whether a given sequence converges or diverges by using graphs, tables, and symbols Find explicit and recursive formulas for sequences Use sequences to analyze realworld situations Determine whether a series is arithmetic or geometric Determine whether a series converges or diverges by using graphs, tables, and symbols Use series to solve real-world problems Construct Taylor polynomials that approximate other functions Construct infinite Taylor series that approximate other functions Determine whether a Taylor series converges or diverges, and describe the interval of convergence
11.2 Series and Convergence
11.3 Taylor Polynomials
11.4 Taylor Series
717
718
CHAPTER 11
11.1 Sequences
Determine whether a given sequence converges or diverges by using graphs, tables, and symbols Find explicit and recursive formulas for sequences Use sequences to analyze real-world situations
It has been reported that as many as 1 in 5 adults has high cholesterol. People with high cholesterol are at a higher risk of heart disease, including heart attack and stroke, than those with healthier levels of cholesterol. Many people take a prescription drug as a way to lower the amount of cholesterol in their bloodstream. In most cases, this type of medicine is considered a maintenance medicineusers take the medicine every day for the rest of their lives. A user may ask, if I take the medicine every day, will the concentration of medicine in my blood continue to increase? Is this safe? Should I take a double dose if I forget to take the medicine at the right time? Is it possible for the amount of medicine in my body to reach unsafe levels? In this section, we will introduce the mathematical topic of sequences. This topic will enable us to address a number of real-life applications, including the use of cholesterol-lowering medicine.
Lipitor is a common cholesterol-lowering medicine taken by about 18 million people in the United States. On average, these people take a 20-milligram dose once per day. Often, such medicines are rated based on their half-life. A medicines half-life is the amount of time it takes for half of the medicine to be eliminated from the body. Lipitor reportedly has a half-life of 14 hours. (Source: www.medguides.medicines.org.uk.) Will the amount of Lipitor in a persons body increase continuously if the person continues to take the medicine every day? We can determine the amount of Lipitor remaining in the body by using the following formula. 1 elapsed time>half-life Percent remaining 5 a b 2 1 24>14 5 a b 2 < 0.305 Since the half-life of Lipitor is 14 hours, roughly 30.5 percent of the dose is left in the body after 24 hours. We assume that when the medicine is taken, it is instantly absorbed into the bloodstream. Under these assumptions, we can compute how much medicine is in the body on the second day, assuming the medicine is taken at the same time each day. Initially, a 20-mg dose is taken. Twentyfour hours later, 30.5 percent of the first 20-mg dose of medicine remains in the body when the second 20-mg dose is taken. 20 ? 0.305 1 20 5 26.1 After the second dose is taken, there will be 26.1 mg of medicine in the body. By the time the third dose is taken, 30.5 percent of the 26.1 mg of medicine remains in the body. 26.1 ? 0.305 1 20 < 27.96
11.1 Sequences
719
After the third dose, there is 27.96 mg of medicine in the body. If we keep track of the amount of medicine in the body each day, we can generate a sequence of numbers. A sequence is a set of numbers given in a particular order. We can think of this set of numbers as a function whose input consists of only natural numbers (1, 2, 3, . . .). In this case, our function can be expressed verbally by saying, Multiply the previous number in the sequence by 0.305 (30.5 percent of the medicine remains in the body) and add 20 (the new dose is 20 mg of medicine). The input for this sequence is the number of the day on which the dose is taken. The output is the amount of medicine remaining in the body at that time. After analyzing the amount of medicine in the body after three doses, we see that the amount of medicine is increasing by less and less each dose. The second dose adds 6.1 mg of medicine to the total amount in the body, while the third dose adds only another 1.86 mg of medicine. This result suggests that, over time, the amount of medicine in the body must reach a limit. This limit is called the maintenance level. In mathematical terms, we say that a sequence that reaches a limit is a converging sequence. A sequence converges if the terms of the sequence get closer and closer to a particular value as the number of terms increases.
CONVERGENCE OF SEQUENCES
520, 26.1, 27.96, c 6
A sequence a1, a2, a3, c , an, c converges (or reaches a maintenance level) if an S L as n S ` . That is, if lim an 5 L, we say that the sequence nS` converges. Otherwise, we say that the sequence diverges. To determine the maintenance level in this case, we first create a formula for the sequence representing the amount of medicine in the body each day. Each number in the sequence is a called a term of the sequence. Terms can be expressed using the notation an.
EXAMPLE 1
Using Sequence Notation

6 , use proper sequence notation to write Given the sequence 520, 26.1, 27.96, c the first three terms of the sequence.
SOLUTION The first term of the sequence corresponds with n 5 1 and is written a1. The second term of the sequence corresponds with n 5 2 and is written a2. In general, the ith term of the sequence corresponds with n 5 i and is written ai. Each term represents the amount of medicine in the body after n doses.
a1 5 20 mg a2 5 26.1 mg a3 5 27.96 mg We can also verbalize the computation used to compute the different terms of the sequence. We multiply the previous term by 0.305 (30.5 percent of the medicine remains in the body) and then add 20 (each dose is 20 mg). We write this in mathematical form as an 5 0.305 ? an 2 1 1 20
720
CHAPTER 11
In this formula, an represents any term of the sequence. Therefore, an 2 1 represents the term right before an. As shown in Example 2, we can better estimate the maintenance level by examining a graph of this formula.
EXAMPLE
Representing Sequences Graphically

Lipitor is taken in 20-mg doses once per day. Assuming that each day 69.5 percent of the medicine is eliminated from the body, use a graph to find the maintenance level and explain what this value represents.
an 28 27 26 25 24 23 22 21 20 0
Amount of drug in blood (in mg)
SOLUTION The formula representing the situation is an 5 0.305 ? an 2 1 1 20. We calculate the first nine terms of the sequence and graph them (Figure 11.1). The graph suggests that the maintenance level is about 28.8 mg of medicine. This means that if 20 mg of Lipitor is taken each day, the amount of medicine in the body will reach 28.8 mg and hold constant at that amount. A Technology Tip detailing the process for creating graphs of sequences is given at the end of this section.
0 1 2 3 4 5 6 7 8 9 n Dose number
FIGURE 11.1
EXAMPLE
Finding the Maintenance Level of a Sequence

An initial dose of 80 mg of Lipitor is taken. After this initial dose, a single 30-mg tablet is to be taken each day. Every 24 hours, 69.5 percent of the medicine is eliminated from the body. Find the maintenance level and explain what it means.
SOLUTION We begin by writing the first four terms of the sequence. The first term is 80 (mg). The following day, only 30.5 percent of the medicine remains in the body when a new dose of 30 mg is taken.
a1 5 80 mg a2 5 0.305 ? 80 1 30 5 54.4 mg This pattern continues for subsequent days. a3 5 0.305 ? 54.4 1 30 < 46.6 mg a4 5 0.305 ? 46.6 1 30 5 44.2 mg We have the first four terms of the sequence. 580, 54.4, 46.6, 44.2, c 6 The sequence suggests that, over time, the amount of medicine in the body will become stable. A graph of the sequence will tell us more. In general, the nth term of the sequence can be expressed as a formula. an 5 0.305 ? an 2 1 1 30
11.1 Sequences
721
Using a1 5 80, we can analyze a graph of the sequence (Figure 11.2).
FIGURE 11.2
The graph shows that the amount of medicine in the body drops very quickly after the initial 80-mg dose. The graph suggests that the maintenance level is about 43.2 mg. We also can determine that because 69.5 percent of 43.2 mg is 30 mg, we have reached a point at which the body eliminates an amount equal to the next dose taken. Therefore, once the level of medicine in the body is 43.2 mg of medicine, the amount eliminated (30 mg) and the amount of the new dose (30 mg) are the same. Therefore, the body is able to maintain a level of 43.2 mg of medicine. An alternative way to find the maintenance level is to tackle the problem symbolically. The maintenance level k will have the property that 0.305k 1 30 5 k Solving this equation for k yields 0.305k 1 30 5 k 30 5 0.695k k5 30 0.695
< 43.2 mg
EXAMPLE
Monitoring Blood Pressure Medication Levels

ATACAND (candesartan cilexetil) is a high blood pressure medicine. According to the technical pamphlet for the drug, the half-life of candesartan is about 9 hours. The typical recommended starting dose is 16 mg once daily. A doctor may recommend that the dose be taken twice daily; however, the total daily dosage should not exceed 32 mg. (Source: www.astrazeneca-us.com.)
722
CHAPTER 11
Suppose that a doctor prescribes that a patient take a 16-mg dose once every 12 hours. (a) What is the maintenance level for the drug? (b) Suppose that after reaching the maintenance level, the patient accidentally misses a dose and takes a double dose the next time he is scheduled to take the medicine. (We do not recommend doing this.) Assuming the drug is instantly absorbed and distributed evenly throughout the bloodstream, how much of the drug will be in the bloodstream immediately after the double dose is taken?
SOLUTION
(a) We first determine how much of the drug remains in the body after 12 hours. 1 elapsed time>half-life Percent remaining 5 a b 2 1 12>9 5 a b 2 < 0.397 We calculate the first eight terms of the sequence representing the amount of medicine in the body. Since 39.7 percent remains after 12 hours,
Amount of drug in blood (in mg) an 26 24 22 20 18 16 0 0 1 2 3 4 5 6 7 8 9 n Dose number
a1 5 16 mg a2 5 0.397 ? 16 1 16 < 22.4 mg a3 5 0.397 ? 22.4 1 16 < 24.9 mg a4 5 0.397 ? 24.9 1 16 < 25.9 mg a5 5 0.397 ? 25.9 1 16 < 26.3 mg a6 5 0.397 ? 26.3 1 16 < 26.4 mg a7 5 0.397 ? 26.4 1 16 < 26.5 mg a8 5 0.397 ? 26.5 1 16 < 26.5 mg It appears as though the maintenance level is approaching 26.5 mg of medicine (Figure 11.3). The graph confirms our conjecture. After quickly increasing, the amount of medicine in the body stabilizes at 26.5 mg of medicine, a level that is in the required, but safe, zone. Taking the symbolic approach, we can say that the maintenance level k will have the property that 0.397k 1 16 5 k Solving this equation for k yields 0.397k 1 16 5 k 16 5 0.603k k5 16 0.603
FIGURE 11.3
< 26.5 mg
11.1 Sequences an 45 40 35 30 25 20 15 10 5 0
723
0 2 4 6 8 10 12 14 16 18 n Scheduled dose number
(b) At a particular time after reaching the maintenance level, 0.397 ? 26.5 1 16 5 26.5 mg of medicine will remain in the body. If a new dose is not taken at the scheduled time, then by the next scheduled time 10.5 mg ( 0.397 ? 26.5 5 10.5 ) of medicine will remain in the body. If the patient takes two doses, she will have 10.5 mg 1 16 mg 1 16 mg 5 42.5 mg of medicine in the body. This amount is substantially higher than the maintenance level and may result in adverse side effects. Additionally, the maximum recommended daily dose is stated to be 32 mg, so the patient is at risk of overdosing if she takes the next 16-mg dose 12 hours later. However, if the patient does not have an adverse reaction, the amount of medicine in the body will return to the maintenance level over time (Figure 11.4).
FIGURE 11.4
Amount of drug in blood (in mg)
Not all sequences reach a maintenance level or a stable limit. Example 5 illustrates a sequence that diverges.
EXAMPLE 5
A Sequence Without a Stable Limit

An employee asks his employer to withhold $1200 for retirement savings during the first year of employment, and to increase the amount withheld by 2 percent each subsequent year. If this pattern continues for 20 years, how much money is withheld in the 20th year?
SOLUTION We calculate the first four terms of the sequence.
a1 5 $1200 a2 5 $1200 1 0.02 ? $1200 5 $1224 a3 5 $1224 1 0.02 ? $1224 5 $1248.48 a4 5 $1248.48 1 0.02 ? $1248.48 5 $1273.44 The sequence representing the amount of money withheld in each of the first four years is 51200, 1224, 1248.48, 1273.44, c 6 So far, the amount of money withheld does not appear to be stabilizing. Using the Technology Tip at the end of this section to graph the sequence an 5 0.02 ? an 2 1 1 an 2 1 5 1.02 ? an 2 1, we determine that the amount withheld in the 20th year will be $1748.17 (Figure 11.5).
an Amount witheld (in dollars) 1700 1600 1500 1400 1300 1200 0 0 2 4 6 8 10 12 14 16 18 n Year of employment
FIGURE 11.5
724
CHAPTER 11
As can be seen from the graph, this sequence does not have a stable limit. Furthermore, if we think about lim ( 1.02 ? an 2 1 ) , we can confirm that 1.02 ? an 2 1 gets larger and larger as n S ` .
nS`
Each of the situations we have seen so far has been a situation in which the formula for the sequence was expressed using a recursive formula. A recursive formula for a sequence is one in which we need to compute all the terms prior to the term we desire. We see this in the notation an 2 1 contained in these formulas. That is, to compute the 20th term of a recursive sequence, we need to first find a20 2 1 5 a19, the 19th term of the sequence. Sometimes it is possible to express the formula for the nth term of a sequence explicitly in terms of the value of n. An explicit formula for a sequence is powerful because it allows us to compute directly any term of the sequence without computing the values of the terms before it.
Just In Time
Exponential Functions
While the situations we will study in this section are discrete in nature, we can apply the same thinking to developing explicit formulas that we applied when studying exponential functions. Exponential functions are characterized as having a common ratio between function values. For example, consider the following table. x 0 2 4 6 8 f (x) 5 5.75 6.6125 7.6044 8.745 5.75 5 1.15 5 6.6125 5 1.15 5.75 7.6044 5 1.15 6.6125 8.745 5 1.15 7.6044 Ratio
With a common ratio of 1.15 for each x-increment of 2, we write an exponential model for this table. f ( x ) 5 5( 1.15 ) x>2
EXAMPLE
Finding an Explicit Formula for a Sequence

In Example 5, we found a recursive formula for the situation in which an employer withheld $1200 for retirement savings during the first year of employment and increased the amount withheld by 2 percent each year. The recursive formula was an 5 0.02 ? an 2 1 1 an 2 1 5 1.02 ? an 2 1
11.1 Sequences
725
Write an explicit formula for the amount of money withheld in the nth year of employment.
SOLUTION We write the first four terms of the sequence, as seen in Table 11.1.
TABLE 11.1
Value of term Number of term
1200 n51
1224 n52
1248.48 n53
1273.44 n54
Next, we notice that we start with the term 1200, multiply it by 1.02, then multiply that answer by 1.02, and finally multiply that answer by 1.02 (Table 11.2).
TABLE 11.2
31.02 31.02 31.02
1200 n51
1224 n52
1248.48 n53
1273.44 n54
We can express the idea of starting with a value (1200) and repeatedly multiplying by another value (1.02) using exponential notation. an 5 1200 ? 1.02n 2 1 We can see that this explicit formula will work if we substitute in values of n. a1 5 1200 ? 1.021 2 1 5 1200 ? 1.020 5 1200 a2 5 1200 ? 1.022 2 1 5 1200 ? 1.021 5 1224 a3 5 1200 ? 1.023 2 1 5 1200 ? 1.022 5 1248.48 a4 5 1200 ? 1.024 2 1 5 1200 ? 1.023 5 1273.44
The explicit formula found in Example 6 also reveals another way to describe the sequence. Each term in the sequence is found by multiplying the previous term in the sequence by the same value (1.02 in this case). Such a sequence is known as a geometric sequence.
DEFINITION: GEOMETRIC SEQUENCE
A geometric sequence is a sequence in which each subsequent term of the sequence is found by multiplying the previous term by a common ratio.
The ratio is called a common ratio because, if we take any term in the sequence and divide it by the previous term (i.e., create a ratio), we get a constant (1.02 in this case). Another type of sequence is formed when we add a common difference. Such a sequence is known as an arithmetic sequence.
726
CHAPTER 11
DEFINITION: ARITHMETIC SEQUENCE
An arithmetic sequence is a sequence in which each subsequent term of the sequence is found by adding a common difference to the previous term.
Formulas may easily be determined for arithmetic and geometric sequences.
EXAMPLE
Writing Recursive and Explicit Formulas for Sequences

A $400 car payment is made each month. Write a recursive formula and an explicit formula for the total amount of money paid out in car payments over time. After making 60 payments (over 5 years), how much money has been paid out?
SOLUTION We begin by writing the first four terms of the sequence, representing a running total of how much money is spent on car payments.
5400, 800, 1200, 1600, c 6 To obtain a particular term of the sequence, we add $400 to the previous term in the sequence. We can express this pattern using a recursive formula for the sequence. an 5 an 2 1 1 400 To write an explicit formula, we organize the sequence as shown in Table 11.3.
TABLE 11.3
400 n51
800 n52
1200 n53
1600 n54
We see that if we multiply the number of the term (n 5 1, n 5 2, and so on) by 400, we obtain the value of the term. Therefore, the explicit formula for this arithmetic sequence is an 5 400n We can use the explicit formula to determine how much money has been paid out in car payments after 60 payments have been made. That is, we can compute a60 5 400 ? 60 5 24,000 This result means that $24,000 has been paid in monthly car payments after 60 payments have been made.
11.1 Sequences
727
Using Technology to Graph a Sequence

The TI83 Plus calculator works nicely in creating a graph of a sequence.
Graphing a Sequence 1. Press MODE . Change the mode to Sequence mode by using the arrows to highlight SEQ and pressing ENTER .
2. Edit the parameters using the Y Editor. The parameter nMin refers to the minimum term number you choose to display. The parameter u(n) represents the formula. The u is accessed by pressing 2nd and the number 7 (note the blue u above the 7 button). The parameter n is accessed by pressing the X,T,U,n key. The parameter u(nMin) represents the first term of the sequence. 3. Edit the WINDOW parameters. The parameters nMin and nMax refer to the number terms in the sequence to be evaluated. PlotStart and PlotStep determine which term to plot first and the incremental value of n. The rest of the parameters are as usual, where x represents the term number and y represents the value in the sequence.
4. Press
GRAPH
(Continued)
728
CHAPTER 11
5. Use TRACE as usual. Note that the term number, n, as well as the sequence value, is displayed.
Graphing a Sequence 1. In column A, write a list of whole numbers from 1 through the number of terms you wish to compute. In cell B2, write the initial term of the sequence. In this case, u( 1 ) 5 20.
2. In cell B3, type 0.305B2+20. This is equivalent to stating that u( n 1 1 ) 5 0.305u( n ) 1 20.
(Continued)
11.1 Sequences
729
3. Copy cell B3 to cells B4 through B11 to display the terms of the sequence.
4. Draw a scatter plot of the data from columns A and B.
35 30 25 20 15 10 5 0 0 1 2 3 4 5 6 7 8 9 10 11
11.1 Summary
In this section, you learned how to create a sequence of numbers given a particular situation. You learned how to express this sequence using a recursive and an explicit formula. You saw that graphing the sequence using the graphing calculator was helpful in answering questions related to sequences.

In Exercises 15, do the following: (a) Determine if an exponential function may be used to represent the table of data. (b) If so, write the exponential function. 1. x 0 1 2 3 4 5 2 2.1 2.205 2.3153 2.431 2.5526 f (x) 2. x 0 5 10 15 20 25 f (x) 10 11 12.1 13.31 14.641 16.105 3. x 0 10 20 30 40 50 f (x) 5 25 45 65 85 105 4. x 0 1 2 3 4 5 f (x) 5 6 9 14 21 30
730
CHAPTER 11
5.
x 0 10 20 30 40 50
f (x) 5 105 405 905 1605 2505
In Exercises 610, write an exponential function to match the given verbal description. 6. The function has an initial value of 50 and increases by a common ratio of 1.4 every 2 days. 7. The function has an initial value of 100 and decreases by 10 percent every year. 8. The function has an initial value of 1000 and increases by 12 percent every 10 years. 9. The function has an initial value of 500 and increases by a common ratio of 1.025 every day. 10. The function has an initial value of 200 and decreases by 50 percent every 10 days.
11.1 Exercises
In Exercises 15, do the following: (a) List the first six terms of the given sequence. (b) Use a graphing calculator to graph the sequence. (c) State whether the sequence appears to converge or diverge. 1. an 5 2. an 5 n 3n 1 1 n 2n 2 1
2
10. 1, 1, 2, 3, 5, 8, 13, 21, . . . 11. 0, 5, 0, 5, 0, . . . 12. 1 1 1 1 ,2 , ,2 ,c 2 4 8 16
13. 0, 50, 87.5, 115.625, . . . 14. 100, 102, 104, 106, . . . 15. 300, 600, 900, 1200, . . . 16.
PharmaceuticalsAspirin Aspirin
1 3. an 5 n 2 4. an 5 4 1 ( 2 1 ) n 1 n 5. an 5 a 1 1 b n 6. Explain what it means for a sequence to converge. Use graphs, numbers, symbols, and words in your explanation. In Exercises 715, find a formula for the general term of the sequence, an, and state whether the sequence converges. Assume that the given pattern continues. 7. 3, 8, 13, 18, 23, . . . 8. 10, 5, 2.5, 1.25, . . . 9. 1 2 3 4 , , , ,c 2 5 10 17
washes out of the bloodstream quite rapidly. The amount of effective aspirin left in the bloodstream is reduced by 50 percent approximately every 3 hours. (Source: www.drugs.com.) Suppose that a patient takes 650 mg of aspirin every 6 hours, as recommended. (Assume that the aspirin is absorbed into the bloodstream immediately.) (a) Generate the first five terms of the sequence that represents this situation. Explain why this situation can be represented using a sequence. (b) Define a sequence giving the aspirin level in the patients bloodstream after each dose is taken and absorbed. (c) Create a graph that shows units of medicine as a function of number of doses. (You may use your graphing calculator.) (d) How many doses can the patient safely take? Support your answer graphically, numerically, symbolically, and verbally.
11.1 Sequences
731
(e) What if the patient misses a dose? Discuss the consequences of missing a dose. Does taking a double dose make up for the missed dose? Explain. 17.
Purchasing a Car As a rule of thumb, a car depreciates in value by 15 percent every year. (Source: www.bankrate.com.) The MSRP (manufacturers suggested retail price) for a 2007 Toyota Camry Hybrid 4-Door Sedan was $25,900 on June 19, 2006. (Source: www.nada.com.) If the car is purchased and kept indefinitely, does the value of the car converge or diverge? If it converges, to what value does it converge? PharmaceuticalsLamisil Lamisil tablets are taken by people with fingernail and toenail fungus problems. Typically, 250-mg tablets are taken once daily for 6 weeks for people with fingernail fungus and for 12 weeks for people with toenail fungus. With a half-life of 36 hours, 37 percent of the medicine is removed from the bloodstream each day. (Source: www.drugs.com.) How much medicine will be in the bloodstream at the end of the treatment period for each of the two treatment groups? How close to the maintenance level does each group get?
safely continue to clean up the spill before the suit must be changed? (d) Re-analysis of the toxicity of the chemicals indicates that the suit can be used safely with up to 40 micrograms of accumulated chemical. Does this make a significant difference in the amount of time the engineer can continue to work on the chemical spill? Explain. 20.
PharmaceuticalsSingulair Singulair is a medicine taken for the long-term management of asthma. Adults aged 15 and over are typically prescribed a single dose of 10 mg of Singulair each day. Children aged 2 to 5 years of age are prescribed a single dose of 4 mg of Singulair each day. With a half-life of 4 hours, about 98.4 percent of the medicine is removed from the bloodstream every day. (Source: www.drugs.com.) Use what you have learned in this section to analyze the effects of the two different dosage levels. Include graphs and formulas and indicate the maintenance level in each case. Forestry Management In its 5-year Urban Forestry Management Plan, the city of Muskego, Wisconsin, planned to increase the number of trees in city parks. According to the plan, there were 500 trees in city parks, and the city planned to increase the number of trees by 50 percent over 5 years. The plan also indicated that the city expected 10 percent of the newly planted trees to die during the 5-year period. Additionally, the city planned to remove 85 trees over the same time period. (Source: City of Muskego Urban
18.
21.
19. Semiconductor Manufacturing In the many different processes for manufacturing semiconductors, highly toxic chemicals are used. When the need arises, environmental engineers wear protective suits designed to neutralize specific toxic chemicals. However, the ability of a suit to neutralize a chemical is usually reduced with prolonged exposure. Suppose an engineer has put on a new protective suit and begins to clean up a chemical spill in the factory. During each hour of cleanup, the suit neutralizes 30 percent of the toxic chemical on it and accumulates an additional 12 micrograms of the chemical. (a) Make a table or spreadsheet showing the amount of unneutralized chemical on the suit during each hour of the cleanup. Describe the growth of these values. (b) Write the first five terms of the sequence related to this situation. Then write a formula and create a graph for the sequence. (c) The suit is safe to use until it has accumulated 35 micrograms or more of the toxic chemical. How long can the engineer
Forestry Management Plan at www.ci.muskego.wi.us.)
Assuming the city will implement the same 5-year plan indefinitely, do the following. (a) Write a sequence of numbers that represents the number of trees in city parks for each of the first four 5-year periods. (b) Based on this sequence, write a rule for the number of trees in city parks after n years. (c) Represent this situation graphically. (d) Does repeated application of the 5-year plan make sense in the long run? Justify your answer. (e) Will the number of trees stabilize? If so, in how many years and with how many trees? Justify your answer.
732 22.
CHAPTER 11
Prilosec is a drug used to eliminate acid reflux disease. Typically, patients take a 20-mg dose once daily. Studies have shown that in healthy people, Prilosec has a half-life of about one-half hour. This is equivalent to having 99.9 percent of the medicine eliminated from the bloodstream every day. The study showed that for people with unhealthy livers, the half-life of Prilosec is 3 hours. This is equivalent to having 99.51 percent of the medicine eliminated every day. (Source: www.drugs.com.) Use what you have learned in this section to analyze the effects of the two different half-lives. Include graphs and formulas and indicate the maintenance level in each case.
PharmaceuticalsPrilosec
25. Create two sequences, both of which diverge. Explain why your sequences diverge and include graphs to support your explanation. 26. Create two sequences, both of which converge. Explain why your sequences converge and include graphs to support your explanation. 27. Consider the general sequence 1, k, k2, k3, k4, c . Determine for which values of k the sequence will converge and for which values of k the sequence will diverge. 28. A sequence is considered strictly monotone if it is either always increasing or always decreasing from term to term. Determine whether or not each of the following sequences is strictly monotone. Make a graphical as well as a numerical argument. Then, determine if the sequence converges or diverges. 1 (a) an 5 n 2 1 (b) an 5 2 n (21)n (c) an 5 n2 29. Using the idea of rate of change, describe the shape of the graph of a sequence that is ever increasing (strictly monotone) but converges. Then, describe the shape of the graph of a sequence that is ever decreasing but converges.
23.
Claritin tablets are taken by people wishing to relieve nasal congestion, runny nose, and watering eyes resulting from allergies. The halflife of Claritin is dependent upon urinary pH. The half-life is approximately 4 hours when the urinary pH is 5, and 12 hours when the urinary pH is 8. Typically, people take a 240-mg dose of Claritin once daily. (Source: www.rxlist.com.) With a half-life of 4 hours, the body will eliminate approximately 98.4 percent of the medicine every day. With a half-life of 12 hours, the body will eliminate approximately 75 percent of the medicine every day. Use what you have learned in this section to analyze the effects of the two different half-lives. Include graphs and formulas and indicate what the maintenance level is in each case.
PharmaceuticalsClaritin
24.
Percocet is a drug taken by people in need of pain relief. It contains oxycodone and acetaminophen, and can be addictive. Furthermore, research has shown that it would be extremely unhealthy if a person were to maintain a level of 15 mg of Percocet in the bloodstream. Dosing depends on the severity of the pain that a person is experiencing. Percocet has a half-life of about 2 hours. (Source: en.wikipedia.org.) Suppose a doctor wants to determine the maximum amount of Percocet she should prescribe to a patient who will take the medicine every 4 hours. What would be the maximum dosage in this situation? Use graphs, formulas, and tables to justify your work.
PharmaceuticalsPercocet
30. Using a calculator, a student starts with a number greater than 1 and repeatedly presses the square root button. If each successive result represents a term of a sequence of numbers, write a formula for the nth term of this sequence. Does the sequence converge? If so, to what value does it converge? 31. Using a calculator, a student starts with a positive number less than 1 and repeatedly presses the square root button. If each successive result represents a term of a sequence of numbers, write a formula for the nth term of this sequence. Does the sequence converge? If so, to what value does it converge?
733
32. The Fibonacci sequence, 1, 1, 2, 3, 5, 8, 13, c , can be defined recursively by the formula an 5 an 2 1 1 an 2 2. In this exercise, you will
12 investigate the ratio an and its relationship to n11 n the ratio an 1 for n $ 1. Describe the relationship 1
for all n $ 1. Determine whether the sequence an 5 n2

n 11
is monotonic. Justify your answer.
35. The following sequence is known as the harmonic sequence. 1 1 1 1 an 5 1, , , , , c 2 3 4 5 Determine whether the ratio of consecutive terms in the harmonic sequence,
an 1 1 an ,
that you find. 33. Investigate the sequence of numbers defined by the formula an 5
4 sin n . n
Does the sequence
converges or
converge? If so, to what value does it converge? 34. A sequence is considered a monotonic sequence if the terms of the sequence always increase (that is, an 1 1 . an) or always decrease (that is, an 1 1 , an)
diverges. Justify your answer.
Determine whether a series is arithmetic or geometric Determine whether a series converges or diverges by using graphs, tables, and symbols Use series to solve realworld problems
In the previous section, we studied and applied sequences to determine the maintenance level of a particular medicine. In this section, we will revisit this idea as we investigate mathematical series.
Azithromycin is a common antibiotic used to fight diseases such as strep throat. A common prescription requires that the patient take two 250-mg tablets initially, followed by a single 250-mg tablet each day for 4 consecutive days. The half-life of azithromycin is reportedly 68 hours. (Source: www.drugs.com.) This means that 50 percent of the medicine is eliminated from the body every 68 hours. To determine what percentage of medicine remains in the body after a 24-hour period, we calculate 1 24>68 a b < 0.78 2 After 24 hours, 78 percent of the dosage remains in the body. (This is equivalent to eliminating 22 percent.) We will investigate this situation to determine how much medicine will be in the body when the final dose is taken. We could study this situation using a recursive sequence formula as before; however, this time we will investigate the situation as an accumulation of medicine. We again assume that the medicine is absorbed into the bloodstream immediately after taking a dose. Initially, 500 mg of medicine is taken. We represent this as M1 5 500 mg
734
CHAPTER 11
On day 2, a second dose of 250 mg is taken. At that time, the amount of medicine in the body is about M2 5 250 1 500( 0.78 ) 5 640 mg This amount includes the 250-mg second dose and the 78 percent of the 500-mg first dose that remains. On day 3, a third dose of 250 mg is taken. At that time, the amount of medicine in the body is M3 5 250 1 640( 0.78 ) < 749 mg This amount includes the 250-mg third dose and 78 percent of the 640 mg that remains from the first two doses. In order to better see a pattern develop, we look at the nonsimplified version of the computations, shown for M3. M3 5 250 1 0.78 ? M2 5 250 1 0.78 ? ( 250 1 500 ? 0.78 ) 5 250 1 250 ? 0.78 1 500 ? 0.782 < 749 mg On day 4, a fourth dose of 250 mg is taken. At that time, the amount of medicine in the body is M4 5 250 1 0.78 ? M3 5 250 1 0.78 ? ( 250 1 250 ? 0.78 1 500 ? 0.782 ) 5 250 1 250 ? 0.78 1 250 ? 0.782 1 500 ? 0.783 < 834 mg We see a pattern developing. Except for the final term of each sum, we see a sum of the form a 1 ar 1 ar 2 1 ar 3 1 c 1 ar n 2 1 Expressions of this form are known as finite geometric series. A finite geometric series is a sum in which each term is found by multiplying the previous term by a constant. In other words, a geometric series is the sum of the terms of a geometric sequence.
FINITE GEOMETRIC SERIES
A finite geometric series with n terms can be expressed as a 1 ar 1 ar 2 1 ar 3 1 c 1 ar n 2 1 where a is the first term of the series and r is the common ratio or constant multiple.
735
Just In Time
Exponential Versus Linear Functions
Recall that exponential functions are characterized by a common ratio between function values for a consistent change in the input variable. Linear functions are characterized by a common difference between function values for a consistent change in the input variable. Consider the following table. x 0 1 2 3 4 f (x) 5 15 45 135 405 g (x) 5 8 11 14 17
The function f ( x ) can be described as having a common ratio of 3 for each increase of 1 of the input variable. The function g ( x ) can be described as having a common difference of 3 for each increase of 1 in the input variable. We can express the functions as follows. f ( x ) 5 5 ? 3x g ( x ) 5 5 1 3x In our medication example, each term (except for the last) is multiplied by 0.78, since 78 percent of the medicine remains in the body upon taking the next dose. We may express this sum as an explicit formula once we recognize the pattern shown in Table 11.4.
TABLE 11.4
250 n51
250(0.78) n52
250( 0.78 ) 2 n53
250( 0.78 ) 3 n54
500( 0.78 ) 4 n55
Each term in the series is expressed as an. a1 5 250 a2 5 250( 0.78 ) a3 5 250( 0.78 ) 2 a4 5 250( 0.78 ) 3 a5 5 500( 0.78 ) 4 We use sigma notation to express the first four terms of this sum.
n21 a 250 ? ( 0.78 ) 4
n51
The Greek letter sigma ( S ) means to create a sum. In this case, the terms of the sum are formed by letting n 5 1, n 5 2, n 5 3, and n 5 4. Of course, in our case, we also have to add the final term. a5 5 500 ? 0.784
736
CHAPTER 11
How much medicine (azithromycin) is in the body after taking the fifth dose?
n21 1 a5 5 250 1 250 ? 0.78 1 250 ? 0.782 1 250 ? 0.783 1 500 ? 0.784 a 250 ? ( 0.78 ) 4
n51
5 900.8 mg
EXAMPLE
Calculating the Future Value of an Investment

A person deposits $2000 every year into an account that earns 8 percent, compounded annually. Using sigma notation, express the amount of money that will be in the account after 8 years. Then determine the amount of money that will be in the account after 8 years.
SOLUTION Let An represent the amount of money in the account immediately after the $2000 is deposited into the account for the nth time.
A1 5 $2000 A2 5 $2000 1 $2000 ? 1.08 5 $4160 Notice that A2, the amount of money in the account after the second years deposit is made, is found by adding the new $2000 deposit to the original $2000 plus the interest earned. We can continue this pattern. After the third years deposit, we will have $2000 plus the previous years deposits and the interest theyve earned. A3 5 $2000 1 $4160 ? 1.08 5 $2000 1 ($2000 1 $2000 ? 1.08 ) ? 1.08 5 $2000 1 $2000 ? 1.08 1 $2000 ? 1.082 5 $6492.80 We will investigate what happens after the deposit is made for year 4, and then look for a pattern that will help us write the sum in sigma notation. A4 5 $2000 1 $6492.80 ? 1.08 5 $2000 1 ($2000 1 $2000 ? 1.08 1 $2000 ? 1.082 ) ? 1.08 5 $2000 1 $2000 ? 1.08 1 $2000 ? 1.082 1 $2000 ? 1.083 5 $9012.22 We see that we have a geometric series that begins with $2000. Each subsequent term is multiplied by the common multiple 1.08. We write the sum using sigma notation.
n21 a 2000 ? 1.08 8
n51
5 $2000 1 $2000 ? 1.08 1 $2000 ? 1.082 1 $2000 ? 1.083 1 c 1 $2000 ? 1.087 5 $21,273.30 After the eighth deposit, the account will be worth $21,273.30.
737
Finding a Formula for the Sum of a Finite Geometric Series

In the previous examples, we needed to add a finite number of terms of a particular geometric series. In the first case, we needed to add a total of five terms. In the second case, we added a total of eight terms. What if we needed to add 100 terms? Our current method would be tedious and inefficient. There must be a better way. To develop a more efficient method, we will begin with a general finite geometric series and use Sn to represent the sum of the first n terms. Sn 5 a 1 ar 1 ar 2 1 c 1 ar n 2 1 Next, we will multiply both sides of this equation by r. r ? Sn 5 ar 1 ar 2 1 ar 3 1 c 1 ar n Pay particular attention to the final term of the series and recall the rule of multiplication for exponents. ar n 2 1 ? r 5 ar n 2 1 ? r 1 5 ar n 2 1 1 1 5 ar n Next, subtract r ? Sn from Sn. Sn 2 r ? Sn 5 ( a 1 ar 1 ar 2 1 c 1 ar n 2 1 ) 2 ( ar 1 ar 2 1 ar 3 1 c 1 ar n 2 1 1 ar n ) 5 ( a 1 ar 1 ar 2 1 c 1 ar n 2 1 ) 2 ar 2 ar 2 2 ar 3 2 c2 ar n 2 1 2 ar n 5 a 1 ar 1 ar 2 1 c 1 ar n 2 1 2 ar 2 ar 2 2 ar 3 2 c2 ar n 2 1 2 ar n 5 a 2 ar n Notice how all of the terms except for the first and last cancel each other out. Since our goal is to find a simplified way to express Sn, we will solve this equation for Sn. Sn 2 r ? Sn 5 a 2 ar n Sn( 1 2 r ) 5 a 2 ar n Sn 5 a 2 ar n 12r
SUM OF A FINITE GEOMETRIC SERIES
The sum of the first n terms of a finite geometric series with first term a and common ratio r is Sn 5 a 1 ar 1 ar 2 1 c 1 ar n 2 1 Sn 5 provided that r 2 1. a 2 ar n 12r
738
CHAPTER 11
EXAMPLE
Finding the Sum of a Finite Geometric Series

Use the formula for the sum of a finite geometric series to confirm that the amount of azithromycin in the body after the fifth dose will be about 901 mg. Recall that 22 percent of the medicine is eliminated from the body each day, and that the prescription calls for an initial dose of 500 mg followed by daily doses of 250 mg.
SOLUTION We first need to find the sum of the first four terms of the finite geometric series, where a 5 250 mg and r 5 0.78 (since 22 percent is eliminated, 78 percent remains).
S4 5
250 2 250 ? 0.784 1 2 0.78
5 715.7 Next, we determine how much of the original dose remains in the body. 500 ? 0.784 5 185.1 We add these results to get the final amount of medicine in the body after the fifth and final dose is taken. 715.7 1 185.1 5 900.8 mg
EXAMPLE
Finding the Sum of a Finite Geometric Series

How much azithromycin will be in the body if the medicine is taken for 10 days? Recall that the prescription calls for an initial dose of 500 mg followed by daily doses of 250 mg, and that 22 percent of the medicine is eliminated from the body each day.
SOLUTION We need to find the sum of the first nine terms of the geometric series, where a 5 250 and r 5 0.78.
S9 5
250 2 250 ? 0.789 1 2 0.78
5 1014.9 mg We also need to consider the amount of the initial 500-mg dose that remains in the body. 500 ? 0.789 5 53.4 mg We add these results to get the final amount of medicine in the body after the tenth and final dose is taken. 1014.9 1 53.4 5 1068.3 mg
EXAMPLE
Using the Sum of a Finite Geometric Series to Find a Future Value

A person deposits $2000 every year into an account that earns 8 percent, compounded annually. Confirm, using the formula for the sum of a finite geometric series, that the amount of money in the account after the eighth deposit is $21,273.30.
739
SOLUTION As before, we have a finite geometric series with a 5 $2000 and r 5 1.08. Recall that the factor of 1.08 is equivalent to 108 percent growth. That is, we have all the money we started with (100%) plus 8 percent interest earned each year.
S8 5
2000 2 2000 ? 1.088 1 2 1.08
5 $21,273.30
EXAMPLE
Using the Sum of a Finite Geometric Series to Find a Future Value

A person deposits $1500 every year into an account that earns 12 percent, compounded annually. How much money will be in this account after 20 years? After 40 years?
SOLUTION We have a finite geometric series with a 5 $1500 and r 5 1.12.
S20 5
1500 2 1500 ? 1.1220 1 2 1.12
5 $108,078.66 After 20 years, the account will be worth over $108,000. We can use the same formula to find the amount of money in the account after 40 years. S40 5 1500 2 1500 ? 1.1240 1 2 1.12
5 $1,150,637.13 After 40 years, the account will contain over 1 million dollars.
Convergence and Infinite Geometric Series

When we look back at the examples involving medicine and saving money, we see that their end behavior differs. While the amount of medicine in the bloodstream stabilized over time, the amount of money in the savings account continued to grow at an increasing rate. If we were to let the number of terms summed in each series go to infinity, we would see very different results. A series whose sum stabilizes (reaches a maintenance level) is said to converge. A series that doesnt converge is said to diverge. We will now consider the sum of an infinite geometric series for different cases of r. Under what circumstances might the infinite sum converge, and under what circumstances might it diverge?
Case 1: r 5 1 Recall that we are working with a geometric series of the form Sn 5 a 1 ar 1 ar 2 1 c 1 ar n 2 1. If r 5 1, then Sn 5 a 1 a 1 a 1 c 1 a
5n?a If n S ` , then Sn S ` and the series diverges.
740
CHAPTER 11
Case 2: r > 1 If r . 1, then subsequent terms in the sequence will get larger and larger (since we are multiplying a by values larger than 1). As n S ` , the sum gets larger and larger and thus diverges. This is what happened in the case of the money account. Because r 5 1.08 . 1, the infinite series diverged. Of course, any investor would hope that an investment would continue to grow and not stabilize! Case 3: r < 21 If r , 2 1, then we multiply a by positive values when the power of r is even and by negative values when the power of r is odd. Since Z r Z . 1, Z r n Z . Z r n 2 1 Z for all values of n. Therefore, the sum will oscillate and become larger and larger in magnitude. Again, the infinite sum diverges. Case 4: 21 < r < 1 The only remaining case is a series with values of r that are between 2 1 and 1, as in the case of the azithromycin, where r 5 0.78. If we analyze the formula for a finite geometric sum, we can understand that this situation will result in a converging infinite series. That is, the series will stabilize.
Sn 5 5 a 2 ar n 12r a( 1 2 r n ) 12r a( 1 2 0 ) 12r a 12r
a . 2r
As n S ` , r n S 0, since 2 1 , r , 1. Therefore,
nS`
lim Sn 5 5
We will refer to the sum of an infinite geometric series as S 5 1
SUM OF AN INFINITE GEOMETRIC SERIES
The sum of the first n terms of an infinite geometric series Sn 5 a 1 ar 1 ar 2 1 c 1 ar n 2 1 converges to S5 a 12r
as n S ` , provided 2 1 , r , 1. Otherwise, the series diverges.
We can verify the formula for the sum of an infinite geometric series if we consider the sum for very large values of n.
741
EXAMPLE
Using an Infinite Sum to Model Drug Concentration

A patient is given a prescription that calls for taking a single 500-mg dose of medicine every day. Each day, 25 percent of the medicine is eliminated from the body. How much medicine is in the body after 10 days? 50 days? 100 days? What if the patient were to take the medicine indefinitely?
SOLUTION We recognize this to be a situation involving a finite geometric series with a 5 500 and r 5 0.75 (because 25 percent of the medicine is eliminated, 75 percent remains). We will use the formula for the sum of a finite geometric series to find the amount of medicine in the body after 10, 50, and 100 days.
S10 5
500 2 500 ? 0.7510 1 2 0.75
5 1887.37 mg After the 10th dose, there will be 1887.37 mg of medicine in the body. We now compute the amount of medicine in the body after 50 days. S50 5 500 2 500 ? 0.7550 1 2 0.75
5 1999.99 mg After the 50th dose, there will be nearly 2000 mg of medicine in the body. It appears as though we are reaching the maintenance level. Lets verify by finding the amount of medicine in the body after the 100th dose. S100 5 500 2 500 ? 0.75100 1 2 0.75
5 2000 mg The amount of medicine is being maintained at 2000 mg. If the medicine were to be taken indefinitely, then we could say that n S ` . We can find the sum of the infinite geometric series with a 5 500 mg and r 5 0.75. S5 500 1 2 0.75
5 2000 mg As predicted, the amount of medicine in the body will stabilize at 2000 mg.
EXAMPLE
Using Infinite Series to Estimate Economic Impact

Attracting large events such as the Super Bowl to a city can boost the local economy. The boost comes in the form of money spent by visitors who pay for lodging, food, transportation, souvenirs, and so on. It is difficult to determine precisely the impact on the economy in terms of dollars, but we estimate that about 70 percent of every dollar introduced into the local economy is spent again locally. Then, 70 percent of that money is spent again locally. The process continues, with 70 percent of the money spent in the community being spent again locally. It is estimated that during the 1998 Super Bowl in San Diego, visitors contributed well over $125 million in direct spending to the region while they were visiting (Source: www.nfl.com). What was the total impact, in dollars, on the local San Diego economy, taking into account the money that was re-spent locally?
742
CHAPTER 11
SOLUTION We begin with the $125 million that was spent by visitors in San Diego. Of that amount, 70 percent was spent again locally. This pattern continues indefinitely. We can express the economic impact in terms of an infinite geometric series S.
S 5 125 1 125 ? 0.70 1 125 ? 0.702 1 125 ? 0.703 1 c 5 a 125 ? 0.70 n 2 1

n51 `
We have an infinite geometric series with a 5 125 and r 5 0.70. S5 125 1 2 0.70
5 416.67 We estimate that the total economic impact of the Super Bowl on the San Diego area was $416,670,000.
Finding a Rational Number Given a Repeating Decimal

Many people might know that the repeating decimal 0.3333 . . . represents the fraction . But what about a repeating decimal such as 0.82222 . . . ? We can use 3 the ideas from this section to find a rational number, or fraction, that is equivalent to a given repeating decimal.
1
EXAMPLE
Writing a Repeating Decimal as a Rational Number

Express the repeating decimal 0.82222 . . . as a rational number.
SOLUTION We notice that the repeating decimal can be rewritten as a sum.
0.8 1 0.02 1 0.002 1 0.0002 1 c After the first term of 0.8, we see a geometric series in which the first term is 0.02 1 and each subsequent term is found by multiplying the previous term by 5 0.1.
10
0.8 1 0.02 1 0.002 1 0.0002 1 c

Geometric series
We can find the sum of the infinite geometric series. S5 5 0.02 1 2 0.1 0.02 0.9
2 100 5 9 10
743
5 5
2 10 ? 100 9 1 45
Notice that we want to keep the calculations in fraction form, since we ultimately want to express the repeating decimal in fraction form. We now add the initial 0.8 to the fraction 45 to obtain the final rational number. 0.82222 c 5 0.8 1 5 5 5 5 5 1 45
1
8 1 1 10 45 1?2 8?9 1 90 90 72 1 2 90 74 90 37 45
We can check our work using the graphing calculator (Figure 11.6).
FIGURE 11.6
11.2 Summary
In this section, you learned how to find the sum of a finite geometric series. You learned that an infinite geometric series in which 2 1 , r , 1 will converge. Otherwise, the series will diverge. You can use the formulas for the sum of a finite or infinite geometric series to solve a variety of problems.
744
CHAPTER 11

In Exercises 16, do the following: (a) Determine if the table represents a linear or an exponential function, and explain how you know. (b) Write a function to represent the data in the table. 1. x 0 1 2 3 4 3. x 0 5 10 15 20 f (x) 10 11.5 13 14.5 16 f (x) 2 4 6 8 10 4. 2. x 0 1 2 3 4 x 0 5 10 15 20 f (x) 10 15 22.5 33.75 50.625 f (x) 2 4 8 16 32 5. x 0 2 4 6 8 f (x) 0.001 0.01 0.1 1 10 6. x 0 2 4 6 8 f (x) 0.001 10.001 20.001 30.001 40.001
In Exercises 710, write a function to match the verbal description. 7. The function has an initial value of 25 and decreases by 5 every day. 8. The function has an initial value of 25 and decreases by 5 percent every day. 9. The function has an initial value of 500 and doubles every 10 years. 10. The function has an initial value of 500 and increases by 2 every 10 years.
11.2 Exercises
In Exercises 15, determine if the series is geometric. Explain how you know.
10 1 n 1. a 500 ? a b 2 n51 10 1 2. a 500 1 ? n 2 n51
8. 2 1
2 2 2 2 1 1>2 1 1>2 1 c 1 1>2 1 c 21>2 3 4 k 1 1 10 1 1 c1 k21 1 c 10 100 10
9. 10 1 1 1
3. a 2n2
n51 5
20
4. a ( 4 1 2n )
n51 n
10. 2 1 4 1 6 1 8 1 c 1 2k 1 c In Exercises 1115, match the given series with the appropriate graph (AE) of the partial sums. A.
y
5. a 4 ? 2
n51
B.
In Exercises 610, find the sum of the first four terms of the series. Then determine a formula for the sum of the first n terms of the series. Determine whether the series converges or diverges. If the series converges, state its sum. 6. 7. 3 3 3 3 c1 3 1 c 0 1 1 1 2 1 3 1 4 4 4 4 4n 2 1 2 3 4 k 1 1 1 1 1 c1 1 c 3 3 3 3 3
1000 800 600 400 200 0 0 2 4 6 8 x
y 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 0 2 4 6 8 10
745
C.
y 30 25 20 15 10 5 0 0 2 4 6 8 10 x
D.
y 5 4 3 2 1 0 0 2 4 6 8 10 x
Suppose that each time this counterfeit money is used, 26 percent of it is seized. From 1999, what is the total impact of the counterfeit money on the economy? 27.
Economic Impact of an Event
E.
y 1 0.75 0.50 0.25 0.25
`
Officials from the city of Indianapolis predicted that up to $45 million would be spent by visitors who were in town for the NCAA Final Four basketball tournament in 2006. It was expected that much of this money would be spent on national chain restaurants and hotels, and therefore might not be re-spent locally. (Source:
www.courierjournal.com.)
8 10 x
Suppose that only 40 percent of the money was re-spent locally. What is the total economic impact of the Final Four Basketball Tournament on Indianapolis? 28. 1 12. a ? ( 2n ) n51 2 14. a ( 2 1 3n )
n51 ` `
Drug Levels in the Bloodstream
1 11. a 4 ? a b 2 n51
` 1 n 13. a a 2 b 4 n51 `
Suppose that 753.3 mg of azithromycin is in the bloodstream after the fifth dose is taken. If the person stops taking the medicine at this time, how long will it take until the medicine is completely eliminated from the bloodstream? The half-life of azithromycin is 68 hours. (Source:
www.drugs.com.)
10 15. a n 5 1 ( n 1 1 )( n 1 2 ) Exercises 1620 give infinite geometric series that may or may not converge. Explain how you can tell if the infinite series will converge just by looking at the expression. Do not determine the sum of the infinite series.
` 1 16. a ? ( 5n ) n51 5 ` 1 n 18. a 5 ? a 2 b 5 n51 ` ` 1 n 17. a 5 ? a b 5 n51 ` 2 n 19. a 1 ? a b 3 n51
29. Create a finite geometric series in which the common ratio is between 2 1 and 1. Explain why the series you created is a geometric series. Find the total sum of your finite geometric series. 30. Explain why the formula for the sum of a finite geometric series includes the stipulation that r 2 1. 31. Explain why the formula for the sum of an infinite geometric series includes the stipulation that 2 1 , r , 1 in order for the infinite series to converge. 32. Create an infinite geometric series that will converge. Explain how you know that the series you created will converge. Find the total sum of your infinite series. 33. Explain why the idea of infinite geometric series can be used to find a rational number that is equivalent to a repeating decimal. Then, write your own repeating decimal and express your number as a rational number.
2 20. a ? ( 1n ) n51 3 In Exercises 2125, express the repeating decimal as a rational number. 21. 0.2222 c 23. 2.242424 c 25. 0.181818 c 26.
Economics According to the Federal Reserve Bank of Boston, $180 million of counterfeit money was put into the economy in 1999. (Source: www.bos.frb.org.)
22. 0.9999 c 24. 1.7373737 c
746
CHAPTER 11
37. Using multiple representations, verify that the

`
34. A ball is dropped from a height of 15 meters. It rebounds one-half of the vertical height from which it just fell. It continues to rebound in this way indefinitely. Find the total distance traveled by the ball. 35. The transcendental number e can be approximated by the infinite series e<1111 1 1 1 1 1 1c 2 6 24
following series converges to 1: a 38. What is (are) the value(s) of p if

n51 22n a ( 2 1 p ) 5 8? `
n51
1 n( n 1 1 )
5 1.
39. Write a discussion revealing the errors in reasoning in the following proof. 05010101c 0 5 (1 2 1) 1 (1 2 1) 1 (1 2 1) 1 c 05121112111211c 0 5 1 1 (21 1 1) 1 (21 1 1) 1 c 0511010c 051 40. Construct an argument that explains why, if the
`
Write this infinite series using sigma notation and verify that the partial sums approach e as the number of terms approaches infinity. 36. The transcendental number p can be approximated by the infinite series 4 4 4 4 4 p< 2 1 2 1 2c 1 3 5 7 9 Write this infinite series using sigma notation and verify that the partial sums approach p as the number of terms approaches infinity.
series a n diverges, then the series a n

n51 n51
1 25
must
also diverge.

Getting Started
Construct Taylor polynomials that approximate other functions
If we use our calculator, we can compute sin( 23) < 0.391. Is there a way to compute sin( 23) < 0.391 without the aid of the trigonometric function keys on a calculator? What computations are involved in doing so? It is clear that we can evaluate polynomial functions (linear, quadratic, cubic, and so on) with or without a calculator. But what about trigonometric, logarithmic, or exponential functions? What computations are involved in evaluating these types of functions without the aid of the trigonometric, logarithmic, or exponential function keys on a calculator?
In this section, we will use Taylor polynomials to approximate non-polynomial functions without the aid of the trigonometric function keys on a calculator. We begin by finding Taylor polynomials Pn( x ) that approximate the function f ( x ) 5 sin( x ) .
Just In Time
Polynomial Functions
For a whole number n, a function of the form y 5 an xn 1 an 2 1 xn 2 1 1 an 2 2 xn 2 2 1 c 1 a1 x 1 a0 with an 2 0 is called a polynomial function of degree n. Each ai xi is called a term. The ai are called the coefficients of the terms. The graphs of polynomial functions are fairly predictable and well behaved. We summarize the appearance of the graphs of polynomial functions of the first through fifth degree in the following table. (Continued)
747
Function Name and Degree Linear 1st degree
Concavity and Inflection Points No concavity No inflection points
Constant Difference First
Sample Graph
y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x
Quadratic 2nd degree
Concave up only or concave down only No inflection points
Second
y 4 3 2 1 4 3 2 11 2 3 4 y 4 3 2 1 4 3 2 11 2 3 4 y 4 3 2 1 4 3 2 11 2 3 4 y 4 3 2 1 4 3 2 11 2 3 4 1 2 3 4 x 1 2 3 4 x 1 2 3 4 x 1 2 3 4 x
Cubic 3rd degree
Concave up and concave down One inflection point
Third
Quartic 4th degree
Concavity changes zero or two times Zero or two inflection points
Fourth
Quintic 5th degree
Concavity changes one or three times One or three inflection points
Fifth
748
CHAPTER 11
One of the simplest polynomial functions is the linear function. To find a linear function that approximates f ( x ) 5 sin( x ) , we need to focus our efforts around a particular value of x. We will use x 5 0 as this value. The linear function that approximates f ( x ) 5 sin( x ) at x 5 0 is the line that is tangent to f ( x ) at x 5 0. We will write this linear function in the form P1( x ) 5 C0 1 C1x. The constant C0 is the value of the y-intercept. This y-intercept needs to be the same for P1( x ) and f ( x ) . Observe that f ( 0 ) 5 sin( 0 ) 50 So the y-intercept value of f ( x ) 5 sin( x ) is 0. The y-intercept value of P1( x ) 5 C0 1 C1x is found by evaluating the function at x 5 0. P1( 0 ) 5 C0 1 C1( 0 ) 5 C0 The y-intercept value of the Taylor polynomial P1( x ) 5 C0 1 C1x is C0. Since the values of the two y-intercepts are to be equal, C0 5 0. Therefore, P1( x ) 5 0 1 C1x. Additionally, we want the slope of the linear function to be equal to the slope of f ( x ) at x 5 0. The derivative can be used to find these slopes. f ( x ) 5 sin( x ) f r( x ) 5 cos( x ) f r( 0 ) 5 cos( 0 ) f r( 0 ) 5 1
y 1 0.50 3
2
P1( x ) 5 C0 1 C1x P1r( x ) 5 C1 P1r( 0 ) 5 C1
P1(x) = 0 + 1x
f (x) = sin x
0.50 1
3 2
FIGURE 11.7
Because we require the slopes to be equal at x 5 0, C1 5 1. Therefore, the linear Taylor polynomial is P1( x ) 5 0 1 1 ? x. A graph of the functions P1( x ) and f ( x ) verifies our findings (Figure 11.7). We see that near x 5 0, the linear Taylor polynomial P1( x ) 5 0 1 1 ? x closely approximates the function f ( x ) 5 sin( x ) . How can we make a better approximation? We can create a polynomial of higher degree. We will create a quadratic polynomial of the form P2( x ) 5 C0 1 C1 x 1 C2 x 2 that approximates the function f ( x ) 5 sin( x ) . Since we have already determined that C0 5 0 and C1 5 1, we need only find the value of C2. We find the value of C2 by using second derivatives. We want the second derivative of f ( x ) to be the same as the second derivative of the quadratic polynomial P2( x ) . Recall that the second derivative of a function can be used to describe the concavity of the graph of the function. It is necessary that the concavity of the function f ( x ) be the same as that of our polynomial P2( x ) . f ( x ) 5 sin( x ) f r( x ) 5 cos( x ) f s( x ) 5 2 sin( x ) f s( 0 ) 5 2 sin( 0 ) f s(0) 5 0 Because we require the second derivatives to be equal at x 5 0, C2 5 0. The quadratic Taylor polynomial would be P2( x ) 5 0 1 1x 1 0x 2 if the coefficient P2( x ) 5 C0 1 C1x 1 C2 x2 P2r( x ) 5 C1 1 2C2 x P2s ( x ) 5 2C2 P2s ( 0 ) 5 2C2
749
on the x 2 term were nonzero. However, this polynomial simplifies to the linear polynomial P1( x ) 5 0 1 1 ? x. That is, we are not able to create a quadratic polynomial that approximates f ( x ) 5 sin( x ) . We will try to create a cubic polynomial of the form P3( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3. Again, we will take advantage of our previous work and use the third derivative to find the value of the constant C3. While the third derivative does not describe something as concrete as the slope of the tangent line (first derivative) or the concavity of the graph (second derivative), it does describe the rate of change of the second derivative. We would like for this rate to be the same for f ( x ) 5 sin( x ) and P3( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3. f ( x ) 5 sin( x ) f r( x ) 5 cos( x ) f s( x ) 5 2 sin( x ) f t( x ) 5 2 cos( x ) f t( 0 ) 5 2 cos( 0 ) 5 21 Because the third derivatives are required to be equal, 6C3 5 2 1 C3 5 2
P3(x) = x 1 x3 6 y 1 0.50 3
2
P3( x ) 5 C0 1 C1x 1 C2 x2 1 C3 x3 P3 r( x ) 5 C1 1 2C2 x 1 3C3 x2 P3 s ( x ) 5 2C2 1 6C3 x P3 t ( x ) 5 6C3 P3 t ( 0 ) 5 6C3
1 6
1 1
Thus the cubic Taylor polynomial is P3( x ) 5 0 1 1x 1 0x 2 2 x 3 5 x 2 x 3. 6 6 A graph confirms that P3( x ) < f ( x ) near x 5 0 (Figure 11.8). We can use the cubic polynomial P3( x ) to approximate the value of sin( 23) < 0.391. Since derivatives of trigonometric functions were used to create the cubic polynomial, we must convert to radian measure. 23 ? We calculate P3 A
23p 180
f (x) = sin x x
0.50 1
3 2
p radians 23p radians 5 180 180
B.
P3 a 23p 23p 1 23p 3 b 5 a b 2 a b 180 180 6 180 < 0.391
FIGURE 11.8
To the nearest thousandth, the cubic polynomial accurately determines that sin( 23) < 0.391. What about other values? Looking back at the graph, we see that if we choose values of x between x 5 22 radians and x 5 2 radians, approximations will be close. Outside of this range, approximations will not be as close. In fact, the farther away we move from this range, the worse the approximation (Figure 11.9).
750
CHAPTER 11
P3(x) = x 1 x3 6
y 1
P3(x) = x 1 x3 6
y 1
0.50 3
2
f (x) = sin x x
0.50 3
2
f (x) = sin x
0.50 1
3 2
0.50 1
3 2
Close approximations when 22 # x # 2 FIGURE 11.9
Less accurate approximations outside this range of values
To obtain good approximations for values outside of this range, we need to calculate a Taylor polynomial of higher order. As we do this, we will look for patterns. We will find a fourth-degree Taylor polynomial of the form P4( x ) 5 C0 1 C1 x 1 C2 x 2 1 C3 x 3 1 C4 x 4 to approximate the function f ( x ) 5 sin( x ) . As we have seen, higher-order derivatives are needed to compute the coefficients of our higher-order polynomials. In this case, we will need to compute the fourth derivative to find the value of C4. f ( x ) 5 sin( x ) f r( x ) 5 cos( x ) f s ( x ) 5 2 sin( x ) f t ( x ) 5 2 cos( x ) f
(4)
P4( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3 1 C4 x 4 P4r( x ) 5 C1 1 2C2 x 1 3C3 x 2 1 4C4 x 3 P4s ( x ) 5 2C2 1 6C3 x 1 12C4 x 2 P 4t ( x ) 5 6C3 1 24C4 x P4(4)( x ) 5 24C4 P4(4)( 0 ) 5 24C4
( x ) 5 sin( x )
f (4)( 0 ) 5 sin( 0 ) f (4)( 0 ) 5 0
Because the fourth derivatives must be equal, C4 5 0. Again, as in the case of the quadratic polynomial (value of C2), we find that the value of C4 is zero. If we look ahead in an effort to recognize patterns, we see that every even-powered polynomial will yield the same result. This is because every other time we take the derivative of f ( x ) , we get either sin( x ) or 2 sin( x ) . When we evaluate these derivatives at x 5 0, we get 0. Consequently, when we then solve for the missing constant, we will get zero. By recognizing this pattern, we learn to focus on the odd-powered polynomials, which will yield nonzero constants. Lets find a fifth-degree Taylor polynomial to approximate f ( x ) 5 sin( x ) . It will be of the form P5( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3 1 C4 x 4 1 C5 x 5. To find the value of C5, we need to compute the fifth derivatives of f ( x ) and P5( x ) . f ( x ) 5 sin( x ) f r( x ) 5 cos( x ) f s ( x ) 5 2 sin( x ) f t ( x ) 5 2 cos( x ) f
(4)
P5( x ) 5 C0 1 C1x 1 C2 x2 1 C3 x3 1 C4 x4 1 C5 x5 P5( x ) 5 0 1 C1x 1 0 1 C3 x3 1 0 1 C5 x5 P 5r( x ) 5 C1 1 3C3 x2 1 5C5 x4 P5s ( x ) 5 6C3 x 1 20C5 x3 P5t ( x ) 5 6C3 1 60C5 x2
( x ) 5 sin( x )
751
f (5)( x ) 5 cos( x ) f (5)( 0 ) 5 cos( 0 ) f (5)( 0 ) 5 1

1
(4) P5 ( x ) 5 120C5 x (5) P5 ( x ) 5 120C5 (5) P5 ( 0 ) 5 120C5
It follows that C5 5 . We find that the fifth-degree polynomial is 120 1 1 5 P5( x ) 5 0 1 1x 1 0x 2 2 x 3 1 0x 4 1 x 6 120 1 1 5 5 x 2 x3 1 x 6 120
1 5 P5(x) = x 1 x 3 + 120 x 6 y
f (x) = sin x
1 0.50
A graph confirms that P5( x ) < f ( x ) near x 5 0 (Figure 11.10). The graph shows that the fifth-degree Taylor polynomial P5( x ) approximates the function f ( x ) 5 sin( x ) over a wider range of x-values than the third-degree Taylor polynomial. For example, lets compare f ( 2 ) 5 sin( 2 ) (where the angle is 2 radians) with P5( 2 ) 5 ( 2 ) 2 ( 2 ) 3 1 6
1 1 ( 2 ) 5. 120
3
2
0.50 1
3 2
f ( 2 ) 5 sin( 2 ) < 0.909 1 1 P5( 2 ) 5 ( 2 ) 2 ( 2 ) 3 1 ( 2 ) 5 < 0.933 6 120 We can make our calculations even more accurate if we construct even higher-order polynomials. Furthermore, the range of values over which the polynomial function will accurately approximate the function f ( x ) 5 sin( x ) will get wider as we construct higher-order polynomials. Finding patterns will help us overcome the tediousness of constructing these higher-order polynomials. One pattern that we have already noted for the Taylor polynomials for y 5 sin( x ) is that the polynomial terms with even powers have coefficients of zero. There is no need to work to find these values because we have seen how and why the coefficients become zero. What about odd-powered polynomials? Are there patterns to be recognized? First, as we compute higher-order derivatives, we find that the terms of the polynomial whose powers are less than the powers of the higher-order derivative we are finding will ultimately become zero. Second, since only the highest-powered term will be left after taking derivatives, we can study this term to see if any patterns develop. Lets analyze the process we used in finding C3. We take the third derivative of the cubic term of the polynomial P3( x ) . T3 5 C3 x3 T 3r 5 3C3 x2 T 3s 5 3 ? 2 ? C3 x T 3t 5 3 ? 2 ? 1 ? C3 T 3t 5 3! ? C3 Recall that 3! (read three factorial) is defined as 3! 5 3 ? 2 ? 1 5 6. Compare this result with our previous value for P 3t ( x ) . The results are the same, but it is
FIGURE 11.10
752
CHAPTER 11
helpful to recognize that the value of the coefficient of the cubic term can be expressed in factorial form. f t ( 0 ) 5 2 cos( 0 ) 5 P2t ( 0 ) 5 6C3 2 1 5 6C3 1 2 5 C3 6 f t ( 0 ) 5 2 cos( 0 ) 5 P2t ( 0 ) 5 3!C3 2 1 5 3!C3 2 1 5 C3 3!
Now lets look back at the work we did to find the value of C5. Again, only the highest term of the fifth-degree polynomial P5( x ) need be considered. T5 5 C5 x5 T5r 5 5 ? C5 x4 T5s 5 5 ? 4 ? C5 x3 T5t 5 5 ? 4 ? 3 ? C5 x2 T5(4) 5 5 ? 4 ? 3 ? 2 ? C5 x T5(5) 5 5 ? 4 ? 3 ? 2 ? 1 ? C5 T5(5) 5 5! ? C5 Armed with this knowledge, we can write as many terms of any higher-order polynomial as we wish. For example, if we want to write P7( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3 1 C4 x 4 1 C5 x 5 1 C6 x 6 1 C7 x 7, we know that the coefficient of the final term will involve 7!. f (7)( 0 ) 5 2 cos( 0 ) 5 P7(7)( 0 ) 5 7!C7 2 1 5 7!C7 2 2 1 5 C7 7!
1 5 C7 5040
Using this information along with our previous work, we can find another approximation for f ( x ) 5 sin( x ) .
y 1 0.50 3
2
P7( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3 1 C4 x 4 1 C5 x 5 1 C6 x 6 1 C7 x 7
f (x) = sin x
1 1 5 1 7 5 0 1 x 1 0 ? x 2 2 x 3 1 0 ? x4 1 x 1 0 ? x6 2 x 6 120 5040 1 1 5 1 7 5 x 2 x3 1 x 2 x 6 120 5040
0.50 1
3 2
1 5 1 7 P5(x) = x 1 x 3 + 120 x 5040 x 6
FIGURE 11.11
We confirm our work graphically and observe that the range of x-values over which the polynomial accurately approximates the function has widened (Figure 11.11). The graph shows that the seventh-degree polynomial more closely approximates the function f ( x ) 5 sin( x ) over a wider interval. We can see that as we construct higher- and higher-order polynomials, the graphs of these polynomials approximate the graph of f ( x ) 5 sin( x ) over a wider and wider range of x-values. The series of graphs in Table 11.5 shows odd-powered polynomials increasing in power as we read from left to right.
753
TABLE 11.5
P1( x ) 5 0 1 1 ? x
y 1 0.5 6 4 2 0.5 1 2 4 6 x
P3( x ) 5 x 2 6 x 3
y 1 0.5 6 4 2 0.5 1 2 4 6 x
P5( x ) 5 x 2 6 x 3 1 120 x 5
y 1 0.5 6 4 2 0.5 1 2 4 6 x
P7( x ) 5 x 2 3! x 3 1 5! x 5 2 7! x 7
y 1 0.5 6 4 2 0.5 1 2 4 6 x
P9( x ) 5 x 2 3! x 3 1 5! x 5 2 7! x 7 1 9! x 9
y 1 0.5 6 4 2 0.5 1 2 4 6 x
P11( x ) 5 x 2 3! x 3 1 5! x 5 2 7! x 7 1 9! x 9 2 11! x 11
y 1 0.5 6 4 2 0.5 1 2 4 6 x
Through the process of taking higher-order derivatives and determining the values of their coefficients, we are able to find a polynomial function that approximates the function f ( x ) 5 sin( x ) . f ( x ) < Pn( x ) sin( x ) < x 2 1 3 1 1 1 1 11 c x 1 x5 2 x7 1 x9 2 x 1 3! 5! 7! 9! 11!
The more terms in the polynomial, the more accurately it approximates the function f ( x ) 5 sin( x ) . Table 11.6 compares function values for f ( x ) 5 sin( x ) and P11( x ) 5 x 2 x 3 1 x 5 2 x 7 1 x 9 2 x 11. 3! 5! 7! 9! 11!
1 1 1 1 1
754
CHAPTER 11
TABLE 11.6
x 24 23 22 21 0 1 2 3 4
f ( x ) 5 sin( x ) 0.7568 2 0.1411 2 0.9093 2 0.8415 0 0.8415 0.9093 0.1411 2 0.7568
P11( x ) 5 x 2 3! x 3 1 5! x 5 2 7! x 7 1 9! x 9 2 11! x 11 0.7668 2 0.1409 2 0.9093 2 0.8415 0 0.8415 0.9093 0.1409 2 0.7668
We see that the polynomial accurately approximates the function f ( x ) 5 sin( x ) and is particularly accurate for values of x near zero. As the value of x gets further and further away from zero, the polynomial approximation becomes less accurate. If we look back at the work we did to compute the terms of the Taylor polynomial centered at x 5 0, we can summarize our work into a single formula. Each term of the polynomial was found by taking the nth derivative. That is, to find the fifth-degree term, we took the fifth derivative. In doing so, we noticed the factorial ( 5! 5 5 ? 4 ? 3 ? 2 ? 1 ) . Solving for the appropriate coefficient required that we divide by this factorial. Putting it all together, to construct a Taylor polynomial, we take the nth derivative, substitute x 5 0, and divide by n!. This yields the coefficient of the nth-degree term.
TERMS OF THE TAYLOR POLYNOMIAL CENTERED AT x 5 0
The nth term of the Taylor polynomial Pm( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x3 1 c 1 Cm x m centered at x 5 0 is given by Tn 5 Cn x n 5a f (n)( 0 ) n bx n!
EXAMPLE
Using a Taylor Polynomial to Estimate a Function Value

Construct a sixth-degree polynomial that can be used to approximate the function g( x ) 5 x sin( x ) , centered at x 5 0. Use this polynomial to evaluate g Q 3 R . Use graphs to check your work.
SOLUTION Once we have Taylor polynomials for the basic functions, we can find Taylor polynomial approximations for combinations of functions without
p
755
going through the full process of finding higher-order derivatives. In this case, we already have a fifth-degree polynomial approximation for f ( x ) 5 sin( x ) . 1 3 1 x 1 x5 3! 5! To construct a Taylor polynomial for g ( x ) 5 x sin( x ) , we multiply both sides of sin( x ) < x 2 sin( x ) < x 2 3! x 3 1 5! x 5 by x. x ? sin( x ) < x ? a x 2 x ? sin( x ) < x 2 2
1 4 1 6 x + 5! x P6 (x) = x2 3! y
1 3 1 x 1 x5b 3! 5!
1 4 1 x 1 x6 3! 5!
3 2 1 4 3 2 1 1 2 3 g (x) = xsin x 1 2 3 4 x
We graph g( x ) along with its sixth-degree Taylor polynomial to verify our solution (Figure 11.12). The graph shows us that the sixth-degree polynomial closely approximates the function g( x ) 5 x sin( x ) on the interval 2 2 # x # 2. We can use the Taylor polynomial function to evaluate g Q 3 R . p p g a b < P6 a b 3 3 p p 2 1 p 4 1 p 6 P6 a b 5 a b 2 a b 1 a b 3 3 3! 3 5! 3 5 0.9072 We can compare this approximation to the estimated value of g Q 3 R given by the calculator. p p p g a b 5 a b ? sin a b 3 3 3 < 0.9069 Not all Taylor polynomials are centered around x 5 0. Example 2 illustrates a Taylor polynomial centered around x 5 1.
p p
FIGURE 11.12
EXAMPLE
Finding a Taylor Polynomial for f( x) 5 ln(x)

Construct a fifth-degree polynomial that can be used to approximate the function f ( x ) 5 ln( x ) , centered at x 5 1. Use this polynomial to evaluate f ( 1.75 ) . Use graphs to check your work.
SOLUTION We want to find a fifth-degree polynomial of the form P5( x ) 5 C0 1 C1x 1 C2 x 2 1 C3 x 3 1 C4 x 4 1 C5 x 5. The coefficient of the nth-degree term is found by taking the nth derivative, substituting x 5 1, and solving for the coefficient. We choose to center the polynomial at x 5 1 because the function f ( x ) 5 ln( x ) is not defined at x 5 0. To find C0, we evaluate f ( x ) 5 ln( x ) at x 5 1 and set the result equal to C0. We can do this because P0( x ) 5 C0.
f ( x ) 5 ln( x ) f ( 1 ) 5 ln( 1 ) 5 P0( 1 ) 5 C0 0 5 C0
756
CHAPTER 11
To find C1, we use the first derivative and a linear (first-degree) polynomial. f ( x ) 5 ln( x ) f r( x ) 5 f r( 1 ) 5 1 x 1 1 P1( x ) 5 C0 1 C1x P1r( x ) 5 C1 P1r( 1 ) 5 C1
f r( 1 ) 5 1 We require that f r( 1 ) 5 Pr 1( 1 ) . Therefore, C1 5 1. To find C2, we use the second derivative and a quadratic (second-degree) polynomial. f ( x ) 5 ln( x ) f r( x ) 5 1 x 1 x2 1 12 P2( x ) 5 C0 1 C1 x 1 C2 x2 P2r( x ) 5 C1 1 2C2 x P2s ( x ) 5 2C2 P2s ( 1 ) 5 2C2
f s(x) 5 2 f s(1) 5 2
5 21 We require that f s ( 1 ) 5 P2s ( 1 ) . Therefore, 2C2 5 2 1 C2 5 2 1 2
As we work through this process, we should be alert to any patterns that are developing. So far it appears as though the coefficients, starting with C1, can be described by 6 n. Lets check this guess by finding C3 using the third derivative and the cubic (third-degree) polynomial. f ( x ) 5 ln( x ) f r( x ) 5 1 x 1 x2 P3( x ) 5 C0 1 C1 x 1 C2 x2 1 C3 x3 P3 r( x ) 5 C1 1 2C2 x 1 3C3 x2 P3 s ( x ) 5 2C2 1 6C3 x P3 t ( x ) 5 6C3 P 3t ( 1 ) 5 6C3
1
f s(x) 5 2 f t(x) 5 f t(1) 5 2 x3 2 13
52
757
We require that f t ( 1 ) 5 P t 3 ( 1 ) . Therefore, 6C3 5 2 C3 5 1 3
It certainly appears as though our conjecture is correct. The coefficients of the polynomial do appear to be of the form 6 , where n represents the degree of the n polynomial. We will complete one more round of differentiation to confirm our conjecture. To find C4, the coefficient of the fourth-degree polynomial, we will use the fourth derivative. f ( x ) 5 ln( x ) f t(x) 5 2 x3 6 x4 6 14 P4( x ) 5 C0 1 C1 x 1 C2 x2 1 C3 x3 1 C4 x4 P4t ( x ) 5 6C3 1 24C4 x
4) P( 4 ( x ) 5 24C4
f (4)( x ) 5 2 f (4)( 1 ) 5 2
(4) P4 ( 1 ) 5 24C4
5 26
(4) We require that f (4)( 1 ) 5 P4 ( 1 ) . Therefore,
24C4 5 2 6 C4 5 2 52 6 24 1 4
We will use pattern recognition to write the fifth-degree polynomial P5( x ) 5 C0 1 C1 x 1 C2 x 2 1 C3 x 3 1 C4 x 4 1 C5 x 5 that approximates the function f ( x ) 5 ln( x ) . 1 1 1 1 ln( x ) < 0 1 1x 2 x 2 1 x 3 2 x 4 1 x 5 2 3 4 5
P5 (x) = x 1 x2 + 1 x3 1 x4 + 1 x5 2 3 4 5 y 3 2 1 2 1 1 2 3 4 x f (x) = ln x
1 1 1 1 ln( x ) < x 2 x 2 1 x 3 2 x 4 1 x 5 2 3 4 5 We can confirm our work graphically (Figure 11.13). The graph clearly shows that the polynomial P5( x ) 5 x 2 x 2 1 x 3 2 x 4 1 x 5 does not accurately ap2 3 4 5 proximate the function f ( x ) 5 ln( x ) . If we closely examine the graph, it appears as though the polynomial function is shifted 1 unit to the left compared with the graph of the function f ( x ) 5 ln( x ) . Recall that we used x 5 1 as the center for our work. We must adjust for this in the polynomial function by shifting the function 1 unit to the right. We accomplish this adjustment through an algebraic transformation. We will replace the x-values on the right-hand side of the equation with the quantity x 2 1. 1 1 1 1 P5( x ) 5 ( x 2 1 ) 2 ( x 2 1 ) 2 1 ( x 2 1 ) 3 2 ( x 2 1 ) 4 1 ( x 2 1 ) 5 2 3 4 5
1 1 1 1
1 2 3
FIGURE 11.13
758
CHAPTER 11
A graph shows that the new function P5( x ) closely approximates the function f ( x ) 5 ln( x ) over the interval 0 # x # 2 (Figure 11.14).
P5 (x) = (x 1) 1 (x 1)2 + 1 (x 1)3 1 (x 1)4 + 1 (x 1)5 2 3 4 5 y 3 2 1 2 1 1 2 3 4 x f (x) = ln x
1 2 3
FIGURE 11.14
We now use our polynomial to approximate f ( 1.75 ) 5 ln( 1.75 ) . 1 1 1 1 P5( 1.75 ) 5 ( 1.75 2 1 ) 2 ( 1.75 2 1 ) 2 1 ( 1.75 2 1 ) 3 2 ( 1.75 2 1 ) 4 1 ( 1.75 2 1 ) 5 2 3 4 5 5 0.5777 We can compare this result to the result obtained by using a calculator to evaluate f ( 1.75 ) 5 ln( 1.75 ) . f ( 1.75 ) 5 ln( 1.75 ) < 0.5596 Our estimate is close to the actual value. To find a Taylor polynomial for a given function that is not centered at x 5 0, we do the same basic work, but we shift the polynomial the appropriate distance left or right.
TERMS OF THE TAYLOR POLYNOMIAL CENTERED AT x 5 a
The nth term of the Taylor polynomial centered at x 5 a is given by Tn 5 Cn( x 2 a ) n 5 a f (n)( a ) b (x 2 a)n n
EXAMPLE
Using a Taylor Polynomial to Approximate h(x) 5 "x

Construct a third-degree polynomial that can be used to approximate the function h( x ) 5 "x, centered at x 5 1. Use this polynomial to evaluate h( 1.3 ) . Use graphs to check your work.
759
SOLUTION We want to find a third-degree polynomial of the form P3( x ) 5 C0 1 C1 x 1 C2 x 2 1 C3 x 3. The coefficient of the nth-degree term is found by taking the nth derivative, substituting x 5 1, and solving for the coefficient. To find C0, we evaluate both functions at x 5 1.
h( 1 ) 5 "1 51
h( x ) 5 "x
P0( x ) 5 C0 P0( 1 ) 5 C0
We require that h( 1 ) 5 P0( 1 ) . Therefore, C0 5 1. To find C1, we use the first derivative and a linear (first-degree) polynomial. h( x ) 5 "x P1( x ) 5 C0 1 C1x P1r( x ) 5 C1 P1r( 1 ) 5 C1 1 2!x
hr( x ) 5 hr( 1 ) 5 5
1 2!1 1 2
We require that hr( 1 ) 5 P 1 r( 1 ) . Therefore, C1 5 2. To find C2, we use the second derivative and a quadratic (second-degree) polynomial. h( x ) 5 "x hr( x ) 5 P2( x ) 5 C0 1 C1 x 1 C2 x2 P2r( x ) 5 C1 1 2C2 x P2s ( x ) 5 2C2 P2s ( 1 ) 5 2C2 1 2!x
hs(x) 5 2 hs(1) 5 2
1 4x3>2 1 4
We require that h s ( 1 ) 5 P2s ( 1 ) . Therefore, 2C2 5 2 C2 5 2 1 4 1 8
Finally, we find C3 using the third derivative and the cubic (third-degree) polynomial. h( x ) 5 "x hr( x ) 5 P3( x ) 5 C0 1 C1 x 1 C2 x2 1 C3 x3 P3 r( x ) 5 C1 1 2C2 x 1 3C3 x2 P3 s ( x ) 5 2C2 1 6C3 x Pt 3 ( x ) 5 6C3 P3 t ( 1 ) 5 6C3 1 2!x
hs(x) 5 2 ht(x) 5 ht(1) 5
1 4x3>2
3 8x5>2 3 8
760
CHAPTER 11
We require that h t ( 1 ) 5 P3t ( 1 ) . Therefore, 6C3 5 C3 5 5 3 8 3 48 1 16
We now can write the third-degree polynomial that approximates h( x ) 5 "x. Remember that we need to shift the polynomial function to the right 1 unit because we used x 5 1 as the center. 1 1 1 "x < 1 1 ( x 2 1 ) 2 ( x 2 1 ) 2 1 ( x 2 1 ) 3 2 8 16 We will use a graph to check our work (Figure 11.15).
1 P3(x) = 1 + 1 (x 1) 1 (x 1)2 + 16 (x 1)3 2 8 y
4 3 2 1 0 0 1 2 3 4 x
h(x) = x
FIGURE 11.15
The graph shows that our polynomial does indeed closely approximate the function h( x ) 5 "x for values of x near x 5 1. We can use the polynomial to evaluate h( 1.3 ) 5 "1.3. 1 1 1 "1.3 < 1 1 ( 1.3 2 1 ) 2 ( 1.3 2 1 ) 2 1 ( 1.3 2 1 ) 3 2 8 16 When we compare this value to the value of h( 1.3 ) 5 "1.3 found using our calculator, we see that the polynomial approximation is good. h( 1.3 ) 5 "1.3 < 1.1402 < 1.1404
11.3 Summary
In this section, you learned how to approximate trigonometric, square root, and logarithmic functions using a Taylor polynomial function. You learned that the Taylor polynomial provides a means by which you can evaluate a variety of non-
761
polynomial functions. Depending on the interval you are working within and the accuracy desired, you can find higher-order polynomials if you wish to increase the accuracy of the approximation.

In Exercises 110, determine whether the function can be labeled a polynomial function. If not, explain why not. 1. y 5 2 3x3 2 7x 1 7 2. y 5 "x 1 x2 3. y 5 2 5 4. y 5 2x 2 5 x2 1 1
2 2
13. f ( x ) 5 2 2x4 2 2x3 1 x2 2 5x 1 2 14. f ( x ) 5 2 x5 1 x2 1 9 15. f ( x ) 5 x6 2 10x5 2 x In Exercises 1619, describe the end behavior of the polynomial function. That is, determine lim f ( x ) and xS` lim f ( x ) .
xS 2`
16. y 5 an xn 1 an 2 1 xn 2 1 1 an 2 2 xn 2 2 1 c1 a1x 1 a0 where an . 0 and n is even. 17. y 5 an xn 1 an 2 1 xn 2 1 1 an 2 2 xn 2 2 1 c1 a1x 1 a0 where an , 0 and n is even. 18. y 5 an xn 1 an 2 1 xn 2 1 1 an 2 2 xn 2 2 1 c1 a1x 1 a0 where an . 0 and n is odd. 19. y 5 an xn 1 an 2 1 xn 2 1 1 an 2 2 xn 2 2 1 c1 a1x 1 a0 where an , 0 and n is odd.
5. y 5 x0.5 1 x23 6. y 5 0.5x5 2 2.5x2 1 1.75x 1 6.3 7. y 5 2x1>3 1 6x21 8. y 5 2x 2 5 9. y 5 2x 10. y 5 sin( 2x ) In Exercises 1115, describe the end behavior of the polynomial function. That is, determine lim f ( x ) and xS` lim f ( x ) .
xS 2`
11. f ( x ) 5 2x2 1 3x 2 2 12. f ( x ) 5 x3 1 4x2 2 5x 1 1
11.3 Exercises
Exercises 12 are designed to help you conceptualize the notion of Taylor polynomials. 1. Suppose you are talking to a friend who missed the class about Taylor polynomials. Write a few sentences explaining what this topic is about and how Taylor polynomials are computed. 2. Describe, in general, the graphical comparisons between a polynomial and the function it is approximating. That is, what if the polynomial is an even function? What if it is an odd function? What needs to happen if you wish to see better and better approximations? In Exercises 312, find the first four nonzero terms of the Taylor polynomial, centered at x 5 0, for each function. Use the process described in this section. 3. f ( x ) 5 cos x 5. f ( x ) 5 "x 1 1 7. f ( x ) 5 cos( 2x ) 9. f ( x ) 5 "3x 1 1 4. f ( x ) 5 e x 6. f ( x ) 5 ln( x 1 1 ) 8. f ( x ) 5 e3x 10. f ( x ) 5 e x
2
11. f ( x ) 5 ln( 3x 1 1 )
12. f ( x ) 5 "( x 1 1 ) 3
762
CHAPTER 11
In Exercises 1318, extend the results of your previous work with Taylor polynomials to find the first four nonzero terms of the Taylor polynomial centered at x 5 0. In each case, graph the Taylor approximation along with the function and estimate the interval over which the approximation is relatively accurate. 13. f ( x ) 5 xe
x
the formula makes sense to you and your peers. The formula for the nth term of the polynomial is Tn 5 a f (n)( 0 ) n bx n!
14. f ( x ) 5 x sin( x ) 16. f ( x ) 5 ln( x2 ) 18. f ( x ) 5 cos( x2 )
15. f ( x ) 5 x cos( x ) 17. f ( x ) 5 "x3
In Exercises 1923, write the first four nonzero terms of the Taylor polynomial centered at the given value, x 5 a. Confirm your work by graphing your polynomial along with the given function. 19. f ( x ) 5 cos( x ) at x 5 21. f ( x ) 5 e x at x 5 4
2
27. Explain the formula used to determine the terms of a Taylor polynomial centered at x 5 a so that the formula makes sense to you and your peers. How is the process different from the process for determining a Taylor polynomial centered at x 5 0? The formula for the nth term of the polynomial centered at x 5 a is Tn 5 a f (n)( a ) b (x 2 a)n n!
p 6
20. f ( x ) 5 e at x 5 2
x
22. f ( x ) 5 ln( x ) at x 5 4
23. f ( x ) 5
1 x24
at x 5 5
28. You can find the first four nonzero terms of a Taylor polynomial centered at x 5 0 for the function f ( x ) 5 e x ? sin x without taking derivatives. Explain how this can be done and then find these nonzero terms. Graph the Taylor polynomial along with the function f ( x ) . 29. We have seen that the higher the order of a Taylor polynomial, the more closely it resembles the function it is meant to approximate. Explain why this is the case. 30. The first four nonzero terms of a Taylor polynomial approximating the function f ( x ) 5 sin x are P7( x ) 5 x 2 x3 x5 x7 1 2 3! 5! 7!
Exercises 24 and 25 will help you better understand how to create and use Taylor polynomials. 24. Find a seventh-degree polynomial that approximates the function f ( x ) 5 tan x at x 5 0. Use a graph to determine the interval over which the x-values reasonably approximate the function f ( x ) 5 tan x. Finally, use your polynomial to approximate f ( 0.5 ) . Compare the Taylor approximation with the actual value of tan( 0.5 ) . 25. Use your work in Exercise 24 to complete the table. Polynomial Approximation of f ( 0.5 )
Imagine using this Taylor polynomial to approximate the value of the function f ( x ) . List the following function values in order from least accurate to most accurate. Explain how you know. P7( 0.5 ) P7( 2 1.5 ) P7( 0 ) P7( 2 )
Degree of Polynomial 1 3 5 7
Value of tan( 0.5 )
31. Use the idea of Taylor polynomials to show that ln( 2 ) 5 1 2 1 1 1 1 1 2 1 2c 2 3 4 5
As you use higher- and higher-degree polynomials to approximate f ( 0.5 ) , what do you notice about how the polynomial approximation compares to the calculator value of tan( 0.5 ) ?
32. Construct a Taylor polynomial with at least four nonzero terms that approximates the function f ( x ) 5 tan21 x, centered at x 5 0, on the interval 2 1 # x # 1. 33. Use your Taylor polynomial from Exercise 32 to show that
26. Explain the formula used to determine the terms of a Taylor polynomial centered at x 5 0 so that
p542
4 4 4 4 1 2 1 2c 3 5 7 9
11.4 Taylor Series
763
34. Use the idea of Taylor polynomials to show that e51111 1 1 1 1 1 1 1 1c 2 6 24 120
35. Construct a Taylor polynomial that approximates the function f ( x ) 5 sin21 x, centered at x 5 0, on the interval 2p # x # p.
11.4 Taylor Series
Construct infinite Taylor series that approximate other functions Determine whether a Taylor series converges or diverges, and describe the interval of convergence
We saw in the last section that it is possible to approximate one function with another. We also saw that higher-order polynomials approximate these functions better. In fact, if we could make a never-ending polynomial to approximate a given function, it would be the best approximation possible. In this section we will explore the idea of an infinite Taylor polynomial, called a Taylor series.
We will begin by revisiting the Taylor polynomial for the function f ( x ) 5 sin( x ) . sin( x ) < x 2 1 3 1 1 1 1 11 x 1 x5 2 x7 1 x9 2 x 3! 5! 7! 9! 11!
Recall that we recognized a pattern. It is clear that the polynomial used to approximate f ( x ) 5 sin( x ) has terms containing only odd powers of x. Furthermore, each odd-powered term with exponent c has coefficient . We also recognize that the c! terms alternate in sign, with the first term being positive, the second negative, and so on. By recognizing these patterns, we are able to express the sum of the terms of the polynomial as an infinite sum.
` 1 sin( x ) 5 a ? x2n 2 1 ? ( 2 1 ) n 2 1 n 5 1 ( 2n 2 1 ) !
5 a
( 2 1 ) n 2 1 ? x2n 2 1 ( 2n 2 1 ) ! n51
`
Notice how we ensure that the terms of the polynomial alternate in sign from positive to negative. By including the factor ( 2 1 ) n 2 1, we ensure that when n is odd (e.g., terms n 5 1, n 5 3, and so on), we will have 2 1 raised to an even power, resulting in a positive 1. When n is even (e.g., terms n 5 2, n 5 4, and so on), we will have 2 1 raised to an odd power, resulting in a negative 1. Recall that the summation notation tells us to begin with n 5 1 and proceed in increments of 1. Each time we evaluate at a particular value of n, we determine that term of the series. Table 11.7 clarifies the connection between the summation notation and the first three terms of the Taylor polynomial that approximates f ( x ) 5 sin( x ) . sin( x ) 5 a ( 2 1 ) n 2 1 ? x2n 2 1 ( 2n 2 1 ) ! n51
`
764
CHAPTER 11
TABLE 11.7
Term number, n Term
n51 (21)121 ? x2 ? 1 2 1 5 x (2 ? 1 2 1)!
n52 (21)221 1 ? x2 ? 2 2 1 5 2 x3 (2 ? 2 2 1)! 3!
n53 1 (21)321 ? x2 ? 3 2 1 5 x5 (2 ? 3 2 1)! 5!
We see that if we continue this process for larger and larger values of n, we can write more and more terms of the Taylor polynomial. Our challenge is to be able to recognize patterns and express the Taylor polynomial as an infinite Taylor series.
EXAMPLE
Constructing an Infinite Taylor Series for f( x) 5 ln( x)

Construct an infinite Taylor series for f ( x ) 5 ln( x ) , centered at x 5 1.
SOLUTION In the last section, we constructed a fifth-degree polynomial to approximate f ( x ) 5 ln( x ) .
1 1 1 1 P5( x ) 5 ( x 2 1 ) 2 ( x 2 1 ) 2 1 ( x 2 1 ) 3 2 ( x 2 1 ) 4 1 ( x 2 1 ) 5 2 3 4 5 We recognize that the terms of this polynomial each contain ( x 2 1 ) raised to a The sign of each term alternates from term to term. If n represents the number of the term in the series, we can express the infinite Taylor series in summation notation.
` 1 ln( x ) 5 a ? ( x 2 1 ) n ? ( 2 1 ) n 2 1 n51 n
power. Furthermore, the coefficient on each of these terms with exponent c is c .
5 a
(21)n21 ? (x 2 1)n n n51

`
You should verify that as n increases from 1 to infinity, we get each particular term of the polynomial.
EXAMPLE
Constructing an Infinite Taylor Series for f( x) 5 e x

Construct an infinite Taylor series for f ( x ) 5 e x, centered at x 5 0.
SOLUTION In the previous section exercises, we found a fourth-degree polynomial that approximated the function f ( x ) 5 e x. The fifth-degree polynomial approximation is
1 1 1 1 5 e x < 1 1 x 1 x2 1 x3 1 x4 1 x 2 6 24 120 In order to best see the pattern that develops in this Taylor polynomial, we recognize that the denominator of each terms coefficient can be expressed using factorials. ex < 1 0 1 1 1 1 1 x 1 x1 1 x2 1 x3 1 x4 1 x5 0! 1! 2! 3! 4! 5!
11.4 Taylor Series
765
When the polynomial is written in this way, we can see various patterns develop. First, we see that the powers of x increase from 0 to 5. Second, we see that the denominator of the coefficient of each term can be written as the factorial of that terms degree. Using these two observations, we can write the infinite Taylor series.
` 1 ex 5 a xn 2 1 n51 (n 2 1)! ` xn 2 1 5 a n51 (n 2 1)!
We can confirm our results by substituting values of n and verifying that we do indeed obtain the sum given by the Taylor polynomial.
Interval of Convergence
We again consider the infinite Taylor series for f ( x ) 5 ln( x ) centered at x 5 1. ln( x ) 5 a (21)n21 ? (x 2 1)n n n51
`
Lets examine the first five terms of the series. 1 1 1 1 ln( x ) < ( x 2 1 ) 2 ( x 2 1 ) 2 1 ( x 2 1 ) 3 2 ( x 2 1 ) 4 1 ( x 2 1 ) 5 2 3 4 5 Now lets see what happens to this sum when x 5 1. 1 1 1 1 (1 2 1) 2 (1 2 1)2 1 (1 2 1)3 2 (1 2 1)4 1 (1 2 1)5 5 0 2 3 4 5 It is clear that when x 5 1, the series converges to 0. This is because the polynomial is approximating the value of f ( x ) 5 ln( x ) at x 5 1. That is, ln( 1 ) 5 0 What happens when x 5 1.5? Lets look at the cumulative sum of the terms when x 5 1.5 and compare it to the value of the function f ( x ) 5 ln( x ) when x 5 1.5 (Table 11.8).
TABLE 11.8
Number of Terms 1 2 3 4 5
ln( 1.5 ) < 0.405 ( 1.5 2 1 ) 1 ( 1.5 2 1 ) 2 ( 1.5 2 1 ) 2 2 1 1 ( 1.5 2 1 ) 2 ( 1.5 2 1 ) 2 1 ( 1.5 2 1 ) 3 2 3 1 1 1 ( 1.5 2 1 ) 2 ( 1.5 2 1 ) 2 1 ( 1.5 2 1 ) 3 2 ( 1.5 2 1 ) 4 2 3 4 1 1 1 1 ( 1.5 2 1 ) 2 ( 1.5 2 1 ) 2 1 ( 1.5 2 1 ) 3 2 ( 1.5 2 1 ) 4 1 ( 1.5 2 1 ) 5 2 3 4 5
Cumulative Sum 0.5 0.375 0.416 0.401 0.407
766
CHAPTER 11
We see that when x 5 1.5, it seems that the series converges to approximately 0.4, as it should. Lets see what happens when x 5 3 (Table 11.9).
TABLE 11.9
Number of Terms 1 2 3 4 5
ln( 3 ) < 1.10 (3 2 1) 1 (3 2 1) 2 (3 2 1)2 2 1 1 (3 2 1) 2 (3 2 1)2 1 (3 2 1)3 2 3 1 1 1 (3 2 1) 2 (3 2 1)2 1 (3 2 1)3 2 (3 2 1)4 2 3 4 1 1 1 1 (3 2 1) 2 (3 2 1)2 1 (3 2 1)3 2 (3 2 1)4 1 (3 2 1)5 2 3 4 5
Sum 2 0 2.6 2 1.3 5.06
It appears that the series does not converge to approximately 1.10 when x 5 3. In Table 11.10, we can compare the graphs of higher- and higher-order polynomials with the graph of f ( x ) 5 ln( x ) to see if we can tell for which values of x the Taylor series will converge to the function values f ( x ) , and for which values it will diverge.
TABLE 11.10
5th-degree polynomial
y 2 1 0 1 2 3 4
y 2 1 0 1 2 3 4
y 2 1 0 1 2 3 4
y 2 1 0 1 2 3 4
y 2 1 0 1 2 3 4
y 2 1 0 1 2 3 4
11.4 Taylor Series
767
We see that even with higher- and higher-degree polynomials, it appears as though the polynomial will converge only when 0 , x , 2. The interval over which a Taylor series converges is known as the interval of convergence. We can determine the interval of convergence for a series by using a symbolic method called the ratio test.
RATIO TEST
For an infinite series for which the limit L 5 lim `

nS`
an 1 1 ` an
exists, the series converges if L , 1 and diverges if L . 1 or if the limit is infinite.
Just In Time
Solving Absolute Value Inequalities
Consider the absolute value inequality 0 2x 1 1 0 , 5. The graph below shows the values of x that make the inequality true.
y 8 6 4 2 4 2 2 2 4 x
Note that Z 2x 1 1 Z , 5 when 2 3 , x , 2. To solve the inequality algebraically, do the following. 0 2x 1 1 0 , 5 2x 1 1 , 5 or 2x , 4 or x , 2 or We write the solution as 2 3 , x , 2. 2x 1 1 . 2 5 2x . 2 6 x . 23
EXAMPLE
Finding the Interval of Convergence for an Infinite Taylor Series

Find the interval of convergence for the infinite Taylor series ln( x ) 5 ` n21 n (21) ? (x 2 1) . a n
n51
768
CHAPTER 11
SOLUTION First, we write the quotient `
(21)n1121 ? (x 2 1)n11 a n11 ` n11 ` 5 n 2 an (21) 1 ? (x 2 1)n n (21)n ? (x 2 1)n11 n11 5 (21)n21 ? (x 2 1)n n 5` 5` 5`
an 1 1 ` and simplify. an
In the numerator, we replace each n with n 1 1.
Simplify n 1 1 2 1 5 n.
(21)n ? (x 2 1)n11 n ? ` n11 (21)n21 ? (x 2 1)n (21) ? (x 2 1) n ? ` n11 1 n ? (2x 1 1) ` n11 lim ` n ? (2x 1 1) ` 5 0 1 ? (2x 1 1) 0 n11
n 11
Invert the denominator and multiply.
Simplify using the laws of exponents.
Next, we take the limit as n S ` .

nS`
To obtain this limit, we recognize that as n S ` , n to converge, L 5 lim `

nS`
S 1. Now, for the series
an 1 1 ` must be less than 1. an Z1 ? (2x 1 1) Z , 1 Z(2x 1 1) Z , 1
2x 1 1 , 1 2x , 0 x.0
or or or
2x 1 1 . 21 2x . 22 x,2
0,x,2 As predicted through the numerical and graphical analysis, the Taylor series converges to the function f ( x ) 5 ln( x ) only when 0 , x , 2. Therefore, this is the interval of convergence.
EXAMPLE
Finding the Interval of Convergence for a Taylor Series

Find the interval of convergence for the infinite Taylor series sin( x ) 5 ` n21 2n 2 1 (21) ?x a ( 2n 2 1 ) ! n51
11.4 Taylor Series
769
SOLUTION First, we write the quotient `
an 1 1 ` and simplify. an
( 2 1 ) n 1 11 ? x 2(n 1 1) 2 1 a 32( n 1 1 ) 2 14 ! ` n11 ` 5 an ( 2 1 ) n 2 1 ? x 2n 2 1 ( 2n 2 1 ) ! ( 2 1 ) n ? x 2n 1 1 ( 2n 1 1 ) ! 5 ( 2 1 ) n 2 1 ? x 2n 2 1 ( 2n 2 1 ) ! 5 ` 5 ` 5 ` 5 ` ( 2 1 ) n ? x 2n 1 1 ( 2n 2 1 ) ! ? ` ( 2n 1 1 ) ! ( 2 1 ) n 2 1 ? x 2n 2 1 ( 2 1 ) n ? x 2n 1 1 ( 2n 2 1 ) ! ? ` ( 2n 1 1 )( 2n )( 2n 2 1 ) ! ( 2 1 ) n 2 1 ? x 2n 2 1 ( 2 1 ) ? x2 ` ( 2n 1 1 )( 2n ) 2 x2 ` ( 2n 1 1 )( 2n ) lim ` 2 x2 ` 50 ( 2n 1 1 )( 2n )

1 ( 2n 1 1 )( 2n )
Next, we take the limit as n S ` .

nS`
To obtain this limit, we recognize that as n S ` , for all values of x because lim ` ( 2n
nS`
S 0 regardless of
the value of x. Therefore, this Taylor series converges to the function f ( x ) 5 sin( x )
2 x2 1 1 )( 2n )
` 5 0, which is less than 1 as required
by the ratio test. We can confirm this by looking at the graphs of higher- and higher-order Taylor polynomials. Table 11.11 shows that as the number of terms in the polynomial increases, the polynomial converges to the function f ( x ) 5 sin( x ) over a larger and larger interval.
TABLE 11.11
y 1 0.5 6 4 2 0.5 1 2 4 6 x
y 1 0.5 6 4 2 0.5 1 2 4 6 x
y 1 0.5 6 4 2 0.5 1 2 4 6 x
(Continued)
770
CHAPTER 11
y 1 0.5 6 4 2 0.5 1 2 4 6 x
y 1 0.5 6 4 2 0.5 1 2 4 6 x
As we can see by this sequence of graphs, higher-degree polynomials approximate the function better. Unlike in Example 3, it appears as though the infinite Taylor series will indeed converge for all values of x. We can say that the interval of convergence is ( 2` , ` ) .
Applications to Derivatives and Integrals

Taylor polynomials can be used to take derivatives and integrals in otherwise difficult or impossible situations.
EXAMPLE
Using Taylor Polynomials to Approximate a Derivative

Using Taylor polynomial approximations, show that
d 3sin( x ) 4 dx
5 cos( x ) .
SOLUTION When we take the derivatives of the first several terms of the Taylor polynomial that approximates the sine function, we obtain the first several terms of the Taylor polynomial that approximates the cosine function.
sin( x ) < x 2
1 3 1 1 1 1 11 x 1 x5 2 x7 1 x9 2 x 3! 5! 7! 9! 11!
d 3 5 7 9 11 10 3sin( x ) 4 < 1 2 x 2 1 x 4 2 x 6 1 x 8 2 x dx 3! 5! 7! 9! 11! cos( x ) < 1 2 1 2 1 1 1 1 10 x 1 x4 2 x6 1 x8 2 x 2! 4! 6! 8! 10!
We see that, indeed, the derivative of the Taylor polynomial that approximates the sine function is the Taylor polynomial that approximates the cosine function.
EXAMPLE
Using a Taylor Polynomial to Approximate an Integral

Use Taylor polynomial approximations to find the integral 3e x dx.
2
11.4 Taylor Series
771
SOLUTION We first need to find a Taylor polynomial that approximates the function f ( x ) 5 e x . To do so, we start with the Taylor polynomial approximation for e x.
2
` xn 2 1 ex 5 a n51 (n 2 1)!
1 1 1 1 5 e x < 1 1 x 1 x2 1 x3 1 x4 1 x 2 6 24 120 To obtain the Taylor polynomial for f ( x ) 5 e x , we substitute x2 for x.

2
1 1 1 1 e x < 1 1 ( x2 ) 1 ( x2 ) 2 1 ( x2 ) 3 1 ( x2 ) 4 1 ( x2 ) 5 2 6 24 120
2
1 1 1 1 10 e x < 1 1 x2 1 x4 1 x6 1 x8 1 x 2 6 24 120
2
` x2n 2 2 ex 5 a n51 (n 2 1)!

2
Now we integrate the first several terms of the Taylor polynomial. 1 4 1 6 1 8 1 10 x 2 3e dx < 3 a 1 1 x 1 2 x 1 6 x 1 24 x 1 120 x b dx
2
1 1 1 1 9 1 11 < x 1 x3 1 x5 1 x7 1 x 1 x 3 10 42 216 1320 <

2
x 1 3 1 5 1 7 1 9 1 1 x 1 x 1 x 1 x 1 x11 0! ? 1 1! ? 3 2! ? 5 3! ? 7 4! ? 9 5! ? 11
` x2n 2 1 x dx 5 e a 3 n 5 1 ( n 2 1 ) !( 2n 2 1 )
While we are unable to find an elementary function for the antiderivative 3e x dx, we are able to express this antiderivative using Taylor polynomials and Taylor series.
2
11.4 Summary
In this section, you learned how to create an infinite Taylor series for a given function. You learned that these series can converge for some values of x and diverge for others. You learned how to use the ratio test to find the interval of convergence for a Taylor series. Finally, you learned how Taylor polynomials and Taylor series can be used in differential and integral calculus.
772
CHAPTER 11

In Exercises 15, use the graph provided to solve the given absolute value inequality.
y 8 6 4 2 4 2 2 2 4 6 8 10 12 x
5.
1. 0 x 2 4 0 , 8
1 0 3x 2 1 0 2 5 . 0 4
y 2 1 8 6 4 2 1 2 3 4 5 2 4 6 8 x
2. 0 4x 1 2 0 2 1 . 9
y 10 8 6 4 2 1 2 3 4 x
In Exercises 610, use algebraic strategies to solve the absolute value inequality. 6. Z x 2 4 Z , 8 7. Z 4x 1 2 Z 2 1 . 9 8. 2Z x Z 1 5 # 2 9. 2 Z x 1 2 Z $ 5 10. 1 Z 3x 2 1 Z 2 5 . 0 4
4 3 2 1 2
3. 2 Z x Z 1 5 # 2
y 6 4 2 2 4 6 8 x
11. Write an absolute value inequality that has no solution. Show what happens graphically and algebraically when an absolute value inequality has no solution. 12. Write an absolute value inequality that has infinitely many solutions. Show what happens graphically and algebraically when an absolute value inequality has infinitely many solutions.
8 6 4 2 2 4 6
4. 2 0 x 1 2 0 $ 5
y 7 6 5 4 3 2 1 1 2 x
6 5 4 3 2 11
11.4 Taylor Series
773
11.4 Exercises
Exercises 12 are designed to help you conceptualize the notion of Taylor Series. 1. Suppose a fellow student was absent on the day on which the concept of the interval of convergence was studied. Describe what is meant by the interval of convergence. 2. Describe how the interval of convergence is determined. In Exercises 310, use sigma notation to write the infinite Taylor series centered at x 5 a. 3. f ( x ) 5 cos( x ) at x 5 0 4. f ( x ) 5 x sin( x ) at x 5 0 5. f ( x ) 5 "x at x 5 1 6. f ( x ) 5 e3x at x 5 0 26. What do we mean when we say that the interval of convergence for a particular Taylor series is (2 `,` )? 27. Near the end of the ratio-test process, you conclude that Z x 2 3 Z , 1. What would be the interval of convergence in this situation? 28. The interval of convergence for the infinite Taylor series
` n51
In Exercises 2125, use Taylor polynomial approximations and Taylor series to integrate the function. 21. 3e3xdx 23. 3ln( x ) dx 25. 3e x dx
2
22. 3cos( 2x ) dx 24. 3x sin( x ) dx
7. f ( x ) 5 ln( 3x 1 1 ) at x 5 0 8. f ( x ) 5 "x 1 1 at x 5 0 9. f ( x ) 5 e x at x 5 0
2
10. f ( x ) 5 cos( 2x ) at x 5 0 In Exercises 1118, determine the interval of convergence for the infinite Taylor series. 11. f ( x ) 5 cos( x ) at x 5 0 12. f ( x ) 5 x sin( x ) at x 5 0 13. f ( x ) 5 "x at x 5 1 14. f ( x ) 5 e3x at x 5 0
(21)
n21
? (x 2 1) 5 ln( x ) n
is 0 , x , 2. Without using the ratio test, find the interval of convergence for the function f ( x ) 5 ln( x 1 1 ) . Explain how you are able to determine this interval without using the ratio test. 29. There is no elementary function that is the antiderivative of the function f ( x ) 5 e x . Explain how Taylor series can be used to find an antiderivative of f ( x ) 5 e x .
2 2
15. f ( x ) 5 ln( 3x 1 1 ) at x 5 0 16. f ( x ) 5 "x 1 1 at x 5 0 17. f ( x ) 5 e x at x 5 0

2
18. f ( x ) 5 cos( 2x ) at x 5 0 In Exercises 1920, use Taylor polynomial approximations and Taylor series to show that the equations are valid. 19. d x 3e 4 5 e x dx 20. 1 d 3ln( x ) 4 5 x dx
30. Explain why each of the following limits is true. 1 50 (a) lim nS `(n 1 1) n2 5` (b) lim nS `(n 1 1) n 51 (c) lim nS `(n 1 1)
774
CHAPTER 11
32. Construct a Taylor series for the function g( x ) 5 cosh( x ) centered at x 5 0. For these exercises, you will use the definitions of the hyperbolic functions. The hyperbolic sine function is defined as f ( x ) 5 sinh( x ) 5 cosh( x ) 5
e x 1 e2x . The 2 e x 2 e2x . 2
33. Construct a Taylor series for the function h( x ) 5 tanh( x ) centered at x 5 0. 34. Use your Taylor series from Exercises 31 and 32 to show that dx 3sinh( x ) 4 5 cosh( x ) .
d
The hyperbolic cosine function is defined as g( x ) 5 hyperbolic tangent function is

sinh( x ) cosh( x ) e x 2 e2x . e x 1 e2x
35. Use your Taylor series from Exercises 31 and 32 to show that dx 3cosh( x ) 4 5 sinh( x ) .
d
defined as h( x ) 5 tanh( x ) 5
31. Construct a Taylor series for the function f ( x ) 5 sinh( x ) centered at x 5 0.
After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What is a sequence, and what does it mean when we say that a sequence converges? (11.1) 2. What is the difference between a geometric sequence and an arithmetic sequence? (11.1) 3. What is the difference between a sequence and a series? (11.2) 4. What does it mean when we say that we can find the sum of an infinite series? (11.2) 5. How can we find the sum of an infinite geometric series? (11.2) 6. What are we trying to accomplish in creating a Taylor polynomial? (11.3) 7. What does it mean to construct a Taylor polynomial centered at x 5 a? (11.3) 8. How are higher-order derivatives used to construct a Taylor polynomial? (11.3) 9. What is the difference between a Taylor polynomial and a Taylor series? (11.4) 10. What does it mean when we say that a Taylor series converges? (11.4) 11. How is the ratio test used to determine the interval of convergence for a Taylor series? (11.4) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
Definitions

Sequence: A set of numbers given in a particular order Maintenance Level: When a sequence reaches a limit as the number of terms tends to infinity
775
Term of a Sequence: Each number in a sequence Recursive Formula: A sequence in which all terms prior to the desired term must be computed Explicit Formula: A sequence in which each term can be computed directly without computing the value of the previous term Geometric Sequence: A sequence in which each subsequent term is found by multiplying the previous term by a common ratio Arithmetic Sequence: A sequence in which each subsequent term is found by adding a common difference to the previous term Geometric Series: A sum in which each term is found by multiplying the previous term by a constant multiple Taylor Polynomial: A polynomial that approximates another function and is written as the sum of n terms of the polynomial Infinite Taylor Series: A series that represents, in general, the terms of a Taylor polynomial Interval of Convergence: The interval over which a Taylor series converges
Convergence of Sequence: A sequence a1, a2, a3, c , an, c converges (or reaches a maintenance level) if an S L as n S ` . That is, if lim an 5 L,
nS`
we say that the sequence converges. Otherwise, we say that the sequence diverges. an 1 1 ` Ratio Test: For an infinite series for which the limit L 5 lim ` n S ` an exists, the series converges if L , 1 and diverges if L . 1 or if the limit is infinite.
Formulas
Finite Geometric Series: A finite geometric series with n terms can be expressed as a 1 ar 1 ar2 1 ar3 1 c1 arn 2 1, where a is the first term of the series and r is the common ratio or constant multiple. Sum of a Finite Geometric Series: The sum of the first n terms of a finite geometric series with first term a and common ratio r is a 2 arn Sn 5 a 1 ar 1 ar2 1 c1 arn 2 1 5 , provided r 2 1. 12r Sum of an Infinite Geometric Series: The sum of the first n terms of an infinite geometric series Sn 5 a 1 ar 1 ar2 1 c1 arn 2 1 will converge to S 5 1
a 2r
as n S ` provided 2 1 , r , 1. Otherwise, the
series will diverge. Terms of a Taylor Polynomial Centered at x 5 0: The nth term of the Taylor polynomial Pm( x ) 5 C0 1 C1x 1 C2x 2 1 C3x3 1 c1 Cm xm f (n)( 0 ) n bx . n! Terms of a Taylor Polynomial Centered at x 5 a: The nth term of the Taylor polynomial centered at x 5 a is given by f (n)( a ) Tn 5 Cn( x 2 a ) n 5 a b ( x 2 a ) n. n! centered at x 5 0 is given by Tn 5 Cn xn 5 a
776
CHAPTER 11

Section 11.1 In Exercises 14: (a) List the first six terms of the given sequence. (b) Use a graphing calculator to graph the sequence. (c) State whether the sequence appears to converge or diverge.
1. an 5 n2 1 1 n 2. an 5 10n 3n bloodstream every 12 hours. A typical prescription calls for the patient to take one 500-mg tablet every 12 hours. (Source: www.drugs.com.) (a) Generate the first five terms of the sequence representing the amount of medicine in the bloodstream every 12 hours. (b) What is the maintenance level in this situation? (c) How long does it take to reach the maintenance level? Confirm your answer graphically.
3. an 5 0.5an 2 1 1 2
4. an 5 1.5an 2 1 1 1
In Exercises 58, find a formula for the general term of the sequence, an, and state whether the sequence converges. Assume that the given pattern continues. 1 5. 2 2, 1, 4, 7, c 6. 18, 6, 2, , c 3 7. 2 4 6 8 , , , ,c 3 5 7 9 9 27 81 8. 2, 3, , , , c 2 4 8
Section 11.2 In Exercises 1114, determine if the series is arithmetic, geometric, or neither. Justify your answer.
11. a 2 1 3n
n51 15 10 5 1 n21 12. a 10 a b 14 n51 5
In Exercises 910, apply your knowledge of sequences to answer the questions. 9.

Heartworm Medicine Dog owners give
13. a n
n51
14. a 2n2 1 1
n51
their dogs heartworm medicine once every month to prevent heartworm disease. One type of heartworm medicine, Interceptor Flavor Tabs, contains 11.5 mg of medicine and has a half-life of about 36 hours. (Source: www.beaconforhealth.org.) (a) Generate the first five terms of the sequence representing the amount of heartworm medicine in the bloodstream every 36 hours. (b) Create an explicit formula representing the amount of medicine in the bloodstream at a given time. (c) Estimate the amount of medicine in the bloodstream at the end of 2 weeks. (d) How much medicine will be in the bloodstream when the next dose is administered one month later? Confirm your answer graphically. 10.
Anti-Inflammatory Medicine Naproxen
In Exercises 1518, express the repeating decimal as a rational number. 15. 0.1111 c 17. 0.1818 c 16. 0.123123 c 18. 0.231 c
In Exercises 1920, apply what you have learned about series to solve the problem. 19.
Pharmaceuticals Suppose that a
maintenance level of 871.1 mg of Naproxen in the bloodstream has been reached. If the person stops taking the medicine at this time, how long will it be before the medicine is eliminated from the bloodstream? The half-life of Naproxen is 15 hours. (Source: www.drugs.com.) 20.
Pharmaceuticals Suppose that a
is an anti-inflammatory medicine often used by patients with arthritis to reduce swelling in joints. With a half-life of 15 hours, about 42.6 percent of the medicine is eliminated from the
maintenance level of 10 mg of Singulair in the bloodstream has been reached. If the person stops taking the medicine at this time, how long will it be before the medicine is eliminated from the bloodstream? The half-life of Singulair is 4 hours. (Source: www.drugs.com.)
777
Section 11.3 In Exercises 2125, write the first four nonzero terms of the Taylor polynomial centered at the given value, x 5 a. Confirm your work by graphing your polynomial along with the given function.
21. f ( x ) 5 sin( x ) at x 5 22. f ( x ) 5 e x at x 5 1 23. f ( x ) 5 e x at x 5 1
2
Section 11.4 In Exercises 2628, find the interval of convergence for the series.
` ( 2n ) ! n 26. a 2 ? x n 5 1 ( n! ) ` (21)n n 27. a ?x n51 n 1 2
p 4
28. a n2( x 2 1 ) n
n51
24. f ( x ) 5 ln( x ) at x 5 2 25. f ( x ) 5 1 at x 5 3 x22
In Exercises 2930, use Taylor polynomial approximations and Taylor series to integrate the function. ex 29. 3 dx x
1
30. 3
0
P R O J E C T
"x 1 1
2
dx
1 1
P R O J E C T
1 1
What to do
1. Choose a drug that you are interested in researching. Determine the half-life of your drug. 2. Write the general term of the sequence describing the amount of medicine in the bloodstream as time passes for your chosen drug. 3. Write the general term of the geometric series describing the amount of medicine in the bloodstream as time passes for your chosen drug. 4. Using graphs, tables, and symbols, provide a full analysis of your drug, including its maintenance level. 5. Describe to a prospective patient the effects of missing a dose. That is, describe several different situations in which a dose is missed and a double dose is taken the next time the medication is to be taken. Does the amount of medicine in the bloodstream drastically surpass the maintenance level? 6. Write an analysis of your chosen drug that could be published on a web page.

Data for this project can be found at the following sites, among others. For half-life data, look under Metabolism, Pharmacokinetics, or Elimination in the technical information section. www.drugs.com www.rxmed.com www.astrazeneca-us.com At these sites, you will also find the type of information that is included for the consumer.
778
12
Chapter
Probability and Calculus

I
n November of 2008, American voters will elect a new president of the United States. Candidates running for this office, like candidates running for most high-profile offices, maintain large campaign staffs that raise money, provide campaign strategies, and monitor polls. One important responsibility of a campaign manager is to monitor the polls and statistics concerning the attitudes and beliefs of the American public. Further, campaign managers work to understand the demographics of American voters. Who are the voters? What are their ages? What are the most important issues to most of these voters? Are there more male or female voters? To which major political party do most voters belong? By finding answers to these questions, the campaign staff and the candidate can develop a platform, write relevant speeches, and create a campaign strategy. In this chapter, we explore how probability and calculus may be used to answer questions such as those posed here.
12.1 Continuous Probability Models

Determine if a function represents a probability density function Construct a probability density function from real-life data Use a probability density function to find the probability of the occurrence of an event
12.2 Uniform and Exponential Distributions
Construct uniform and exponential probability density functions given the appropriate information Given a real-life situation that may be modeled using a uniform or an exponential probability density function, find the probability of the occurrence of an event
12.3 Expected Value, Variance, and Standard Deviation
Compute and interpret the expected value, variance, and standard deviation of discrete and continuous random variables Compute and interpret formulas for the expected value, variance, and standard deviation of uniform and exponential probability density functions Use a normal probability density function to find the probability of the occurrence of an event Given a probability, find the range of values of a random variable that produces that probability
12.4 Normal Distributions
779
780
CHAPTER 12
Determine if a function represents a probability density function Construct a probability density function from reallife data Use a probability density function to find the probability of the occurrence of an event
Politicians are interested in knowing voting statistics so that they can plan election and re-election campaigns. By analyzing voting data, politicians can better understand who is voting and how they are voting. Knowing this information, they can target their efforts toward appropriate groups. In this section, we will investigate how continuous probability models can help us forecast human behavior, including voting patterns.
Analysis of data from the 2004 presidential election can provide us with insight about the ages of voters. Table 12.1 shows the number of people, by age, who voted in the 2004 election. The table also shows the percentage of voters in each age range. These values were found by taking the number of voters in a particular categoryfor example, 2.4 million in the 1820 age rangeand dividing by the total number of voters. For the purpose of the model, we grouped all voters over 64 years old into the 6580 group.
TABLE 12.1
2004 Presidential Election Age Range 1820 2124 2534 3544 4564 6580 Number of Voters (millions) 2.4 3.6 10.2 15.7 34.4 18.9 Percentage of Voters 2.81% 4.26% 11.94% 18.48% 40.36% 22.15%
The number of voters, in millions, can take on a number value X for any age range, up to the number of people registered to vote in that age range. The value of X in any given election is random. That is, we cannot say with certainty how many people from any age range will vote in any given election. We call such a variable a random variable. If a random variable takes on only integer values, we call it a discrete random variable. If a random variable can take on real number values, we call it a continuous random variable.
CONTINUOUS RANDOM VARIABLE
A continuous random variable is a variable that can take on any value in some interval of real numbers.
We can create a visual of the voting situation by constructing a bar graph (Figure 12.1).
781
0.45 Fraction of total number of voters 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 1820 2124 2534 35 44 4564 6580 Age range
FIGURE 12.1
Another way to represent this information is to create a histogram in which each vertical bar is extended so that the area of the resulting rectangle represents the percentage of voters in that age group (Figure 12.2). We accomplish this by taking the percentage of voters for each age range and dividing by the number of years in that age range, as shown in Table 12.2.
2.2 2.0 1.8 Fraction of voters by age 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1820 2124 2534 35 44 4564 Age range 6580
360080_la_12_002
2004 Presidential Election Age Range 1820 2124 2534 3544 4564 6580 Percentage of Voters 2.81% 4.26% 11.94% 18.48% 40.36% 22.15% Width of Histogram Bar 3 4 10 10 20 16 Height of Histogram Bar
2.81%> 3 5 0.937% 11.94%> 10 5 1.194% 4.26%> 4 5 1.065%
22.15%> 16 5 1.384%
40.36%> 20 5 2.018%
18.48%> 10 5 1.848%
782
CHAPTER 12
The height of the vertical bar represents the average fraction of voters by age. For example, 11.94 percent of voters were in the 2534 age range. By dividing by 10 (there are 10 ages in that age range), we are saying that 1.194 percent of the voters were 25, 1.194 percent were 26, and so on. Most importantly, the area of the rectangle represents the total percentage of voters in any particular age range. This histogram enables us to use geometric intuition to determine the percentage of voters who were in a given age range. By computing the area of the rectangle representing a particular age group, we can find this percentage.
EXAMPLE
Using a Histogram
In the 2004 presidential election, what percentage of the voters were younger than age 45?
SOLUTION We sum the areas of the first four rectangles in order to find the percentage of voters younger than age 45 (Figure 12.3).
2.2 2.0 1.8 Fraction of voters by age 1.6 1.4 1.2 1.0 4.26% 11.94% 0.8 2.81% 0.6 0.4 0.2 0.0 1820 2124 2534 35 44 4564 Age range 6580 18.48%
FIGURE 12.3
2.81% 1 4.26% 1 11.94% 1 18.48% 5 37.49% In 2004, approximately 37.5 percent of all voters were under age 45.
EXAMPLE
Using a Histogram
In the 2004 presidential election, what percentage of the voters were teenagers?
SOLUTION If we assume that the percentage of voters was distributed evenly over the 1820 age range, we can say that the area of two-thirds of the first rectangle represents the number of voters who are 18 or 19 years old (teenagers). See Figure 12.4.
783
360080_la_12_004
FIGURE 12.4
We can find the area of two-thirds of the first rectangle by taking the entire area and multiplying it by . 3 2.81% ? 2 5 1.873% 3
2
In the 2004 presidential election, about 1.9 percent of all voters were teenagers.
Density Function
Imagine that rather than knowing the number of voters (and therefore the percentage, or fraction, of voters) in a particular age range, we are given the number of voters for smaller and smaller age ranges. We can visualize what this might look like by connecting the midpoints of the tops of the bars in the previous histogram (see Figure 12.5).
FIGURE 12.5
784
CHAPTER 12
If we smooth out this function and extend it to the endpoints of our interval, we have what is called a density function. The area under this density function for any range of ages tells us the fraction of voters in the 2004 presidential election in that age range (Figure 12.6).
2.2 2.0 1.8 Fraction of voters by age 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 18 20 25 35 44 Age 64 80
FIGURE 12.6
Although it will be a rough estimate, we can use a graphing calculator to find a regression equation to model this density function.
EXAMPLE 3
Using a Regression Model to Approximate a Density Function

Find a regression model to approximate the density function in Figure 12.6 showing the fraction of voters by age.
SOLUTION We use the regression feature of a graphing calculator to find a cubic regression model for the points determined by the midpoints of each interval. The data table for these points is shown in Figure 12.7. We chose a cubic model because it appeared to approximate the data well.
FIGURE 12.7
785
The cubic function that best models the data is V( x ) 5 2 0.0000258 x 3 1 0.002452 x 2 2 0.0351 x 1 0.881 percent of voters, where x is the age of the individual.
EXAMPLE
Using a Regression Model to Forecast Voter Distribution

Use the cubic regression model from Example 3 to estimate the percentage of voters in the 2004 presidential election who were younger than age 45. Then determine the percentage of voters who were teenagers. Compare your results with the results obtained previously.
SOLUTION We found earlier that the area under the density curve was used to represent the percentage of voters in a particular age range. If we compute the value of the integral over the interval 1845, we will have determined the percentage of voters younger than age 45 (Figure 12.8).
45
3 V( x ) dx < 37.8
18
FIGURE 12.8
About 37.8 percent of all voters were younger than age 45. Previously, we found that 37.5 percent of all voters were younger than age 45. The discrepancy can be attributed to the error in the model used to approximate the fraction of voters by age. If we compute the value of the integral over the interval 1820, we will have determined the percentage of voters who were younger than 20 (teenagers). See Figure 12.9.
20
3 V( x ) dx < 1.8
18
About 1.8% of all voters were younger than age 20. Previously, we found that 1.9% of all voters were teenagers.
FIGURE 12.9
Despite the imperfection of our density model, it appears to do a remarkably good job of estimating the distribution of voters.
Just In Time
Computing Probabilities
Recall the basic principle of probability. Let a sample space S consist of equally likely outcomes. Let event E be a subset of S. The probability that event E will occur is given by P( E ) 5 n( E ) n( S )
For any event, 0 # P( E ) # 1. Also, if the probability of event E occuring is P( E ) 5

n( E ) , n( S )
then the probability of event E not occurring is 1 2 P( E ) .
786
CHAPTER 12
Probability Density Function

We will now focus our attention on what is referred to as a probability density function. Rather than work to create a density function that represents the fraction of voters by age, we will start with the original data showing the number of voters in each age range and approximate a probability density function. A function that describes a continuous probability distribution is called a probability density function. For the data we have been investigating, the continuous probability distribution will be in terms of the percentage of voters by age. As we have seen, the area under the graph of this function over a given interval will represent the percentage of voters in that age range. Since this percentage is a random variable that can take on values between 0 and 1 (0 percent to 100 percent), the total area under the probability density function must be 1 (e.g., the percentage, or probability, of voters between ages 18 and 80 must be 1). Furthermore, the probability density function must always have function values greater than 0 (i.e., no negative percentages).
PROBABILITY DENSITY FUNCTION
A function P is a probability density function of a random variable x over the interval 3a, b4 if P( x ) $ 0 for all x in 3a, b4 , and if 3 P( x ) dx 5 1.
a b
We can create a probability density function for the data showing the number of voters in given age ranges for the 2004 presidential election. We will take the midpoint of each age range and say that the number of voters per year of age is the corresponding value given previously, as shown in Table 12.3.
TABLE 12.3
2004 Presidential Election Midpoint of Age Range 19 22.5 29.5 39.5 54.5 72.5 Number of Voters (millions) 2.4 3.6 10.2 15.7 34.4 18.9 Number of Voters per Year of Age (millions) 2.4 5 0.8 3 3.6 5 0.9 4 10.2 5 1.02 10 15.7 5 1.57 10 34.4 5 1.72 20 18.9 5 1.18 16
Age Range 1820 2124 2534 3544 4564 6580
Years in Range 3 4 10 10 20 16
787
We next find a regression equation to model the number of voters per year of age as a function of age. We can use the graphing calculator to do this (Figure 12.10).
FIGURE 12.10
The equation is V( x ) 5 2 0.0000221x 3 1 0.00211x 2 2 0.0307x 1 0.760. This function, however, is not a probability density function. Note that the area under this curve on the interval 318, 804 is not equal to 1. In fact, the area under this curve can be interpreted as the total number of voters because the product of the vertical units (number of voters in millions per age in years) and the horizontal units (age in years) is given in units of million voters (Figure 12.11).
80
3 V( x ) dx < 83.2 million voters

FIGURE 12.11
18
We estimate that 83.2 million voters younger than 80 years old voted in the 2004 presidential election. This model is not 100 percent accurate, but it does give us a general idea of what is happening. (The Statistical Abstract of the United States, 2006, Table 405, p. 263 indicates that the total number of voters in the 2004 presidential election was actually about 85 million voters.) However, we can use this model to approximate a probability density function and approximate the percentage (probability) of voters for different age ranges. To obtain this probability density function, we need to adjust the function V( x ) so that the total area under the curve is 1. That is, we need to find the value of c such that the area under the curve over the interval 318, 804 is 1.
80
3 c ? V( x ) dx 5 c ? 83.21075 5 1
18
Solving for c gives c 5
1 . 83.21075
By dividing V( x ) by the total number of voters,
83.21075, we are finding the percentage of voters in a given age range.
788
CHAPTER 12
EXAMPLE
Showing that a Function Is a Probability Density Function

Show that the function y 5 c ? V( x ) 5 ( 2 0.0000221x3 1 0.00211x2 2 83.21075 0.0307x 1 0.760 ) , which gives the percentage of voters of age x, is a probability density function.
SOLUTION A probability density function must have positive function values for all x in the interval 318, 804 . A graph of the function shows this to be the case (Figure 12.12). A probability density function also must have a total area under the curve of 1 over the given interval (Figure 12.13). Again, a graph verifies this area.
1
FIGURE 12.12
FIGURE 12.13
80
3 c ? V( x ) dx 5 1
18
EXAMPLE
Using a Probability Density Function

Use the probability density function c ? V( x ) from Example 5 to find the percentage of voters who were under 45 years of age. Then find the percentage of voters who were teenagers. Compare your results with those found previously.
SOLUTION To find the percentage of voters who were under 45 years of age,
45
we must compute the integral 3 c ? V( x ) dx (Figure 12.14).

18 45
3 cV( x ) dx < 0.387

FIGURE 12.14
18
Approximately 38.7 percent of all voters were under age 45. This is reasonably close to the estimates found previously. To find the percentage of voters who were teenagers, we must compute the
20
integral 3 c ? V( x ) dx (Figure 12.15).

18 20
3 c ? V( x ) dx < 0.019
18
FIGURE 12.15
Approximately 1.9% of all voters were teenagers. This is very close to the estimates found previously.
789
We have seen that we can create probability density functions from a data set. We have also seen that the area under a probability density function over a particular interval represents the probability that a random variable will be within that interval. The process used to find such probabilities can be summarized in the following steps.
Create a Probability Density Function To create a probability density function, do the following: 1. Create a scatter plot of the data showing the frequencies of outcomes of an experiment. Often, the data will involve intervals. Use the midpoint of each interval as the data point for that interval. 2. Determine a regression model P( x ) for the data. 3. Adjust the function P( x ) by finding the value of c such that 3 c ? P( x ) dx 5 1 over the desired interval 3a, b4 . 4. To compute probabilities, use the function c ? P( x ) to find the area under the function over desired subintervals on 3a, b4 . We will apply these steps as we explore another application in the next example.
a b
EXAMPLE
Creating a Probability Density Function

Imagine that you surveyed a group of 100 people and asked them to consider the last time they visited a local restaurant on a Friday night. You asked them to indicate how long they had to wait for a table on that previous visit. They could choose a wait time of 0 to 45 minutes, given in 5-minute increments. The results of your survey are given in Table 12.4.
TABLE 12.4
Wait Time (minutes) Frequency
0 3
5 5
10 8
15 12
20 18
25 22
30 15
35 6
40 7
45 4
Total: 100 Find a probability density function that describes the probability that a person will have to wait for a given amount of time. Then find the probability that someone will have to wait for less than 10 minutes. Find the probability that someone will have to wait for at least 20 but less than 30 minutes.
SOLUTION We begin by creating a scatter plot of the data. We plot frequency as a function of wait time so that we can use the graphing calculator to find a regression equation to model the data (Figure 12.16).
790
CHAPTER 12
FIGURE 12.16
We will use a quartic regression model to approximate these data (Figure 12.17).
FIGURE 12.17
We will call the regression model W( x ) . To obtain a probability density function, we need to adjust the function W( x ) so that the total area under the curve is 1. That is, we need to find the value of c such that the area below the curve on the interval 30, 454 is 1 (Figure 12.18).
45
3 c ? W( x ) dx 5 c ? 475.75503 5 1
0
FIGURE 12.18
c5
1 475.75503
We now have a probability density function, as shown in Figure 12.19.
FIGURE 12.19
12.1 Continuous Probability Models Y
791
1 . This allows Note from Figure 12.19 that we define the equation Y2 5 475.75503
us to use the regression equation in Y1 and quickly create the probability density function. The area under this curve over the interval 30, 454 is 1, as it should be. We can now compute the probabilities requested. The probability that the wait time will be less than 10 minutes can be found by computing the integral
10
3 c ? W( x ) dx (see Figure 12.20).

0
FIGURE 12.20
10
3 c ? W( x ) dx < 0.09
0
Approximately 9 percent of the time, the wait time is less than 10 minutes. A look at the original data shows that a total of 8 percent of the time, the wait time is 0 or 5 minutes. The probability that the wait time will be at least 20 minutes but less than
30
30 minutes can be found by computing the integral 3 c ? W( x ) dx (Figure 12.21).

20 30
3 c ? W( x ) dx < 0.377
20
FIGURE 12.21
Approximately 38 percent of the time, the wait time is more than 20 minutes but less than 30 minutes. A look at the original data verifies that this result is reasonable.
12.1 Summary
In this section, you learned how to create a probability density function and how to use this function to find the probability (or percentage of the time) that a particular event will occur. You also learned about continuous random variables and the role they play in probability density functions.

In Exercises 16, use the basic principle of probability to answer each question. 1. A bag contains five red, three blue, and six white marbles. What is the probability of blindly drawing a red marble? 2. A bag contains five red, three blue, and six white marbles. What is the probability of blindly drawing a blue or white marble? 3. A pair of dice is rolled. What is the probability of rolling doubles (both dice showing the same number)? 4. In a game, a red die and a blue die are rolled. The probability that the dice pair shows a sum of 7 or 11 is
8 . 36
What is the probability that the
pair does not show a sum of 7 or 11? 5. In Yahtzee, the highest-scoring dice roll occurs when all five dice rolled show the same number. The probability that all five dice rolled will show the same number on a single roll is
1 . 1296
What is
the probability that not all five dice will show the same number on a single roll? 6. Suppose it is reported that the probability of snow on a particular day is 30 percent. What is the probability of it not snowing?
792
CHAPTER 12
12.1 Exercises
In Exercises 110, determine whether the function is a probability density function. Explain. 1. f ( x ) 5 x 1 1 on 30, 14 2 21. Abstract of the United States gives the number of high school dropouts for different age ranges. (Source: Statistical Abstract of the United States, 2006, Table 259, p. 173.) The age of a high school dropout, d, varies between the ages of 16 and 24 and can be modeled by the following probability density function. p( d ) 5 2 0.017996d 2 1 0.708169d 2 6.743966 (a) Verify that p( d ) is a probability density function. (b) Find the probability that a high school dropout selected at random is younger than 18 years old. (c) Find the probability that a high school dropout selected at random is 18 years old or older. 22. Abstract of the United States shows the number of men in the labor force in 2004 for different age ranges. (Source: Statistical Abstract of the United States, 2006, Table 577, p. 387.) The age of a male in the labor force, m, varies between 16 and 80 and can be modeled by the following probability density function. p( m ) 5 0.61791 3 1026m3 2 0.00011202m2 1 0.0058587m 2 0.06798 (a) Verify that p( m ) is a probability density function. (b) Find the probability that a male in the labor force selected at random is younger than 22 years old. (c) Find the probability that a male in the labor force selected at random is at least 25 years old but less than 40 years old. (d) Find the probability that a male in the labor force selected at random is 55 years old or older. 23. The Statistical Abstract of the United States shows the number of women in the labor force in 2004 for
Civilian Labor Force Civilian Labor Force The Statistical High School Dropouts The Statistical
2. g( x ) 5 4 2 5x on 30, 0.54 3. h( x ) 5 3 2 6 on 31, 24 x 1 13 13
2 4. g( t ) 5 t 2 1 1 on 31, 34 11 1 1 5. r( t ) 5 t 3 1 t 2 1 on 30, 14 3 2 6. p( t ) 5 1 3 1 1 t 1 t 1 on 30, 24 12 6 6
7. r( t ) 5 0.007t 4 2 0.048t 3 1 0.085t 2 2 0.002t 1 0.19 on 30, 44 8. f ( x ) 5 3e 9. f ( x ) 5 10. f ( x ) 5

23x
on 30, ` )
2
"2p
e2(1>2) x on ( 2 ` , ` )
1 on 30, 64 6
In Exercises 1120, find the value of c that will make the function a probability density function over the given interval. 11. f ( x ) 5 cx on 30, 14
2
12. g( x ) 5 c"x on 31, 24 13. h( x ) 5 cx3 on 31, 34 14. f ( t ) 5 c( t 2 1 t 2 2 ) on 31, 24 1 15. g( t ) 5 c at 4 1 2t 3 1 t 2 2 t 1 1 b on 31, 44 2 16. M( x ) 5 c on 30, 84 17. P( x ) 5 2e2cx on 30, ` ) 18. R( t ) 5 ce25t on 30, ` )
2 2
19. f ( x ) 5 ce2x on ( 2 ` , ` ) 20. f ( x ) 5 ce2(1>2)x on ( 2 ` , ` )
793
different age ranges. (Source: Statistical Abstract of the United States, 2006, Table 577, p. 387.) The age of a female in the labor force, w, varies between 16 and 80 and can be modeled by the following probability density function. p( w ) 5 0.45449 3 1026 w 3 2 0.00009035w 2 1 0.005002w 2 0.05802 (a) Verify that p( w ) is a probability density function. (b) Find the probability that a female in the labor force selected at random is younger than 25 years old. (c) Find the probability that a female in the labor force selected at random is at least 25 years old but less than 50 years old. (d) Find the probability that a female in the labor force selected at random is 55 years old or older. 24.
Number of Hours Worked Consider the data given in the table. (Source: Statistical Abstract of the United States, 2006, Table 590, p. 395.)
(c) Use the histogram to find the probability that a person chosen at random works 40 hours or more each week. (d) Use the histogram to find the probability that a person chosen at random works 2 hours or less each week. What assumption must be made in responding to this exercise? 25.
Number of Births The table shows the
number of births (in thousands) in the United States by age of the birth mother.
(Source: Statistical Abstract of the United States, 2006, Table 74, p. 65.)
Age Interval Under 20 2024 2529 3034 3539 4044 4549
Midpoint of Interval 16 (estimate) 22 27 32 37 42 47
Number of Births (thousands) 433 1,022 1,060 951 454 96 5
The table shows the number of persons (in thousands) in the United States who work the indicated number of hours. Number of Hours at Work per Week 1 to 4 5 to 14 15 to 29 30 to 34 35 to 39 40 41 to 48 49 to 58 60 or more Number of People (in thousands) 1,337 4,868 15,763 9,832 9,026 54,347 13,039 13,835 9,590
(a) Create a scatter plot showing the number of births as a function of the midpoint of each age interval. (b) Use the regression feature of your graphing calculator to find a cubic model B( x ) for these data. (c) Find the value of c such that the function is a probability density function describing the probability that a baby is born to a mother in a particular age range. Use the overall age range of at least 16 years old but less than 45 years old to determine this
45
value. Then verify that 3 cB( x ) dx 5 1.

16
(a) Create a histogram in which each vertical bar is extended so that the area of the resulting rectangle represents the percentage of people in that work-hours interval. (b) Use the histogram to find the probability that a person chosen at random works less than 40 hours each week.
(d) Use the function cB( x ) to determine the probability that the age of a mother chosen at random is at least 29 years old. (e) Use the function cB( x ) to determine the probability that the age of a mother chosen at random is less than 25 years old. (f ) Use the function cB( x ) to determine the probability that the age of a mother chosen at random is in her thirties.
794 26.
CHAPTER 12
Cardiovascular Disease The table
27.
shows the number of deaths (in thousands) caused by a major cardiovascular disease, by age, for the year 2000. (Source: Statistical Abstract of the
United States, 2006, Table 108, p. 86.)
The table shows the number of deaths (in thousands) caused by motor vehicle accidents for different age groups for the year 2002.
Motor Vehicle Accidents
(Source: Statistical Abstract of the United States, 2006, Table 108, p. 86.)
Age Interval 1524 2534 3544 4554 5564 6574 7584 85+
Midpoint of Interval 19.5 29.5 39.5 49.5 59.5 69.5 79.5 92.5 (estimated)
Number of Deaths (thousands) 1 4 17 44 77 156 296 340
Age Interval 1524 2534 3544 4554 5564 6574 7584 85+
Midpoint of Interval 19.5 29.5 39.5 49.5 59.5 69.5 79.5 92.5 (estimated)
Number of Deaths (thousands) 11 7 7 6 4 3 3 1
(a) Create a scatter plot showing the number of deaths as a function of the midpoint of an age interval. (b) Use the regression feature of your graphing calculator to find a logistic model D( x ) for these data. (c) Find the value of c such that the function cD( x ) is a probability density function describing the probability that a person who dies of a major cardiovascular disease is in a specific age range. Use the overall age range of at least 15 years old but less than 92.5 years old to determine this value. Then
92.5
verify that 3
15
cD( x ) dx 5 1.
(a) Create a scatter plot showing the number of deaths as a function of the midpoint of an age interval. (b) Use the regression feature of your graphing calculator to find an exponential model D( x ) for these data. (c) Find the value of c such that the function cD( x ) is a probability density function describing the probability that a person who dies of a major cardiovascular disease is in a specific age range. Use the overall age range of at least 15 years old but less than 92.5 years old to determine this value. Then
92.5
(d) Use the function cD( x ) to determine the probability that a person chosen at random who dies of a major cardiovascular disease is at least 44 years old. (e) Use the function cD( x ) to determine the probability that a person chosen at random who dies of a major cardiovascular disease is younger than 35 years old. (f ) Use the function cD( x ) to determine the probability that a person chosen at random who dies of a major cardiovascular disease is at least 75 years old but less than 84 years old.
verify that 3
15
cD( x ) dx 5 1.
(d) Use the function cD( x ) to determine the probability that a person chosen at random who dies in a motor vehicle accident is at least 15 years old but less than 35 years old. (e) Use the function cD( x ) to determine the probability that a person chosen at random who dies in a motor vehicle accident is at least 20 years old. (f ) Use the function cD( x ) to determine the probability that a person chosen at random who dies in a motor vehicle accident is a teenager.
795
28. Return to the data given for Exercise 25. Use the data to find the percentage of mothers who were 30 years old or older when they gave birth to a child. Compare this percentage to the probability computed in Exercise 25. Discuss why the percentage is similar to the probability. What could account for any differences in these values? 29. Return to the data given for Exercise 26. Use the data to find the percentage of people who were age 45 or older when they died of a major cardiovascular disease. Compare this percentage to the probability computed in Exercise 26. Discuss why the percentage is similar to the probability. What could account for any differences in these values? 30. Return to the data given for Exercise 27. Use the data to find the percentage of people 15 years old or older, but less than 35 years old, who died in a motor vehicle accident. Compare this percentage to the probability computed in Exercise 27. Discuss why the percentage is similar to the probability. What could account for any differences in these values?
35. Given the probability density function h( x ) 5

3 2 x 13 6
1 13 on 31, 24 , explain what

c
information 3
1
A 13 x 2 1 13 B dx provides.
3 6
36. Explain why the graph of the function shown represents a probability density function. Then use the graph to find the probability that x , 1.
y
0.50
0.25
37. A cumulative distribution function is a function that represents the probability up to a particular number x. The cumulative distribution function is defined as F( x ) 5 3 f ( t ) dt, where f ( t ) is a probability
C x
31. Write an explanation telling why the area under a probability density function over a specific interval measures the probability that a random variable lies within that specific interval. You may use graphs, tables, and symbols to make your explanation clear. 32. Explain why the total area under the graph of a probability density function must equal 1. 33. Given the probability density function
3 6
density function. The integral represents the probability that a random variable is at least C but less than x. Find F( C ) and explain what it means. 38. The graph shown is the graph of a probability density function. Sketch a graph of its corresponding cumulative distribution function.
y
0.50
mean when we say that the probability that x is greater than 1 and less than 1.6 is about 52 percent. 34. Given the probability density function
3 6
h( x ) 5 13 x 2 1 13 on 31, 24 , explain what we
0.25
1
x
h( x ) 5 13 x 2 1 13 on 31, 24 , find the value of c such that the probability that a random variable is greater than 1 but less than c is 75 percent.
39. Suppose that F( x ) 5 3 f ( t ) dt is a cumulative

50
distribution function for the heights (in inches) of the students in your class. Explain, in this context, the meaning of F( 62 ) .
796
CHAPTER 12
Construct uniform and exponential probability density functions given the appropriate information Given a real-life situation that may be modeled using a uniform or an exponential probability density function, find the probability of the occurrence of an event
You have seen that we are able to construct probability density functions for particular situations and use these functions to determine probabilities. Specifically, it is the area under these probability density functions that corresponds to the probability we desire. In practice, it is more typical to use a special probability density function that is already well known. That is, depending on the situation, we will assume that the probability is distributed in a certain way. In this section, we will study two such distributions: uniform and exponential distributions.
Uniform Distributions
y
10
FIGURE 12.22
When visiting Disneyland, people often take a shuttle bus from their hotel to the park. These hotel shuttles leave the hotel at advertised intervalssay, every 15 minutes. When a family arrives at the shuttle departure area, they dont know how long they will need to wait. They only know that the wait time will be anywhere between 0 minutes and 15 minutes, with each wait time being equally likely. The wait time X is a continuous random variable for which 0 # X # 15. Since each wait time in this interval is equally likely to occur, the probability density function can be modeled using a uniform probability density function, as shown in Figure 12.22. What should be the value of the constant for this uniform probability density function? Recall that a probability density function has the property that 3 f ( x ) dx 5 1. Therefore, we need to define the constant function f ( x ) such that
a b
this property holds true. We will do this in Example 1.
EXAMPLE
Determining a Wait-Time Probability

If the wait time X for the shuttle to Disneyland is uniformly distributed over the interval 0 # X # 15, what is the uniform probability density function for this situation?
SOLUTION Since f ( x ) is a constant function, we need to find a constant such
b
that 3 f ( x ) dx 5 1.
a 15 15
3 f ( x ) dx 5 3 cdx
0 0
5 cx `
15 0
5 15c 2 0c 5 15c
797
Since this integral must have a value of 1, we can now find the value of c. 15c 5 1 c5
1
1 15
Therefore, if f ( x ) 5 15, it is a uniform probability density function for shuttle wait times on the interval 30, 154 .
We can generalize this result by formally defining a uniform probability density function.
UNIFORM PROBABILITY DENSITY FUNCTION
A function P is a uniform probability density function of a random variable x on the interval 3a, b4 if P( x ) 5 b
P (x) 1 ba 1 2a
for all x in 3a, b4 .
EXAMPLE
Determining a Wait-Time Probability

Find the probability that the family waiting for the shuttle bus in Example 1 will have to wait 10 minutes or more.
SOLUTION Recall that the area under a probability density function represents the probability of the event occurring as described. In this case, the area under the
function f ( x ) 5 15 on the interval 310, 154 will provide the probability we desire.
1
1 1 15 3 15 dx 5 15 x ` 10 10 5 1 1 ( 15 ) 2 ( 10 ) 15 15 2 3
15
512 5
1
1 3
The probability is 3 that the family will have to wait 10 minutes or more for the shuttle bus. We could also say that 33% of the time, a family that shows up randomly for the shuttle bus will have to wait 10 minutes or more.
798
CHAPTER 12
Just In Time
Graphing Exponential Functions
Graphs of exponential functions of the form y 5 aebx either always increase or always decrease. The values of the parameters a and b determine whether the function increases or decreases.
a . 0 and b . 0
y 7 6 5 4 3 2 1 3 2 1 0 1 2 3 x
a . 0 and b , 0
y 7 6 5 4 3 2 1 3 2 1 0 1 2 3 x
a , 0 and b . 0
y 1 3 2 11 2 3 4 5 6 7 1 2 3 x
a , 0 and b , 0
y 1 3 2 11 2 3 4 5 6 7 1 2 3 x
Exponential Distributions
An exponential probability density function is derived from situations involving exponential decay. For example, electronic components commonly fail in an exponential-decay pattern. That is, since there are no moving parts to cause the electronic component to become worn out over time, these components fail randomly based on an exponential distribution.
EXAMPLE
Modeling Light Bulb Failure

Consider the following hypothetical situation. We purchase 100 light bulbs. The packaging claims that each bulb has a 750-hour bulb life. We turn on each of the 100 bulbs simultaneously and wait for the first bulb to burn out. Then we record the number of bulbs that remain burning each day. We will call the day prior to the day that the first bulb burns out the start date of the experiment. We assume the bulbs will burn out at a continuous rate of 3 percent per day after the start date. Create an exponential decay model for the number of light bulbs that remain burning as a function of the number of days since the experiment began.
799
SOLUTION We started with 100 light bulbs. A continuous exponential decay model of the form y 5 ( initial amount ) ? e2rate ? x will be used. A function f ( x ) representing this situation can be determined by substituting 100 for the initial amount and 0.03 (3 percent) for the rate.
f ( x ) 5 100e20.03x In this situation, f ( x ) represents the number of light bulbs that remain burning x days after the start date of the experiment. We now have a function representing the number of light bulbs that remain burning as a function of the number of days since the start date. We now focus our attention on the number of light bulbs that have burned out. Since we started with 100 light bulbs, we can find the number of light bulbs that have burned out by subtracting the number of light bulbs that remain burning. We will call this function g ( x ) . g ( x ) 5 100 2 100e20.03x Next, we focus our attention on the percentage of light bulbs that have burned out. Since we started with 100 light bulbs, we take the number of light bulbs that have burned out, g ( x ) , and divide by 100. We will call this percentage function h( x ) . h( x ) 5 100 2 100e20.03x 100
5 1 2 e20.03x We now have a way to determine the probability, expressed as a percentage, that a randomly chosen light bulb will burn out x days after the start date of the experiment. We also know that if X is a continuous random variable representing the number of days (after the start date) it takes for a light bulb to fail, the area under a probability density function will tell us the probability of a light bulb burning out within x days after the experiment start date. Therefore, we can make the following connection. Probability that a light bulb burns out 5 3 P( t ) dt 5 1 2 e20.03x
0 x
The function P( t ) represents the probability density function for this situation. If we solve this equation for P( t ) , we can determine this probability density function, known as an exponential probability density function.
20.03x 3 P( t ) dt 5 1 2 e
d d a 3 P( t ) dt b 5 ( 1 2 e20.03x ) dx 0 dx P( x ) 5 0.03e20.03x
EXAMPLE
Proving that a Function Is an Exponential Probability Density Function

Show that the function P( x ) 5 0.03e20.03x is a probability density function on the interval 30, ` ) .
800
y 0.030 0.025 0.020 0.015 0.010 0.005 0 0
CHAPTER 12
SOLUTION A probability density function must be nonnegative on the indicated
interval and must have the property that 3 P( x ) dx 5 1. In this case, we can see graphically that P( x ) is nonnegative on the interval 30, ` ) . See Figure 12.23. We now check to see if the area under the curve on the interval 30, ` ) is indeed 1.
20.03x dx 5 2 e20.03x ` 3 0.03e
` 0 b 0
20
40
60
80
FIGURE 12.23
5 lim ( 2 e20.03x ) `
bS`
5 lim ( 2 e
bS`
20.03b
1 e20.03 ? 0 ) 1 e20.03 ? 0 b
5 lim a 2
bS`
1 e
0.03b
5011 51 We can generalize this result by formally defining an exponential probability density function.
EXPONENTIAL PROBABILITY DENSITY FUNCTION
A function P is an exponential probability density function of a random variable x on the interval 30, ` ) if P( x ) 5 ae2ax.
P
a P(x)
EXAMPLE
Determining the Probability of Light Bulb Failure

Use the exponential probability density function P( x ) 5 0.03e20.03x to determine the probability that a light bulb chosen at random will burn out in the first 5 days after the start date of the experiment. (Recall that we defined the start date to be the day prior to the day that the first bulb burned out.)
801
SOLUTION The area under the function P( x ) 5 0.03e20.03x on the indicated in-
terval gives us the probability that we desire.

20.03x dx 5 2 e20.03x ` 3 0.03e
5 0
5 2e
20.03 ? 5
2 ( 2 e20.03 ? 0 )
5 2 e20.03 ? 5 1 1 < 2 0.861 1 1 5 0.139 There is an approximately 13.9 percent chance that a randomly selected light bulb will burn out in the first 5 days after the start date. We can confirm this result using the graphing calculator (Figure 12.24).
FIGURE 12.24
12.2 Summary
In this section, you learned that there are special probability density functionsthe uniform probability density function and the exponential probability density function. Each is useful in certain situations, and each can be used to find probabilities by finding the area under the particular function over an interval. Uniform distributions are used to model situations in which a random variable has an equally likely chance of occurring. Exponential distributions are used to model situations involving continuous exponential decay. A common application involves the failure rate of electronic components.

1. The function f ( x ) 5 2ex is an ever-increasing function. Write a convincing explanation telling why it is an increasing function. 2. The function f ( x ) 5 4e22x is an ever-decreasing function. Write a convincing explanation telling why it is a decreasing function. 3. Consider the function f ( x ) 5 2ex. Rank the following values in increasing order. Explain your thinking. 2 2, f ( 0 ) , f ( 2 2 ) , f ( 2 ) , 0 4. Consider the function f ( x ) 5 4e22x. Rank the following values in increasing order. Explain your thinking. 2 2, f ( 0 ) , f ( 2 2 ) , f ( 2 ) , 0 5. Write the equation of an exponential function of the form f ( x ) 5 aebx such that the function is always increasing and f ( x ) , 0 for all x.
802
CHAPTER 12
12.2 Exercises
In Exercises 15, create a uniform probability density function over the indicated interval. Show all the integration steps necessary to create this function. 1. 30, 54 3. 32, 84 2. 30, 204 4. 315, 374 15.
Twister Spinner The game Twister uses a spinner that is divided into four equal regions. Each of the four regions is divided into four colors: blue, yellow, red, and green.
5. 32 5, 54
6. Write several sentences explaining how you can create a uniform probability density function on the interval 3a, b4 without performing any computations. In Exercises 711, create an exponential probability density function for a function with the given continuous failure rate (such as the rate at which light bulbs burn out). Show all the steps necessary to create this function. 7. 6 percent 9. 12 percent 11. 50 percent 12. Write several sentences explaining how you can create an exponential probability density function with a continuous failure rate of a percent without performing any computations. 13.
Bus Wait Time Arizona State University offers its students FLASH bus service, which provides transportation around the very large campus. FLASH service is available approximately every 10 minutes at any FLASH bus stop. (Source: www.asu.edu.) Assuming the wait times are uniformly distributed, what is the probability that a random student arriving at the bus stop will have to wait 5 minutes or less for the next bus?
8. 10 percent 10. 15 percent 16.
If X represents the angle formed by the spinner and the positive x-axis, what is the probability that the spinner will land in the green area in the first quadrant?
Twister Spinner The game Twister uses a spinner that is divided into four equal regions. Each of the four regions is divided into four colors: blue, yellow, red, and green. If X represents the angle formed by the spinner and the positive x-axis, what is the probability that the spinner will land in any blue region? Twister Spinner The game Twister uses a spinner that is divided into four equal regions. Each of the four regions is divided into four colors: blue, yellow, red, and green. Using the idea of the area under a probability density function, explain why the probability of landing in any particular color region is the same. Call Center Help Ministries, Inc.,
17.
18.
14.
Arizona State University offers its students FLASH bus service, which provides transportation around the very large campus. FLASH service is available approximately every 10 minutes at any FLASH bus stop. (Source: www.asu.edu.) Assuming the wait times are uniformly distributed, what is the probability that a random student arriving at the bus stop will have to wait 8 minutes or more for the next bus?
Bus Wait Time
operates a call center that receives phone calls very frequently. (Source: www.helpministries.org.) Suppose an operator has just ended a call. Assuming that calls are uniformly distributed on the interval 30, 64 minutes, what is the probability that the next call will come in within the next 2 minutes? 19.
Call Center Help Ministries, Inc.,
operates a call center that receives phone calls very frequently. (Source: www.helpministries.org.) Suppose an operator has just ended a call. Incoming
803
calls are uniformly distributed on the interval 30, 64 minutes. Determine any time interval over which there is a 50 percent probability that a phone call will come in during that time interval. 20.
Call Center Help Ministries, Inc.,
operates a call center that receives phone calls very frequently. (Source: www.helpministries.org.) Suppose an operator has just ended a call. Incoming calls are uniformly distributed on the interval 30, 64 minutes. Determine the longest time interval until the next call such that there is a 10 percent probability that an incoming call will occur in this interval. 21.
Computer Hard Drive Failure Hard
within 5 years of the cancers detection. (Source: www.cancer.org.) Use this information to create an exponential probability density function to model the situation. Use this function to determine the probability of surviving 10 or more years after detection. 26.
Cancer Survival If colon and rectal
cancer is detected early, 10 percent of patients die within 5 years of the cancers detection. (Source: www.cancer.org.) Use this information to create an exponential probability density function to model the situation. Use this function to determine the probability of surviving 20 or more years after detection.
disk drives in the Maxtor MaXLine Pro 500 series are designed to last about 114 years, on average, before failing. (Source: www.maxtor.com.) An exponential probability density function for this situation is P( x ) 5 0.009e20.009x. These hard drives have a warranty period of 5 years. What is the probability that one of these hard drives will fail during the warranty period? 22.
Computer Hard Drive Failure Hard
27. Explain the concept of a uniform probability density function. Use graphs and symbols to support your explanation. 28. Explain the concept of an exponential probability density function. Use graphs and symbols to support your explanation. 29. Show why, for the uniform probability density function P( x ) 5 b 3 P( x ) dx 5 1.
a b
disk drives in the Maxtor MaXLine Pro 500 series are designed to last about 114 years, on average, before failing. (Source: www.maxtor.com.) An exponential probability density function for this situation is P( x ) 5 0.009e20.009x. What is the probability that one of these hard drives will fail sometime between 5 and 10 years? 23.
Cancer Survival Stage I lung cancer
1 2a
on 3a, b4 , we can say that
patients who undergo radiation therapy survive an average of 21 months. (Source: patient.cancerconsultants.com.) An exponential probability density function for this situation is P( x ) 5 0.05e20.05x, where x is measured in years. What is the probability that a stage I lung cancer patient will survive more than 2 years after diagnosis? 24.
Cancer Survival Stage I lung cancer
30. Show why, for the exponential probability density function P( x ) 5 ae2ax on 30, ` ) , we can say that 3 P( x ) dx 5 1.
a b
31. Explain why, if P( x ) 5 be2ax on 30, ` ) is an exponential probability density function, it is necessary that a 5 b and a 2 0, b 2 0.
patients who do not undergo radiation therapy survive an average of 14 months. (Source: patient.cancerconsultants.com.) An exponential probability density function for this situation is P( x ) 5 0.07e20.07x, where x is measured in years. What is the probability that a stage I lung cancer patient will survive more than 2 years after diagnosis if the patient does not undergo treatment? 25.
Cancer Survival If breast cancer is
32. A cumulative distribution function is a function that represents the probability up to a particular number x. The cumulative distribution function is defined as F( x ) 5 3 f ( t ) dt, where f ( t ) is a
C x
probability density function. The integral represents the probability that a random variable is at least C but less than x. For a uniform
detected early, 2.1 percent of patients die
804
CHAPTER 12
probability density function f ( t ) on 32, N4 , find the cumulative distribution function F( x ) . 33. A cumulative distribution function is a function that represents the probability up to a particular number x. The cumulative distribution function is defined as F( x ) 5 3 f ( t ) dt, where f ( t ) is a
C x
is modeled by the exponential probability density function f ( t ) 5 0.25e20.25t on the interval 30, ` ) . What is the probability that you will not have to wait in line to place your order? Explain your answer clearly, using graphs, symbols, and words. 35. Suppose that the time, in minutes, spent waiting in line at a fast-food restaurant during the lunch hour is modeled by the exponential probability density function f ( t ) 5 0.25e20.25t on the interval 30, ` ) . What is the probability that you will have to wait in line for 30 minutes or more before placing your order? Explain your answer clearly, using graphs, symbols, and words.
probability density function. The integral represents the probability that a random variable is at least C but less than x. For an exponential probability density function f ( t ) on 30, ` ) , find the cumulative distribution function F( x ) . 34. Suppose that the time, in minutes, spent waiting in line at a fast-food restaurant during the lunch hour
Compute and interpret the expected value, variance, and standard deviation of discrete and continuous random variables Compute and interpret formulas for the expected value, variance, and standard deviation of uniform and exponential probability density functions
As we explore probability density functions and use them to determine the probability of the occurrence of an event, we may also be interested in the nature of the random variable represented by the density function. Specifically, we may want to find a single value to represent that random variable, and we may want to describe the extent to which the potential values of the random variable vary or deviate from the mean. In this section, we will investigate the concepts of expected value, variance, and standard deviation used to describe a random variable. To help make sense of these ideas, we will first investigate a situation involving discrete random variables.
Expected Value of a Discrete Random Variable

Recall that a discrete random variable is a random variable that can take on only certain values in a given interval. For example, the state of Arizona has a lottery game called Fantasy 5 (www.arizonalottery.com). In this game, players choose five numbers between 1 and 35, inclusive. A player wins the game if the five numbers drawn match three, four, or five of the players numbers. Depending on whether the player matches all five numbers, four numbers, or three numbers, a different dollar amount is awarded. If all five numbers match, the player wins $50,000. If four numbers match, the player wins $500; and if three numbers match, the player wins $5. In this situation, the random variable X can take on the discrete values $50,000, $500, $5, or $0. That is, X represents the amount of money won in playing the game. The website www.arizonalottery.com lists the probability of winning in each case (Table 12.5).
805
TABLE 12.5
Fantasy 5Arizona State Lottery Number of Matching Numbers 5 4 3 Probability of Occurrence 1 324,632 1 2164 1 75 Prize $50,000 $500 $5
Note that no money is won if less than three numbers are matched. A lottery player may want to determine the long-term average winnings per game played. For example, the probability
1 324,632
for matching all five numbers can be inter-
preted to mean that, in the long run, a player will match all five numbers only 1 time in 324,632 games played. On average, this player would win
$50,000 , 324,632 games played
or about $0.15 per play. The total expected value is the sum of
all such average amounts earned per play.
EXAMPLE
Finding the Fantasy 5 Expected Value

Find the total expected value for a player playing the Arizona State Lottery Fantasy 5 game.
SOLUTION We already determined that, in the long run, a player can expect to win $50,000 only 1 time for every 324,632 games played, for an average of $0.15 per play. We next look at the average amount won per play for matching four numbers. In the long run, a player will match four numbers 1 time in 2164 plays. On
average, this player would win

$5
$500 , 2164 games played
or about $0.23 per play. In the
long run, a player will match three numbers 1 time in 75 plays. On average, this player would win 75 games played, or about $0.07 per play. The expected value is the sum of these outcomes. Average winnings 5 $50,000 ? 1 1 1 1 $500 ? 1 $5 ? 324,632 2164 75
5 $0.15 1 $0.23 1 $0.07 5 $0.45 In the long run, a Fantasy 5 player can expect to win 45 cents each time the game is played. A Fantasy 5 game ticket costs the player $1.00. So, in the long run, a player loses 55 cents each time the game is played! We can express the expected value computation in words. Expected value is computed by taking the particular value of the discrete random variable (prize
806
CHAPTER 12
amount in this case) and multiplying it by the probability of that event occurring, to obtain an average. Summing up these averages yields the expected value. The expected value of a random variable is the sum of the probability of each possible outcome of the experiment multiplied by its value.
Just In Time
Summation notation is a shorthand way of expressing the sum of many terms of a sequence. For example, if we wish to express the sum 1 1 2 1 3 1 4 1 5 using summation notation, we write a n. The idea is to start
n51 5
with n 5 1, substitute integers up to 5 for n, and then add the results. For
n51
a 2n 1 5, we obtain 7 1 9 1 11 1 13 5 40.
EXPECTED VALUE OF A DISCRETE RANDOM VARIABLE
A random variable taking on values x1, x2, c , xn with corresponding probabilities P1, P2, c , Pn has an expected value E( X ) 5 a xi ? Pi
i51 n
In words, we multiply each value of the random variable by its corresponding probability and then sum the results.
Expected Value of a Continuous Random Variable

We can extend the idea of expected value to a continuous random variable. Recall that the area under a probability density function determines the probability of an event occurring on the indicated interval. The continuous random variable X can take on any value in that interval. Therefore, we can find the expected value of a continuous random variable by using integration.
EXPECTED VALUE OF A CONTINUOUS RANDOM VARIABLE
If x is a continuous random variable with probability density function P( x ) on the interval 3a, b4 , then the expected value is E( X ) 5 3 x ? P( x ) dx
a b
807
EXAMPLE
Finding the Expected Age of a Typical Voter in the 2004 Election

Find the expected value of the random variable x that represents the ages of voters, between 18 and 80 inclusive, in the 2004 presidential election. The associated probability density function is P( x ) 5 1 ( 2 0.0000221x3 1 0.00211x2 2 0.0307x 1 0.760 ) 83.232
SOLUTION In this case, the expected value is the typical age of a voter in the 2004 presidential election. We will use technology to compute the expected value (Figure 12.25).
80
E( X ) 5 3 x ? P( x ) dx
18
FIGURE 12.25
According to our probability density function model, the expected age of a voter in the 2004 presidential election is just over 50 years of age.
Variance and Standard Deviation of a Discrete Random Variable

For a particular situation, we will be interested in determining how spread out the random variable is from the expected value. That is, we might want to express how a random variable varies or deviates from the expected value. Lets explore this idea by returning to the Arizona State Lottery Fantasy 5 game. We determined previously that the expected value for the Fantasy 5 game was $0.45. That is, over the long run, we expect to win 45 cents for every ticket that we buy. Recall that if all five numbers match the drawn numbers, the lottery player wins $50,000. This value deviates from the expected value by $50,000 2 $0.45 5 $49,999.55. In other situations, a random variable might deviate from the mean in such a way that the difference is negative (if the random variable is less than the expected value). To avoid negative values when computing these differences, we will square the differences. In the above scenario (matching all five numbers), we will compute the square of the deviation. ( $50,000 2 $0.45 ) 2 5 ( $49,999.55 ) 2 As with the expected value, we find the average deviation by interpreting the probability of matching all five numbers,
1 , 324,632
to mean that in 1 time out of
324,632, the square of the deviation will be ( $50,000 2 $0.45 ) 2 5 ( $49,999.55 ) 2.
808
CHAPTER 12
On average, the deviation will be
( $49,999.55 ) 2 . 324,632 games played
(Recall that 324,632 is
the theoretical number of times you would have to play to win $50,000.) The variance is the sum of all the squares of the differences between the value of the random variable and the expected value. For the case of matching four numbers, we compute the square of the deviation as before. ( $500 2 $0.45 ) 2 5 ( $499.55 ) 2 On average, the square of the deviation will be
( $499.55 ) 2 . 2164 games played
Finally, we com-
pute the square of the deviation for the case of matching three numbers. ( $5 2 $0.45 ) 2 5 ( $4.55 ) 2 On average, the square of the deviation will be found by adding these averages together. Var( x ) 5 ( $49,999.55 ) 2 ( $499.55 ) 2 ( $4.55 ) 2 1 1 324,632 games played 2164 games played 75 games played
( $4.55 ) 2 . 75 games played
The variance is
< 7700.89 1 115.32 1 0.28 < 7816.49 The units for the variance are square dollars, which is difficult to interpret. We define the standard deviation to be the square root of the variance. In this case, the units for the standard deviation are dollars. s( x ) 5 "7816.49 < $88.41
So, on average, a players winnings will deviate from the expected value by $88.41. The variance and standard deviation are not very valuable by themselves, but they are valuable when used as tools of comparison. That is, a variance of 7816.49 and a standard deviation of $88.41 are very large (i.e., the random variable deviates from the expected value by a large amount) when compared with a variance of, say, 9 and a standard deviation of $3. We summarize these results.
VARIANCE AND STANDARD DEVIATION OF A DISCRETE RANDOM VARIABLE
A random variable taking on values x1, x2, c , xn with corresponding probabilities P1, P2, c , Pn has variance Var( x ) 5 a 3xi 2 E( x ) 4 2 ? Pi
i51 n
In words, we subtract the expected value from the value of the random variable and square the result (to avoid negative values). The square of the difference is multiplied by the probability of that random variable occurring, and the sum of the products is determined. The standard deviation is found by taking the square root of the variance. s( x ) 5 "Var( x )
809
As with the expected value, the variance of a continuous random variable is found in essentially the same way except that we must use integration to account for the continuous nature of the random variable.
VARIANCE AND STANDARD DEVIATION OF A CONTINUOUS RANDOM VARIABLE
If x is a continuous random variable with probability density function P( x ) on the interval 3a, b4 , then the variance is Var( x ) 5 3 3( x 2 E( x )) 2 ? P( x ) 4 dx
a b
In words, we subtract the expected value from the value of the random variable and square the result (to avoid negative values). The square of the difference is multiplied by the probability of that random variable occurring, and the sum of the products is determined. The standard deviation is found by taking the square root of the variance. s( x ) 5 "Var( x )
EXAMPLE
Finding the Variance and Standard Deviation of a Continuous Random Variable

Find the variance and standard deviation of the random variable x that represents the ages of voters, between 18 and 80 inclusive, in the 2004 presidential election. The associated probability density function is P( x ) 5 1 ( 2 0.0000221x3 1 0.00211x2 2 0.0307x 1 0.760 ) 83.232
SOLUTION The variance measures the overall deviation from the expected value, or the spread of the values of ages of voters in the election. A relatively large variance tells us that the values are widely spread. We will use technology to evaluate the integral (Figure 12.26).
FIGURE 12.26
810
CHAPTER 12
1 Var( x ) 5 3 c ( x 2 50.5 ) 2 ? ( 2 0.0000221x3 1 0.00211x2 2 0.0307x 1 0.760 ) d dx 83.232 18 Var( x ) < 253.5 < 15.9 s( x ) 5 "253.5 With a variance of 253.5, the standard deviation of ages is 15.9 years.
80
Expected Value, Variance, and Standard Deviation of a Uniform Distribution

We have explored the meaning and computation of the expected value, variance, and standard deviation of a probability density function. We now focus our attention on the special distributions studied in the previous section. We will apply our methods for computing expected value, variance, and standard deviation to the uniform probability density function.
EXAMPLE
Finding the Expected Value of a Uniform Probability Density Function

Calculate the expected value of the general uniform probability density function P( x ) 5 b
1 2a
on 3a, b4 .
SOLUTION
1 E( x ) 5 3 x ? dx b2a a 5 1 2 1 b ?x ? ` 2 b2aa
1 1 1 1 5 a ? b2 ? b 2 a ? a2 ? b 2 b2a 2 b2a 5 5 1 1 ? ( b2 2 a2 ) 2 b2a 1 1 ? ? (b 2 a) ? (b 1 a) 2 b2a
1 5 (b 1 a) 2 The symbolic integration required in computing variance can sometimes be problematic. Therefore, it is often helpful to use an alternate formula for variance. Var( x ) 5 3 3( x 2 E( x )) 2 ? P( x ) 4 dx
a b
5 3 53x 2 2 2xE( x ) 1 ( E( x )) 24 ? P( x ) 6 dx
a
811
5 3 3P( x ) x2 2 P( x ) 2xE( x ) 1 P( x )( E( x )) 24 dx
a
5 3 P( x ) x2dx 2 2E( x ) 3 xP( x ) dx 1 ( E( x )) 2 3 P( x ) dx

a a a
Lets closely examine each of the three integrals. The middle integral, 3 xP( x ) dx,
a
is the definition of expected value, E( x ) . The third integral, 3 P( x ) dx, is the

a
integral of a probability density function and has a value of 1. Making these two substitutions allows us to create a simplified version of the variance formula. Var( x ) 5 3 P( x ) x2dx 2 2E( x ) 3 xP( x ) dx 1 ( E( x )) 2 3 P( x ) dx
a a a b b b
5 3 P( x ) x2dx 2 2E( x ) ? E( x ) 1 ( E( x )) 2 ? 1
a
5 3 P( x ) x2dx 2 2( E( x )) 2 1 ( E( x )) 2
a
5 3 P( x ) x2dx 2 ( E( x )) 2
a
EXAMPLE
Finding the Variance and Standard Deviation of a Uniform Probability Density Function
Calculate the variance and standard deviation of the general uniform probability density function P( x ) 5 b
1 2a
on 3a, b4 .
SOLUTION We will use the alternate form of the variance formula to determine the variance of a uniform probability density function.
Var( x ) 5 3 P( x ) x2dx 2 ( E( x )) 2
a
2 1 2 1 5 3 x dx 2 c ( b 1 a ) d 2 a b 2 a
1 1 b 1 ? x3 ` 2 ? ( b 1 a ) 2 b2a 3 a 4 1 1 1 1 1 ? b3 2 ? a3 b 2 a ? ( b 1 a ) 2 b b2a 3 b2a 3 4
5 a 5
a3 1 b3 2 2 ? (b 1 a)2 3( b 2 a ) 3( b 2 a ) 4
812
CHAPTER 12
5 5 5 5 5 5 5 5
b3 2 a3 1 2 ? (b 1 a)2 3( b 2 a ) 4 ( b 2 a )( b2 1 ab 1 a2 ) 1 2 ? (b 1 a)2 3( b 2 a ) 4 b2 1 ab 1 a2 (b 1 a)2 2 3 4 b2 1 ab 1 a2 b2 1 2ab 1 a2 2 3 4 4( b2 1 ab 1 a2 ) 3( b2 1 2ab 1 a2 ) 2 12 12 3b2 1 6ab 1 3a2 4b2 1 4ab 1 4a2 2 12 12 b2 2 2ab 1 a2 12 (b 2 a)2 12
(b 2 a)2 . The 12
The variance is Var( x ) 5
standard deviation is found by taking the
square root of the variance. s( x ) 5 5 The standard deviation is s( x ) 5 (b 2 a)2 12 b2a "12
b2a . ! 12
EXPECTED VALUE, VARIANCE, AND STANDARD DEVIATION OF A UNIFORM PROBABILITY DENSITY FUNCTION
If x is a random variable with a uniform probability density function P( x ) 5 b

1 2a
on the interval 3a, b4 , then
1 E( x ) 5 ( a 1 b ) 2 Var( x ) 5 s( x ) 5 (b 2 a)2 12 b2a "12
813
EXAMPLE
Analyzing Shuttle Bus Wait Times

Shuttle buses arrive at a hotel every 15 minutes to take visitors to Disneyland. Compute and interpret the expected value for this situation. Then calculate the variance and standard deviation.
SOLUTION Since the wait time is a uniform continuous random variable on the interval 30, 154 , we will use the uniform probability density function
P( x ) 5 15
1 20
5 15 to compute the expected value. 1 E( x ) 5 ( 0 1 15 ) 2 5 15 2
5 7.5 minutes On average, a visitor would expect to wait 7.5 minutes for the shuttle bus. Next we compute the variance. Var( x ) 5 5 5 5 The variance is ( 15 2 0 ) 2 12 152 12 225 12 75 4
75 75 . The average square of the distance from the expected value is 4 . 4
We now calculate the standard deviation. s( x ) 5 75 4
< 4.33 minutes The standard deviation is 4.33 minutes.
Expected Value, Variance, and Standard Deviation of an Exponential Distribution

We will determine formulas for the expected value, variance, and standard deviation of an exponential probability density function of the form P( x ) 5 ae2ax on the interval 30, ` ) .
814
CHAPTER 12
EXAMPLE
Finding the Expected Value of an Exponential Probability Density Function

Find a formula for the expected value of an exponential probability density function.
SOLUTION
`
E( x ) 5 3 x ? ae2axdx
0 `
5 a 3 x ? e2axdx
0
` 1 1 5 a a 2 x e2ax 2 2 e2ax b ` a a 0 ` 1 5 a 2 xe2ax 2 e2ax b ` a 0 b 1 5 lim c a 2 xe2ax 2 e2ax b ` d a bS` 0
5 lim a 2
bS`
b x 1 ax 2 ax b ` e ae 0
5 lim a 2
bS`
b 1 0 1 2 ab b 2 a 2 a ? 0 2 a ? 0 b eab ae e ae
1 5 (0 2 0) 2 a 0 2 b a 5 1 a
EXAMPLE
Finding the Variance and Standard Deviation of an Exponential Probability Density Function
Find a formula for the variance and standard deviation of an exponential probability density function.
SOLUTION
Var( x ) 5 3 P( x ) x 2dx 2 ( E( x )) 2
a
1 2 5 3 ae2axx 2dx 2 a b a a 1 2 5 a 3 e2axx 2dx 2 a b a 0

` 1 1 1 1 5 a a 2 x 2 e2ax 2 2x 2 e2ax 2 2 3 e2axb ` 2 2 a a a a 0
5 a 2 x 2e2ax 2
` 2x 2ax 2 1 e 2 2 e2ax b ` 2 2 a a a 0

b 2x 2ax 2 1 e 2 2 e2ax b ` d 2 2 a a a 0
815
5 lim c a 2 x 2e2ax 2
bS`
5 lim a 2 b2e2ab 2
bS`
2b 2ab 2 e 2 2 e2ab b a a
2 ? 0 2a ? 0 2 1 e 2 2 e2a ? 0 b 2 2 a a a 2 1 5 (0 2 0 2 0) 2 a 0 2 0 2 2 b 2 2 a a 5 5 2 1 2 2 a2 a 1 a2
2 a 2 02e2a ? 0 2
We can now compute the standard deviation by taking the square root of the variance. s( x ) 5 "Var( x ) 5 5 1 a2 1 a
EXPECTED VALUE, VARIANCE, AND STANDARD DEVIATION OF AN EXPONENTIAL PROBABILITY DENSITY FUNCTION
If x is a random variable with an exponential probability density function P( x ) 5 ae2ax on the interval 30, ` ) , then E( x ) 5 Var( x ) 5 s( x ) 5 1 a 1 a2 1 a
EXAMPLE
Finding the Expected Value, Variance and Standard Deviation for Light Bulb Failure
Find the expected value, variance, and standard deviation of the exponential probability density function P( x ) 5 0.03e20.03x for the situation introduced in Section 12.2 involving the failure of light bulbs.
816
CHAPTER 12
SOLUTION We first compute the expected value. In this case, the expected value is the expected number of days, after the experiment start date, until a light bulb chosen at random will burn out.
E( x ) 5
1 0.03
1 5 33 days 3 According to the model, the typical light bulb, if left on continuously, will burn out 333 days after the experiment start date. We now compute the standard deviation. Because this is an exponential probability density function, the standard deviation will be equal to the expected value. We will verify this result mathematically. Var( x ) 5 1 0.032 1 9
1
5 1111
1
The variance is 11119. The standard deviation is found by taking the square root of the variance. s( x ) 1 1 2 5 0.03 0.03 1 5 33 days 3 The standard deviation of 333 days is equal to the expected value, as we had predicted.
1
12.3 Summary
In this section, you learned how to compute the expected value, variance, and standard deviation of probability density functions. You also learned how to interpret these values. The expected value is the single number representing the typical value of a discrete or continuous random variable. The variance and standard deviation (the square root of the variance) measure the extent to which the random variable deviates from the expected value. The greater the variance, the greater the values of the random variable deviate from the expected value. That is, the values of the random variable are more widely spread out for large variance values. You also learned that the variance and standard deviation are most useful when used as comparison tools. For example, a random variable with a variance of 2 is more greatly spread out than a random variable with a variance of 4.
1
817

In Exercises 14, expand the expression and find the sum. 1. a ( n2 2 1 )
n51 5 6 5 1 n21 2. a 4 a b 2 n51 4
In Exercises 59, write the sum using summation notation. 5. 2 1 4 1 6 1 8 1 10 6. 1 1 3 1 9 1 27 1 81 7. 1 1 4 1 9 1 16 1 25 1 36 1 49 1 64 1 81 8. 6 1 7 1 8 1 9 1 10 9. 3 1 8 1 15 1 24
3. a sin( np )
n50
4. a en
n50
12.3 Exercises
In Exercises 110, for each probability density function over the given interval, find the following. (a) The expected value (b) The variance (c) The standard deviation 1. f ( x ) 5 x 1 2. h( x ) 5 1 on 30, 14 2 In Exercises 1620, for each exponential probability density function, find the following. (a) E( X ) (b) Var( X ) (c) s( X ) 16. f ( x ) 5 0.02e20.02x on 30, ` ) 18. h( x ) 5 3e23x on 30, ` ) 17. g( x ) 5 0.25e20.25x on 30, ` ) 1 19. q( x ) 5 e2x>3 on 30, ` ) 3 20. q( x ) 5 e2x on 30, ` ) 21.
Arizona State Lottery The Arizona
3 2 6 on 31, 24 x 1 13 13
1 1 1 3. p( t ) 5 t 3 1 t 1 on 30, 24 12 6 6 4. h( x ) 5 1 2 x21>2 on 31, 44 6. r( x ) 5 2x23 on 31, ` ) 5. r( t ) 5 0.007t 4 2 0.048t 3 1 0.085t 2 2 0.002t 1 0.19 on 30, 44 7. w( x ) 5 3x24 on 31, ` ) 9. K( x ) 5 6x27 on 31, ` )
8. G( x ) 5 5x26 on 31, ` ) 10. f ( x ) 5 ax 2(a 1 1) on 31, ` ) , for a . 1 In Exercises 1115, if X is a uniform random variable on the given interval, find the following. (a) The probability density function (b) E( X ) (c) Var( X ) (d) s( X ) 13. 35, 204 11. 30, 104 12. 32, 84 14. 31, 24
State Lottery has a game called The Pick in which players choose six numbers between 1 and 41, inclusive. The table summarizes the winning possibilities and their associated probabilities. (Source: www.arizonalottery.com.) The PickArizona State Lottery Number of Matching Numbers 6 5 4 3 Probability of Occurrence 1 4,496,388 1 22,041 1 504 1 34
Prize $1,000,000 $1000 $40 $2
15. 32 5, 54
818
CHAPTER 12
(a) Compute the expected value and explain what it means in this context. (b) Compute the variance and standard deviation and compare them to the variance and standard deviation values for the Fantasy 5 game (see the beginning of this section). (c) Explain whether or not, in the long run, it pays to play The Pick. 22.
Minnesota State Lottery The Minnesota State Lottery has a game called Northstar Cash in which players choose five numbers between 1 and 31, inclusive. The table summarizes the winning possibilities and their associated probabilities. (Source:
(a) Find E( d ) and interpret its meaning. (b) Find Var( d ) . (c) Find s( d ) . 24. Abstract of the United States shows the number of men in the labor force in 2004 for different age ranges. (Source: Statistical Abstract of the United States, 2006, Table 577, p. 387.) The age of males in the labor force, m, varies between the ages of 16 and 80 and is modeled by the following probability density function. p( m ) 5 0.61791 3 1026m3 2 0.00011202m2 1 0.0058587m 2 0.06798 (a) Find E( m ) and interpret its meaning. (b) Find Var( m ) . (c) Find s( m ) . 25. Abstract of the United States shows the number of women in the labor force in 2004 for different age ranges. (Source: Statistical Abstract of the United States, 2006, Table 577, p. 387.) The age of females in the labor force, w, varies between the ages of 16 and 80 and is modeled by the following probability density function. p( w ) 5 0.45449 3 1026w3 2 0.00009035w2 1 0.005002w 2 0.05802 (a) Find E( w ) and interpret its meaning. (b) Find Var( w ) . (c) Find s( w ) . 26. Compare the variance and standard deviation values found in Exercises 24 and 25. What information does this comparison give you about the ages of men and women in the civilian labor force in 2004? 27.
Bus Wait Time Arizona State University Civilian Labor Force The Statistical Civilian Labor Force The Statistical
www.lottery.state.mn.us.)
Northstar CashMinnesota State Lottery Number of Matching Numbers 5 4 3 2 Probability of Occurrence 1 169,911 1 1307 1 52 1 7
Prize $25,000 $50 $5 $1
(a) Compute the expected value and explain what it means in this context. (b) Compute the variance and standard deviation and compare them to the variance and standard deviation values for the Fantasy 5 game (see the beginning of this section). (c) Explain whether or not, in the long run, it pays to play Northstar Cash. 23. Abstract of the United States shows the number of high school dropouts for different age ranges. (Source: Statistical Abstract of the United States, 2006, Table 259, p. 173.) The age of a high school dropout, d, varies between 16 and 24 years and is modeled by the following probability density function. p( d ) 5 2 0.017996d 2 1 0.708169d 2 6.743966
High School Dropouts The Statistical
offers its students FLASH bus service, which provides transportation around the very large campus. FLASH service is available approximately every 10 minutes at any FLASH bus stop. (Source: www.asu.edu.) Assuming the wait times are uniformly distributed, what is the expected wait time for a random student arriving at the bus stop? 28.
Call Center Help Ministries, Inc., operates a call center that receives phone calls every 8 minutes. (Source: www.helpministries .org.) Suppose an operator has just ended a call.
819
Assuming calls are uniformly distributed on the interval 30, 64 , what is the expected wait time until the next call? 29.
Call Center Help Ministries, Inc., operates a call center for which the time between calls is a uniform random variable on the interval 30, 64 . (Source: www.helpministries.org.) Suppose that at a different call center, the time between calls is a uniform random variable on the interval 30, 84 . Find the variance and standard deviation for each situation. Compare your results and describe what information these values tell you with regard to the call centers.
33. Explain what the formula for expected value tells us to do, and explain what the result means. E( X ) 5 a xi ? Pi
i51 n
Then explain how we can assert that expected value also can be found using the formula E( X ) 5 3 x ? P( x ) dx
a b
30.
Hard disk drives in the Maxtor MaXLine Pro 500 series are designed to last about 114 years, on average, before failing. (Source: www.maxtor.com.) An exponential probability density function that models this situation is P( x ) 5 0.009e20.009x. Explain why the probability density function is P( x ) 5 0.009e20.009x. That is, why is the value 0.009 used in this model?
Hard Drive Failure Cancer Survival Stage I lung cancer
34. Explain what the formula for variance tells us to do, and explain what the result means. Var( x ) 5 a 3xi 2 E( x )4 2 ? Pi
i51 n
Then explain how we can assert that variance also can be found using the formula Var( x ) 5 3 3x 2 E( x ) 4 2 ? P( x ) dx
a b
31.
patients who undergo radiation therapy survive an average of 21 months. (Source: patient.cancerconsultants.com.) An exponential probability density function for this situation is P( x ) 5 0.05e20.05x. Patients who do not undergo radiation therapy survive an average of 14 months. An exponential probability density function for this situation is P( x ) 5 0.07e20.07x. Find the variance and standard deviation for each density function. Compare the variance and standard deviation values and explain what information this comparison provides. 32.
Alcohol-Related Accident Fatalities
35. When computing variance, we square the difference between the expected value of the random variable and the value of the random variable itself. Explain why these differences are squared. 36. Suppose we know that the emergency response time (for example, fire or police) in one city has an expected value of 5.5 minutes and a standard deviation of 1.25 minutes. Suppose we also know that the emergency response time in another city is 7 minutes, with a standard deviation of 0.5 minute. Compare and contrast these statistics for each city, and be sure to explain precisely what information each statistic provides. 37. Suppose that a particular light bulb brand claims that its bulbs have an expected life span of 2000 hours, with a standard deviation of 120 hours. Another light bulb brand claims that its bulbs have an expected life span of 1700 hours, with a standard deviation of 45 hours. Compare and contrast these statistics for each light bulb brand, and be sure to explain precisely what information each statistic provides.
(Source: Modeled from www.nrd.nhtsa.dot.gov.)
A probability density function that models the age of a driver involved in a fatal car accident (in 2004) in which alcohol was involved is F( x ) 5 0.0776e20.0303x on the interval 325, 844 .
(a) Show that F( x ) is a probability density function. (b) What is the expected age of a driver in a fatal car accident? (c) Compute the variance and standard deviation.
820
CHAPTER 12
38. Suppose that f ( x ) is a probability density function for the shoe size of the adult male population of the United States. The owners of a new bowling center are deciding how many pairs of shoes to order in each size. What is the meaning of
10
40. Suppose that f ( x ) is a probability density function for the shoe size of the adult male population of the United States. The owners of a new bowling center are deciding how many pairs of shoes to order in each size. Write an expression for the variance of the density function f ( x ) and explain how this information might be useful to the owners of the bowling center. 41. In this section, we found formulas for the expected value (mean), variance, and standard deviation of uniform and exponential probability density functions. Imagine that you must explain the derivations of these formulas to a student who needs to understand these derivations. Study the derivations and prepare an explanation for your fellow student.
3 f ( x ) dx, and how might this value be useful to

8
the owners of the bowling center? 39. Suppose that f ( x ) is a probability density function on 35, 144 for the shoe size of the adult male population of the United States. The owners of a new bowling center are deciding how many pairs of shoes to order in each size. Write an expression for the mean of the density function f ( x ) and explain how this information might be useful to the owners of the bowling center.
Use a normal probability density function to find the probability of the occurrence of an event Given a probability, find the range of values of a random variable that produces that probability
Intelligence quotient (IQ) tests were originally developed to determine if young children in school were capable of performing at their particular grade or age level. The term intelligence quotient was created because, if a child scored as well on the test as a child of an older age, the two ages were divided (quotient) and the result was multiplied by 100. Since IQ tests are now given more often to adults than to children, this technique is no longer used. However, the term IQ is still used by convention. (Source: www.IQTest.com.) In this section, we will discuss normal distributions. Normal distributions effectively model a wide variety of phenomena, including IQ scores.
It is assumed that IQ scores are distributed in such a way that most people will score around the expected value, or average, of 100. A few people will attain scores higher than 100, and a few people will attain scores lower than 100. We say that the IQ scores are distributed in a way known as a normal distribution. The shape of the normal distribution is shown in the following graph and is dependent upon the expected value (notated as m) and the standard deviation s. The derivation of the probability density function associated with a random variable that is normally distributed is beyond the scope of this textbook; however, the definition of the normal probability density function is given next.
821
NORMAL PROBABILITY DENSITY FUNCTION
A function P is a normal probability density function of a random variable x on the interval ( 2` , ` ) if P( x ) 5 e2(x 2m) >(2s ) for all x in ( 2` , ` ) , s ! 2p and
1
2 2
m 5 E( x )
y
s2 5 Var( x )
EXAMPLE
Computing the Area Under the Normal Probability Density Function

For an expected value m 5 0 and standard deviation s 5 1, show that the function P( x ) 5 s
2 2 1 e2(x 2m) >(2s ) meets the characteristics of a probability density function. ! 2p
SOLUTION Recall that a probability density function must meet two charac`
teristics: P( x ) $ 0 for all x on the interval ( 2` , ` ) and 3 P( x ) dx 5 1. Clearly, P( x ) $ 0 for all x because an exponential function multiplied by a positive constant is always positive. We will use technology to perform the integration. In this case, P( x ) 5
2 1 e2(x) >(2). ! 2p
2`
Since the TI-83 Plus cannot compute integrals on in-
finitely large intervals, we will get a sense of this integral by investigating the probability density function on the interval 32 5, 54 (Figure 12.27).
FIGURE 12.27
We can see that the area under the probability density function is close to 1 on the interval used. The web site www.IQTest.com reports that the expected value, or mean score, on an IQ test is 100 and the standard deviation is 15. However, there are
822
CHAPTER 12
different IQ tests with different mean scores and standard deviations. Nevertheless, for our purposes we will use m 5 100 and s 5 15, as reported on the web site.
EXAMPLE
Determining the Probability of a Particular IQ Score

Calculate the probability that a person chosen at random will have an IQ score between 85 and 115. What is the probability that the IQ score will be between 70 and 130?
SOLUTION Using the fact that m 5 100 and s 5 15, we can determine the normal probability density function modeling this situation.
P( x ) 5 5
15"2p 15"2p 1
e2(x 2 100) >(2 ? 15 )

2 2
e2(x 2 100) >450

2
We can use technology to evaluate the integrals needed to calculate the requested probabilities (Figure 12.28).
115
3
85
15"2p
e2(x 2 100) >450dx < 0.683

2
FIGURE 12.28
About 68% of people will have an IQ score between 85 and 115, inclusive (Figure 12.28). We next calculate the probability of an IQ score between 70 and 130.
130
3
70
15"2p
e2(x 2 100) >450dx < 0.954

2
About 95% of all people will have IQ scores between 70 and 130, inclusive (Figure 12.29).
FIGURE 12.29
Notice that the integrals found in Example 2 involved the area under the normal probability density function 1 standard deviation on either side of the mean and 2 standard deviations on either side of the mean. Regardless of the situation, the probabilities will be as found in Example 2. That is, the probability that a random variable will be in the interval 3m 2 s, m 1 s4 will always be approximately 68 percent. The probability that a random variable will be in the interval 3m 2 2s, m 1 2s4 will always be approximately 95 percent.
823
Technology can be used to create graphs of normal probability density functions. The Technology Tip at the end of the section details the process for doing so.
EXAMPLE
Predicting Mens Heights

According to the National Health and Nutrition Examination Survey (NHANES) administered by the Centers for Disease Control and Prevention, the mean height of 19-year-old men is 69.6 inches, with a standard deviation of 4.97 inches. Assuming the data are normally distributed, find the probability that a 19-yearold man chosen at random is between 5 feet 5 inches and 6 feet tall.
SOLUTION We first determine the normal probability density function corresponding to this situation.
P( x ) 5 5
4.97"2p 4.97"2p 1
e2(x 2 69.6) >(2 ? 4.97 )

2 2
e2(x 2 69.6) >49.4

2
To find the desired probability, we find the area under this probability density function on the given interval. We will have to determine the interval in inches. Since 5 feet 5 inches is 65 inches and 6 feet is 72 inches, we will find the area under the probability density function on the interval 365, 724 (Figure 12.30).
FIGURE 12.30
72
3
65
4.97"2p
e2(x 2 69.6) >49.4dx < 0.508

2
About 51 percent of 19-year-old men are between 65 inches and 72 inches tall. Technology can be used to find the area under a normal curve. A Technology Tip detailing the process for doing so can be found at the end of this section.
EXAMPLE 4
Determining Cholesterol Levels

In a study of 40 Italian men aged 46 to 80, it was found that the mean total cholesterol level was m 5 185.7 mg> dL, with a standard deviation of s 5 44.7. (Source: www.biomedcentral.com.) Assuming the cholesterol level of men is normally distributed, determine the percentage of men in this study who had an unhealthy total cholesterol level of greater than 200 mg/dL.
824
CHAPTER 12
SOLUTION We will use technology to determine this percentage (Figure 12.31).
FIGURE 12.31
Note that in this case there is no upper bound. To use the ShadeNorm command, we select an upper limit that is high enough to provide a good approximation for the percentage. In this case we chose 350 mg/dL as the upper limit. The result tells us that about 37 percent of the men in the study had unhealthy cholesterol levels.
EXAMPLE
Determining Cholesterol Levels

In a study of 40 Italian men aged 46 to 80, it was found that the mean total cholesterol level was m 5 185.7 mg> dL, with a standard deviation of s 5 44.7. (Source: www.biomedcentral.com.) Assuming the cholesterol level of men is normally distributed, determine the percentage of men in this study who had a healthy total cholesterol level of less than 200 mg/dL.
SOLUTION We will use technology to determine this percentage (Figure 12.32).
FIGURE 12.32
Note that in this case there is no lower bound. To use the ShadeNorm command, we select a lower limit that is low enough to provide a good approximation for the percentage. In this case we chose 0 mg/dL as the lower limit. Also note that since the entire area under the curve is 1, we could have subtracted the result of Example 4 from 1 to get 1 2 0.37 5 0.63. We find that about 63 percent of the men in the study had healthy cholesterol levels. We have seen how the normal probability density function can be used to determine probabilities or percentages in various situations. Sometimes it is useful to be able to work the process in reverse. That is, what if we wished to find the value of a random variable that marked the beginning of the top 10 percent of a distribution? Recall that IQ tests typically have an expected value, or mean score, of 100 and a standard deviation of 15. What IQ test score marks the beginning of the
825
highest 10 percent of all scores? Put another way, we need to determine the solution to the following equation, where Z represents the IQ score in question.
`
0.10 5 3
15"2p
e2(x 2 100) >450dx

2
Since there is no elementary antiderivative for this probability density function, we will use the process of guess-and-check to determine the value of Z. We know that about 95 percent of all scores are within 2 standard deviations of the mean. In this case, that includes scores between 70 and 130. Since we want to find the value of Z such that 90 percent of the scores are less than Z, we will guess that the score is 115. That is, we will guess that scores of 115 and above account for the top 10 percent of all scores (Figure 12.33).
FIGURE 12.33
Because a graphing calculator will not accept infinity as an upper limit, we choose the maximum x-value our window setting will allow. In this case, our Xmax value is 130. We see that a score of 115 or higher represents the top 13.6 percent of the scores. Therefore, we will have to choose a score slightly higher than 115. We choose 120 (Figure 12.34). Only about 7 percent of the scores are scores of 120 or above. We will try one more time to find the cut-off for the top 10 percent of scores. This time we choose 117 (Figure 12.35).
FIGURE 12.34
FIGURE 12.35
826
CHAPTER 12
The graphing calculator verifies that our guess of 117 is a good guess. We estimate that a score of about 117 represents the cut-off for the top 10 percent of all IQ test scores. Fortunately, rather than proceed through a tedious guess-and-check process, we can use technology to find the value of a random variable that marks the cutoff for a given probability. A Technology Tip detailing this process can be found at the end of this section.
EXAMPLE
Using the Inverse Normal Feature of a Calculator

Use the Inverse Normal feature of your graphing calculator to find the interval of IQ test scores representing the middle 50 percent of all scores. That is, find a score less than the mean of 100 and a score greater than the mean of 100 such that the mean is in the middle of these two scores and the area under the normal curve on this interval is 0.5, or 50 percent. Recall that the standard deviation is 15 points.
SOLUTION We will first find the lower value of the IQ score interval. Since we are looking for the middle 50 percent, we know that 25 percent of all scores lie below this lower value (Figure 12.36). We now find the upper value of the IQ score interval. Again, since we are looking for the middle 50 percent, we know that 75 percent of all scores lie below this upper value (Figure 12.37). Our results tell us that the middle 50 percent of all IQ test scores are scores between about 90 and 110. We can check our work using the ShadeNorm command (Figure 12.38).
FIGURE 12.36
FIGURE 12.37
FIGURE 12.38
Using Technology to Graph a Normal Probability Density Function

We can use technology to create the graphs of normal probability density functions if we know the expected value and standard deviation associated with the density function.
827
Graphing a Normal Probability Density Function 1. We will graph a normal probability density function with expected value 10 and standard deviation 2. To begin, press Y5 .
2. Press
2nd
VARS
and choose
1: normalpdf(.
3. The arguments required for the normalpdf option are the variable x, the expected value m, and the standard deviation s. In this case, we will enter normalpdf(x, 10, 2).
4. We need to set the WINDOW appropriately for the situation. In this case, since the expected value is 10 and the standard deviation is 2, we can select an Xmin of 4 (3 standard deviations less than 10) and an Xmax of 16 (3 standard deviations greater than 10). Next, press ZOOM and select 0:ZoomFit to obtain appropriate Ymin and Ymax values.
5. The graph will appear.
828
CHAPTER 12
Using Technology to Find the Area Under a Normal Curve

We can use technology to find the area under a normal probability density function.
Finding the Area Under a Normal Probability Density Function 1. Set the WINDOW appropriately for the problem situation. We will use a window setting suitable for the problem described in Example 3.
2. On the home screen, press 2nd VARS and press the right arrow key to select DRAW. Choose 1: ShadeNorm(.
3. The ShadeNorm option allows us to input the interval values 3a, b4 , the mean, and the standard deviation. Using the data from Example 3, we enter
ShadeNorm(65, 72, 69.6, 4.97)
and then press
ENTER
4. Notice that we get the same results as we did in Example 3. The advantage of the ShadeNorm command is that we dont need to enter the probability density function.
Using Technology to Find a Value for a Given Probability

We can use technology to find the value of a random variable that marks the cutoff for a given probability.
829
Finding a Value for a Given Probability 1. On the home screen, press 2nd and choose 3: invNorm(.
VARS
2. The invNorm option allows us to input the probability, or percentage, of the area under the normal curve to the left of score for which we are looking. For example, if we are looking for the top 10 percent of scores, 90 percent of the scores are less than the desired score. Next we enter the expected value (mean) and the standard deviation. Using the IQ score data from this section, we input invNorm(0.9, 100, 15) and press ENTER . 3. The actual score marking the top 10 percent of IQ scores is just over 119. This result is slightly different from our guess-and-check result because we used a score of 130 as our upper limit in the guess-and-check approach rather than infinity.
Graphing a Normal Probability Density Function 1. Record the mean in cell B2 and the standard deviation in cell C2. Then create column labels for the x and y columns.
(Continued)
830
CHAPTER 12
2. Start the x column at a numerical value 3 standard deviations below the mean (in this case, x 5 4) and continue to 3 standard deviations above the mean (in this case, x 5 16).
3. In cell B5, type =NORMDIST(A5,$B$2,$C$2,FALSE). The parameters of the normal distribution function are (x, mean, standard deviation, FALSE).
(Continued)
831
4. Copy cell B5 and paste it to the cells below it.
5. Graph the resultant values.
0.25 0.2 0.15 0.1 0.05 0 0 5 10 15 20
12.4 Summary
In this section, you learned about another common probability density function, the normal probability density function. This density function is dependent on the mean and standard deviation. The area under the normal probability density function 1 standard deviation on either side of the mean will be about 0.68. The area 2 standard deviations on either side of the mean will be about 0.95. We can use the normal probability density function to analyze situations that can be assumed to be normally distributed (e.g., the data follow the bell-curve shape).
832
CHAPTER 12
12.4 Exercises
In Exercises 110, use your graphing calculator to graph the normal probability density function with the given mean and standard deviation on the specified interval. Record the shape of the graph on paper. 1. m 5 5, s 5 0.25; 30, 104 2. m 5 5, s 5 0.5; 30, 104 3. m 5 5, s 5 1; 30, 104 5. m 5 5, s 5 2; 30, 104 4. m 5 5, s 5 1.5; 30, 104 6. m 5 0, s 5 1; 32 3, 84 7. m 5 1, s 5 1; 32 3, 84 8. m 5 2, s 5 1; 32 3, 84 10. m 5 4, s 5 1; 32 3, 84 9. m 5 3, s 5 1; 32 3, 84 18. 15.
0.8 0.6 0.4 0.2 0 0 2 y 0.8 0.6 0.4 0.2 0 0 2 4 6 8 x 4 6 8 x y
16.
0.8 0.6 0.4 0.2 0
17.
11. Referring to the graphs in Exercises 110, write several sentences describing what happens to the shape of a normal probability density function for a specific expected value m when the standard deviation s is changed. 12. Referring to the graphs in Exercises 110, write several sentences describing what happens to the shape of a normal probability density function for a specific standard deviation s when the expected value m is changed. In Exercises 1317, match the graph of the normal probability density function with the correct expected value and standard deviation given in AE. Explain how you were able to complete the task. A. m 5 5, s 5 0.5 C. m 5 3, s 5 1.5 E. m 5 6, s 5 1 13.
0.8 0.6 0.4 0.2 0 0 2 4 6 8 x y
Typically, IQ test scores are based on an expected value (mean score) of 100 and a standard deviation of 15. (Source: www.IQtest.com.) Assuming IQ test scores are normally distributed, what is the probability that a person selected at random will score 143 or higher on an IQ test?
IQ Test Scores
19.
Typically, IQ test scores are based on an expected value (mean score) of 100 and a standard deviation of 15. (Source: www.IQtest.com.) Assuming IQ test scores are normally distributed, what is the probability that a person selected at random will score between 40 and 59 on an IQ test?
IQ Test Scores
B. m 5 5, s 5 1 D. m 5 3, s 5 2 14.
0.8 0.6 0.4 0.2 0 0 2 4 6 8 x
20.
Explain, without doing any computations, how you could determine that approximately 68 percent of people taking an IQ test would attain scores between 85 and 115, inclusive. (Source: www.IQtest.com.)
IQ Test Scores
21.
Imagine that you are giving an IQ test to a large group of people. Assuming an expected value, or mean score, of 100 and a standard deviation of 15 (Source: www.IQtest.com), what middle range of scores do you predict 50 percent of the group will attain? That is, find a score below the mean and a score above the mean such that 50 percent of all scores will fall between those two scores. Explain how you computed your result.
IQ Test Scores
833
22.
Less than 1 percent of all IQ test takers can be classified as genius. (Source: www.IQTest.com.) A score of 145 or greater is required to be classified as genius. Assuming the test scores are normally distributed with a mean of 100 and a standard deviation of 15, verify that less than 1 percent of people can be classified as genius.
IQ Test Scores SAT Reasoning Test The mean combined score on the SAT Reasoning Test is 1029 (out of a possible 1600), with a standard deviation of 113. (Source: www .collegeboard.com.) To be admitted to Arizona State University (ASU), a combined score of 1040 or better is required. (Source: www.asu.edu.) Assuming SAT test scores are normally distributed, what percentage of students taking the SAT test would qualify to be admitted to ASU? SAT Reasoning Test The mean score on the math portion of the SAT Reasoning Test is 520 (out of a possible 800), with a standard deviation of 114. (Source: www.collegeboard.com.) Harvard University does not have strict scoring guidelines for admission, but reports that successful candidates typically have a score of 600 on the math portion of the SAT. (Source: admissions.college.harvard.edu.) Assuming SAT test scores are normally distributed, what percentage of students nationally would qualify to be admitted to Harvard University based on their SAT math score? Weights of 19-Year-Old Men
weights of 50 percent of all 19-year-old men fall between those two weights. Explain how you computed your result. 27.
Heights of NBA Players The mean height of an NBA player during the 20042005 season was 79.26 inches (about 6 feet 7 inches), with a standard deviation of 0.48 inch. (Source: www.nba.com.) Assuming NBA player heights are normally distributed, what percentage of NBA players are 7 feet tall or taller? What percentage of NBA players are 6 feet tall or shorter? Discuss the validity of these results. Heights of NBA Players The mean height of an NBA player during the 20042005 season was 79.26 inches (about 6 feet 7 inches), with a standard deviation of 0.48 inch. (Source: www.nba.com.) Assuming NBA player heights are normally distributed, what percentage of NBA basketball players are between 64 and 75 inches tall? Discuss the validity of these results.
23.
28.
24.
29. Explain why, if P( x ) 5

` 2`
s"2p
e2(x 2m) >(2s )

2 2
then 3 P( x ) dx 5 1. 30. For the normal probability density function P( x ) 5

2 1 e2x >2, ! 2p
25.
According to the National Health and Nutrition Examination Survey (NHANES) conducted by the Centers for Disease Control and Prevention, the mean weight of 19-year-old men is 172.1 pounds, with a standard deviation of 46 pounds. (Source: www.cdc.gov.) Assuming mens weights are normally distributed, what percentage of 19-year-old men weigh over 200 pounds? 26.
Weights of 19-Year Old Men
verify each of the following
integrals and explain what information each provides. 1 2x >2 (a) 3 e dx < 0.683 21 " 2p
2
1 2x >2 (b) 3 e dx < 0.954 " 2p 22

2
According to the National Health and Nutrition Examination Survey (NHANES) conducted by the Centers for Disease Control and Prevention, the mean weight of 19-year-old men is 172.1 pounds, with a standard deviation of 46 pounds. (Source: www.cdc.gov.) Assuming mens weights are normally distributed, what interval of weights accounts for the weights of 50 percent of 19-year-old men? That is, find a weight below the mean and a weight above the mean such that the
1 2x >2 (c) 3 e dx < 0.997 " 2p 23

2
31. Explain why, for the normal probability density function P( x ) 5

1
2 1 e2x >2, ! 2p
we can say that

1
2
1 2x >2 1 2x >2 2? 3 e dx 5 3 e dx 0 " 2p 21 "2p

2
Use graphs to support your explanation.
834
CHAPTER 12
32. For the normal probability density function P( x ) 5

2 1 e2x >2, ! 2p
show that
`
2
value of the function at this value of x. You may use a computer algebra system to assist you. 35. Find the value of x for which the normal probability density function P( x ) 5 s
2 2 1 e2(x 2m) >(2s ) ! 2p
3
2
e2x >2dx 5 3 e2x >2dx "2p 22 " 2p

2
is
and explain what these integrals represent. 33. Explain how we can find the interval into which 50 percent of the values of a random variable fall, given that the random variable is normally distributed.
maximized for constant values of m and s. Find the value of the function at this value of x. You may use a computer algebra system to assist you. 36. Find the inflection points of the normal probability density function P( x ) 5 (with m 5 0 and s 5 1). You may use a computer algebra system to assist you.
2 1 e2x >2 ! 2p
34. Find the value of x for which the normal probability density function P( x ) 5 (with m 5 0 and s 5 1) is maximized. Find the
2 1 e2x >2 ! 2p
37. Find the inflection points of the normal probability density function P( x ) 5 s
2 2 1 e2(x 2m) >(2s ). You ! 2p
may use a computer algebra system to assist you.

After working through this chapter, you should be able to answer the following questions, which are focused on the big ideas of the chapter. 1. What is a continuous random variable? (12.1) 2. What are the characteristics of a probability density function, and what does the area under a probability density function over a given interval represent? (12.1) 3. Given a particular data set, how is a probability density function created? (12.1) 4. What is a uniform probability density function, and for what types of situations would it make sense to use a uniform probability density function as a model? (12.2) 5. What is an exponential probability density function, and for what types of situations would it make sense to use an exponential probability density function as a model? (12.2) 6. How is expected value computed and what information does it provide? (12.3) 7. How is variance computed and what information does it provide? (12.3) 8. How is standard deviation computed and what information does it provide? (12.3) 9. What is a normal probability density function, and for what types of situations would it make sense to use a normal probability density function as a model? (12.4) 10. In what types of situations do we use the inverse normal function? (12.4) You should also be familiar with the following definitions; procedures, properties, and tests; and formulas that were emphasized in this chapter.
835
Definitions
Continuous Random Variable: A variable that can take on any value in some interval of real numbers. Probability Density Function: A function P is a probability density function of a random variable x on the interval 3a, b4 if P( x ) $ 0 for all x in 3a, b4 , and if 3 P( x ) dx 5 1.
a b
P( x ) 5 b
Uniform Probability Density Function: A function P is a uniform probability density function of a random variable x on the interval 3a, b4 if Exponential Probability Density Function: A function P is an exponential probability density function of a random variable x on the interval 30, ` ) if P( x ) 5 ae2ax. Expected Value: The average value produced by an experiment over the long run. If we multiply the probability of each possible outcome of an experiment by its value and then sum the results, we get the expected value. Variance: A measure of how dispersed (spread out) a random variable is from the expected value. Standard Deviation: The square root of the variance. Standard deviation describes the amount of deviation from the expected value. Normal Probability Density Function: A function P is a normal probability density function of a random variable x on the interval ( 2 ` , ` ) if P( x ) 5 s ! 2p e2(x 2m) >(2s ) for all x in ( 2 ` , ` ) and E( x ) 5 m,
1
2 2
1 2a
for all x in 3a, b4 .
Var( x ) 5 s2.
Creating a Probability Density Function sity function:
To create a probability den-
1. Create a scatter plot of the data. Often, the data will involve intervals. Use the midpoint of each interval as the data point for that interval. 2. Determine a regression model P( x ) for the data. 3. Adjust the function P( x ) by finding the value of c such that 3 c ? P( x ) dx 5 1 over the desired interval 3a, b4 . 4. To compute probabilities, use the function c ? P( x ) to find the area under the function over desired subintervals on 3a, b4 .
a b
Formulas
Expected Value of a Discrete Random Variable: E( X ) 5 a xi ? Pi

i51
Expected Value of a Continuous Random Variable: E( X ) 5 3 x ? P( x ) dx

a b
836
CHAPTER 12
Variance of a Discrete Random Variable: Var( x ) 5 a 3xi 2 E( x ) 4 2 ? Pi

n i51
Standard Deviation of a Discrete Random Variable: s( x ) 5 "Var( x ) Variance of a Continuous Random Variable: Var( x ) 5 3 3( x 2 E( x )) 2 ? P( x ) 4 dx
a b
Standard Deviation of a Continuous Random Variable: s( x ) 5 "Var( x ) Expected Value of a Uniform Probability Density Function: 1 E( x ) 5 ( a 1 b ) 2 Variance of a Uniform Probability Density Function: Var( x ) 5 ( b 2 a )2 12
Standard Deviation of a Uniform Probability Density Function: b2a s( x ) 5 "12
Expected Value of an Exponential Probability Density Function: 1 E( x ) 5 a Variance of an Exponential Probability Density Function: 1 Var( x ) 5 2 a Standard Deviation of an Exponential Probability Density Function: 1 s( x ) 5 a

Section 12.1 In Exercises 14, determine whether the function is a probability density function. Explain.
1. f ( x ) 5 x 1 1 on 30, 14 2 In Exercises 58, find the value of c that will make the function a probability density function on the given interval. 5. f ( x ) 5 cx 2 on 31, 54 6. g ( x ) 5 ce x on 30, 24 7. h( x ) 5 cx 4 on 30, 24 8. f ( t ) 5 c( t 3 1 t ) on 30, 24
2. g( x ) 5 4 2 5x on 30, 0.54 3. h( x ) 5 3 2 6 x 1 on 31, 24 13 13
4. r( t ) 5 3"t 2 1 on 30, 14
837
In Exercises 910, apply your knowledge of continuous probability functions to answer the questions. 9.
Alcohol-Related Car Accidents The table shows the number of deaths from alcohol-related car accidents, by age, in 2004.
Age Interval 1620 2124 2534 3544 4564
Midpoint of Interval 18 23 29.5 39.5 54.5
Number of Deaths 74 145 285 391 550
(Source: www.nhtsa.dot.gov.)
Age Interval 2534 3544 4554 5564 6574
Midpoint of Interval 29.5 39.5 49.5 59.5 69.5
Number of Deaths 6855 6383 5985 3813 2741
(a) Create a scatter plot showing the number of deaths as a function of the midpoint of each age interval. (b) Using the regression feature of your graphing calculator, find a quadratic model D( x ) for these data. (c) Find the value of c such that the function cD( x ) is a probability density function describing the probability that a driver in a particular age range died in an alcohol-related car accident in 2004. Use the overall age range of 25 to 74 to determine this value.
74
(a) Create a scatter plot showing the number of deaths as a function of the midpoint of each age interval. (b) Using the regression feature of your graphing calculator, find a quadratic model P( x ) for these data. (c) Find the value of c such that the function cP( x ) is a probability density function describing the probability that an intoxicated pedestrian in a particular age range died in a car accident in 2004. Use the overall age range of 16 to 64 to determine this value.
64
Then verify that 3 cP( x ) dx 5 1.

16
Then verify that 3 cD( x ) dx 5 1.

25
(d) Use the function cP( x ) to determine the probability that the age of an intoxicated pedestrian involved in a car accident in 2004 was between 16 and 25. (e) Use the function cP( x ) to determine the probability that an intoxicated pedestrian involved in a car accident in 2004 was older than 60 years old.
(d) Use the function cD( x ) to determine the probability that the age of a driver involved in a fatal car accident in 2004 was between 30 and 39. (e) Use the function cD( x ) to determine the probability that a driver involved in a fatal car accident in 2004 was older than 50 years old. 10.
Alcohol-Related Pedestrian Deaths
Section 12.2 In Exercises 1112, create a uniform probability density function over the indicated interval. Show all the integration steps necessary to create this function.
11. 31, 64 12. 32 8, 24
The table shows the number of intoxicated pedestrian deaths, in 2004, resulting from an accident with a motor vehicle. (Source:
www.nhtsa.dot.gov.)
In Exercises 1314, create an exponential probability density function for a function with the given continuous failure rate (such as the rate at which light bulbs burn out). Show all the integration steps necessary to create this function. 13. 3.2 percent 14. 24 percent
838 15.
CHAPTER 12
New York City Subway Wait Time
During peak hours, New York City subway riders should expect the time between subway trains to be uniformly distributed over the interval 30, 44 . (Source: www.mta.nyc.ny.us.) Find the probability that a subway rider will wait less than 2 minutes for a subway train. 16.
23. H( x ) 5 0.85e20.85x on 30, ` ) 24. g( x ) 5 0.859e20.6x on 30, 24 25. k( x ) 5 1 on 30, 154 15
26. Q( x ) 5 0.15x 2 1 0 . 15x 1 0.15 on 30, 24 27. M( x ) 5 0.75e20.75x on 30, ` ) 28. F( x ) 5 1.077e20.15x on 30, 14 29. P( x ) 5 30. 1 on 30, 504 50
During peak hours, New York City subway riders should expect the time between subway trains to be uniformly distributed over the interval 30, 44 . (Source: www.mta.nyc.ny.us.) Determine the interval of time that represents the shortest 5 percent of all wait times. 17.
New York City Subway Time to Failure
Data reveal that for March 2005, the average distance traveled by a New York City subway train before breaking down was 200,651 miles. (Source: www.mta.nyc.ny.us.) An exponential probability density function representing this situation is T( x ) 5 0.005e20.005x, where x is measured in thousands of miles. Determine the probability that a subway train will break down in the first 100,000 miles of travel. 18.
During peak hours, New York City subway riders should expect the time between subway trains to be uniformly distributed over the interval 30, 44 . (Source: www.mta.nyc.ny.us.) (a) Find the expected wait time. (b) Find the variance of the wait time. (c) Find the standard deviation of the wait time. 31.
Data reveal that for March 2006, the average distance traveled by a New York City subway train before breaking down was 172,641 miles. (Source: www.mta.nyc.ny.us.) An exponential probability density function representing this situation is T( x ) 5 0.0058e20.0058x, where x is measured in thousands of miles. Determine the probability that a subway train will break down after traveling more than 200,000 miles.
Data reveal that for March 2005, the average distance traveled by a New York City subway train before breaking down was 200,651 miles. (Source: www.mta.nyc.ny.us.) An exponential probability density function representing this situation is T( x ) 5 0.005e20.005x, where x is measured in thousands of miles. (a) Find E( x ) and explain what it means. (b) Find Var( x ) . (c) Find s( x ) . 32.
Section 12.3 In Exercises 1929, for each probability density function on the given interval, find the following. (a) The expected value (b) The variance (c) The standard deviation
19. f ( x ) 5 1 on 30, 84 8
Data reveal that for March 2006, the average distance traveled by a New York City subway train before breaking down was 172,641 miles. (Source: www.mta.nyc.ny.us.) An exponential probability density function representing this situation is T( x ) 5 0.0058e20.0058x, where x is measured in thousands of miles. (a) Find E( x ) and explain what it means. (b) Find Var( x ) . (c) Find s( x ) . 33. Compare the variance and standard deviation values found in Exercises 31 and 32. Explain what information your comparison provides in the context of the situation.
20. h( x ) 5 0.15e0.15x on 30, ` ) 21. r( t ) 5 3"t 2 1 on 30, 14 22. g( x ) 5 4 2 5x on 30, 0.54
839
Section 12.4 In Exercises 3438, sketch graphs of normal probability density functions having the given expected values (means) and standard deviation values. Describe the qualitative differences between the two graphs.
34. A. m 5 75, s 5 20 35. A. m 5 10, s 5 0.5 36. A. m 5 0, s 5 a 37. A. m 5 50, s 5 5 38. A. m 5 M, s 5 a 39. B. m 5 75, s 5 5 B. m 5 10, s 5 2 B. m 5 0, s 5 3a B. m 5 25, s 5 5 B. m 5 4M, s 5 a
41.
Batting Averages Data for the 2005
major league baseball season show that the mean batting average of the top 784 batters was 0.247, with a standard deviation of 0.103. Assuming batting averages are normally distributed, what is the probability that a randomly chosen player will have a batting average greater than 0.400? 42.
Human Pregnancy According to a
study conducted by Kieler, et al., the mean length of human pregnancy is 280.6 days when calculated using ultrasonic measurement of the fetal biparietal diameter (BPD) in the second trimester. According to the study, the standard deviation is 10.5 days. (Source: H Kieler, O
Axelsson, S Nilsson, U Waldenstr (1995). The length of human pregnancy as calculated by ultrasonographic measurement of the fetal biparietal diameter, Ultrasound in Obstetrics and Gynecology 6(5), 353357.) Premature births are successful when
major league baseball season show that the mean batting average of the top 784 batters was 0.247, with a standard deviation of 0.103. Assuming batting averages are normally distributed, what is the probability that a randomly chosen player will have a batting average between 0.280 and 0.350? 43.
major league baseball season show that the mean batting average of the top 784 batters was 0.247, with a standard deviation of 0.103. Assuming batting averages are normally distributed, what is the probability that a randomly chosen player will have a batting average of less than 0.200? 44.
the length of the pregnancy is at least 182 days. Assuming pregnancy lengths are normally distributed, determine the percentage of pregnancy lengths that are greater than 182 days but less than 282 days. 40.
Human Pregnancy According to a
study conducted by Kieler, et al., the mean length of human pregnancy is 280.6 days when calculated using ultrasonic measurement of the fetal biparietal diameter (BPD) in the second trimester. The standard deviation is 10.5 days.
(Source: H Kieler, O Axelsson, S Nilsson, U Waldenstr (1995). The length of human pregnancy as calculated by ultrasonographic measurement of the fetal biparietal diameter, Ultrasound in Obstetrics and Gynecology 6(5), 353357.) Assuming pregnancy
major league baseball season show that the mean batting average of the top 784 batters was 0.247, with a standard deviation of 0.103. Assuming batting averages are normally distributed, what is the batting average that represents the lowest average in the top 90 percent of all batting averages? That is, what is the cut-off batting average for the top 90 percent of all batting averages? 45.
lengths are normally distributed, what interval of pregnancy lengths accounts for 50 percent of all pregnancy lengths? That is, find a pregnancy length below the mean and a pregnancy length above the mean such that 50 percent of all pregnancy lengths fall between these two values. Explain how you computed your result.
major league baseball season show that the mean batting average of the top 784 batters was 0.247, with a standard deviation of 0.103. Assuming batting averages are normally distributed, what interval of batting averages accounts for the middle 50 percent of all batting averages? That is, find a batting average below the mean and a batting average above the mean such that 50 percent of all batting averages fall between these two values. Explain how you computed your result.
P R O J E C T
1 2
What to do
1. Choose a situation that you are interested in researching for which probability density functions may be constructed. Consider the data sets given in this chapter for ideas and for examples of the types of data sets that will work best. 2. Once you have identified a data set, construct a probability density function. Depending on the situation, you may or may not be able to use one of the special probability density functions. 3. Use your probability density function to find probabilities that are interesting to you. Also find the expected value, variance, and standard deviation. 4. Repeat this process for an additional two or three data sets related to the topic you are researching. 5. Write a technical report describing the situation, the associated probability density function, and how it was constructed. Write an explanation of the expected value, variance, and standard deviation.
Where to find data

Data for this project can be found at the following sites, as well as others: www.census.gov/statab/www/ www.cancer.org www.nrd.nhtsa.dot.gov www.IQTest.com www.google.com www.nba.com www.mlb.com www.nfl.com
840
Answers to Odd-Numbered Exercises

CHAPTER 1 Algebra Flashback 1.1 (page 14) 1. a will be a negative number and b will be a positive number. 3. t 5. It means to find f ( 4 ) . 7. f ( 2 2 ) 5 4( 2 2 ) 2 1 3( 2 2 ) 2 1 5 16 2 6 2 1 59 Section 1.1 (page 14) 1. Your weight is a function of your age, since at any instant in time you have only one weight. 3. Temperature is a function of the time of day. 5. The number of salmon in a catch is not a function of the number of fish caught. 7. C( 4 ) 5 159.80 The total cost of four pairs of shoes is $159.80. 9. H( 2 ) 5 56 Two seconds after he jumped, the cliff diver is 56 feet above the water. 11. E( 4 ) 5 0.06 In the fourth quarter since December 1999, shares in the tortilla company earned $0.06 per share. 13. P( 4 ) < 18.72 On November 8, 2001, the closing stock price of the computer company was approximately $18.72. 15. It appears that near x 5 1, the graph goes vertical. If it does so, a vertical line drawn at that point would touch the graph in multiple locations. However, if the graph doesnt actually go vertical near x 5 1, then it is a function. One drawback of reading a graph is that it is sometimes difficult to tell if the graph goes vertical or not. 17. The graph is a function. 19. The graph is a function. 25. y < 4; y 5 4 27. y < 2; y is undefined. 29. The domain is all real numbers. 31. The domain is all real numbers. 33. The domain is all real numbers except t 5 1. That is, 5t 0 t 2 16 . 35. The domain is all real numbers greater than or equal to 2 1. That is, 5a 0 a $ 2 16 . 21. 23.
37. The domain is the set of real numbers greater than or equal to 2 3. That is, 5x 0 x $ 2 36 . 39. The domain is the set of whole numbers. That is, 5n 0 n is a whole number6 .
41. Using our current calendar system, the domain of the function is the set of whole numbers between 1 and the current year. 43. They both indicate that when the input value to the function is 3, the output value is 8. 45. The independent variable is used car sales and the dependent variable is new car sales. 47. Yes. To be a function, each input (credits) must have only one output (tuition cost). Since two students taking the same number of credits at the same school have different tuition costs, the same input (credits) has more than one output (tuition cost). 49. The function f ( x ) 5 x2
x21 21
is undefined
whenever the denominator is equal to zero. The domain of the function is all real numbers except x 5 2 1 and x 5 1. 51. Yes. Even though the domain value of x 5 1 is listed twice in the table, it is linked with the same range value, y 5 6. A1
A2
Algebra Flashback 1.2 (page 30) 1. If we let x1 5 a, x2 5 b, y1 5 f ( a ) , and y2 5 f ( b ) , then the formulas are equivalent. 3. 2 mx 1 y 5 b 5. If b 5 d, the line is horizontal. 7. The term no slope may be interpreted as zero slope or an undefined slope. For clarity, it is better to use the mathematically precise terms undefined slope and zero slope. Section 1.2 (page 30) 1. m 5 2 1 3. m 5 2 0.2 5. m 5 0 7. y -intercept: ( 0, 10 ) x -intercept: ( 2 2, 0 ) 9. y -intercept: ( 0, 11 ) x -intercept: (25.5, 0) 11. y -intercept: ( 0, 2 4 ) x -intercept:
21. y 2 4 5 0.5( x 2 2 )
y
4
y 4 = 0.5(x 2)
2
2 2
23. 2x 2 3y 5 5
y
2
2 2 4
2x 3y = 5
25. y 5 2 3 x 1 3
y
2 1
13. The slope-intercept form of the line is y 5 2 x 1 7. The standard form of the line is x 1 y 5 7. A point-slope form of the line is y 2 5 5 2 1( x 2 2 ) . 15. The slope-intercept form of the line is y 5 2 0.2x 1 3.64. The standard form of the line is 5x 1 25y 5 91. A point-slope form of the line is y 2 3.4 5 2 0.2( x 2 1.2 ) . 17. The slope-intercept form of the line is y 5 0x 1 2 and is commonly written as y 5 2. The standard form of the line is 0x 1 y 5 2 and is also often written as y 5 2. A point-slope form of the line is y 2 2 5 0( x 1 2 ) . 19. y 5 4x 2 2
y
6 4 2 2 1 2 4 6 8 10 1 2
4 , 3
0B
y =2x + 4 3 3
1 1
27. The slope-intercept form of the line is y 5 3 x 1 3. 29. The equation of the line is x 5 3. 31. The data represent a linear function. m 5 993.2 year 2000 dollars per year Between 1989 and 1999, the U.S. average personal income (in year 2000 dollars) increased by an average of $993.20 per year. 33. The data represent a linear function. m 5 1183.01 dollars per month Between September 2001 and October 2001, the employees take-home pay increased at a rate of $1183.01 per month. 35. The data represent a linear function. m 5 $0.0375 per pound It costs an average of $0.0375 per pound to dispose of clean wood.
2
y = 4x 2
A3
37. In order to consume 8 grams of fiber, you would need to eat 3 servings
along with the large banana.
1 24
cups B of Wheaties
5. We have y 5 kx, with the y units being dollars and the x units being hours. Observe that k 5 Therefore, the units of the slope, k, are dollars per hour. 7. Since 4 $ 4, f ( 4 ) 5 10( 4 ) 1 1 5 41. Section 1.3 (page 52) 1. (a) The scatter plot shows that the data are nearlinear.
y dollars . x hours
39. The slope of a vertical line is undefined, and both the slope-intercept and point-slope forms require a defined slope. 41. No. A linear function has a constant rate of change. To be a linear function, the annual income would have to increase by a constant dollar amount each year. In this case, the firstyear increase is ( 0.03 )( 29,450 ) 5 $883.50. So the salary for the second year is 29,450 1 883.50 5 $30,333.50. The increase for the second year is ( 0.03 )( 30,333.50 ) 5 $910.01. Since this increase differs from the first-year increase, the function is not linear. 43. No. A constant dollar increase is not equivalent to a constant percentage increase. Tax (with 5% increase per year) 4000 4200 4410 4630.50 Tax (with $200 increase per year) 4000 4200 4400 4600
(b) The equation of the line of best fit is y 5 2 6.290x 1 110.9. (c) m 5 2 6.290 means that the Harbor Capital Appreciation Fund share price is dropping at a rate of $6.29 per month. The y -intercept means that in month 0 of 2000, the fund price was $110.94. Since the months of 2000 begin with 1, not 0, the y intercept does not represent the price at the end of January. It could, however, be interpreted as being the price at the end of December 1999. (d) This model is a useful tool to show the trend in the stock price between October and December 2000. Since stock prices tend to be volatile, we are somewhat skeptical of the accuracy of data values outside of that domain. (Answers may vary.) 3. (a) The scatter plot shows that the data are nearlinear.
Year 0 1 2 3
45. Both equations are linear functions. Since two points define a line, we need only show that the graphs of the equations share two points in common. The points ( 0, 3 ) and ( 4, 0 ) are on both lines. 47. 884,210 people 49. b 2 c and m 5 n. 51. In 2010 (or any other year) the world population is expected to be increasing at a rate of 81.3 million people per year. 53. No. 55. a must be equal to 12.
Algebra Flashback 1.3 (page 52) 1. The student reversed the order of the x terms in the denominator. The correct slope is m 5 3. 3. The student incorrectly stated that 0 5 0. The slope is actually undefined, since 0 is undefined.
9 9 2
(b) The equation of the line of best fit is y 5 1165.9x 1 80,887. (c) m 5 1165.9 means that Washington State public university enrollment is increasing by about 1166 students per year. The y -intercept means that, according to the model, Washington State public university enrollment was 80,887 in 1990.
A4
(d) The model fits the data extremely well, as shown by the graph of the line of best fit.
This model could be used by Washington State legislators and university administrators in budgeting and strategic planning. (Answers may vary). 5. (a) The scatter plot shows that although the data are not linear, they are nonincreasing.
For values of x $ 780, C( x ) is directly proportional to x . (b) $15.25 (c) $23.26 (d) The cost per pound is lowest when 780 or more pounds of trash are disposed of. As a construction company, we would try to keep the weight of our trash deliveries at or above 780 pounds. (Answers may vary.) 9. (a) C( t ) 5 e 29.99 0 # t # 200 0.4t 2 50.01 t . 200 (b) For the Qwest plan, we have C( 300 ) 5 $82.49 For the Sprint plan, we have C( 300 ) 5 $69.99 The Sprint plan is the better deal for a customer who uses 300 anytime minutes.
(b) The equation of the line of best fit is y 5 2 3.4x 1 39.6. (c) m 5 2 3.4 means that the per capita income ranking of North Carolina is changing at a rate of 3.4 places per year. That is, the state is moving up in the rankings by about 3 places per year. The y -intercept means that in 1995 (year 0), North Carolina was ranked 40th out of the 50 states. (Only positive whole-number rankings make sense.) (d) This model is not a highly accurate representation of the data, as shown by the graph of the line of best fit, so it should be used with caution.
11. (a) Let t be the production year of a Toyota Land Cruiser 4-Wheel Drive and V be the value of the vehicle in 2001. V 5 4505t 2 8,966,350 (b) $7610 (c) The model substantially underestimated the value of a 1992 Land Cruiser. 13. (a) Let F be the number of grams of fat in x Chef Salads. We have F 5 8x. The number of fat grams is directly proportional to the number of Chef Salads. (b) Less than 10 salads (c) Up to 7.75 salads (d) Up to 5.875 salads 15. Let s be the number of segments and C be the maximum total cost of the ticket (in dollars) from Seattle to Phoenix. We have C 5 10s 1 222 Note that s $ 1. 17. If the companys expenses E( x ) are directly proportional to the number of items produced x, then E( x ) 5 kx for some constant k. The term kx represents the variable cost. There is no fixed cost. Therefore, if the company does not have any fixed cost, the companys expenses are directly proportional to the number of items produced. 19. Although a mathematical model predicts what may happen in the future, it does not determine what will happen.
However, an incumbent government official could use the model in a 2000 reelection campaign as evidence that the states economy had improved during his or her tenure in office. The official might also use the model to claim that the trend of improvement will continue if he or she is reelected. (Answers may vary.) 7. (a) 15.25 C( x ) 5 0.04274x 1 1.036 0.04409x 0 # x # 320 340 # x # 760 780 # x
A5
21. If the scatter plot of a data set appears to have two or more linear or near-linear pieces, a piecewise linear function may be an appropriate model to use. 23. a is the cost of making a cup of coffee, b is the cost of making a bagel, c is the cost of making a muffin, and d is the amount of the fixed costs. Fixed costs may include rent on the building, electricity costs, equipment costs, etc. 25. Production costs are not typically directly proportional to the number of items produced. This is because fixed costs such as building rent, equipment, and so on are usually associated with production. 27. Since the line passes through the origin, we know the model equation may be written as y 5 mx. Furthermore, since the correlation coefficient is equal to 2 1, the linear model has a negative slope and fits the data perfectly. Thus, the output of the original data is directly proportional to the input of the original data. 29. The piecewise linear function is 2x 1 2, 0#x#2 y5 e x 1 4, 2,x#6 CHAPTER 1 REVIEW EXERCISES Section 1.1 (page 58) 1. C( 2 ) 5 99.90 The cost to buy two pairs of shoes is $99.90. 3. H( 2 ) 5 36 The cliff diver is 36 feet above the water 2 seconds after he jumps from a 100-foot cliff. 5. The domain is all real numbers. 7. The domain is all real numbers except r 5 1 and r 5 2 1. 9. At t 5 2, P < 21.03. The stock price of the computer company at the end of the day two days after December 16, 2001, was about $21.03. 11. The graph represents a function. Section 1.2 (page 59) 13. m 5 2 3 15. m 5 2 11
19. y 5 2 1x 1 7 In standard form, x 1 y 5 7. Section 1.3 (page 59) 21. (a) Let x be the number of super size orders of French fries consumed. The amount of fat consumed is given by F( x ) 5 29x grams. (b) Let y be the number of Big N Tasty sandwiches consumed. The amount of fat consumed is given by G( y ) 5 32y grams. (c) Only one combination meal may be eaten. 23. (a) Let t be the production year of the MercedesBenz Roadster two-door SL500 and v( t ) be the value of the car in 2001. v( t ) 5 7312.5t 2 14,558,975 (b) $58,712.50 (c) The linear model was extremely effective at accurately predicting the value of the 1999 Mercedes-Benz Roadster two-door SL500. The $212.50 difference between the predicted value of the model and the NADA guide average value was negligible. CHAPTER 2 2.1 Algebra Flashback (page 80) 1. In compliance with order of operations, the calculator performs exponentiation before multiplication. 2 12 is equivalent to 2 1 ? 12 5 2 1 ? 1 5 2 1. 3. a 5 3, b 5 2 6, c 5 4 5. The graph does not have any x-intercepts. 7. The Quadratic Formula is x 5 . 2a It calculates the x-coordinates of the x-intercepts. Section 2.1 (page 81) 1. Concave up, y-intercept ( 0, 1 ) , vertex ( 1, 0 ) 3. Concave up, y-intercept ( 0, 0 ) , vertex ( 2 0.5, 2 0.75 ) 5. Concave up, y-intercept ( 0, 2.1 ), vertex (0.25, 1.925) 7. y 5 2 x2 1 4x 2 1 9. y 5 x2 2 10x 1 5
2 b 6 " b2 2 4ac
17. The y -intercept is ( 0, 18 ) . The x -intercept is ( 6, 0 ) .
A6
11. (a) W( t ) 5 36.92t2 2 43.60t 1 2072 (b) 19. (a) (c) As shown in the scatter plot, the model appears to fit the data well. It seems reasonable that womens basketball game attendance will continue to increase; consequently, we believe this model is good. 13. (a) D( t ) 5 896.67t2 1 723.33t 2 540 (b)
increase, we anticipate that the model will be relatively accurate in forecasting per capita personal incomes between 1980 and 2010.
(b) The scatter plot does not seem to resemble a parabola. Consequently, we do not think that a quadratic function will fit the data well. 21. (a) (See part d.) (b) The scatter plot roughly resembles a parabola, so we will model the data with a quadratic function. (c) The quadratic model is M( t ) 5 0.02628t2 1 0.3340t 1 47.46 (d)
(c) As shown in the scatter plot, the model appears to fit the data well. Since the scatter plot fits the data well and we expect that DVD shipments will continue to increase, we believe this model is a good model. 15. (a) C( v ) 5 2 0.03550v2 1 85.451v 2 47,197 (b)
(c) As shown in the scatter plot, the model fits the data perfectly. Although the scatter plot fits the data perfectly, we believe that the model has some definite limitations. According to the model, as the number of blank VHS tapes drops below 1195, the number of CD-R disks will also begin to decrease. This doesnt seem reasonable. We anticipate that CD-R disk sales will continue to increase even as VHS tape sales decrease. 17. (a) (See part d.) (b) Since the data set consists of three nonlinear points, a quadratic model will fit the data perfectly. (c) The quadratic model is I( t ) 5 2 10.31t2 1 972.4t 1 22,530 (d)
(e) Since the model fits the data and since it seems reasonable to assume that the upward trend in conventional mortgages will continue, we anticipate that the model will be relatively accurate in forecasting the percentage of new privately owned one-family houses financed with a conventional mortgage between 1977 and 2005. (Note: 2005 is arbitrarily selected as a year near the last data point.) (f) When t < 5.4, M( t ) 5 50. Since t is the number of years since the end of 1970, we estimate that in mid-1976 ( t 5 5.4 ), 50 percent of mortgages were conventional mortgages. 23. (a)
(b) The scatter plot does not seem to resemble a parabola. Consequently, we do not think that a quadratic function will fit the data well. 25. (a) (See part d.) (b) The scatter plot seems to resemble a portion of a parabola. We think that a quadratic model will fit the data. (c) The quadratic model is H( t ) 5 2 0.01293t2 1 1.398t 1 57.60
(e) Since the model fits the data and since we expect personal incomes to continue to
A7
(d)
(e) We believe the model is a good model because it seems reasonable to assume that the percentage of homes with a garage will continue to increase. (f) Since H( 33.6 ) < 90, we estimate that in mid-2004 ( t 5 33.6 ) , 90 percent of new homes had garages. 27. (a) (See part d.) (b) Although the scatter plot appears to be near-linear, we can still construct a quadratic model. When quadratic models are used to model near-linear functions, the value of a is relatively close to zero. (If we had our choice of which type of model to construct, we would construct a linear model because it is simpler.) (c) The quadratic model is P( t ) 5 2 0.02143t2 1 8.279t 1 68.27 (d)
31. (a) (See part d.) (b) The scatter plot looks like a concave up parabola, so we believe that a quadratic model will fit the data well. (c) The quadratic model is S( t ) 5 29.23t2 1 79.33t 1 177.4 (d)
(e) Since the model fits the data extremely well, and since we expect Starbucks sales to continue to increase, we believe we have a good model. (f) Since S( 11.6 ) < 5000, we expect that midway through fiscal year 2005 ( t 5 11.6 ) , Starbucks sales will reach $5000 million. 33. (a) (See part d.) (b) The scatter plot looks more or less concave up and increasing. A quadratic model will probably be a good choice for the data set. (c) The quadratic model is P( t ) 5 0.0007346t2 1 0.009701t 1 0.7803 (d)
(e) Since the model fits the scatter plot fairly well, and since home sales prices typically increase over time, we believe that our model will relatively accurately predict future home prices in the northeastern United States. 29. (a) (See part d.) (b) The scatter plot looks concave up. Consequently, we expect that a parabola will fit the data. (c) The quadratic model is P( t ) 5 0.0417t2 1 0.06833t 1 2.455 (d)
(e) Since the model fits the data relatively well, and since we expect the price of chicken to continue to increase (because of inflation), we believe we have a good model. 35. (a) (See part d.) (b) The scatter plot is generally concave up and increasing. We believe that a quadratic model will fit the data well. (c) The quadratic model is S( t ) 5 0.9195t2 1 23.25t 1 2652 (d)
(e) Since the model fits the scatter plot fairly well, and since auto leasing appears to be increasing in popularity, we believe that our model will relatively accurately predict the percentage of households leasing vehicles. (f) Since P( 7 ) < 5, we estimate that at the end of 1996, 5 percent of households were leasing vehicles.
(e) Since the model fits the data relatively well, and since it seems reasonable to anticipate that private college enrollment will continue to increase, we believe that the model is a good model.
A8
37. (a) (See part d.) (b) The scatter plot appears concave up and increasing. A quadratic model may fit the data well. (c) The quadratic model is P( t ) 5 123.73t2 1 445.58t 1 22,847 (d)
(c) The quadratic model is F( t ) 5 12.66t2 1 48.52t 1 2102 (d)
(e) Since the model fits the data relatively well, and since it seems reasonable to anticipate that personal incomes will continue to increase, we believe the model is a good model. 39. (a) (See part d.) (b) The scatter plot appears near-linear. Nevertheless, we can come up with a decent quadratic model for the data. (c) The quadratic model is P( t ) 5 2 14.470t2 1 883.18t 1 17,620 (d)
(e) Since the model fits the data well, and since it seems reasonable that the payroll will continue to grow, we believe that the model is a good model. 45. (a) (See part d.) (b) The scatter plot appears more or less concave down and increasing. A quadratic model may fit the data well. (c) The quadratic model is F( t ) 5 2 0.04079t2 1 1.453t 1 401.6 (d)
(e) Since the model fits the data relatively well, and since it seems reasonable to anticipate that personal incomes will continue to increase, we believe that the model is a good model. 41. (a) (See part d.) (b) The scatter plot appears slightly concave down and increasing. A quadratic model should fit the data well. (c) The quadratic model is P( t ) 5 2 32.71t2 1 778.1t 1 2341 (d)
(e) Since the model fits the data well, and since it seems reasonable that the number of private-school teachers will continue to increase, we believe that the model is a good model. 47. (a) (See part d.) (b) The scatter plot appears more or less concave up and increasing. A quadratic model may fit the data well. (c) The quadratic model is W( t ) 5 0.003197t2 1 0.1333t 1 4.606 (d)
(e) Since the model fits the data well, and since it seems reasonable that the companys profits will continue to increase, we believe that the model is a good model. (f) Since P( 6.5 ) < 6000, we estimate that midway through 2004 ( t 5 6.5 ) , Johnson & Johnson profits reached $6000 million. 43. (a) (See part d.) (b) The scatter plot is concave up and increasing. A quadratic function may fit the data well.
(e) Since the model fits the data well, and since it seems reasonable that wages will continue to increase, we believe that the model is a good model. 49. (a) (See part d.) (b) The scatter plot appears concave up. A quadratic model may fit the data well. (c) The quadratic model is A( t ) 5 9.396t2 2 11.58t 1 1063 (d)
A9
(e) Since the model fits the data well, and since it seems reasonable that advertising expenditures will continue to increase, we believe that the model is a good model. (f ) Since A( 10.6 ) < 2000, we estimate that in mid-2001 ( t 5 10.6 ) , billboard advertising expenditures reached $2000 million ($2 billion). 51. The x-intercepts of a parabola occur at x5
2 b 6 " b2 2 4ac 2a 2 b 2 " b2 2 4ac . 2a
3. Cubic functions have exactly one inflection point, whereas quadratic functions have no inflection points. 5. For both types of functions, if y S ` as x S ` , then y S ` as x S 2` . Similarly, if y S 2` as x S ` , then y S 2` as x S 2` . 7. At most two bends Section 2.2 (page 107) 1. The quartic function that best fits the data is given by S( t ) 5 2 0.53891t4 1 15.050t3 2 103.86t2 1 48.261t 1 12,393. S(16 ) 5 12,904 means that there were 12,904 thousand students enrolled in high school in 2000. 3. The quadratic function that best fits the data is R( t ) 5 1.874t2 2 1.425t 1 2165. R( 21.5 ) < 3000 means that the number of public elementary and secondary teachers reached 3 million (3000 thousand) in mid-2002 ( t 5 21.5 ) . 5. The cubic function that best fits the data is C( t ) 5 2 0.5278t3 1 5.905t2 2 4.234t 1 4.048. C( 7 ) < 83 means that 83 percent of publicschool classrooms had Internet access in 2001. C( 8 ) < 78 means that 78 percent of publicschool classrooms had Internet access in 2002. The 2001 estimate seems reasonable, since we expect Internet access in schools to become increasingly common. The 2002 estimate does not seem reasonable, since it forecasts a decline in Internet access. 7. The cubic model that best fits the data is C( t ) 5 0.1563t3 2 2.268t2 1 14.84t 1 164.0. C( 15 ) < 404 means that there will be 404 million debit cards by the end of 2005. Our model estimate is substantially higher than the consulting firms 270 million card projection. 9. The quartic function that best fits the data is given by S( t ) 5 4.807t4 2 98.56t3 1 541.3t2 1 2936t 1 10,610. S( 25 ) < 760,000 means that 760 billion shares were traded in 2005. Because of the rapid growth in the market in recent years, this estimate seems reasonable. However, the recession in the early 2000s may have affected the number of shares sold. 11. The quartic function that best fits the data is W( t ) 5 100t4 2 1518.5t3 1 8244.4t2 2 17,294t 1 42,667. W( 6.5 ) < 40,000, meaning that in mid-1999 ( t 5 6.5 ), the average annual
and x2 5
To determine the x value halfway between the two x-intercepts, we calculate the average of the two values. 2 b 1 "b2 2 4ac 2 b 2 "b2 2 4ac 1 x1 1 x2 2a 2a 5 2 2 2 b 1 "b2 2 4ac 2 b 2 "b2 2 4ac 2a 5 2 2 2b 5 4a b 52 2a Since we know that x 5 2 2a is the x coordinate of the vertex, we agree that our classmate is right. 53. Any function with a positive slope is said to be increasing. This means that as x gets bigger, y gets bigger. Concavity describes the rate at which the slope is changing. If the rate of change of the slope of a function is positive, we say that the function is concave up. The graph of a concaveup function curves upward. 55. The graph of a quadratic function is a parabola. Since 2 2 , 0, the parabola is concave down everywhere. Thus our classmate is incorrect. 57. No. Consider the points ( 0, 0 ) , ( 0, 1 ) , and ( 0, 2 ) . Each of these points lies on the x-axis. Since a quadratic function has at most two x-intercepts, no quadratic function has a graph that passes through all three points. (Answers may vary.) 59. It is an x-intercept of the graph. 2.2 Algebra Flashback (page 106) 1. A polynomial is a function of the form y 5 anxn 1 an 2 1xn 2 1 1 c 1 a1x 1 a0 .
b
A10
wage per worker for computer and equipment retailers first reached $40,000. 13. The quadratic function of best fit is W( t ) 5 1421.4t2 2 2321.4t 1 56,429. W( 7.6 ) < 121,000, meaning that in mid-2000 ( t 5 7.6 ) , the average wage per worker in the prepackaged software development industry reached $121,000. 15. The cubic model that best fits the software wholesalers wages is W( t ) 5 113.89t3 2 251.19t2 1 277.78t 1 52,571. We estimate that at the end of 2002, the average wage of both software wholesalers and software retailers was $144,000. 17. The quartic model that best fits the data is W( t ) 5 51.515t4 2 701.52t3 1 3119.7t2 2 2450.4t 1 41,563. W( 7 ) < 60,000, meaning that the television broadcasting average wage will first exceed $60,000 at the end of 1999. This estimate may be a bit high, but it is not unreasonable. Since t 5 7 is close to the end of the data set ( t 5 6 ) and $60,000 is somewhat close to $54,600, we are comfortable with the estimate. 19. The cubic model that best fits the data is W( t ) 5 2 0.005860t3 1 0.2430t2 2 0.06623t 1 6.541. C( 32 ) < 61.2, meaning that 61.2 percent of TV homes had cable at the end of 2002. We would generally expect the percentage of homes with cable to continue to increase, as it has over the 29 years of the data set. For this reason, the estimate seems unreasonable. However, if people are replacing cable with satellite dishes or some other technology, it is possible that the percentage of homes with cable could decline. 21. A function with an increasing rate of change is concave up. 23. The graph appears to change concavity twice. A quartic function may best fit the data. 25. A quartic function can have zero or two inflection points.
y
6 4 2 1 2 1 2 3 4 5 x
then x 2 c is a factor of p( x ) . Therefore, p( x ) may be written in the form p( x ) 5 ( x 2 c1 )( x 2 c2 ) c( x 2 ck ) q( x ) where the ci are the x values of the x-intercepts of p( x ) and q( x ) is the polynomial that remains after one factor for each x-intercept is factored out of p( x ) . The product of the k factors of the form x 2 ci will be of the form xk 1 ak 2 1xk 2 1 1 c 1 a1x 1 a0 , where the ai are constants. If q( x ) 5 b (b is a constant), then the degree of p( x ) is k. If q( x ) is a polynomial of degree m, then the degree of p( x ) is k 1 m. Since the degree of a polynomial with k x-intercepts is at least k, the number of x-intercepts of a polynomial function can never exceed the degree of the polynomial. 29. y 5 x3 2 x 2.3 Algebra Flashback (page 131) 1. A linear function has a constant rate of change, whereas an exponential function has a constant percentage change. 3. A horizontal asymptote of a function f is a line y 5 b with the property that as x S ` or x S 2` , f ( x ) S b. 5. An exponential function is of the form y 5 abx, whereas a polynomial function is of the form y 5 anxn 1 an 2 1xn 2 1 1 c1 a1x 1 a0 . 7. No. This is a power function. An exponential function must be of the form y 5 abx. Section 2.3 (page 131) 1. The graph is decreasing and concave up. The y-intercept is (0, 4).
y
6 4 2 1 0 1 2 3 4 5
3. The graph is increasing and concave up. The y-intercept is (0, 0.5).
y
12 8
27. No. Let p( x ) be a polynomial with k x-intercepts. If ( c, 0 ) is an x-intercept of the polynomial p( x ) ,
4 1 0 1 2 3 4
A11
5. The graph is increasing and concave up. The y-intercept is (0, 0.4).
y
160 120 80 40 1 0 1 2 3
29. p( t ) 5 59.90( 1.28 ) t where p is the Dave Matthews Band concert tickets price and t is the number of years since September 2004. A concert ticket that costs $59.90 in September 2004 is expected to cost $98.14 in September 2006. 31. x 5 8.827 33. x 5 14.21 35. x 5 2 37. bx is always positive, so the graph of y 5 abx can never cross the x-axis. 39. The data must be increasing at a constant percentage rate. 41. (a) V( t ) 5 1000( 1.0867 ) t 1 2000( 1.0693 ) t , where V is the investment value (in dollars) and t is the number of years from now. (b) 12,913 dollars 43. (a) At the end of 2002, the value of the Social Choice account was $970.20 and the value of the Stock account was $792.70. (b) The combined value of the accounts is projected to reach $2000 in August 2004 ( t < 1.6 ) . 45. It will take 18.71 years for the accounts to attain the same value. At that time, each account will be worth $5950.79. 2.4 Algebra Flashback (page 146) 1. A logarithm is an exponent. 3. log2( 12 ) 5 log2( 22 ? 3 ) 5 log2( 22 ) 1 log2( 3 ) 5 2 1 log2( 3 ) 5. Since 52 5 25 and 53 5 125, log5( 30 ) lies between 2 and 3. Since 30 is closer to 25 than to 125, log5( 30 ) will be closer to 2 than to 3. 7. ln( 8 ) ln( 23 ) 5 ln( 2 ) ln( 2 ) 3 ln( 2 ) 5 ln( 2 ) 53
7. The graph is decreasing and concave down. The y-intercept is ( 0, 2 1.2 ) .

1 5 10 15 20 25 1 2 3
9. The graph is decreasing and concave up. The y-intercept is (0, 3).
y
3 2 1 5 0 5 10 15 20 25 x
11. y 5 2( 3 ) x 1 x 15. y 5 64 a b 2 1 x 19. y 5 256 a b 2
13. y 5 5( 2 ) x 17. y 5 0.25( 4 ) x
21. I( t ) 5 82.67( 1.061 ) t (rounded model) I( 25 ) 5 364.5 (using unrounded model) At the end of 2005, dental prices are expected to be 364.5 percent of the 1984 price. 23. I( t ) 5 106.1( 0.9657 ) t (rounded model) I( 25 ) 5 44.4 (using unrounded model) At the end of 2005, the price of a television set is expected to be 44.4 percent of the 1984 price. 25. I( t ) 5 86.75( 1.051 ) t (rounded model) I( 25 ) 5 302.7 (using unrounded model) At the end of 2005, the price of admission to an entertainment venue is expected to be 302.7 percent of its 1984 price. 27. B( t ) 5 235( 1.0232 ) where B is the balance (in dollars) and t is the number of years from now. In just over 2 years and 8 months ( t < 2.7 ) , the balance is projected to reach $250.
t
Section 2.4 (page 146) 1. y 5 2 3. y 5 6

y
1 0 1 2 3 2 4 6
5. y 5 2 2
7. y 5 log4( x ) is concave down and increasing.
A12
9. y 5 log0.7( x ) is concave up and decreasing.

y
8 4 0 4 2 4 6
11. x 5 6 17. x 5 1 16
13. x 5 2 1 19. x 5 243
15. x 5 3 21. 4 log( 2 )
23. log( 2x4 ) 27. log a 1 216 x b 324
25. log( jam ) 29. 2 log( x ) 35. ln( 27 )
By the end of 1998 ( t 5 14 ) , 84.6 percent of TV homes had VCRs. We need to increase this percentage by 5.4 percentage points in order to reach the desired 90 percent. Mid-2000 is just 1.5 years after the end of 1998. To attain the desired percentage, the percentage of VCRs would need to increase by 3.6 points per year. Looking at the raw data, the last time the annual increase in percentage reached or exceeded 3.6 percentage points was in 1989. (From 1989 to 1990, the percentage increased by 4 percentage points.) So, even though the correlation coefficient is extremely close to 1 ( r 5 0.9955 ) , we expect that our model prediction is a bit low. That is, we believe it will take more than 1.5 years to reach the desired 90 percent. (Answers may vary.) 53. When x , 1, logb( x ) . logc( x ) . 55. c 5 b2 2.5 Algebra Flashback (page 160) 1. Linear, quadratic, cubic, quartic, exponential, logarithmic, and logistic 25 3. 21 5 1 1 17e20.25x 20.25x 21( 1 1 17e ) 5 25 25 1 1 17e20.25x 5 21 25 17e20.25x 5 21 21 4 17e20.25x 5 21 4 e20.25x 5 17( 21 ) 4 e20.25x 5 357 4 2 0.25x 5 ln a b 357 4 x 5 2 4 ln a b 357 5. Linear 7. On a calculator, set y1 5 3.95 and y2 5 2.17t3 2 9.55t 1 6.84. Then use the calculators INTERSECT feature to determine the point of intersection of the two functions. The corresponding function input values are the solutions to the equation.
31. 2 ln( x ) 37. ln a 1 b 64
33. 2 ln( x )
39. ln( 729x5 )
41. T( i ) 5 2 73.69 1 16.71 ln( i ) T( 125 ) 5 7.0 According to the model, the Consumer Price Index reached 125 seven years after 1980. That is, at the end of 1987, the price of dental services was 125 percent of their 1984 price. 43. T( i ) 5 129.8 2 27.76 ln( i ) T( 125 ) 5 2 4.2 According to the model, the Consumer Price Index was at 125 4.2 years before the end of 1980. That is, in late 1976, the price of a television set was 125 percent of its 1984 price. 45. T( i ) 5 2 88.48 1 19.84 ln( i ) T( 125 ) 5 7.3 According to the model, the Consumer Price Index was at 125 7.3 years after the end of 1980. That is, in early 1988, the price of admission to an entertainment venue was 125 percent of its 1984 price. 47. The classmate is correct. y 5 logb( x ) means that y is the exponent placed on b in order to get x. In other terms, by 5 x. 49. The graph should be concave down and increasing with a vertical asymptote at x 5 0. 51. (a) V( t ) 5 21.70 1 24.93 ln( t ) (b) t 5 e(V 2 21.70)>24.93 (c) Ninety percent of TV homes were expected to have VCRs by mid-2000 (15.48 years after the end of 1984).
A13
Section 2.5 (page 161) 1. S( t ) 5 8.451t 2 111.9t 1 470.0t 1 965.0 Since S( 12 ) < 5099, we forecast that gaming hardware sales will be $5099 million in 2002.
3 2
19.
0.1x if 0 # x # 7000 T( x ) 5 0.15x 2 350 if 7000 , x # 28,400 0.25x 2 3190 if 28,400 , x # 68,800
3. C( t ) 5 0.3214t2 1 0.2500t 1 37.36 Since C( 8 ) < 60, we forecast that the 200506 tuition cost per credit will be $60. 5. S( t ) 5 1165.9t 1 80,887 Since S( 20 ) < 104,200, we forecast that the 2010 public university enrollment will be 104,200. 7. C( t ) 5 1106 1 1 0.1259e0.5491t Since S( 13 ) < 7, we forecast that there will be 7 pediatric AIDS cases in the United States in 2005. 38.56 1 16 1 1 14.52e20.8032t Since A( 10 ) < 54, we forecast that there were 54 air carrier accidents in 2002.
21. Mathematical models cannot determine the future or the past. They may predict unknown results, but they do not determine unknown results. 23. These statistics indicate how well the model fits the data. The closer the absolute value of r or r2 is to 1, the better the model fits the data. 25. Your classmate is incorrect. Mathematical models cannot determine the future or the past. They may predict unknown results, but they do not determine unknown results. 27. A cubic model or a logistic model may be used to model the function on the interval 30, 84 , and a linear function with negative slope may be used to model the function on the interval 38, 154 . Two different functions that exhibit the indicated behavior are given below. 2 x 3 1 15x 2 1 3x 1 3 0 # x # 8 2 95x 1 1235 8 , x # 15 100 0#x#8 1 1 ( e2 ) e20.40x g( x ) 5 2 x 1 84.852 8 , x # 15 (Answers may vary.) f (x) 5 e 29. You might want to model the average price over a particular time period (e.g., 90 days or 180 days). Modeling the average price rather than the daily price minimizes the effects of daily fluctuations. (Answers may vary.) CHAPTER 2 REVIEW EXERCISES Section 2.1 (page 168) 1. The quadratic model that best fits the data is A( t ) 5 42.293t2 1 158.32t 1 6574.5 Since A( 12 ) < 14,564, we estimate that $14,564 million was spent on magazine advertising in 2002. 3. Early in the second quarter of 2001 ( t < 10.3 ) . 5. A( t ) 5 34.992t2 1 94.909t 1 8963.2 Since A( 11.8 ) < 15,000, we estimate that $15 billion was spent on Yellow Pages advertising in the one-year period that ended in late 2002 ( t 5 11.8 ) .
9. A( t ) 5
11. C( t ) 5 0.60( 1.03 ) t dollars, where t is the number of years from now. 13. P( t ) 5 3t 1 87 dollars, where t is the number of years from now. 15. The logistic model for monthly product sales is S( t ) 5 200.67 1 1 25.85e20.4973t
17. The logistic model for the data is 6930 1 2600 1 1 166.2e20.4019t The maximum projected population of the city is 6930 1 2600 5 9530. According to the logistic model, the 2003 population was 6258, since P( 13 ) < 6258. The exponential model that best fits the aligned data is P( t ) 5 P( t ) 5 67.59( 1.377 ) t 1 2600 According to the exponential model, the 2003 population was 6925, since P( 13 ) < 6925. Although this estimate is still below the actual 2003 population (7480), it is substantially better than the logistic model estimate. (Answers may vary.)
A14
Section 2.2 (page 169) 7. A( t ) 5 2 32.085t 1 656.16t 2 1640.7t 1 31,914 Since A( 11 ) < 50,557, we estimate that $50,557 million was spent on newspaper advertising in 2001 ( t 5 11 ) .
3 2
Section 2.4 (page 172) 25. y 5 2 1 27. The graph of y 5 log0.4( x ) is decreasing and concave up.
y
3 2 1 1 1 2 1 2 3 4 x
9. A( t ) 5 10.222t4 2 221.51t3 1 1616.4t2 2 2487.8t 1 26,671 Since A( 11.0 ) < 50,000, we estimate that $50 billion was spent on broadcast TV advertising in 2001 ( t 5 11 ) . Broadcast TV annual advertising expenditures beyond 2001 are expected to exceed $50 billion. 11. E( t ) 5 0.001228t3 2 0.03338t2 1 0.6439t 1 9.844 Since E( 23 ) < 21.94, we estimate that average hourly earnings of a manufacturing industry employee in Michigan in 2003 ( t 5 23 ) was $21.94. 13. The manufacturing wage in Florida is roughly 65 percent of the Michigan manufacturing wage. Thus labor costs in Florida will be substantially less than labor costs in Michigan. For this reason, a new manufacturing business may prefer to start up in Florida as opposed to Michigan. However, additional factors such as business tax rate, infrastructure, availability and cost of raw materials, etc., should also be considered. (Answers may vary.) Section 2.3 (page 172) 15. The function y 5 2 0.3( 2.8 ) x is concave down and decreasing. The y-intercept is ( 0, 2 0.3 ) .
2 1 2 4 6
29. x 5 4 33. log( 2x )

2
31. x 5 0.04 35. 2 log( 3 )
37. F( t ) 5 49.96 1 5.112 ln( t ) . Since F( 10 ) < 61.7, we estimate that in 1999, 61.7 percent of auto accident fatalities were not alcohol-related. Section 2.5 (page 173) 39. P( t ) 5 57.81 1 1 0.5443e0.3900t
CHAPTER 3 3.1 Algebra Flashback (page 183) 1. m 5 y2 2 y1 x2 2 x1
3. The average temperature is the temperature obtained by averaging a number of temperatures recorded at equally spaced time intervals. The units of the average temperature are equal to the units of the temperature (e.g., degrees). The average rate of change in temperature is expressed as
units of temperature units of time
y = 0.3(2.8) x
1 2
and represents the
average change in temperature over the interval per unit of time (e.g., 4 degrees per hour). 5. The units are the units of output divided by the units of input. 7. The slope of the line is undefined and x1 5 x2. Section 3.1 (page 183) 1. 2 7. 3. 2 1 5. 0 9. "2 < 0.2357 6
17. y 5 9( 2 )
19. N( t ) 5 22,925( 1.0727 ) t N( 10 ) < 46,248 means that there will be 46,248 McDonalds restaurants in 2007. 21. E( t ) 5 4500( 1.023 ) t E( 5 ) < 5042 means that five years from now, my monthly household expenses are expected to be $5042. 23. x < 1.723
ln( 5 ) < 0.0805 20
A15
11.
7 degrees per hour 4
37. < 2 427 people per month 39. < 407 people per month 41. f ( b ) 2 f ( a ) represents the change in the output values and b 2 a represents the change in the input values. The difference quotient gives the change in output per one unit change in input. 43. The average rate of change of a function f over an interval 3a, b4 is the slope of the secant line through the points ( a, f ( a )) and ( b, f ( b )) . 45. The statement means that if the driver had driven at a constant rate of exactly 50 miles per hour between Phoenix and Salt Lake City (without stopping), the trip would have taken the same amount of time. In actuality, the drivers speed varied throughout the trip. On interstate freeways, he may have traveled at 75 miles per hour. While stopping for gas or food, his speed was 0 miles per hour. 47. 2a 1 h
y
6 5 4 3 2 1
13. Between 1996 and 1998, 1366.55 points per year; between 1997 and 1999, 1794.40 points per year 15. 30 words per minute per quarter 17. 2 0.378 newspaper subscription per cable TV subscriber 19. 1.375 people below poverty level per unemployed person 21. f ( x ) 5 2x
y
16 12 8 4 0 0 1 2 3
f (3) 2 f (1) 53 321 23. f ( x ) 5 5
49. Slope of secant 5 h 2 0 5 h. So as h gets smaller, so does the slope of the secant line. Both values approach 0.
2 x
h2 2 0
3.2 Algebra Flashback (page 202) 1. Any of the following formulas may be used. f ( a 1 h ) 2 f ( a ) f ( b ) 2 f ( a ) y2 2 y1 D y , , , x2 2 x1 D x h b2a 3. It means that as h gets close to 0, the value of the function f gets close to 10. 5. f ( x 1 h ) 5 2 3( x 1 h ) 2 1 7( x 1 h ) 2 1 5 2 3( x 2 1 2hx 1 h2 ) 1 7x 1 7h 2 1 5 2 3x 2 2 6hx 2 3h2 1 7x 1 7h 2 1 7. 35( 2 1 1 h ) 3 1 9( 2 1 1 h ) 4 2 35( 2 1 ) 3 1 9( 2 1 ) 4 h 5( 2 1 1 3h 2 3h2 1 h3 ) 1 ( 2 9 1 9h ) 2 ( 2 5 2 9 ) h 2 5 1 15h 2 15h2 1 5h3 2 9 1 9h 1 14 h 24h 2 15h2 1 5h3 h 24 2 15h 1 5h2
2 1
f (3) 2 f (1) 50 321 25. f ( x ) 5 ( x 2 2 ) 2

y
4 3 2 1 0 0 1 2 3
f (3) 2 f (1) 50 321 27. f (3) 2 f (1) 52 321 29. f (3) 2 f (1) 5 21 321 5 5 5 5
31. < 147 people per month 33. < 2013 people per month 35. 2300 people per month
A16
Section 3.2 (page 202) 1. h 0.1 0.01 0.001 Difference Quotient 4.1 4.01 4.001
23.
h 0.1 0.01 0.001
Difference Quotient 2 1.7305 2 1.7388 2 1.7396
Instantaneous rate < 2 1.740 25. h 0.1 0.01 0.001 Difference Quotient 14,198.385 9,588.682 9,231.052
Instantaneous rate < 4.0 3. h 0.1 0.01 0.001 Difference Quotient 2 65.6 2 64.16 2 64.016
Instantaneous rate < 9,200 27. h 0.1 0.01 0.001 Difference Quotient 0.4879 0.4988 0.4999
Instantaneous rate < 2 64.0 5. h 0.1 0.01 0.001 Difference Quotient 4 4 4
Instantaneous rate < 0.50 29. h 0.1 0.01 0.001 Difference Quotient 70.271 61.170 60.347
Instantaneous rate < 4.0 7. h 0.1 0.01 0.001 Difference Quotient 0 0 0
Instantaneous rate < 60.0 31. $1339 per year. In 2000, teacher salaries are increasing at the rate of $1339 per year. 33. $14.3 billion per year. In 2005, the amount of money spent on prescription drugs will be increasing at the rate of $14.3 billion per year. 35. $0.871 billion per year. Video-game sales in the first quarter of 2007 will be increasing at a rate of $0.871 billion per year. 19. 1070 37. $31.91 per year. In 2001, the cost of full-time resident tuition at Green River Community College was increasing at a rate of $31.91 per year. 39. 242.5 million disks per year. In 2002, the demand for CD-R disks was increasing at a rate of 242.5 million disks per year. 41. < 2 51 feet per second
Instantaneous rate < 0 9. h 0.1 0.01 0.001 Difference Quotient 1,120 1,075 1,070.5
Instantaneous rate < 1,070 11. 4 21. 13. 2 64 h 0.1 0.01 0.001 15. 4 17. 0
Difference Quotient 2 0.06940 2 0.07358 2 0.07402
Instantaneous rate < 2 0.074
A17
43. < 2 12 feet per second 45. < 2 15 feet per second 47.
f (a 1 h) 2 f (a) h
5. lim
xS9
ln( x 2 8 ) 51 x29 2x 2 4 2 b 54 x22
is the more useful form. Since h represents the distance between the endpoints of the interval, letting h S 0 allows us to calculate the instantaneous rate of change at the point ( a, f ( a )) .
7. lim a
xS5
9. The limit does not exist. 11. lim 22x 5 0

xS`
49. It means that as t approaches 2 from the right or from the left, the function g( t ) approaches the same finite value. 51. The derivative is lim difference
f (x 1 h) 2 f (x) , whereas h hS0 f (x 1 h) 2 f (x) quotient is . h
13. lim ( ex 1 200 ) 5 0

x S 2`
the
ln x 50 3x 2 4 3 17. lim 21 5 0 x S 2` x 1 x 19. lim ( 4x 29 1 x 4 ) 5 ` 15. lim

xS` xS`
53. Not necessarily. Consider f ( x ) 5 x3 . Since f r( x ) 5 3x2 , f r( 1 ) 5 f r( 2 1 ) 5 3. However, 1 2 2 1. Therefore, different values of x may have the same derivative. 55. This function will have the form f ( x ) 5 c, where c is an arbitrary constant. There are two cases: (1) c 5 0 1 infinitely many x-intercepts (2) c 2 0 1 no x-intercepts 3.3 Algebra Flashback (page 215) x 1. , x 2 25 x11 3. As x approaches 5 from the right, the expression 5 5 approaches 5 1 1 5 6. 5. x 2 1 9x 1 14 x12 5 , x 2 27 2 x 2 49 x27 As x nears 7 through values less than 7, the numerator of the expression is positive but the denominator is negative and approaching zero. Therefore, the function approaches 2 ` . 7. For large values of x, x 2 1 1 behaves like x 2 5 x . So the expression approaches zero as x goes to infinity. Section 3.3 (page 215) 1. lim
xS1
21. Continuous 23. Discontinuous at x 5 3 25. Continuous 27. Continuous 29. Discontinuous at x 5 0 31. lim I( d ) 5 5617 means that as net domestic sales
dS`
go to infinity, net international sales will approach 5617 million dollars (according to the model). 33. lim 1194 20.6872t 1 2100 5 3294 t S ` 1 1 17.04e The maximum number of prescriptions will be about 3294 million (according the model). r( 40 ) r( 50 ) r( 60 ) r( 1000 ) < < < < 9.070 9.012 9.002 9.000
35.
x11
The death rate will be 9.070 deaths per 100,000 people in 2010; 9.012 deaths per 100,000 people in 2020; 9.002 deaths per 100,000 people in 2030. The long-term death rate is projected to be 9 deaths per 100,000 people. 37. Imagine that the 1 in the numerator represents one cookie and the x in the denominator represents the number of people sharing the cookie. As the number of people sharing the cookie increases, the portion given to each person decreases. If we try to share the cookie with infinitely many people, the portion size will be reduced to an infinitely small piece (near zero).
2x 2 4 2( 1 ) 2 4 5 52 x22 122 x2 2 4 5 24 x12
3. lim
xS 22
A18
39. We mean that the graph has no breaks (holes, jumps) in it. It may be drawn without lifting the pencil off the paper. 41. No. Let f ( x ) 5 x 2 and h( x ) 5 x 4. Then g ( x ) 5
xS0 xS0
13. 20.27 square feet per year; 2238.5 square feet 15. Increasing at the rate of 0.128 student per teacher per year; 15.157 students per teacher 17. When the number of cassette tapes shipped is 250 million, the value of the tapes is increasing at a rate of 7.775 million dollars per million cassette tapes shipped. The estimated shipment value when 251 million cassette tapes are shipped is about 2134 million dollars. 19. The rate of change in private college enrollment when 12,752 public students are enrolled is 0.340 thousand private college enrollments per thousand public college enrollments. (In other words, 340 private enrollments per 1000 public enrollments.) We estimate that when 12,753 public college students are enrolled, 3879 private college students are enrolled. 21. $1750 per year. In 1997, the annual wage was increasing at a rate of $1750 per year. 23. 2 13.1 newspapers per year. In 1995, the number of different daily newspapers in the United States was decreasing at a rate of about 13 newspapers per year. 25. $26.50 per year; $34 per year. The rate of increase is itself increasing.
1 x2 1 x4
5 x 2. Although
xS0
lim f ( x ) 5 ` and lim h( x ) 5 ` ,
lim g ( x ) 5 0. 5, x 2 3 1, x 5 3
43. f ( x ) 5 e
3.4 Algebra Flashback (page 229) 1. You use the difference quotient. 3. From 2 3 to 0 and from 2 to 4. 5. They are equal. 7. The slope is 0. Section 3.4 (page 229) 1.
2 1 11 2 3 4 1 2 3 4 5 x
y = x 2 4x = f (x) y
3.
g (x) = x 2 + 2x + 1 y
4 3 2 1 (0, 1) 1 1 1 2
(1, 3)
Tangent line: y = 2x 1
Tangent line: y = 2x + 1 y
27. 1.1 million people per year. In 1997, the attendance at American League baseball games was increasing at a rate of 1.1 million people per year. 29. Increasing at a rate of 5.1 million people per year 31. It represents the slope of the secant line through ( a, f ( a )) and ( a 1 h, f ( a 1 h )) . 33. Let A have coordinates ( a, f ( a )) . Two points horizontally equidistant from this point are ( a 2 h, f ( a 2 h )) and ( a 1 h, f ( a 1 h )) . The average rate of change between these two points is f (a 1 h) 2 f (a 2 h) f (a 1 h) 2 f (a 2 h) 5 (a 1 h) 2 (a 2 h) 2h f (a 1 h) 2 f (a) 1 f (a) 2 f (a 2 h) 5 2h ( f ( a 1 h ) 2 f ( a )) 1 ( f ( a ) 2 f ( a 2 h )) 5 2h ( f ( a 1 h ) 2 f ( a )) 1 ( f ( a 1 h ) 2 f ( a )) < 2h
5.
g (x) = x 2 4x 5 y
2 12 4 6 8 10 1 2 3 4 5 x
7.
6 4 (0, 0) 2 2 2 4 6
h (x) = x 3
2
(4, 5) Tangent line: y = 4x 21 f (x) = (x 3)2
Tangent line: y =0
9.
y
10 8 6 4 2 2 0
(3, 0) Tangent line: y =0

2 4 6 x
11. 2.356 billion gallons per year; 55.75 billion gallons
Assuming the vertical distance between f ( a 1 h ) and f ( a ) is roughly the same as the distance between f ( a 2 h ) and f ( a )
A19
<
2( f ( a 1 h ) 2 f ( a )) 2h f (a 1 h) 2 f (a) < h
15.
x 1 3 5
hr( x ) 0 0 0 Sr( x ) 4 16 28
17.
x 1 3 5
Wr( x ) 24 24 24
If we let h S 0, we get f r( a ). Therefore, the average rate of change between ( a 2 h, f ( a 2 h )) and ( a 1 h, f ( a 1 h )) is approximately equal to f r( a ). 35. The tangent line passes through ( a, f ( a )) and is near the graph of f for values of x near a (see figure).
y
19.
x 1 3 5
21. Pr( x ) < 1.099( 3x )
23. Cr( x ) < 2 4.162( 4x ) 25. Rr( x ) < 2 0.102( 0.98x )

Tangent a x
27. 2000 31. 1999
29. Northeast
37. Possible pairs: (0, 0) and (4, 8) or ( 1, 2 1 ) and (3, 3). Note that for each pair, the point (2, 4) is horizontally equidistant. That is, the point (2, 4) lies horizontally in the middle of each pair. 39. gr( 3 ) 5 2 3.5 Algebra Flashback (page 239) 1. g ( x 1 h ) 5 3( x 1 h ) 2 1 2( x 1 h ) 1 1 5 3( x 2 1 2hx 1 h2 ) 1 2x 1 2h 1 1 5 3x 2 1 6hx 1 3h2 1 2x 1 2h 1 1 3. g ( x 1 h ) 2 g ( x ) 5 ( x 1 h ) 3 2 x 3 5 x 3 1 3hx 2 1 3h2x 1 h3 2 x 3 5 3hx 2 1 3h2x 1 h3 5. g ( x 1 h ) 5 2 3( x 1 h ) 2 1 7( x 1 h ) 1 2 5 2 3( x 2 1 2hx 1 h2 ) 1 7x 1 7h 1 2 5 2 3x 2 2 6hx 2 3h2 1 7x 1 7h 1 2 7. g ( x 1 h ) 5 2x 1 h Section 3.5 (page 240) 1. f r( x ) 5 2x 2 4 5. gr( x ) 5 2x 2 4 9. f r( t ) 5 2t 2 6 11. x 1 3 5 gr( x ) 5 13 21 13. x 1 3 5 f r( x ) 0 4 8 3. gr( x ) 5 2x 1 2 7. hr( x ) 5 3x2
33. Since t 5 19 is the end of 2009, we estimate that in mid-2010 per capita prescription drug spending will be increasing at a rate twice that of the 2000 rate. 35. 5.517 thousand students per year 37. The left-hand and right-hand limits of the difference quotient differ as h S 0. Therefore, the limit of the difference quotient as h S 0 does not exist. Thus the derivative does not exist. 39. The shortcuts work only for certain types of functions; however, the limit definition of the derivative can be used for all functions with derivatives. 41. f r( x ) 5 e Also, f r( x ) 5 43. At x 5 1 45. The result of Exercise 44 indicates the x-values ( x 5 61 ) at which the graph of f has a horizontal tangent, a possible location for a local minimum or maximum value.
y 50 f(x) = x3 3x 40 30 20 10 1 10 1 2 3 x
2 1, x # 0 1, x . 0
Zx Z x
A20
3.6 Algebra Flashback (page 245) 1. The cost of ten 3.5-pound bags of Boston Baked Beans is $73.35. 3. 212 degrees Fahrenheit is equivalent to 100 degrees Celsius. 5. The cost of 20 gallons of gasoline is $64.38. 7. The volume of a cube with a surface area of 6 square units is 1 cubic unit. Section 3.6 (page 245) 1. According to the model, the weight of a 10-yearold girl is 76.64 pounds. A 10-year-old girl will gain weight at a rate of 8.24 pounds per year. An 11-year-old girl will weigh about 84.88 pounds. 3. According to the model, the annual per capita income in Washington state in 1984 was $14,204. The income in 1984 was increasing at a rate of $178.40 per year. The income in 1985 was about $14,382.40. 5. According to the model, the average daily volume of the Nasdaq market in 1998 was 918 million shares. In 1998, this volume was growing at the rate of 299 million shares per year. The volume in 1999 was about 1217 million shares. 7. According to the model, the annual volume of the New York Stock Exchange in 1998 was 168,659 million shares. This volume was growing at the rate of 38,760 million shares per year. This volume in 1999 was about 207,419 million shares. 9. According to the model, the average annual wage of an employee in the radio broadcasting industry in 1998 was $34,086. The wage in 1998 was increasing at a rate of $2750 per year. The wage in 1999 was about $36,836. 11. According to the model, the amount of money spent on billboard advertising in 2000 was $1887 million. This amount was increasing at a rate of $176 million per year in 2000. 13. According to the model, when the net number of CDs shipped is 900 million, the value of the CDs is $12,261.9 million. When the net number of CDs shipped is 900 million, the value of the CDs is increasing at a rate of $17.9 million per million CDs shipped.
15. According to the model, the number of women in the Summer Olympics in 1992 was 2662 and was growing at a rate of 150 women per year. 17. According to the model, the population of the United States in the year 2000 was 284 million people and was growing at a rate of 2.70 million people per year. 19. According to the model, the number of homicides resulting from an alcohol-related brawl in the year 2000 was 185. In 2000, this number was decreasing at a rate of 14 homicides per year. 21. According to the model, the gross revenue from sales of Johnson & Johnson and its subsidiaries at the end of 2001 was $32,317 million and was increasing at a rate of $3810 million per year. 23. According to the model, the operating profit of Frito-Lay North America at the end of 2001 was $2056 million and was increasing at the rate of $93.5 million per year. 25. According to the model, the flow of the Snoqualmie River at noon on November 24, 2002, was 300 cubic feet per second and was decreasing at 45 cubic feet per second per 24-hour period. The river rafting company could predict when the flow was fast, slow, safe, unsafe, and so on. 27. According to the model, the per capita personal income of Colorado at the end of 1998 was $28,634 and was increasing at a rate of $1700.10 per year. 29.
Projected Growth in Income (dollars) 791.7 1,700.1 1,998.6 Projected Per Capita Income in 1999 (dollars) 24,503.7 30,334.1 38,765.6
Per Capita Income in 1998 (dollars) California Colorado Connecticut 23,712 28,634 36,767
31. If the derivative is zero, the billing rate is remaining constant. If the derivative is negative, the billing rate is decreasing. From a consumers perspective, a decreasing billing rate is a good thing.
A21
33. The odometer is the distance function. The speedometer is the derivative of the distance function. It shows the instantaneous rate at which the distance function is changing. 35. We know we get
f (x 1 h) 2 f (x) h
seventh-place runners time is increasing at the rate of 0.0687 second per place (rank). Section 3.3 (page 252) 15. 0 17. 0 19. Continuous
< f r( x ) . Letting h 5 1,
Section 3.4 (page 253) 21. y 5 2x 2 2

y f (x) = 2x 2 2x
f (x 1 1) 2 f (x) < f r( x ) 1 f ( x 1 1 ) 2 f ( x ) < f r( x ) f ( x 1 1 ) < f ( x ) 1 f r( x ) 37. As time increases, the slope of the tangent line is getting steeper. Hence the annual sales income was increasing most rapidly at t 5 9, or during the year 2002. 39. Since f ( a ) # f ( x ) for all x, f ( a ) is a minimum value for f. In addition, f is a continuous, smooth function on the interval ( 2 ` , ` ) , so the tangent line at ( x, f ( a )) must be horizontal. Hence f r( a ) 5 0. CHAPTER 3 REVIEW EXERCISES Section 3.1 (page 251) 1. 6 3. $230.2 per year 5. 12
10 8 6 4 2 (1, 0) 1 2 1 2
Tangent line: y = 2x 2
23. < $36.33 per year; < $996.12 25. 2 649.2 fatalities per year. In 1985, the number of highway fatalities was decreasing at a rate of about 649 fatalities per year. Section 3.5 (page 254) 27. f r( x ) 5 4x 29. 8 at x 5 1; 20 at x 5 3; 32 at x 5 5 31. Pr( x ) < 0.693( 2x ) Section 3.6 (page 254) 35. According to the model, McDonalds sales were $27,015 million in 2002 and were increasing at a rate of $1233.5 million per year. CHAPTER 4 4.1 Algebra Flashback (page 263) 1. 15x 2 1 8x 5 x( 15x 1 8 ) 3. 216x 22 1 33x 2 1 8x 5. x 4>3 7. 5x 23>2 33. 1999
Section 3.2 (page 252) 7. h 0.1 0.01 0.001 Difference Quotient 4.2 4.02 4.002
Instantaneous rate < 4.0 9. 4 11. h 0.1 0.01 0.001 Difference Quotient 0.9777 0.9767 0.9766
Section 4.1 (page 264) 1. 0 7. 2 4t 2 2 11. y 5 0.5x 1 0.5 15. y 5 551.23t 2 1203.9 17. y 5 3.8n 1 102.2 19. y 5 3.66x 2 12.74 21. 140.05 thousand acres per year 3. 15t 2 1 20t 23 5. 22 1 12x 2 9. 3t 2 1 6t 1 6 13. y 5 27x 2 1.4
Instantaneous rate < 0.977 13. T r( 7 ) < 0.0687 second per rank. In the 100-meter mens race in the 2000 Olympics, the
A22 23. Yes 27. No
25. 9-year-old girl 29. 2000
Section 4.2 (page 272) 1. 2( 3x 1 4 ) 1 3( 2x ) 3. 2( 10t 1 5 ) 1 10( 2t 2 6 ) 5. 6n( n3 2 n ) 1 ( 3n2 2 1 )( 3n2 1 8 ) 7. ( 2t 1 2 )( t 2 1 4t 1 4 ) 1 ( 2t 1 4 )( t 2 1 2t 1 1 ) 9. 3x 2( x 4 2 x 3 ) 2 ( 4x 3 2 3x 2 )( x 3 2 1 ) (x4 2 x3)2 1 2 13. 25.5 15. 56 19. 2 4
31. The equation yr 5 2x may be used to calculate the slope of the graph of y 5 x 2 at any point. 33. d means find the derivative with respect to x dx dy whereas means the derivative of y with dx respect to x.
35. n 2 1 37. See Chapter 3 Review, Exercise 31. If f ( x ) 5 2x, then f r( x ) < 0.693( 2x ) 2 x ? 2x 2 1. 39. f ( x ) 5 k ? g( x ) f (x 1 h) 2 f (x) f r( x ) 5 lim hS0 h k ? g( x 1 h ) 2 k ? g( x ) 5 lim hS0 h k 3g( x 1 h ) 2 g( x ) 4 5 lim hS0 h 3g( x 1 h ) 2 g( x ) 4 5 k ? lim hS0 h 5 k ? gr( x ) 4.2 Algebra Flashback (page 272) 1. 4x( 5x 2 1 2x 1 7 ) 1 ( 10x 1 2 )( 2x 2 ) 5 20x 3 1 8x 2 1 28x 1 20x 3 1 4x 2 5 40x 3 1 12x 2 1 28x 5 4( 10x 3 1 3x 2 1 7x ) 3. ( 14x 1 12 )( x 3 2 1 ) 1 ( 7x 2 1 12x 2 4 )( 3x 2 ) 5 14x 4 1 12x 3 2 14x 2 12 1 21x 4 1 36x 3 2 12x 2 5 35x 4 1 48x 3 2 12x 2 2 14x 2 12 5. 6x( x 2 2 1 ) 2 2x( 3x 2 ) 6x 3 2 6x 2 6x 3 5 2 2 (x 2 1) (x2 2 1)2 26x 5 2 (x 2 1)2 2 x 22( x 2 1 5x 1 1 ) 2 ( 2x 1 5 )( x 21 ) x 2 1 5x 1 1 21 2 5x 21 2 x 22 2 2 2 5x 21 5 x 2 1 5x 1 1 23 2 10x 21 2 x 22 x 2 5 ? 2 x 2 1 5x 1 1 x 3x 2 1 10x 1 1 52 2 2 x ( x 1 5x 1 1 )
11. 2
17. 589
21. < 1504.35 million dollars per year 23. < 178.74 million dollars per year 25. < 1369.96 million dollars per year 27. d f (x) f r( x ) g( x ) 2 gr( x ) f ( x ) c d 5 dx g( x ) 3g( x ) 4 2 f r( x ) g( x ) gr( x ) f ( x ) 5 2 2 3g( x ) 4 3g( x ) 4 2 21 5 f r( x ) 3g( x ) 4 2 3g( x ) 4 22 gr( x ) f ( x ) The Quotient Rule supports the classmates claim. 29. It is a matter of personal preference. We think it is more efficient to multiply out the product and then find the derivative. y 5 ( 3x 1 5 )( 2x 2 1 ) 5 6x2 1 7x 2 5 yr 5 12x 1 7 y 5 ( 3x 1 5 )( 2x 2 1 ) yr 5 3( 2x 2 1 ) 1 2( 3x 1 5 ) 5 6x 2 3 1 6x 1 10 5 12x 1 7
7.
31. f r( 25 ) 5 1 means that when 65 trees per acre are planted, planting an additional tree will increase the overall yield by about 1 bushel of apples per acre. f r( 30 ) 5 0 means that when 70 trees per acre are planted, planting an additional tree will not change the overall yield of apples per acre. f r( 35 ) 5 21 means that when 75 trees per acre are planted, planting an additional tree will decrease the overall yield by about 1 bushel of apples per acre. 33. Yes. The functions in both Exercise 31 and Exercise 32 are quadratic, so they both have a maximum value. Furthermore, for each function, the derivative changes from a positive value to a negative value, which indicates that the function goes from increasing to decreasing, implying that
A23
f ( x ) achieves a maximum somewhere in between. It appears that the maximums occur at x 5 30 and x 5 45 for Exercises 31 and 32, respectively. 35. $0.575 per pound 4.3 Algebra Flashback (page 283) 1. 4( 2x 1 1 ) ( 2 ) 5 8( 2x 1 1 )
3 3
35. The Chain Rule is used to find the derivative of the composition of functions. The derivative itself also contains a composition of functions. That is, d ( f ( g( x )) 5 f r( g( x )) gr( x ) . dx 37. The Leibniz form is useful in a real-world context dy dx because we may know and from the context dx dt of the problem even if we dont know the function equations for y and x. The Leibniz notation form dy enables us to determine . dt 39. g( t ) 5 ( t 2 1 1 ) 3( t 2 1 3 ) 4 gr( t ) 5 3( t 2 1 1 ) 2( 2t )( t 2 1 3 ) 4 1 4( t 2 1 3 ) 3( 2t )( t 2 1 1 ) 3 5 2t( t 2 1 1 ) 2( t 2 1 3 ) 3 33( t 2 1 3 ) 1 4( t 2 1 1 ) 4 5 2t( t 2 1 1 ) 2( t 2 1 3 ) 3( 7t 2 1 13 ) 4.4 Algebra Flashback (page 292) 1. 2x ln 5( x 2 1 4 ) 3. 3x 2 1 4 ln 10( x 3 1 4x 2 7 )
3. The expression is already simplified. 7. 3 32( 5 ) 2 1 7( 5 ) 2 114 2 34( 5 ) 1 74 5 3( 50 1 35 2 11 ) 2 ( 20 1 7 ) 5 3( 74 ) 2 ( 27 ) 5 443,556 Section 4.3 (page 283) 1. 2x 2 1 11 5. 3x 6 2 11x 3 7. f ( x ) 5 x 2 2 6x 1 9; g( x ) 5 x 1 1 9. f ( x ) 5 2 ln( x ) ; g( x ) 5 x 1 4
2
5. 4 32( 2 2 ) 1 14 3 ( 2 ) 5 8( 2 3 ) 3 5 2 216
3. 7x 2 2 14x 1 3
11. 30( 3x 1 1 ) 13. 12( t 3 2 t 2 1 1 ) 3( 3t 2 2 2t ) 15. 20( n2 1 2n 1 1 )( n 1 1 ) or 20( n 1 1 ) 3 17. 4( 2 80x 2 1 42x 2 5 )( 2 80x 1 21 ) 19. ( 3t 1 2 )( 5t 1 2 )( 12t 1 1 ) 21. 2 13 4 23. 13 81 25. 2 11,280 29. y 5 14,641
5. 32t 32 ln 3( t 2 2 1 ) 1 2t4
7. 2 ln 3( 32(0) ) 3( 0 ) 2 2 14 1 32( 0 ) 4 ( 32(0) ) 5 2 ln 3( 1 )( 2 1 ) 1 ( 0 ) 5 ln 3 Section 4.4 (page 292) 1. ln( 4 ) 4x 5. 5e x 1 1 x 3. 4 t
5 1 27. y 5 t 1 8 16 31. 2.124 inches per year 33.
1 7. 5e x c ln( x ) 1 d x
9. ln 2( 2x25x )( 2x 1 5 ) 11. 15. 6 5 ln 10 3 ln 3 e 17. 3 log 5 13. log 2 19. 15e2
dg is the rate of change in gross pay as a function dx of the number of hours worked. The units are dollars per hour. dt is the rate of change in take-home pay as a dg function of gross pay. The units are take-homepay dollars per gross-pay dollar. dt is the rate of change in take-home pay as a dx function of the number of hours worked. The units are dollars per hour.
21. y 5 3 ln 2( t ) 2 3 ln 2 1 1 23. y < 9.862x 2 4.443 27. 23.63 dollars per year 29. < 1.63 percentage points per year. In 1999, the percentage of homes with a VCR was increasing at a rate of 1.63 percentage points per year. 25. y 5 1 e
A24
31. The derivative of an exponential function y 5 a x is yr 5 ln a( a x ) . Let k 5 ln a . Then yr 5 k( a x ) . Therefore, the derivative of an exponential function is exponentially proportional to the function itself. 33. The derivative of g( x ) 5 ln x is gr( x ) 5 . So x gr( 1 ) 5 1, gr( e ) 5
1 , e 1
13.
dy 50 dx y x2y
15. 19.
y2 1 2 xy y xy 2 2x
17. 2
and gr( a ) 5
1 . a
21. ( 0, 1 ) and ( 0, 2 1 ) 25. ( 1, 2 2 ) and ( 2 1, 2 )
1 23. ( 0, 0 ) and a 0, b 2
35. The derivative of f ( t ) 5 logb t is f r( t ) 5 ln b
1 ?
. t
Since b . 1, ln b . 0. Since f ( t ) 5 logb t is defined only for t . 0 and since ln b . 0, t S ` , f r( t ) S 0, so the rate of change is decreasing. Thus f is concave down. 37. dy 1 5 2x B 1n 2 ln( 2x ) 1 R x dx
t
27. When the dependent variable cannot be isolated, it is necessary to use implicit differentiation to find the derivative. 29. The correct answer may take on many different forms. 31. 33. 35. dy x dx y 5 and 5 . They are reciprocals. y x dx dy dy e y 2 e xy 5 x dx e 2 xe y dy y 2 xy 2e xy 5 2 xy dx x ye 2 x
f r( t ) 5 ln b
1 ?t
. 0. Thus f is increasing. As
39. hr( t ) 5 ln 3( 316 ) ln 16( 16t ) 4.5 Algebra Flashback (page 299) 1. 2( x 1 y )( 1 1 d ) 5 0 11d5 0 2( x 1 y ) 0 2( x 1 y )
CHAPTER 4 REVIEW EXERCISES Section 4.1 (page 302) 1. 0 3. 21t 2 1 12t 24 5. 1999: Decreasing at a rate of 12.773 thousand people per year 2000: Decreasing at a rate of 7.648 thousand people per year Section 4.2 (page 302) 7. 18x 2 2 24x 9. 2 6t 2 1 16t 1 4 11. < 24,422 prisoners per year Section 4.3 (page 302) 13. 2 6x 2 1 3x 1 4 15. 54( 9x 1 5 ) 2 17. 2219.31 million dollars per year Section 4.4 (page 303)
d 5 21 1 3. ln 3( 3 ) d 5 1
y
1 d5 ln 3( 3y ) 5. 2x 1 4yd 5 d 2x 5 d 2 4yd 2x 5 d( 1 2 4y ) 2x d5 1 2 4y 7. de x 1 e xy 5 y 2 1 2xyd e xy 2 y 2 5 2xyd 2 de x e xy 2 y 2 5 d( 2xy 2 e x ) d( 2xy 2 e x ) 5 e xy 2 y 2 e xy 2 y 2 d5 2xy 2 e x Section 4.5 (page 299) 1. Undefined 5. 2 9. 5 3 2 3. 2 2 7. 11. 1 3 1 2y
19. ln 9( 9 x )
21.
2x x 13
2
23. 20.1 thousand people per year
A25
Section 4.5 (page 303) 25. 2 1 21 27. 2 4 29. 22 ln 2 1 1 6 ln 2 2 3
21.
f ' (x)
min 2
x CHAPTER 5 5.1 Algebra Flashback (page 324) 1. x 5 2 2, x 5 5 3. x( x 1 12 )( x 2 15 ) 5 0 so x 5 0, x 5 2 12, x 5 15 5. When x , 2 2, x 1 2 , 0 and x 2 5 , 0. Since a negative times a negative is positive, y . 0 for x , 2 2. When 2 2 , x , 5, x 1 2 . 0 and x 2 5 , 0. Since a positive times a negative is negative, y , 0 for 2 2 , x , 5. When x . 5, both factors are positive. Since a positive times a positive is positive, y . 0 for x . 5. Thus y is positive for x , 2 2 or x . 5. 7. Interval x , 2 12 2 12 , x , 0 x . 15 x x 1 12 x 2 15 y 5 x( x 1 12 ) ( x 2 15 ) + + + + + ( 2 )( 2 )( 2 ) 5 2 ( 2 )( 1 )( 2 ) 5 1 ( 1 )( 1 )( 2 ) 5 2 ( 1 )( 1 )( 1 ) 5 1 x 23 21 0 1 y is positive on the intervals 2 12 , x , 0 and x . 15. Section 5.1 (page 324) 1. f r( x ) 5 0 at ( 1, 2 1 ) 3. f r( x ) 5 0 at ( 0, 1 ) 5. f r( x ) 5 0 at ( 2, 0 ) x 22 21 1 2 19. p 5 61; stationary 7. f r( x ) is undefined at ( 2, 4 ) 9. f r( x ) 5 0 at ( 0, 200 ) and ( 4, 2 56 ) 11. x 5 2; stationary 15. x 5 0; singular x 5 61; stationary 17. x 5 61; stationary 13. t 5 0, 4; stationary 3 27.
h' (x) min 1 +
21 2 5 23.
h' (t) + max 0
f (x) on 32 1, 54 5 24 5
min + 4
Absolute maximum Relative and absolute minimum Absolute maximum
t 21 0 4 7 25.
g' (x) min + 1 0
h(t) on 32 1, 74 27 0 2 32 49
+ max 1
Relative maximum Relative and absolute minimum Absolute maximum
g(x) on 32 3, 34 2 1.327 0 2 1.327

max 1
0 , x , 15 +
22
h(x) on 32 2, 24 2 0.4 2 0.5 0.5 0.4

A26 29.
C' (p)
+ max 1
min + 1
45. 59.457 million 47. All relative extrema occur at critical values. 49. When the derivative is positive, the graph of the original function is increasing. When the derivative is negative, the graph of the original function is decreasing. When the derivative changes from positive to negative, the graph of the original function changes from increasing to decreasing. 51. Not necessarily. The given values are for particular points only, not intervals, and are indicated in the following chart:
f' (x) + 2 0 3 4
p 23 21 1 3 31.
f ' (x) + max 4 3
C( p) on 32 3, 34 2 65 21 79
min 4 3 +
15
x 23
4 2 3 4 3
f (x) on 32 3, 34 2 15
< 3.08 < 2 3.08 15

min 1 e +
It is possible, for example, to have f r( x ) , 0 immediately before x 5 3 and f r( x ) . 0 immediately after x 5 3. In such a case, x 5 3 would represent a relative minimum point for f ( x ) . 53. f ( x ) 5 x 1>3 55. A relative minimum occurs at x 5 2 and a relative maximum occurs at x 5 5. 5.2 Algebra Flashback (page 347)
33.
h' (t)
1. A is the area and r is the radius. 3. The volume V is the product of the length l, the width w, and the height h. That is, V 5 lwh. 5. x 1 4y 5 60, so x 5 60 2 4y. Thus V 5 xy 2 is equivalent to V 5 ( 60 2 4y ) y 2.
Relative and absolute minimum Absolute maximum
t 0.1
1 e
< 2 0.2303 < 2 0.3679 < 3.2958
h(t) on 30.1, 34
7. The domain is r . 0. Section 5.2 (page 347) 1. (a) P( t ) 5 564.3t 2 2 613.7t 1 3658.9 (b) 2000 3. (a) The maximum profit of $4984.1 million occurs when 66.87 million iPods are sold. (b) The price that will maximize profit is $109.53.
35.
g' (x)
min 0.567
x 0.1 0.567 3 37. $25 43. Highest: 1999 Lowest: 1990
g(x) on 30.1, 34 < 3.408 < 2.330 < 18.987

Relative and absolute minimum Absolute maximum
39. $40
41. From about 1983 through 1988
5. (a) s( t ) 5 17.8( 1.04 ) t 2 2 billion unit cases [equivalent to s( t ) < 16.5( 1.04 ) t billion unit cases] (b) Rr( s ) 5 2 0.8058s 1 14.44 dollars per case sold (c) 17.92 billion cases (d) R( s( t ) ) 5 2 0.4029 317.8( 1.04 ) t 2 24 2 1 14.44 317.8( 1.04 ) t 2 24 2 109.2 (e) Early 2002
A27
7. (a) 2002 (b) C( t ) 5 2 120.5t 2 1 623.5t 1 6553 million dollars; t is the number of years since the end of 1999. (c) 2002 (d) They are reasonable, since the maximum number of sales could incur additional costs. 9. 2000 11. 30 trees 13. $0.19 per pound 19. 1200 books 15. $116.92 million per year 17. < 27.5 feet 3 18.2 feet 21. 500 books 25. 43 vehicles 27. Limit the number of cars sold to the amount that maximized revenue. 29. 250 pounds 31. With calculus, we may use the derivative to find relative extrema. 33. We have V 5 pr 2h and A 5 2prh 1 2pr 2, where V is the volume, A is the surface area, r is the radius, and h is the height. Since the surface area is 360 p , we can write 360p 5 2prh 1 2pr 360 5 2rh 1 2r 2 180 5 rh 1 r 2 rh 5 180 2 r 2 180 2 r 2 h5 r Thus V 5 pr 2h 180 2 r 2 5 pr 2 a b r 5 180pr 2 pr 3
2
5.3 Algebra Flashback (page 372) 1. Because the function is increasing, we know that f r . 0. Since f r is decreasing, f s is negative. Thus the graph of f is concave down. 3. Because the function is decreasing, we know that f r , 0. Since f r is decreasing, f s is negative. Thus the graph of f is concave down. 5. x 0 1 2 3 4 f ( x ) First Difference Second Difference 2 3 6 11 18 1 3 5 7 2 2 2
23. 50 vehicles
7. From 2 2 to 0, the graph is concave down. From 0 to 2, the graph is concave up. Section 5.3 (page 373) 1. f ( x ) 5 x 3 2 3x 2
y
10 5 1 5 10 15
Inflection point x
Concave up
Concave down
3. f ( x ) 5 x 4 2 12x 2
y
40 20 3 2 1 20
Concave down
1 2 3
35. No. The width of the barn, which forms one side of the corral, is 50 feet, so the side opposite the barn cannot be longer than 50 feet. Thus the solution of 60 feet does not work in this real-world situation. 37. dx 5 0 when t < 4.15 or 1.85. dt dy 5 0 when t 5 3. dt
Concave up
Concave up Inflection point
5. y 5 x 4 2 8x 3 1 18x 2
Concave y up
60 40 20 1 1 2
Concave up
Concave down Inflection point x 3 4
39. Since speed is the absolute value of velocity, we seek the minimum of Z xr( t ) Z . From the sign chart in Exercise 38, this minimum is 0 meters per second. 41. Yes, the speeds are relatively slow.
A28
7. y 5 x 4 2 2x 3 1 12x 2
Inflection point y
50 3 2 1 50 Concave 2 3
33.
y A
3 2 1 E
B x C
2 1 1
x G
down
100 150
Concave down
Concave up
9. f ( x ) 5
x2 x 11
2
35.
y Concave 0.8 up B
16 E 12 8 4 D
Concave down
0.6 0.4 0.2
Concave down x
432 4 8 12 C 16
1 2 3 4x
Inflection point
11. ( 1, 2 2 ) 13.
15. ( 1, 11 ) ; ( 3, 27 ) 19.
A 2 "2, 2 20 B ; A "2, 2 20 B A
1 1 2 ,4 3
21. ( 0, 0 ) relative maximum; ( 2, 2 4 ) relative minimum 23.
B; A
1 1 , 3 4
17. ( 2 2, 48 ) ; ( 1, 9 )
37.
3 2 20 40 60 80 100 120 A 140
A( 2 4, 16 ) absolute maximum on 32 4, 44 B( 2 3.464, 0 ) x-intercept C( 2 2, 2 16 ) relative and absolute minimum on 32 4, 44 D( 0, 0 ) x- and y-intercept; inflection point E( 2, 16 ) relative and absolute maximum on 32 4, 44 F( 3.464, 0 ) x-intercept G( 4, 2 16 ) absolute minimum on 32 4, 44
E
A( 2 2, 3 ) absolute maximum on 32 2, 24 B( 2 1.532, 0 ) x-intercept C( 2 1, 2 1 ) relative and absolute minimum on 32 2, 24 D( 2 0.347, 0 ) x-intercept E( 0, 1 ) inflection point and y-intercept F( 1, 3 ) relative and absolute maximum on 32 2, 24 G( 2, 2 1 ) absolute minimum on 32 2, 24
1 2 3 4 5 x
C D
25. (0, 0) relative minimum; (3, 27) inflection point
A 2 "6, 2 36 B relative minimum; ( 0, 0 ) relative maximum; A "6, 2 36 B relative minimum
27. (0, 0) relative minimum; ( 2 3.312, 83.967 ) relative maximum; ( 1.812, 16.721 ) relative maximum 29. (0, 0) relative minimum 31.
y
6 4 2
F B E x
A A 2 2, 2 6 3 B absolute
2
39.
y
2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2
2 2 1 2 3 4 4 C D 6
minimum on 32 2, 44 B( 0, 0 ) relative maximum; x- and y-intercept C A 1, F A 4, D A 2, 2 3 B relative minimum

4 2 23
A( 2 2, 1.636 ) B( 0, 1 ) x-intercept; relative and absolute minimum on 32 2, 24 C( 2, 2.614 ) absolute maximum on 32 2, 24

2
A( 2 3, 2 129.6 ) absolute minimum on 32 3, 54 B( 0, 0 ) x- and y-intercept; relative and absolute maximum on 32 3, 54 C( 3, 2 32.4 ) inflection point D( 4, 2 51.2 ) relative minimum E( 5, 0 ) x-intercept; absolute maximum on 32 3, 54
B inflection point
B
0 1
2 1
41. Minimum increase: 1975, 2 $5.765 per year Maximum increase: 2000, $1.465 per year 43. Wages should continue to increase to compensate for diminishing wages during the years directly after 1975.
E( 3, 0 ) x-intercept 32 2, 44
1 53
B absolute maximum on
A29
45. 1995; 2000 49. 1991; 299
47. 1995; 1998 51. 1 3.
53. srr( t ) 5 2 32 feet per second squared, acceleration due to the force of gravity 55. 3.9 seconds 57. (23.16, 12.56); in the 23rd week, the box office sales were increasing most rapidly at about 12.56 million dollars per week. 59. Although revenue is increasing, it is increasing at a decreasing rate. That is, it is not going up as quickly as it was before. 61. Because g s ( c ) is defined and c is a critical value, the derivative is 0 at x 5 c. This means that a relative extremum may occur at x 5 c. Since g s ( c ) . 0, the function g is concave up at x 5 c. Thus a relative minimum occurs at x 5 c. 63. x s ( t ) 5 ( 6t 2 18 ) meters per second squared y s ( t ) 5 2 2 meters per second squared 65. The point (3, 15) is the only inflection point for x on [0, 6]. Since y s 5 2 2 , 0 for t-values on [0, 6], y has no inflection points. 67. Analogous to Exercise 66, the slowest horizontal acceleration of the train occurs at time t 5 0, where x s ( 0 ) 5 6( 0 ) 2 18 5 2 18 meters per second squared. 5.4 Algebra Flashback (page 384) 1. A 5 pr 2 3. A 5 p( 3 ) 2 5 9p; A 5 p( 5 ) 2 5 25p. The area changes by 16p square inches. 5. The volume of a spherical balloon is V 5 pr 3, 3 where r is the radius of the balloon. 7. A rate is a ratio of two quantities that shows how one quantity changes in relation to the other. Section 5.4 (page 384) 1. dA dl dw 5 w1 l dt dt dt dA 5 change in area with respect to time in dt square inches per minute dl 5 change in length with respect to time in dt inches per minute
4
dw 5 change in width with respect to time in dt inches per minute dV dl dw dh 5 wh 1 lh 1 lw dt dt dt dt dV 5 change in volume with respect to time in dt cubic inches per minute dl dw , , same as in Exercise 1. dt dt dh 5 change in height with respect to time in dt inches per minute. dA dl dw dw dh 5 2a w 1 l1 h1 w dt dt dt dt dt dl dh 1 h 1 lb dt dt Physical meanings same as in Exercises 1 and 3 dA dr 5 8pr dt dt dA , same as in Exercise 1 dt dr 5 change in radius with respect to time in dt inches per minute dA dr 5 2pr dt dt Physical meanings same as in Exercise 7
5.
7.
9.
11. 3.782 inches per hour; 12.69 hours 13. 0.26 inch per second or 0.6773 centimeter per second 15. 0.1014 foot per hour 17. 8.83 degrees per hour 19. 4.20 degrees per hour 21. We mean that a change in one of the rates has a direct impact on the other rate. 23. Leibniz notation helps us quickly determine the units of each of the rates of change. 25. In related-rate problems, there are multiple derivatives. By keeping track of units, we can make sense of the meaning of each of the derivatives.
A30
CHAPTER 5 REVIEW EXERCISES Section 5.1 (page 387) 1. f r( x ) 5 0 at ( 1.5, 2.25 ) and ( 2.5, 2.25 ) f r( x ) undefined at ( 2, 2 ) 3.
17. a 2
19. (0, 0.9) stationary point and relative and absolute minimum 21. 4; 17.72 miles per day Section 5.4 (page 389) 23. dA ds 5 10s dt dt dA 5 change in area with respect to time in dt square centimeters per second ds 5 change in side length with respect to time in dt centimeters per second 25. 0.32 inch per minute 29. 0.0159 foot per hour CHAPTER 6 6.1 Algebra Flashback (page 400) 1. 0.5x 21 3. 0.75x 4 5. They have the same slope. 7. Example: f ( x ) 5 4x 2 1 9 and g ( x ) 5 4x 2 1 8 27. 0.20 inch per minute
10 10 , 0.925 b , a , 0.925 b 3 3
5. (0, 4) 7.
f ' (x)
12 , < 15.95 B ; (0, 10); A , < 4.051 B A 2 12 5 5
0 12
5
0 0
0
12 5
x 22 2 12 5 0 12 5 2 9.
y' + 0 0
f (x) on 32 2, 24 10 < 15.95 10 < 4.051 10
Relative and absolute maximum Relative and absolute minimum
x 23 0 3 11. 469.4 miles
y on 32 3, 34 2 32 4 2 32
Absolute minimum Relative and absolute maximum Absolute minimum
Section 6.1 (page 401) 1. 2x 1 C 5. t2 1 20t 1 C 4 3. ln | t | 2 7. 11. 3t 2 1C 2
Section 5.2 (page 388) 13. (a) (b) (c) (d) 2002 C( t ) 5 2 88.5t 2 1 410.5t 1 3019 2002 The same year is reasonable because more sales could generate more costs.
t4 3t 2 2 t3 1 2t1C 4 2 3x x4 2 1C ln ( 3 ) 4
9. 2.3 ln | x | 1 x 1 C 13. 15.
24 1 2 ln | t | 1 t 1 C t 2( 3x ) 3( 2x ) 500 2 1C 1 C 17. 2000 ln | p | 1 ln ( 3 ) p ln ( 2 ) 2t 3x 2 1 C 21. 2 5x 1 C ln ( 2 ) 2
Section 5.3 (page 389) 15.

y
0.96 0.94 0.92
Concave up
19. 2 5t 21 1
Concave 0.90 down
Concave down
23. t 2 2 ln | t | 1 C 25. 2 3x 21 2 x 4 1 2x 1 C
x 4 3 2 1 0 1 2 3 4 Inflection point
A31
27. 2 5t 21 2 16
t3 2 9t 1 C 3
13.
4 3t e 1C 3
15. 3ln ( x ) 4 2 1 C 19. 2 ln | x 3 2 9 | 1 C
29. 5u 2 4 ln | u | 2 u21 1 C 31. 2 400x 21 1 200 ln | x | 1 50x 1 C 33. 4 ln | x | 1 5x 21 1 C 35. 2t 2 1 4t 1C ln ( 4 )
17. ln | t 2 2 t 1 2 | 1 C 21. ln | ln ( x ) | 1 C 23. 25.
6 4 ( x 2 2 ) 5>2 1 ( x 2 2 ) 3>2 1 C 5 3 2 2 ( x 1 1 ) 3>2 1 C 3 27. 1 2 ( t 2 1 ) 3>2 1 C 3
37. 0.013p 1 5 dollars, p 5 number of pages 39. 2 0.53325t 4 1 1.333t 3 2 0.07665t 2 1 0.3167t 41. No. Trying to do so leads to a zero in the denominator. Additionally, 3x 21dx 5 ln Z x Z 1 C. 43. The derivative of a linear function y 5 mx 1 b is yr 5 m, where m is the slope of the function. Consequently, the antiderivative of a constant function f ( x ) 5 m must be a linear function F( x ) 5 mx 1 C. The Constant Rule of Integral states that the antiderivative of a constant function is a linear function. 45. Since F1( x ) and F2( x ) are specific antiderivatives of f ( x ) , they must differ by a constant. Therefore, Z F1( x ) 2 F2( x ) Z is a constant that represents the vertical shift from one graph to the other. 47. In order for a polynomial F r( x ) to have degree n, F( x ) must have degree n 1 1, which stipulates that F( x ) will have at most n 1 1 x-intercepts. 49. F( x ) $ 0 for x $ 0. 6.2 Algebra Flashback (page 409) 1. vu3 5. uv Section 6.2 (page 410) 1. 5. (x2 1 3)6 1C 6 (ex 1 1)2 1C 2 3. 2x 3 1 5x 2 1 C 3. v (u)3 2
29. ln | x 3 2 x 2 1 x 1 1| 1 C 31. 33. 1 ( 2x 2 1 1 ln | 2x 2 1| ) 1 C 4 1 ln | 4x 2 2 5 | 1 C 4 35. t 4 2 3t 3 1 t 2 1 C
37. ln ( 2p2 1 1 ) 2
2 ( 2p2 1 1 ) 3>2 1 C 3
39. 2 ln | t 2 2 t | 1 C 41. The purpose is to write the integrand in a way that allows us to use basic integration rules to integrate the function. 43. u 5 x 1 k for some constant k. 45. Integration by substitution undoes what the Chain Rule does. The Chain Rule is used to find the derivative of a composition of functions. Integration by substitution takes the derivative of a composition of functions and returns the original function. 47. Yes, as follows: Let u 5 x ln ( x ) . Then 1 du 5 c ( 1 ) ln ( x ) 1 x d dx x ln ( x ) 1 1 3 f ( x ) dx 5 3 x ln ( x ) dx du 5 3 u 5 3ln ( x ) 1 14 dx
7. 4v( 2u )
7. 0.4x 3 2 1.2x 2 1 0.6x 1 C 9. 1 x2 2 4x e 1C 2 11. 1 3x2 2 6 e 1C 2
A32
49. Yes, as follows: Let u 5 ln ( x ) . Then du 5 f (x) 5 5 5 3 f ( x ) dx 5 5 1 dx x ln ( x 2 ) 4x 2 ln ( x ) 4x ln ( x ) 2x ln ( x ) 3 2x dx 1 u du 23
5. LHS 5 89.0
y
20 15 10 5 0 0 3 4 5 6 7
7.
n 2 4 10
LHS 0.0625 0.1406 0.2025
9.
n 2 4 10
LHS 9.333 8.167 7.564 LHS 50.00 61.00 67.92 17.75
11.
n 2 4 10
LHS 6.119 6.779 7.151 LHS 27.20 21.41 18.41 2.39 2.14 2.27 60.00
13.
6.3 Algebra Flashback (page 431) 1. A 5 lw, where l is length and w is width. 3. A 5 ( 3 1 4 1 5 1 6 )( 0.5 ) 5 9 square centimeters 5. D x 5 7. 421 3 5 8 8
n 2 4 10
15.
n 2 4 10
17.
LHS
19.
LHS RHS Avg.
RHS 23.75 Avg. 20.75 21. LHS RHS Avg. 25. LHS RHS Avg. 29. 3. LHS 5 5.121
y
5 4 3 2 1 0 0 0.5 1.0 1.5 2.0
cubic feet feet ? feet ? feet 5 second second 1 yard 1 yard 1 yard ? ? ? 3 feet 3 feet 3 feet 1 cubic yards 60 seconds 5 ? 27 second 1 minute 60 5 cubic yards per minute 27 5 2 cubic yards per minute 9
2
2.00 2.00 2.00 90.00 102.00 96.00 0.339
23.
LHS
RHS 72.00 Avg. 66.00 27. LHS RHS Avg. 31. LHS 1.40 2.15 1.78 0.776
Section 6.3 (page 431) 1. LHS 5 8

y
2.0 1.5 1.0 0.5 0 0 1 2 3 4 5
LHS
RHS 0.686 Avg. 0.512 33. LHS RHS

x
RHS 1.205 Avg. 0.990 35. LHS RHS Avg. 324.69 441.86 383.28
2.67 3.36 3.01
Avg.
A33
37. (a) 3600F( t ) cubic feet per hour (b) LHS RHS Avg. 685,115 cubic feet < 5,125,037 gallons 597,275 cubic feet < 4,467,946 gallons 641,195 cubic feet < 4,796,492 gallons
Dx 5
420 8 5 0.5 A 5 ( 21.0 )( 0.5 ) 5 10.5
(c) More hydroelectric power is needed at 5:45 a.m. than at 7:45 a.m. (Answers may vary.) 39. About 1,399.686 million gallons. That is, roughly 1.4 billion gallons. 41. The area of a rectangle is easy to calculate. By ensuring that one of the upper corners of the rectangle is on the graph, we are able to use the area of the rectangle to approximate the area between the graph of the function and the horizontal axis. 43. Increasing the number of rectangles reduces the width of each individual rectangle. The smaller the width of the rectangle, the more accurately the rectangle estimates the area between the graph and the horizontal axis on the given interval. 45. miles ? hours 5 miles hour
49. f ( x ) 5 Z 2 2x 1 2 Z . Note the graph for the exact area if we calculate the area as the sum of the areas of two triangles: A 5 2 A 2 B bh 5 2 A 2 B ( 1 )( 2 ) 5 2
1 1
y
2 1 0 0 1 2
For n = 1, A 5 4. For n = 2, A 5 2. 20 For n = 3, A 5 . 9 For n = 4, A 5 2. For n = 5, A 5 2.08. For n = 6, A 5 2. Since the actual area is 2 units, the cases in which n = 2, 4, or 6 (even integers) give the actual area. 6.4 Algebra Flashback (page 446) 1 1. A 5 bh where b is the length of the base and 2 h is the height. 3. A 5 a 5. 16 Section 6.4 (page 447) 1. 21 7. 2.71875 11. Area 5 2
y
1.5
47. No. If we compute the area using eight rectangles, we get the following results. x 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 f (x) 0.000 1.750 3.000 3.750 4.000 3.750 3.000 1.750 0.000 Height Sum Height Sum Average of Sums = 21.0 21.0 21.0 Left * * * * * * * * * * * * * * * * Right
b1 1 b2 b h where b1 and b2 are the lengths 2 of the bases and h is the height. 7. 28.5
3. 2 3
5. 36 9. 6.875 13. Area 5 8

y
4 3 2 1 1 1 1 2 3 4
1.0 0.5 0 0 1 2 3 4 5 6 7
A34
15. Area 5 2
6 5 4 3 2 1 11
17. Area 5 4
y
5 4 3 2 1
47. A definite integral has limits of integration, whereas an indefinite integral does not. 49. It means that the sum of the areas below the horizontal axis is subtracted from the sum of the areas above the horizontal axis.
1
1 1
19. Area 5 4
6 5 4 3 2 1 2 1 1 1
51. No. See Exercise 47 in Section 6.3, where the lefthand and right-hand sums are the same but the actual area is different. 53. This inequality is equivalent to left-hand sum # 3 f ( x ) dx # right-hand sum
b a
which is false for a decreasing function. 55. Two possible choices are f ( x ) 5 0 or f ( x ) 5 e x . 29. 2 23 6.5 Algebra Flashback (page 461) 1. F( 2 ) 2 F( 1 ) 5 6 2 0 5 6 3. F( 0.25 ) 2 F( 0 ) 5 0 2 0 5 0 5. F( e ) 2 F( 1 ) 5 1 2 0 5 1 7. F( 1 ) 2 F( 2 1 ) represents the vertical distance between the two points ( 1, F( 1 )) and ( 2 1, F( 2 1 )) .
21. 10
3
23. 10
25. 4
27. 3
31. 3 ( 2 x 2 1 5x 2 6 ) dx
0 4
33. 3 ( 2 x 1 4x 1 20 ) dx
3
22 4
35. 3 ( 2 | x 2 3 | 1 3 ) dx
22
37.
y
1 0 1 2 1 2 3
39.
x
y
2.5 2.0 1.5 1.0 0.5 2 1 0 1 2
y f (x) = x3 1
1 1 2 1 2
41. 10.5 gallons per person. From the end of 1990 to the end of 1999, the consumption of bottled water increased by 10.5 gallons per person. 43. 5.06 pounds per person. From the end of 1980 to the end of 1995, the consumption of mozzarella cheese increased by 5.06 pounds per person. 45. According to the model, the number of students participating in the AB Calculus exam increased by roughly 80,690 students from the end of 1992 to the end of 2002. That is, the number of students taking the exam in 2002 is calculated to be 80,690 students higher than the number of students taking the exam in 1992.
y
10,000 8000 6000 4000 2000 0 2000
Section 6.5 (page 461)

7 4
1. 3 1 dx 5 2
5 2
3. 3 x dx 5 8
0
5. 3 ( 2 3x 2 1 5 ) dx 5 2 2
1 1
7. 3 ( 8x 3 1 2x ) dx 5 0 1 9. 3 a b dx 5 ln( 3 ) < 1.099 x 1

5 21 3
11. 3 ( 2t ) dt < 40.40

2 4
13. 3 ( 5 # 2x ) dx < 113.61

t
5 10 15 20 25 30 35
2x b dx 5 ln( 2 ) < 0.693 15. 3 a 2 x 11 0
22 1

2
A35
4 17. 3 ( x 2 2 1 )( x 3 2 3x ) 3 dx 5 < 1.33 3 0

21
3. H( x ) 5 2x 2 2 3 ln | x | 1 C 5. x 4 2 5 2 x 2x1C 2 7. t 1 1 ln | t | 1 C 2
19. 3 ( 2 2x 1 1 ) dx 5 4
22
21. Area 5 78
1 < 78.08 12 25. Area 5 2 29. Area 5 16 < 1.83

5
9. 0.018p 1 1.30 dollars, where p is the number of pages Section 6.2 (page 466) 11. 1 2 (x 2 9)6 1 C 12 1 t1C 2 13. 3ln ( x ) 4 2 1 C
1 < 0.08 23. Area 5 12 27. Area < 5.28

43
15. 3 ln | t | 1
31. 3 ( 315.70t 2 770.64 ) dt < 112,260

33
Section 6.3 (page 466) 17. n 2 4 10 21. LHS RHS Avg. 25. LHS RHS Avg. LHS 48.0 40.0 35.2 0.906 1.156 1.031 2.34375 1.59375 1.96875 23. 19. n 2 4 10 LHS RHS Avg. LHS 6.417 6.954 7.276 3.5095 3.8845 3.697
According to the model, 112,260 more people will take the exam in 2012 than in 2002.
1.5
33. 3 V( t ) dt < 1.318 miles

0 19
35. 3 ( 0.0009778t 3 2 0.02958t 2 1 0.2392t 2

10
0.06311 ) dt 5 2.3 pounds per person Between the end of 1990 and the end of 1999, the per capita mozzarella cheese consumption increased by 2.3 pounds per person. 37. Yes. The original limits of integration apply only to the original variable. If you rewrite the integrand in terms of a new variable, you must adjust the limits of integration. 39. The trapezoid formed by the lines x 5 2, x 5 4, y 5 0, and y 5 f ( x ) has area 9. Therefore,
4
F( 4 ) 5 F( 2 ) 1 3 f ( x ) dx 5119 5 10 41. F( 2 ) 5 16
1 2
27. (a) 3600 f ( t ) (b) The total amount of water that flowed past the Priest Rapids Dam between 5:00 p.m. and 9:45 p.m. on November 22, 2002, is estimated to be 1,584,925,729 cubic feet. (c) 15,849 swimming pools Section 6.4 (page 467) 29. 15 31. Area 5 21
y
10 8 6 4 2 12 1 2 3 4 5
43. 3 v ( t ) dt 5 33
0.5
During the second half-hour of the trip, the car traveled 33 miles. 45. f ( x ) 5 x will work. F( a ) 5 F( 2 a ) . CHAPTER 6 REVIEW EXERCISES Section 6.1 (page 466) 1. G( x ) 5 2 x 2 1 5x 1 C
A36
33. Area 5 9
y
3 2 1 1 1 1 2 3 4 5 6
35. Area 5 4
y
2 1
11. 15. 19. 21. 25.
ln|3t 2 1| 1C ln( 3 ) 1 4 x 2 9x 2 1 C 2
13. 3xe x 2 3e x 2 x 3 1 C 17. 1 6 t 2 t3 1 C 2
x x 1 0 1 2 3 4 5 6
3x 2x 2 a x2 2 1 2 3b 1 C ln( 3 ) ln( 3 ) 3ln( 3 ) 4 2 1 3x2 e 1C 2 2ln(x) 1C ln( 2 ) 23. 1 23 et 1C 3
37. 12 41.
y
0.5 0.4 0.3 0.2 0.1 0
39. 1
27. < 280.05 million dollars

x 0 0.5 1.0 1.5 2.0 2.5
29. < 3773.7 million dollars 31. The Product Rule. If the derivative of a function is found by using the Product Rule, then the original function may be found from the derivative by using integration by parts. 33. The integrand on the left is obtained by taking the derivative of the function on the right. If we let v 5 x2 and u 5 ln( x ) , then dv 5 2x dx and du 5 x dx. In terms of u and v, we may rewrite the equation as 3( dv u 1 v du ) 5 v( u ) 3u dv 1 3v du 5 uv
1
43. (a) 2 1.385 gallons per person. From the end of 1993 to the end of 1999, the per capita consumption of milk decreased by 1.385 gallons. (b) Answers may vary. Section 6.5 (page 468) 45. 8.5 47. 0 16 3 49. 0.693 51. 132.5
53. 4.605 55.
57. < 5821.04 dollars per person
CHAPTER 7 7.1 Algebra Flashback (page 479) 1. e x( x2 2 2x 1 2 ) 3. e2 x( x 2 1 ) 5. e2 x 1 1( 2x 2 1 ) 2 7. 4t 1.5 c ln( t ) 2 d 3 Section 7.1 (page 479) 1. 2 e ( x 1 1 ) 1 C
2x
3u dv 5 uv 2 3v du 35. Yes. Let u 5 ln( x ) and dv 5 x dx. Then du 5 x dx and v 5 ln( x ) . Using integration by parts, we have 3u dv 5 uv 2 3v du 1 1 3ln( x ) a x dx b 5 3ln( x )43ln( x )4 2 3ln( x ) a x dx b 1 2 3ln( x ) a dx b 5 3ln( x ) 4 2 x
1 1
3 3 3. t 2 ln( t ) 2 t 2 2 5t ln( t ) 1 5t 1 C 2 4 1 1 5. t 4 ln( t ) 2 t 4 1 C 4 16 7. e2 x 1 1( 2x 2 1 ) 1 C 9. e2x( x 2 1 ) 1 C
1 3ln( x ) 4 2 ln ( x ) a dx b 5 1C 3 x 2
37. f ( t ) 5 e t( t 2 2 2t 1 2 )
A37
39. The Product Rule tells us that

d ( uv ) dx
du # v dx
dv # u. dx
Integrating both sides
25. 773 million dollars. According to the model, the annual profit increased by 773 million dollars from the end of 1990 through 2001.
y
800 600 400 200 0 0 1 2
yields d du dv 3 dx ( uv ) dx 5 3 dx v dx 1 3 dx u dx uv 5 3v du 1 3u dv 3v du 5 uv 2 3u dv. This result is the integration by parts method of integration. 7.2 Algebra Flashback (page 495) 1. f ( x ) 2 g( x ) 5 x2 1 4x 2 ( 7x 2 8 ) 5 x2 2 3x 1 8 3. f ( x ) 2 g( x ) 5 ln( x ) 2 ex 5. They intersect at x 5 0 and x 5 1. 7. f ( x ) is above g( x ) on the intervals ( 2 1, 0 ) and ( 1, ` ) . Section 7.2 (page 495) 1. 1 6 3. 64 3 5. 1 6 7. 10 3 13. 1 19. < 0.0858
2
r (t ) = 402t + 806 s (t ) = 355t + 372.5 x
27. Equilibrium point: ( 10, 49.02 ) Consumer surplus: $50.5 Producer surplus: $94 29. Equilibrium point: ( 2512.97, 0.778 ) Consumer surplus: $255.94 Producer surplus: $788.77 31. The definite integral with integrand f ( x ) 2 g( x ) may be used to calculate the area between f and g, provided that appropriate limits of integration are used. 33. For even functions, 3 3 f ( x ) 2 g( x )4 dx 5 2 3 3 f ( x ) 2 g( x )4 dx
2a 0 a a
35. < 16.240 37. < 2.030 39. < 3.9672 7.3 Algebra Flashback (page 507) 1. f (0) f ( 99 ) f ( 9999 ) f ( 999,999 ) f (0) f ( 100 ) f ( 1000 ) f ( 10,000 ) f (1) f ( 0.1 ) f ( 0.01 ) f ( 0.001 ) f (1) f ( 100 ) f ( 1000 ) f ( 10,000 ) 51 < 0.000102 < 0.00000 < 0.0000
9. 5980.50 15. <0.0975
11. < 7.647 17. < 3.460
21. < 40,804.7 million dollars

y
30,000 25,000 20,000 15,000 10,000
R(t ) = 665t 2 + 1150t + 27,357
C(t ) = 103t 2 + 315t + 8539
3.
0 0.5 1.0 1.5 2.0 t
23. 3,569,382 people. According to the model, the number of adults in prison, in jail, on probation, or on parole increased by 3,569,382 more people than the number of students enrolled in private colleges from the end of 1980 through 1998.
y
250 200 150 100 50 0 0
< 1.2857 < 0.6594 < 0.6659 < 0.6667
5.
50 5 25 5 2 100 5 2 1500 5 undefined < 2 0.2171 < 2 0.1448 < 2 0.1086
r (t ) = 238.1
7.
s (t ) = 1.839t + 23.25
5 10 15 20 t
A38
Forcible rapes (per 100,000)
Section 7.3 (page 507) 1. (a) 1 3 x 1 2 dx < 0.8473 1 1 3 x 1 2 dx < 1.3863 1 1 3 x 1 2 dx < 2.8526 1
100 50 10 5
27. (a)
F
0.025 0.020 0.015 0.010 0.005 0 0 2000 4000 6000 Population (100,000) `
f (x) = 0.2678 e 0.0008 p
(b) The improper integral 3

2628
F( P ) dP represents
3
1
1 dx < 3.5264 x12
the total number of forcible rapes if the population continued to increase indefinitely from 2628 (hundred-thousand).
`
(b)
y
0.4 0.3 0.2 0.1 0 0 10 20 30 40 x
(c) 3
2628
F( P ) dP < 40.894. This result tells us
f (x) =
1 x+2
(c) The improper integral does not appear to converge because the values of the integral get larger and larger as the upper limit increases. 3. Converges to 1 3 29. (a)
that as the population continues to increase, the total number of forcible rapes will approach 4,089,400. We might question the validity of this result because it depends on an ever-increasing population and on a mathematical model that may not be valid for this ever-increasing population. There must be an upper population limit.
Pediatric AIDS cases
P
1000 800 600 400 200 0 0 10 20 30 40 t Years (since the end of 1992) `
f (x) = 1276.3e 0.2776 t
5. Converges to 2 1 7. Converges to ln( 2 ) 9. Diverges 11. Diverges 13. Converges to 4 15. Diverges 17. Converges to 1 1 19. Converges to 5 21. Diverges
`
(b) The improper integral 3 P( t ) dt represents

1
the total number of pediatric AIDS cases reported after 1992.

`
(c) 3 P( t ) dt < 3482. The model predicts that

1
23. 3
1
1 dx xp
converges if p . 1.
`
there will be a total of 3482 reported cases of pediatric AIDS after 1992. We would have to question the validity of the model over an indefinitely long time period. 31. Answers will vary. Here is an example of a
`
25. When p . 1, 3
1
1 dx xp
5p
1 . 21
possible student response. The integral 3 e22xdx

0
is considered improper since the upper limit of
A39
integration is infinity. Whenever the interval over which we are integrating is unbounded on one end of the interval, we have an improper integral.
5
CHAPTER 7 REVIEW EXERCISES Section 7.1 (page 511) 1. 2 2e2 x( x 1 1 ) 1 C 3. e x( 3x 2 2 6x 1 6 ) 1 C 7. 5. 1 3ln( x ) 4 4 1 C 4
Another example is the integral 3

2
1 x22
dx.
Since the function
1 x22
is undefined when x 5 2,
`
the integral is considered improper. A final example is the integral 3 xe2x dx, which is
2
5 2 5 x ln( x ) 2 x 2 2 3x ln( x ) 1 3x 1 C 2 4
2`
unbounded on both ends of the interval. 33. The phrase it gets small enough fast enough in the context of improper integrals provides a way to think about the convergence or divergence of the integral. The pronoun it refers to the function or output values of the function being integrated. When these function values get small enough fast enough, we are thinking about how steeply the function is approaching the x-axis. If function values approach 0 quickly, then the improper integral might converge. 35. The bumper sticker says Calculus is as easy as p
`
1 9. e4x a x 2 b 1 C 4 Section 7.2 (page 511) 11. 72 13. <3.46 15. < 29.264 17. According to the model, the accumulated change in profit from the end of 1999 through 2001 is 3918 million dollars.
y
5000 4000 3000 2000 1000 0 0
r (t ) = 1,330t + 1,150
c (t ) = 206t + 315
1 2
since 3
2` `
1 x2 1 1
dx 5 p.
` 1 dx 5 arctan ( x ) ` 3 x2 1 1 2` 2`
19. The gap between public- and private-school teachers is predicted to increase by 116.15 thousand teachers between the year 2000 and the year 2010.
b
5 lim lim arctan( x ) `

bS` dS`
y
d
5 lim lim ( arctan( b ) 2 arctan( d ) ) bS` dS` p 2p 5 2 2 2 p p 5 1 2 2 5p 2 37. F( s ) 5 3 s 41. F( s ) 5

` 2`
16 12 8 4 0
g (t ) = 1.4242t + 16.11
p (t ) = 0.8158t + 1.453
s2 39. F( s ) 5 2 s 11
Section 7.3 (page 513) 21. Diverges 25. Converges to 1 ln( 5 ) 23. Diverges 27. Diverges
1 s21
1
43. 3 f ( x ) dx 5 3 1dx
0
Movie ticket sales (in millions of dollars)
29. (a)
s
200 150 100 50 0 0 5 10 15 t Weekend number
5 x` 5 1 2 0
1 0
s (t) = 200e 0.5044 t
51
`
45. 3 2e22xdx 5 1
0
(b) The indefinite integral represents the total movie ticket sales as time goes toward infinity.
A40
`
(c) 3 200e20.5044t dt < 239.44 (million dollars).

1
z = exy z
This result tells us that over time, the total box office sales will be about $239,440,000. We might question the validity of this result because it depends on the movie remaining in theaters for an infinite amount of time, and on a mathematical model that most likely will not be valid for this infinite amount of time. CHAPTER 8 8.1 Algebra Flashback (page 525) 1. A < 250 A 1 1 3. P < 6228.70
0.08 12
Note: In the graph, each square of the grid has dimensions 4 3 4 . 7. f ( x, y ) 5 x2 1 1 y2 1 1 x T 22 0 2

2 z = x2 + 1
yS 21 2.5 0.5 2.5 0 5.0 1.0 5.0 1 2.5 0.5 2.5
B 12 ? 4 5 343.92
z
y +1
5. V 5 80p cubic inches 7. V 5 360 cubic inches Section 8.1 (page 525) 1. z 5 3x 2 2y x T 22 0 2 yS 21 24 2 8 0 26 0 6 1 28 22 4
x y z = 3x 2y z
Note: In the graph, each square of the grid has dimensions

1 4
3 4.
9. z 5 ( x 1 y ) 3 x T 22 0 2 yS 21 2 27 21 1
z
0 28 0 8
1 21 1 27
3. z 5 4xy 1 4 x T 22 0 2 yS 21 12 4 24 0 4 4 4 1 24 4 12
x y z = 4xy + 4 z
z = (x + y)3
Note: In the graph, each square of the grid has dimensions 3 . 4 4 11. 6889 thousand vehicles 13. $4938 15. 504 square inches 17. A 187-pound man who is 71 inches tall has body mass index < 26.1. (Answers will vary.) 19. 2 15F 21. A multivariable function is a function that has more than one independent variable.
1 1
5. f ( x, y ) 5 e2xy x T 22 0 2 yS 21 0.135 1.00 7.39 0 1.00 1.00 1.00 1 7.39 1.00 0.135
A41
23. A multivariable function may have any number of independent variables but only one dependent variable. 25. The equation tells us that if we slice the threedimensional shape defined by f ( x, y ) 5 x 2 2 y 2 with the plane y 5 1, the cross section will look like the parabola z 5 x 2 2 1. 27. The domain of f is 5( x, y ) Z x $ 2 and Z y Z $ 26 . The range of f is 33, ` ) . 8.2 Algebra Flashback (page 542) 1. f ( 3, 5 ) 5 2 2 3. y 2 4 5 4( x 2 3 ) 5. y 5 2 4x 1 26 7. y 2 3 5 4( x 2 1 ) Section 8.2 (page 542) 1. fx 5 fy 5 fxx 5 fxy 5 fyy 5 3y 3x 0 3 0 3. fx fy fxx fxy fyy 5 4y 5 4x 2 2y 50 54 5 22
13. WT 5 0.6215 1 0.4275V 0.16 WV 5 0.16V 20.84( 2 35.75 1 0.4275T ) WT ( 27, 5 ) < 1.2 means that when the temperature is 27 F and the wind speed is 5 mph, increasing the temperature by 1 F while leaving the wind speed unchanged will increase the wind chill temperature by about 1.2 F. WV ( 27, 5 ) < 2 1.0 means that when the temperature is 27 F and the wind speed is 5 mph, increasing the wind speed by 1 mph while leaving the temperature unchanged will decrease the wind chill temperature by about 1 F. 15. BH 5 2 1406WH 23 BW 5 703H 22 BH ( 67, 155 ) < 2 0.7 means that when a person is 67 inches tall and weighs 155 pounds, increasing the persons height by 1 inch while leaving the weight constant will decrease the body mass index by about 2 0.7 index point. BW ( 67, 155 ) < 0.2 means that when a person is 67 inches tall and weighs 155 pounds, increasing the persons weight by 1 pound while leaving the height constant will increase the body mass index by about 0.2 index point. 17. RE 5 2 PE22 RP 5 E21 RE( 4, 32 ) 5 2 2 means that when the earnings are $4 per share and the price is $32 per share, increasing the earnings per share by $1 while leaving the share price constant will decrease the price-to-earnings ratio by about 2 units. RP( 4, 32 ) 5 0.25 means that when the earnings are $4 per share and the price is $32 per share, increasing the share price by $1 while leaving the earnings per share constant will increase the price-to-earnings ratio by about 0.25 unit. 19. According to the model, increasing the number of employees from 26,424 to 26,425 while leaving the amount of money spent on advertising constant would increase sales by about $0.17 million. According to the model, increasing advertising spending from $519.2 million to $520.2 million while leaving the number of employees constant would decrease sales by about $0.37 million. 21. The tangent line in the x-direction is z 5 0x 2 4. The tangent line in the y-direction is z 5 8y 1 4. 23. The tangent line in the x-direction is z 5 x 1 1. The tangent line in the y-direction is z 5 0y 1 1.
5. Wt 5 v 0.5 Wv 5 0.5tv 20.5 Wtt 5 0 Wtv 5 0.5v 20.5 Wvv 5 2 0.25tv 21.5 9. zx 5 ln( xy ) 1 1 zy 5 xy 2 1 1 zxx 5 x 1 zxy 5 y zyy 5 2 xy 22 11. Vr 5 2prh Vh 5 pr 2
7. Vr 5 2prh Vh 5 pr 2 Vrr 5 2ph Vrh 5 2pr Vhh 5 0
Vr( 3, 2 ) 5 12p means that when the height of the cylinder is 3 inches and the radius is 2 inches, increasing the radius by 1 inch while leaving the height unchanged will increase the volume by about 12p cubic inches. Vh( 3, 2 ) 5 4p means that when the height of the cylinder is 3 inches and the radius is 2 inches, increasing the height by 1 inch while leaving the radius unchanged will increase the volume by about 4p cubic inches.
A42
25. The tangent line in the x-direction is z 5 0x 1 e. The tangent line in the y-direction is z 5 2 2ey 2 e. 27.
'h 'r
9. f ( x, y ) 5 x 3 2 6x 1 xy 2 y 2 has a saddle point at

3 99 , 2 4, 16 B . A23 2 2 140 , , 2 27 B and a relative maximum at A4 3 3
is the rate of change in the heat index with respect to the relative humidity when the air temperature is held constant.
'r 't
is the rate of
11. f ( x, y ) 5 13. f ( x, y ) 5
x2 1 1 y2 1 1 21 x2 1 1
has a saddle point at ( 0, 0, 1 ) . 1 y 4 2 4y 2 has a saddle point
change in the relative humidity with respect to the air temperature when the heat index is held constant.
'h 't
is the rate of change in the heat index
with respect to the air temperature when the relative humidity is held constant. 29. Partial derivatives are similar to derivatives in that they represent rates of change. However, partial derivatives are used for multivariable functions, whereas derivatives are used for single-variable functions. 31. The rate of change in the x-direction is e 2 per unit and the rate of change in the y-direction is 1 2 e 2 per unit. 33. f ( x, y ) 5 x 4 1 y 4 1 x 2 1 y 2 1 xy 1 1 (Answers may vary.) 35. a 5 2, b 5 0, c 5 2 3 8.3 Algebra Flashback (page 558) 1. x 5 2 1 and x 5 6 3. x 5 61 5. x 5 0, x 5 12, and x 5 2 7. x 5 0 and x 5 62 Section 8.3 (page 558) 1. f ( x, y ) 5 x 2 1 4y 2 2 10 has a relative minimum at ( 0, 0, 2 10 ) . 3. f ( x, y ) 5 2 x 2 2 y 3 1 3y 2 x has a relative
1 maximum at 2 2, 1 3 2 2, 2 1, 2 14 . 1
15. f ( x, y ) 5 x 3 2 3x 1 y 3 2 12y has a relative maximum at ( 2 1, 2 2, 18 ) , a relative minimum at ( 1, 2, 2 18 ) , and saddle points at ( 2 1, 2, 2 14 ) and ( 1, 2 2, 14 ) . 17. f ( x, y ) 5 x 2 2 x 2 3y 2 1 3y has a saddle point at
1 1 , , . A1 2 2 2B
at ( 0, 0, 2 1 ) and relative minima at A 0, 2 "2, 2 5 B and A 0, "2, 2 5 B .
19. f ( x, y ) 5 ( 0, 0, 1 ) . 21. f ( x, y ) 5
x2 1 y2 1 1 2 2 e x y x2 2 y2 2 2 e x y
has a relative maximum at
and a saddle point at ( 0, 0, 0 ) . at A 1, , e9>4 B . 2

1
1 , relative minima at A 0, 61, 2 B , A 61, 0, 1 eB e
has relative maxima at
23. f ( x, y ) 5 e2x 2 1 2x 2 y2 1 y 1 1 has a relative maximum
25. f ( x, y ) 5 x 3y 1 xy 3 2 4xy has relative maxima at ( 2 1, 1, 2 ) and ( 1, 2 1, 2 ) , relative minima at ( 2 1, 2 1, 2 2 ) and ( 1, 1, 2 2 ) , and saddle points at ( 2 2, 0, 0 ) , ( 0, 2 2, 0 ) , ( 0, 0, 0 ) , ( 0, 2, 0 ) , and ( 2, 0, 0 ) . 27. fx( a, b ) 5 0 and fy( a, b ) 5 0 29. You may evaluate the function at points surrounding the critical point. 8.4 Algebra Flashback (page 573) 1 1. l 5 2 lw 2 8 3. l 5 2 hl 9 5. h 5 0 7. y 5 0
5. f ( x, y ) 5 x 2y 2 y 2 1 x has a saddle point at
1,
1 24
B and a saddle point at
7. f ( x, y ) 5 x 2 4x 1 y 2 y has a saddle point at ( 1, 0, 2 3 ) and a relative minimum at

4 3 2
A 2 1, 2, 2 4 B .
1 3 2 85
A 1, 3, 2 27 B < ( 1, 0.667, 2 3.15 ) .
A43
Section 8.4 (page 574) 1. A relative minimum occurs at ( 2, 9, 2 14 ) . 3. A relative maximum occurs at
5. A relative maximum occurs at (4, 4, 32), and a relative minimum occurs at (0, 0, 0). 7. A relative minimum occurs at
A 8, 8, 8 B .
3 1 1 1 , 5
5. For Apple Computer, we have A 5 $3950 million. For General Electric, we have A < $59,887.5 million. Microsofts debt/equity ratio of 0 means that all of the companys assets are financed through equity. The company doesnt have any debt. Section 8.2 (page 579) 7. fx 5 4xy 2 4y 2 fy 5 2x 2 2 8xy fxx 5 4y fyy 5 2 8x fxy 5 4x 2 8y 9. fx 5 y ln( y ) fy 5 x ln( y ) 1 x fxx 5 0 x fyy 5 y fxy 5 ln( y ) 1 1 11. Vr( h, r ) 5 2prh Vr( 4, 5 ) 5 40p When the height is 4 inches and the radius is 5 inches, increasing the radius by 1 inch will increase the volume by about 40p < 126 cubic inches. Vh( h, r ) 5 pr 2 Vh( 4, 5 ) 5 25p When the height is 4 inches and the radius is 5 inches, increasing the height by 1 inch will increase the volume by about 25p < 79 cubic inches. 13. The tangent line in the x-direction is z 5 x 1 3. The tangent line in the y-direction is z 5 2 y 1 2. Section 8.3 (page 580)
9. A relative minimum occurs at A2 , , < 21.99 B , 51 51

50 53
2 1,
4 25
B.
and a relative maximum occurs at ( 2 2, 2 1, 31 ) .
11. The box whose dimensions are 30 inches 3 28.24 inches 3 28.24 inches has maximum volume. The volume of this box is 23,917 cubic inches. 13. According to the model, Ford should spend $18,232 million on labor and $21,768 million on capital in order to produce the maximum number of vehicles (10,680 thousand) while remaining within the $40,000 million budget. 15. According to the model, Coca-Cola should spend $44 million on labor and $873 million on capital in order to produce the maximum number of unit cases (18.9 billion) while remaining within the $917 million budget. 17. It is a plane. 19. Yes CHAPTER 8 REVIEW EXERCISES Section 8.1 (page 578) 1. z 5 4x 2 5y x T 22 0 2 yS 21 23 5 13 0 28 0 8 1 2 13 25 3
x y z
15. f ( x, y ) 5 x 2 2 xy 1 4y 2 1 10 has a relative minimum at (0, 0, 10). Section 8.4 (page 580) 17. A relative maximum occurs at ( 1, 2, 6 ) . 19. A relative maximum occurs at (6.60, 4.33, 86.6). The dimensions of the garden that maximize the area subject to the constraint are shown in the figure.
3. z 5 xy 1 4y x T 22 0 2 yS 21 22 24 26 0 0 0 0 1 2 4 6
x y z
8.66ft
6.60ft
arc length =13.6ft
A44
CHAPTER 9 9.1 Algebra Flashback (page 599) 1. Answers will vary 5. No solutions 9. u 5 0, 13. u 5 p 2 p 5p , p, 3 3 3. Answers will vary 7. u 5 0, p 11. u < 1.498, 5.927
daylight hours increases to a maximum of about 13.75 daylight hours and decreases to a minimum of about 10.5 daylight hours. (e) The horizontal shift of 80 days to the right on the graph aligns the model in such a way that the maximum number of daylight hours (for example) occurs on about the 180th day of the year (mid-to-late June). 13. (a) H 5 5.835 sin c (b)
H
16 12 8 0 0 100 200 300
2p ( t 2 80 ) d 1 12.335 365
Section 9.1 (page 600) 1. Yes, it would be reasonable to model this situation with a periodic function. We might expect the population to rise to a maximum in the spring and then fall to a minimum in the winter. This pattern may repeat from year to year. 3. No, it does not seem that this situation is periodic. Rather, the population continues to increase. 5. Yes, a periodic function would be appropriate, assuming that the satellites orbit is elliptical so that there is a maximum distance and a minimum distance. The satellites distance will periodically change from the minimum distance to the maximum distance and then back again. 7. Amplitude = 1; period = 2p; horizontal shift = p; vertical shift = 0 9. For the sine function, amplitude 5 4, period 5 2, horizontal shift 5 , and vertical shift 5 22. 2 11. (a) H 5 1.625 sin c (b)
H
13 12 11 10 0 0 100 200 300
2p ( t 2 80 ) d 1 12.125 365
(c) The amplitude tells us that the number of hours of daylight in Juneau varies by a total of 11.67 hours from the least number of daylight hours to the greatest number of daylight hours. (d) The midline value for the number of daylight hours in Juneau is 12.335 hours. From this midline value, the number of daylight hours increases to a maximum of about 18.17 daylight hours and decreases to a minimum of about 6.5 daylight hours. (e) The horizontal shift of 80 days to the right on the graph aligns the model in such a way that the maximum number of daylight hours (for example) occurs on about the 180th day of the year (mid-to-late June). 15. The number of daylight hours depends on the location of the city. The amplitude for the model representing the number of daylight hours for Miami (closer to the equator) is smaller than that for Juneau (farther from the equator). Since Sydney is in the southern hemisphere, the seasons are opposite those in the northern hemisphere, so the value of A is negative. 17. In each case, the model must be shifted 80 units (days) to the right in order to align with the data. Note that 80 days into the year would be the beginning of spring. This ensures that, except for Sydney, the graph is increasing at the greatest rate in the spring and decreasing at the greatest rate in the fall. In Sydney, the opposite is true.
(c) The amplitude tells us that the number of hours of daylight in Miami varies by a total of 3.25 hours from the least number of daylight hours to the greatest number of daylight hours. (d) The midline value for the number of daylight hours in Miami is 12.125 hours. From this midline value, the number of
A45
19. (a) V( t ) 5 165 sin( 120pt ) (b) Amplitude = 165; period =
1 ; 60
31. Answers will vary. vertical

y
6 4 y = sin x 2 2 2 4 6 2 x
shift = 0; horizontal shift = 0. The amplitude is 165 because the current varies between 165 and 2 165 volts. The period is cycle is completed in 21. (a) (b)
1 60 1 60
because
there are 60 cycles per second. That is, one of a second.
y = sin x + x
(c) (d)
p A( d ) 5 1209.5 sin c ( d 2 210 ) d 110 1 405,288.5 Amplitude = 1209.5; period = 220 days; vertical shift = 405,288.5; horizontal shift = 210 days. The amplitude tells us that the distance varies by a total of 2419 kilometers from minimum to maximum. The period tells us that the apogee distance completes a cycle every 220 days. The vertical shift indicates the midline, which in this case is 405,288.5 kilometers. The horizontal shift aligns the model with the data by shifting the model 210 days to the right. This ensures that the moments of greatest increase and decrease occur on the appropriate days. 406,155.9 kilometers May 2 (122nd day), July 7 (188th day), and December 8 (342nd day)
33. Assuming the rider begins at the bottom of the Ferris wheel, the model y 5 2 125 cos a 2p t b 1 141 5
models the height of the rider above the ground at time t. A 5 125: The Ferris wheel travels 125 feet above and below the midline value of 141 feet. B5
2p : 5
By completing a full revolution in
5 minutes we must take the regular period of 2p and divide it into 5 complete cycles. C 5 1.25: By shifting the graph to the right 1.25 minutes, we align the model so that its height is increasing at the greatest rate 1.25 minutes into the ride. D 5 141: The midline value, or the center of the Ferris wheel, is 141 feet above the ground. 35. t 5 0.98, 4.01, 5.98, 9.02 minutes 9.2 Algebra Flashback (page 615) 7 1 1. C 3. E 5. D 7. y 5 x 2 6 3 9. y 5 x 1 2 11. y 5 7 Section 9.2 (page 616) 1. Answers will vary. 3. f r( x ) 5 3 cos x 5. dr( t ) 5 2 6 cos( 3t 1 6 ) 7. yr 5 5 sec2 x 9. Br( t ) 5 8.6p p sin a t 2 6 b 12 12
p 23. y 5 2 sin a x b 1 1 2 25. y 5 2 sin c 2p 5 ax 2 b d 1 4 5 4
27. Answers will vary. 29. We know that R( t ) 5 cos( 2t ) has a period of
2p 2
5 p, indicating that it repeats its output values
each time we add p to the input values. On the other hand, S( t ) 5 cos( 4t ) has a period of
2p 4
5 2 , indicating that it repeats more
frequently than R( t ) 5 cos( 2t ) . We see that S( t ) 5 cos( 4t ) is compressed horizontally compared to R( t ) 5 cos( 2t ) .
1 0.75 0.50 0.25 2 0.50 0.75 1
S (t)
11. f r( x ) 5 4 sin x ? cos x 13. rr( u ) 5 4 tan u ? sec2 u 15. yr 5 9x2 ? 3cos( x3 2 1 ) 4 2 ? 3sin( x3 2 1 ) 4
R(t)
A46
17. zr( u ) 5 cos2 u 2 sin2 u 5 2 cos2 u 2 1 19. yr 5 2 csc x

2
1 radian increase in the angle of impact, the horizontal distance will change by dr( u ) feet. (b) dr Q 18 R 5 2 210.1 feet per radian. At an angle of 18 radian, the horizontal distance traveled after 5 seconds will decrease by 2 210.1 feet for every increase of 1 radian in the angle of impact. 35. Because f r( x ) 5 2 sin x 2 1, we know that f r( x ) # 0. We know this is true because output values for 2 sin x are between 2 1 and 1. Subtracting 1 from any of these output values will produce values that are less than or equal to 0. The graph of f r( x ) 5 2 sin x 2 1 confirms this analysis.
y
0.5 2 1 1.5 2 2 x
21. Since every fourth derivative cycles the result back to sin x, we know that by the 50th derivative we will cycle back to sin x 12 times (the 48th derivative). Therefore, we would have to take the derivative two more times. That is, y s 5 2 sin x. 23. y 5 2 x 1 p 2 25. y 5 x 3.25p 2p cos c ( t 2 80 ) d 365 365
27. (a) Mr( t ) 5
(b) Mr( 90 ) 5 0.0276 additional daylight hour per day. (c) On the 90th day of the year, the number of daylight hours is increasing by 0.0276 hour (1.656 minutes) per day. That is, the days are getting longer by 1.656 minutes each day, which makes sense because the 90th day is around the end of March. In the spring, the days do get longer until the first day of summer. 11.5p 2p 29. (a) Jr( t ) 5 cos c ( t 2 80 ) d 365 365 (b) Jr( 90 ) 5 0.0975 additional daylight hour per day. (c) On the 90th day of the year, the number of daylight hours is increasing by 0.0975 hour (5.85 minutes) per day. That is, the days are getting longer by 5.85 minutes each day, which makes sense because the 90th day is around the end of March. In the spring, the days do get longer until the first day of summer. p 31. (a) dr( t ) 5 242 cos a b 2 147.4 < 80 9 The baseballs horizontal position is changing at a rate of about 80 feet per second. p (b) dr( 5.3 ) 5 242 cos a b 2 147.4 < 80 9 At the moment the ball has been traveling for 5.3 seconds, its horizontal speed is about 80 feet per second. 33. (a) dr( u ) 5 2 1210 sin u The baseballs horizontal position after 5 seconds is changing at a rate of dr( u ) horizontal feet per radian. That is, for every
y = sin x 1
37. If Fr( x ) 5 2 sin x ? cos x, then F( x ) 5 sin2 x 1 C. 39. lim

hS0
sin( x 1 h ) 2 sin( x ) h sin( x ) cos( h ) 1 cos( x ) sin( h ) 2 sin( x ) 5 lim hS0 h sin( x ) cos( h ) 2 sin( x ) 1 cos( x ) sin( h ) 5 lim hS0 h sin( x ) 3cos( h ) 2 14 1 cos( x ) sin( h ) 5 lim hS0 h cos( h ) 2 1 sin( h ) 5 sin( x ) lim 1 cos( x ) lim hS0 h hS0 h 5 sin( x ) ? ( 0 ) 1 cos( x ) ? ( 1 ) 5 cos( x ) p sin x 5 < 0.01745 x 180
sin (h) hS0 h
41. When x is measured in degrees, lim

xS0
43. Looking back at the solutions to Exercises 39 and 40, we see that when we compute lim
p
when h
is in degrees, the result is 180 rather than 1. Therefore, dx 3sin x4 5 180cos x and
d p
d 3cos x4 dx
5 2 180sin x.
A47
9.3 Algebra Flashback (page 627) 1.

y
4 3 2 1 2 1 1 1 2 x
41. (a)
y
1 Cost (in dollars)
3.
0.5 0.5
5.
y
0.5 0.25 4 2 0.25 2 4 x
280 260 240 220 200 180 160 140 120 100 80 60 0
0 1 2 3 4 5 6 7 8 9 10 11 12 t Months
Section 9.3 (page 627) 1. F( x ) 5 2 2 cos x 1 C 5. V( t ) 5 2 3 ln Z cos t Z 1 C 7. S( t ) 5 2 24 p cos c ( t 2 4 ) d 1 5t 1 C p 12 1 3. R( t ) 5 sin t 1 C 3
(b) $1901.09 is the total amount spent on electricity for this year. (c) $158.42 is the average cost of electricity each month. 43. (a) A( d ) 5 1209.5 sin S 110( d 2 210 ) T 1 405,288.5 (b) The average apogee distance is 405,432.5 kilometers. 45. 32.3F 47. Since 3cos x dx represents an antiderivative, we are looking for a function whose derivative is yr 5 cos x. The function y 5 sin x is the function whose derivative is yr 5 cos x. Remember, the derivative of a constant is 0. Therefore, an arbitrary constant could be added to the function y 5 sin x and still have a derivative function of yr 5 cos x. Therefore, 3cos x dx 5 sin x 1 C. Said another way, if y 5 sin x 1 C, yr 5 cos x.
p 0
9 2 1 9. H( x ) 5 sin a x b 1 x 2 1 C 2 3 2 1 11. P( x ) 5 2 cos x 2 1 C 2 1 13. F( t ) 5 ( sin t ) 2 1 C 2 1 15. H( x ) 5 2 ln Z 1 1 cos( 2x ) Z 1 C 2 17. Y 5 x 2 sin x 1 2x cos x 2 2 sin x 1 C 19. Y 5 ln Z sin t Z 1 C 5 23. 2 ln Z cos( 4x ) Z 1 C 4 27. e
sin t
21. 2 cos( 2x ) 1 C 25. 2 sin( 2 2 x ) 1 C
49. 3 Z cos x Z dx 5 3 cos x dx 1 3 5 sin x ` 5 a sin

p>2 0
p>2
p p>2
2cos x dx
1C
29. 2"1 1 cos u 1 C 33. 1 2 tan x 1 C 2
2 sin x `
p p>2
31. 2 35.
15 p cos c ( x 1 2 ) d 1 C p 5
1 2 1 t 2 sin( 2t ) 1 C 2 2
p p 2 sin 0 b 2 a sin p 2 sin b 2 2 5 (1 2 0) 2 (0 2 1) 5111 52

y
1 0.75 0.50 0.25 0 0 2
37. 12.125 hours of daylight, on average, each day. 39. 12.25 hours of daylight, on average, each day.
A48
51. 3e x sin x dx 5
ex sin x 2 e x cos x 1C 2
53. 3cot x dx 5 ln Z sin( x ) Z 1 C 55. 3sec x dx 5 ln Z sec( x ) 1 tan( x ) Z 1 C CHAPTER 9 REVIEW EXERCISES Section 9.1 (page 632) 1. Amplitude = 2; period = 2; horizontal shift = 2 2; vertical shift = 1 3. Amplitude = 2; period < 6; horizontal shift < 0.7; vertical shift 5 2 5. Amplitude < 1; period = 2p; horizontal shift = 1; vertical shift = 2 7. x 5 0, 1, 2, 3, 4, 5, 6 9. x 5 p 2p , 3 3
year was decreasing by 1.43 million pounds per year. (c) We really have no way to know if the sign makes sense. Why would the crab harvest be decreasing in 1988? There is no obvious answer. Students should make their own conjectures. Section 9.3 (page 634) 23. S( t ) 5 2 25. F( t ) 5 6 p cos c ( t 1 2 ) d 2 t 1 C p 6
1 3sin( 2t ) 4 2 1 C 4
27. cos( cos x ) 1 C 29. 3 cos a 31. (a)

Population
p 2 t b 1 5t 1 C 2
y
300 200 100 0 0 8 16 24 x Years (since 1949)
p 11. (a) P( t ) 5 22.5 sin c ( t 2 10 ) d 1 55.5 2 (b) According to the model, the population of the mallard duck in 1980 was 78. (c) t < 1.52, so the mallard duck population was 40 ducks midway between 1966 and 1967, according to the model. Section 9.2 (page 633) 13. dr( t ) 5 20 cos( 4t 2 8 ) 15. cr( x ) 5 17. y 5 2 1 "sin( 2x ) cos( 2x )
32
(b) 3 L( t ) dt < 6549. Since L( t ) can be

0
interpreted as the population for a given year,

32
3 L( t ) dt represents the total number of lynx

0
that ever lived in Minnesota between 1949 and 1981. (c) The average annual Canadian lynx population in Minnesota between 1949 and 1981 was about 205 lynx. CHAPTER 10 Section 10.1 (page 645) 1. D 3. F 5. B 7. (a)
1400 1200 1000 800 600 400 200 0
19. (a) Gr( t ) 5 106.77 cos( 0.121739t 2 2.03472 ) (b) Gr( 84 ) < 235.34 This tells us that during the 84th week, the amount of natural gas in storage is decreasing at a rate of 35.34 billion cubic feet per week. (c) The negative result makes sense because t 5 84 represents mid-November, 1998. We would expect natural gas storage to decrease during the winter months, when natural gas usage is on the rise. 21. (a) Dr( t ) 5 7.149 cos( 1.05723t 2 0.342318 ) (b) Dr( 2 ) < 21.43 This tells us that in 1988, the number of pounds (in millions) of crab harvested each
10
20
30
Answers to Odd-Numbered Exercises dP dt
A49
(b) When t 5 0, P 5 2 and
is relatively small.
As t increases, P increases at an increasing rate until P is around 600. At that point, P continues to increase but at a decreasing rate until it approaches the carrying capacity of 1200. That is, as t continues to increase, P approaches 1200. 9. (a)
M
30,000 20,000 10,000 0 0 2 4 6 8
(d) The slope field provides a visual image of the scenario described in part (b). That is, we can see the slopes getting steeper and steeper as time increases, confirming our description of the population of the United States increasing at an ever-increasing rate. (e) P
3 108 2 108 1 108 0 0 50 100 150 200 t
15.
T
90 80 70 0 0 10 20
(b) Initially,
increases, M increases indefinitely at an increasing rate. That is, as time goes on, the amount of money continues to grow at an ever-increasing rate. 11. Graph A. Since the rate of salt water entering the tank is greater than the rate of salt water leaving the tank, we would expect the amount of salt S to increase as t increases. However, if there initially was a large amount of salt, we would expect the amount of salt to decrease because the rate of water into the tank is only slightly greater than the rate out. Graph A supports this description. 13. (a) P 5,000,000 50,000,000 100,000,000 150,000,000 200,000,000 250,000,000 dP dt 357,770.9 1,131,370.8 1,600,000.0 1,959,591.8 2,262,741.7 2,529,822.1
dM dt
is small but positive. As t
When T is large, T 2 70 is relatively large, so is very negative. As T decreases, T 2 70 gets smaller, which causes
dT dt dT dt
dT dt
to get less and less
negative. If we continue this pattern indefinitely, will approach 0 as T approaches 70F.
17.
y
80 60 40 20 0 0 10 20
The rate of increase in the number of publicschool classrooms with Internet access is small for years near 1994. As the number of classrooms with Internet access increases,
dy dt
increases until
(b) As the population of the United States increases, the rate at which it increases also increases. That is, the population is growing faster and faster as time increases. (c) P
3 108 2 108 1 108 0 0 50 100 150 200 t
we see the greatest rate of increase near the middle of the graph. Just past this midway portion of the graph, we see the rate of change begin to decrease. That is, the number of classrooms with Internet access continues to increase, but at a decreasing rate. This continues until the number of classrooms reaches about 86, where
dy dt
is 0. More
concisely, the number of public-school classrooms with Internet access increases slowly, then more quickly, and again more slowly as time increases from 1994.
A50 19.
S
300,000 250,000 200,000 150,000 100,000 50,000 0
Answers to Odd-Numbered Exercises dy dt dy dt
5 ( 1 )( 1 ) ( 1 ) 5 ( 1 ) . That is, when y . 3, . 0.
Section 10.2 (page 658)

0 20 40
Initially, the news spreads relatively quickly. As more people hear the news, the number of people who hear the news continues to increase, but at a decreasing rate. As the number of people who have heard the news gets near 350,000 people, the population of the city, the rate of increase gets closer and closer to 0. This is because once the entire population has heard the news, the news can no longer spread. 21. Answers will vary. 23. Since the differential equation was only in terms of y, the slopes depend only on y. So, a solution curve through ( 0, 4 ) would look like the solution curve given, mostly, just shifted to the left. That is, if we know the slope at y 5 4 from the given graph, we know the slope at y 5 4 for the new graph. See the graph below.
y
9 8 7 6 5 4 (0, 4) 3 2 1 0 0 1 2 3 4 5 6 7 8 9 10 11 t
1. y( 4 ) 5 4 3. y( 4 ) 5 8.228 5. y( 4 ) 5 1.875 7. y( 4 ) 5 8.837 9. The student is correct. Since Eulers Method follows the exact solution by using straight line segment approximations, these approximations will always be greater than the y-values of the actual solutions if the actual solution is concave down. 11. Approximately midway through the 23rd year 13. After approximately 8 minutes 15. Approximately 19 minutes 17. Eulers Method approximates y values by following the solution curve in linear chunks, so to speak. As we make the step size smaller, y values are approximated over smaller and smaller linear chunks. Therefore, smaller step sizes more closely follow the actual solution curve. Note that the actual solution curve is usually not known when using Eulers Method. For there to be no error in the result obtained using Eulers Method, the step size would have to be infinitesimally small. 19. Eulers Method predicts that the U.S. population was 300 million sometime between 1978 and 1979. 21. Eulers Method predicts that 80 percent of publicschool classrooms had Internet access in mid1997. 23. Eulers Method predicts that the number of students served was 200,000 in 2001 or 2002. 25. Eulers Method predicts that the temperature of the coffee will be 100F after 36 or 37 minutes. 27. Answers will vary. 29. Since Eulers Method over- and underestimates, as output ( y) values near y 5 4, we will obtain values that are either just greater than or just less than y 5 4. The differential equation
dy dt
25. For the differential equation

dy dt
dy dt
5 y( y 1 1 )( y 2 3 ) ,
5 0 when y 5 0, y 5 2 1, and y 5 3. When

dy dt
y , 2 1, the signs of the factors in this differential equation are 5 ( 2 )( 2 )( 2 ) 5 ( 2 ) . That is,
dy dt
when y , 2 1,
dy dt
, 0. When 2 1 , y , 0, the
signs of the factors in this differential equation are 5 ( 2 )( 1 )( 2 ) 5 ( 1 ) . That is, when
dy dt
2 1 , y , 0,
dy dt
. 0. When 0 , y , 3, the signs
of the factors in this differential equation are 5 ( 1 )( 1 ) ( 2 ) 5 ( 2 ) . That is, when

dy dt
0 , y , 3,
, 0. Finally, when y . 3, the signs
of the factors in this differential equation are
5 2 y( y 2 4 )
Answers to Odd-Numbered Exercises dy dt
A51
shows us that when y . 4, when y , 4 and

dy dt
, 0 and Eulers
Section 10.3 (page 671) 1. y 5 6"x2 1 C 5. y 5 9. y 5 x 1 1 Cx 4Ce4x 1 1 Ce4x 4e4x 7 1 e4x
Method will output values less than 4. However, . 0, Eulers Method will output
2 x 3. y 5 a 1 C b 2
values greater than 4. As Eulers Method continues to over- and underestimate, it will produce output values just greater than and just less than y 5 4. 31. x 0 3 y ynew dy 5 ycurrent 1 ? 0.5 dx
7. y 5 6Ce0.25x 1 4 11. y 5 "2x 2 1 14 15. y 5 2 5e25x 1 5 19. y 5 100e2x 50e2x 2 49
13. y 5 2 e3x 17. y 5
ynew 5 3 1 0 ? 0.5 5 3 ynew 5 3 1 2 0.25 ? 0.5 5 2.875 ynew 5 2.875 1 2 0.522 ? 0.5 5 2.614 ynew 5 2.614 1 2 0.861 ? 0.5 5 2.184
0.5 3 1.0 2.875 1.5 2.614 2.0 2.184 33. x 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
21. (a) T( t ) 5 76 2 41.7e20.01344t (b) 0.49F/min; 0.29F/min 23. <19.08 minutes 25. About 5 hours 8 minutes 27. About 5 hours 24 minutes 29. y 5 1000e0.1t 31. A( t ) 5 2500e0.17293t 33. Function y5t y 5 3t 1 2 y 5 5t 1 4 t11 3t 1 3 5 3( t 1 1 ) 5t 1 5 5 5( t 1 1 ) y11 y11 t11 1 3 5 dy dt 1 3 5
y 3 3 2.995 2.985 2.97 2.95 2.924 2.894 2.857 2.815 2.767
x 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
y 2.713 2.652 2.584 2.509 2.425 2.332 2.23 2.115 1.988 1.844
Any function of the form y 5 k( t 1 1 ) 2 1 will have the property 35. Observe that
dy dt y11 t11
2
5 k. Similarly,
dy dt
5 k.
5 2tet 5 2ty. So the separable

dy dt
35. y( 2 ) 5 "3 < 1.73 We see that as the step size gets smaller, Eulers Method approximates a result that is closer and closer to the actual result, y( 2 ) 5 "3. 10.3 Algebra Flashback (page 671) 1. e M 1 N 5 e M ? e N 3. e2 2 K 5 e2 ? e2K 5. x 25 ? x 5 5 x 25 1 5 5 x 0 5 1 7. C 5 ln 5 9. x 5 5 1 ln 1 2
differential equation is
5 2ty. Solving this

2
equation yields y 5 6eCet . To make this look like our equation, we require that y( 0 ) 5 1. This forces 6eC to be 1 and the resultant equation to be y 5 et .
2
37. Separating the differential equation yields y23 dy 5 x dx y 22 1 5 x2 1 C 22 2 1 1 2 2 5 x2 1 C 2y 2 If y 5 0, the left-hand side of the equation is undefined. So y 2 0 and thus y( 0 ) 5 0.
A52 39.
2y 1 yrx2 5 5 dy 2 x 5 5 2 2y dx dy dx 5 2 5 2 2y x dy dx 3 5 2 2y 5 3 x 2 ln Z 5 2 2y Z 5 2 x21 1 C 22 ln Z 5 2 2y Z 5 2x21 1 C Z 5 2 2y Z 5 e2>x 1 C Z 5 2 2y Z 5 e2>xeC Z 5 2 2y Z 5 Ae2>x, where A 5 eC There are two branches to the solution. First branch: 5 2 2y 5 Ae 2 2y 5 Ae2>x 2 5 1 5 y 5 2 Ae2>x 1 2 2 Second branch: 2 ( 5 2 2y ) 5 Ae2>x 2 5 1 2y 5 Ae2>x 2y 5 Ae2>x 1 5 1 5 y 5 Ae2>x 1 2 2
2>x
(d) Yes. It appears that the number of new cases will continue to drop. 9. (a) y 5 85.88 . This equation 1 1 29.65e20.9249t represents the percentage of public schools with access to the Internet, where t is the number of years since the end of 1994. (b) Increasing at the rate of 7.4 percentage points per year dN 5 0.02311N( 27.80 2 N ) dt (b) Increasing at $4.03 per year and $1.98 per year, respectively
11. (a)
13. 57.1 minutes; 2124 people per minute 15. 22 homicides per year 17. (a) Week 1 (b) The model predicts that the cumulative gross box office sales will level out at about 242 million dollars after Week 51. Therefore, new sales are diminishing. 19. About 69.75%; about 0.15 percentage point per year 21. If y , M, then 0 , M 2 y. This means that
dy dt
5 0.05( M 2 y ) must be positive for all values
of M greater than y. Therefore, the graphs of the solution curves will be increasing. Notice, however, that as y nears M, the value M 2 y nears
dy dt
Section 10.4 (page 686) 1. (a) 14.56 cm (b) 16.25 cm (c) Female 3. (a) 2.86 cm/year (b) 4.921 cm/year (c) Male; male is 20 cm at t < 1.994 years and female is 20 cm at t < 1.555 years 5. (a) 84.00 cm (b) 11.438 cm/year 7. (a) 927 dN (b) 5 2 0.000084335N( 919.9 2 N ) dt (c) < 2 5.86 cases per year. According to the model, the number of new pediatric AIDS cases was dropping at a rate of approximately 6 cases per year at the end of 2000.
0, which means that
S 0. So the solution curve
flattens out as y nears M. As M varies, the solution curves will be horizontal shifts of one another because the differential equation depends only on y. 23. Since M is the maximum value of the function y, y # M and 0 # M 2 y. Therefore, k( M 2 y ) $ 0. Thus the derivative 0 # y , M.
dy dt
is always increasing for
25. A limited growth curve is increasing and concave down. As the function approaches its maximum value, the rate of change approaches 0. An increasing logistic growth curve starts out concave up but turns concave down. The rate of change is initially increasing as the function approaches its maximum value. However, after reaching the inflection point, the rate of change begins to decrease, approaching 0 as the function nears its maximum value.
Answers to Odd-Numbered Exercises M , 1 Se2kMt
A53
27. We must prove that if y 5 1

dy dt
then
M . 1 Se2kMt
15. t 5
5 ky( M 2 y ) . We know that y 5 1
( 2x ) 3>2 C 1 x 12x tet 2 et 1 K t C esin x
d d M (y) 5 a b dt dt 1 1 Se2kMt dy d 5 3M( 1 1 Se2kMt ) 214 dt dt 5 2 M( 1 1 Se2kMt ) 22( Se2kMt )( 2 kM ) kM 2Se2kMt 5 ( 1 1 Se2kMt ) 2 M MSe2kMt 5 ka b 2kMt b a 1 1 Se 1 1 Se2kMt Add 0 M 1 MSe2kMt 2 M 5 ky a b ( M 2 M ) to 1 1 Se2kMt numerator M( 1 1 Se2kMt ) 2 M 5 ky a b 1 1 Se2kMt M b 1 1 Se2kMt 5 ky( M 2 y ) 5 ky a M 2 10.5 Algebra Flashback (page 704) 1. 3. 5. (x 1 3)3 1 5 5 5 (x 1 3) 1 2 (x 1 3) (x 1 3)2 x5 2 x3 1 x 2 5 1 5 5 x3 2 x 1 2 2 2 x x x
( w 1 2) 2 1 3 ? ( w 1 5 ) 1 1 w16
17. C 5
19. y 5 1 1
21. y 5 2e x 1 2e2x 1 7 23. y 5 x 2 1 2 2x 25. S 5 27. (a) (b) 4t( 30 1 t ) 1 75 15 1 t dC C 5 0.015 2 dt 150
C
2.0 1.5 1.0 0.5 0 0 20 40 60 80 100 t
The chlorine level increases at a decreasing rate as time goes on, until it reaches an upper limit of about 2.25 parts per million of chlorine. (c) After about 127.1 minutes, or about 2 hours and 7 minutes, the chlorine level will reach the desired level of 1.5 parts per million. 29. (a) dM 5 0.025M 1 2600 dt (b) $59,888.62
cannot be simplified in
a helpful way. Section 10.5 (page 704) 1. Reverse Product Rule 3. Reverse Product Rule 5. Separation of variables 1 7. y 5 e2x 1 Ce22x 4 3 C 9. N 5 t 2 1 2 2 t 11. y 5 2 2x 2 1 C x21
31. The goal of the integrating factor method is to make one side of the differential equation equal to the result of a product rule. By creating and recognizing this situation, we can readily integrate both sides of the equation and solve for the dependent variable. 33. Answers will vary. A sample student response follows. When dealing with situations where one quantity is going in (e.g., water into a tank, money into an account) and one quantity is going out (e.g., water leaking from the tank, money being withdrawn from the account), we can determine the rate of change of the quantity by finding the difference rate in rate out. For example, if we deposit $500 into an account each week but withdraw $250, Rate in 2 rate out 5 $500 2 $250
1 1 C 13. v 5 t ln t 2 t 1 2 4 t
A54
This result tells us that the net effect is that we are adding $250 to the account each week. 35. When the rate in and the rate out are equal, the term containing the variable t will be eliminated from the differential equation. Recall that in developing a differential equation for the salt-water tank situation, we can focus on the denominator of the rate out portion of the equation. In this denominator, we consider the net change in the quantity. If the net change is 0, then the term containing the variable t will be eliminated. dy 5 0.12y 2 500 37. dt Answers will vary but may involve some initial investment growing at a rate of 12 percent but with a quantity of 500 units being taken away per unit of time. 39. dy 5 0.085y dt Answers will vary but should involve continuous growth at 8.5 percent.
(b) The population of the community initially increases but then decreases. That is, the population fluctuates over time. The greater the population, the greater the maximum population reached before falling to the minimum population. Section 10.2 (page 710) 11. y( 4 ) 5 9.016 13. y( 4 ) 5 1.568 15. y( 4 ) 5 2 1.227 17. By Eulers Method we estimate that the number of hours of Internet use per person per year will be 200 hours around 2002 or 2003. 19. By Eulers Method we estimate that the population will never reach P 5 3. Rather, the population oscillates, reaching a lesser and lesser maximum value with each cycle. Section 10.3 (page 711) (x 1 C)3 27 21. y 5 6"x 2 1 1 1 2C 2 1 23. y 5
CHAPTER 10 REVIEW EXERCISES Section 10.1 (page 709) 1. C 3. A 5. G 7. (a)

H
200 150 100 50 0 0 2 4 6 8 10 t
25. y 5 6"3x 2 2 2 27. y 5 2e2x 1 1 e2x
29. It takes a little over 8 seconds for the height of the water in the tank to be 5 feet. 31. It will take approximately 53.47 minutes for the lemonade to cool to 50F. Section 10.4 (page 712) 33. (a) dN 5 0.01473N( 121.8 2 N ) dt (b) 41.91; 14.87 million units per year
(b) The number of hours of Internet use per person per year increased at an increasing rate for years near 1995. As the number of hours of use increases, the rate at which use increases decreases and reaches an upper limit of about 232 hours per person per year. 9. (a)
P
10 8 6 4 2 0 3 2 5 t
2 2
35. (a) 69.07 cm (b) 9.3 cm per year 37. 0.134 homicide per 100,000 people per year (about 1.34 homicides per million people per year) Section 10.5 (page 713) 39. Separation of variables 41. y 1 ln Z y Z 5 6 ln Z x Z 1 C C 1 43. y 5 x 2 1 x 3
A55
45. y 5 47. S 5
3 1 Ce22x 4
(c) The sequence converges to approximately 2.71 ( e ) . 7. an 5 3 1 5( n 2 1 ) , where a1 5 3. This sequence does not converge. 9. an 5 n2 11. an 5 e
7 n , 11
24t 1 3t 2 1 8 41t
dM 5 0.03M 1 5200 49. (a) dt (b) $47,969.95 51. It will take 12.5 minutes for the tank to fill. When full, the tank will contain approximately 61.3 pounds of salt. CHAPTER 11 11.1 Algebra Flashback (page 729) 1. (a) Yes. The data have a common ratio of 1.05 (b) y 5 2( 1.05x ) 3. (a) No. The data are linear with a common difference of 20. 5. (a) No. The data are quadratic with a common second difference of 200. 7. y 5 100( 0.90 ) x 9. y 5 500( 1.025 ) x Section 11.1 (page 730) 1 2 3 4 5 6 1. (a) an 5 , , , , , , c 4 7 10 13 16 19 (b)
where a1 5 2.
This sequence converges to 0. 0, if n is odd 5, if n is even This sequence does not converge.
3
13. an 5 4an 2 1 2 4an 2 2, where a1 5 0 and a2 5 50. This sequence converges to 200. 15. an 5 300 1 300( n 2 1 ) , where a1 5 300. This sequence does not converge. 17. an 5 25,900( 0.85 ) n The value of the car converges to 0, assuming that the car depreciates by 15% indefinitely. 19. (a)
0 1 2 3 4 5
0 0.70 ? 0 1 12 5 12 0.70 ? 12 1 12 5 20.4 0.70 ? 20.4 1 12 5 26.28 0.70 ? 26.28 1 12 5 30.396 0.70 ? 30.396 1 12 5 33.28
(b) 0, 12, 20.4, 26.28, 30.396 c an 5 0.70 ? an 2 1 1 12 (c) The sequence converges to approximately 0.328. 1 1 1 1 1 1 3. (a) an 5 , , , , , , c 2 4 8 16 32 64 (b) (c) The suit should be changed after 6 hours. (d) It appears as though the suit will never accumulate 40 micrograms of chemical. Note that 0.70 ? 40 5 28. So once the suit accumulates 40 micrograms of chemical, each hour 30% will be neutralized, leaving 28 micrograms. Then an additional 12 micrograms are added, making the total 40 micrograms again. 21. (a) 500, 640, 843, 1138, c (b) an 5 1 ? 45an 2 1 2 85, where a0 5 500 and n represents the number of 5-year periods.
(c) The sequence converges to 0. 5. (a) an 5 2, 2.25, 2.37, 2.44, 2.49, 2.52, c (b)
A56 (c)

an
30,000 20,000 10,000 0 0 4 8 12 16 n
35. The sequence converges to 1. 11.2 Algebra Flashback (page 744) 1. (a) The table represents a linear function with a common difference of 1.5. (b) y 5 1.5x 1 10 3. (a) The table represents a linear function with a common difference of 5. 2 (b) y 5 x 1 2 5 5. (a) The table represents an exponential function with a common ratio of 10 for every increase of 2 in x. (b) y 5 0.001( 10 ) x>2 7. y 5 2 5x 1 25 9. y 5 500( 2 ) x>10 Section 11.2 (page 744) 1. This is a geometric series with a constant ratio 1 of 2. That is, each subsequent term of the series is found by multiplying the previous term by 2.
1 2
(d) No. As the 5-year plan is repeated, the number of trees increases without bound. (e) No, the forest size does not stabilize. 23. 4-hour half-life: an 5 0.016an 2 1 1 240
With such a short half-life, the maintenance level of 243.9 mg is quickly reached. 12-hour half-life: an 5 0.25an 2 1 1 240
In this case, the maintenance level of 320 mg is reached more slowly than in the 4-hour half-life situation. 25. Answers will vary. 27. The sequence 1, k, k2, k 3, k 4, c will converge when 2 1 , k # 1. Otherwise, the sequence diverges. 29. For a strictly increasing sequence to converge, it must increase at a decreasing rate. The graph of such a sequence would resemble a function that is concave down. If increasing and concave down, the sequence will continue to increase but at a lesser and lesser rate, causing the sequence to converge. For a strictly decreasing function to converge, it must decrease in magnitude at a decreasing rate. The graph of this sequence would resemble a function that is concave up. If decreasing and concave up, the sequence will continue to decrease but at a lesser and lesser rate, causing the sequence to converge. 31. an 5 a1>2n 2 1, where 0 , a , 1. The sequence converges to 1. 33. The sequence converges to 0.
3. This series is not geometric. The series does not have a constant ratio. 5. This is a geometric series with a constant ratio of 2. That is, each subsequent term of the series is found by multiplying the previous term by 2. 7. S4 5
10 ; 3
Sn 5 a 3
k51
The series diverges because the terms of the series get larger as n increases. 9. S4 5 11.11; Sn 5
100 9
The series converges to 11. B 13. E 15. D
100 1 9 10 100 . 9
A B
17. This series will converge because r 5 5, which is between 2 1 and 1. 19. This series will converge because r 5 3, which is between 2 1 and 1. 21. 0.222 c 5 2 9
2
A57
23. 2.24 5 25. 0.18 5
222 99 2 11
39. Answers will vary, but students should realize that grouping a finite number of terms of an infinite series as shown in the proof is problematic. 11.3 Algebra Flashback (page 761) 1. Polynomial 3. Polynomial 5. Not a polynomial because the exponents are not whole numbers 7. Not a polynomial because the exponents are not whole numbers 9. Not a polynomial because the variable is in the exponent 11. lim f ( x ) 5 `
xS` x S 2`
27. The Final Four tournament will have an estimated $75 million impact on the local economy of Indianapolis. 29. Answers will vary. 31. An infinite geometric series will not converge unless the common ratio r is a value between 1 and negative 1 ( 2 1 , r , 1 ) . This is because, when r . 1 or r , 2 1, the terms of the series will become larger and larger in magnitude. If the terms become larger and larger in magnitude, the series cannot converge. 33. A repeating decimal can be written as the sum of a infinite geometric series, where 2 1 , r , 1. Therefore, finding the sum of this infinite geometric series enables us to find the equivalent rational number. Student examples will vary. 35. e 5 1 1 1 1
` 1 e5 a n! n50
lim f ( x ) 5 `
13. lim f ( x ) 5 2`
xS` x S 2`
lim f ( x ) 5 2`
15. lim f ( x ) 5 `
xS` x S 2`
1 1 1 1 1 1c 2 6 24
lim f ( x ) 5 `
17. lim f ( x ) 5 2`
xS` x S 2`
1 1 1 e<1111 1 1 1c 2 6 24 < 2.708 1 51 37. a n ( n 1 1) n51 1 1 1 1 1 1 1 1c 1( 1 1 1 ) 2( 2 1 1 ) 3( 3 1 1 ) 4( 4 1 1 ) 5 1 1 1 1 1 1 1 1c 2 6 12 20

`
lim f ( x ) 5 2`
19. lim f ( x ) 5 2`
xS` x S 2`
lim f ( x ) 5 `
Section 11.3 (page 761) 1. Answers will vary. 1 1 1 6 x 3. P6( x ) 5 1 2 x2 1 x4 2 2 24 720 1 1 1 5. P3( x ) 5 1 1 x 2 x2 1 x3 2 8 16 1 2 7. P6( x ) 5 1 2 2x2 1 x4 2 x6 3 45 3 9 27 9. P3( x ) 5 1 1 x 2 x2 1 x3 2 8 16
1 1 1 1 1 1 1 5 0.8 2 6 12 20
y
1 0.8 0.6 0.4 0.2 0
0 20 40 60 80 x
9 81 11. P4( x ) 5 3x 2 x2 1 9x3 2 x4 2 4
A58
1 1 13. P4( x ) 5 x 1 x2 1 x3 1 x4 2 6
y
40 30 20 10 8 6 4 2 0 2 x
23. P3( x ) 5 1 2 ( x 2 5 ) 1 ( x 2 5 ) 2 2 ( x 2 5 ) 3
y
5 4 3 2 1 0 1 2 3
Interval of convergence: ( 2 2, 2 ) 1 1 1 7 15. P7( x ) 5 x 2 x3 1 x5 2 x 2 24 720

an
3 2 1 1 2 3
25. Degree of Polynomial 1 3 5 7 Polynomial Approximation of f ( 0.5 ) 0.5000 0.5417 0.5458 0.5463 Value of tan( 0.5 ) 0.5463 0.5463 0.5463 0.5463
Interval of convergence: ( 2p, p ) 1 1 1 17. P3( x ) 5 1 1 ( x3 2 1 ) 2 ( x3 2 1 ) 2 1 ( x3 21 ) 3 2 8 16

y
3 2 1 0 0 1 2
As the degree of the polynomial gets greater and greater, the polynomial approximation of f ( 0.5 ) approaches the actual value of tan( 0.5 ) . 27. Answers will vary. 29. Answers will vary. 31. We know that 1 1 ln( x ) 5 ( x 2 1 ) 2 ( x 2 1 ) 2 1 ( x 2 1 ) 3 2 3 1 1 4 2 (x 2 1) 1 (x 2 1)5 2 c 4 5 Therefore, 1 1 ln( 2 ) 5 ( 2 2 1 ) 2 ( 2 2 1 ) 2 1 ( 2 2 1 ) 3 2 3 1 1 4 2 (2 2 1) 1 (2 2 1)5 2 c 4 5 1 1 1 1 ln( 2 ) 5 1 2 1 2 1 2 c 2 3 4 5 p 21 33. Since tan ( 1 ) 5 , 4 p 1 1 1 1 3 5 1 2 ( 1 ) 1 ( 1 ) 5 2 ( 1 ) 71 ( 1 ) 9 2 c 4 3 5 7 9 p 1 1 1 1 512 1 2 1 2c 4 3 5 7 9 Multiplying both sides by 4 gives p542 4 4 4 4 1 2 1 2c 3 5 7 9
Interval of convergence: ( 0.5, 1.5 ) 19. P3( x ) 5
"3 1 p "3 p 2 2 ax 2 b 2 ax 2 b 2 2 6 4 6 1 p 3 1 ax 2 b 12 6
y
1 2 1 1 2 1 2 3 4 x
21. P3( x ) 5 e16 1 8e16( x 2 4 ) 1 33e16( x 2 4 ) 2 280 16 1 e (x 2 4)3 3

8 10 7 6 10 7 4 10 7 2 10 7 0 2 10 7
1 2 3 4
A59
1 3 5 7 x 35. P7( x ) 5 x 1 x3 1 x5 1 6 40 112 11.4 Algebra Flashback (page 772) 1. 2 4 , x , 12 3. x # 2 3 or x $ 3 1 5. x , 2 6 or x . 7 3 7. x , 2 3 or x . 2 9 1 9. x # 2 or x $ 2 2 11. Answers will vary. One answer is Z x 2 1 Z 1 3 , 2 1. Section 11.4 (page 773) 1. Answers will vary.
` 1 x2n 2 2( 2 1 ) n 1 1 3. a ( 2 n 2 2 ) ! n51 ` 1 1 5. 1 1 ( x 2 1 ) 1 a n ( x 2 1 ) n 2 1( 2 1 ) n 2 n53 2 ` 3n 7. a xn( 2 1 ) n 1 1 n51 n
1 1 1 25. 3ex dx 5 x 1 x3 1 x5 1 x7 1 c 1 C 3 10 42
2
27. The interval of convergence is 2 , x , 4, also written as ( 2, 4 ) . 29. We already know the Taylor series for y 5 ex. We can find the Taylor series for y 5 ex by replacing x with x 2 in the Taylor series for y 5 ex. By expanding the Taylor series for several terms, we can integrate this Taylor expansion term by term. The resulting series can then be expressed as an infinite Taylor series.
2
31. sinh( x ) 5 33. tanh( x ) 5
` ex 2 e2x 1 5 a x 2n 2 1 2 ( 2 n 2 1)! n51 ` ex 2 e2x x 5 x 2x a e 1e n 5 1 2n 2 1
1 1 1 6 c 35. cosh( x ) 5 1 1 x2 1 x4 1 x 1 2 24 720 d 1 1 5 c 3cosh( x ) 4 5 0 1 x 1 x3 1 x 1 dx 6 120 5 sinh( x ) CHAPTER 11 REVIEW EXERCISES Section 11.1 (page 776) 5 10 17 26 37 1. (a) an 5 2, , , , , , c 2 3 4 5 6 (b)
1 9. a x2n n 5 1 n! 11. ( 2` , ` ) 13. ( 0, 2 ) 1 1 15. a 2 , b 3 3 17. ( 2` , ` ) 19. ex 5 1 1 x 1 1 2 1 1 x 1 x3 1 x4 1 c 2! 3! 4! d x 1 1 3e 4 5 0 1 1 1 x 1 x2 1 x4 1 c dx 2! 3!
(c) The sequence seems to diverge. Each subsequent term in the sequence is larger than the preceding term. 3. (a) an 5 50, 27, 15.5, 9.75, 6.875, 5.4375, c (b)
3 3 9 21. 3e3xdx 5 x 1 x2 1 x3 1 x4 1 c 1 C 2 2 8 1 1 23. 3ln( x ) dx 5 ( x 2 1 ) 2 2 ( x 2 1 ) 3 2 6 1 1 1 (x 2 1)4 2 (x 2 1)5 12 20 1 c1 C
(c) The sequence converges to 4. 5. an 5 3n 2 5 The sequence diverges.
A60 7. an 5
2n 2n 1 1 The sequence converges to 1.
9. (a) 11.5, 5.75, 2.875, 1.4375, 0.71875 1 n21 (b) an 5 11.5 a b 2 (c) 0.022 mg (d) 0.0000219 mg Section 11.2 (page 776) 11. This is an arithmetic series because each subsequent term in the series is found by adding 3 to the preceding term. 13. This is an arithmetic series because each subsequent term in the series is found by adding 1 to the preceding term. 15. 0.1 5 1 9
1 23. P3( x ) 5 e 1 e( x2 2 1 ) 1 e( x2 2 1 ) 2 2 1 1 e( x2 2 1 ) 3 6
y
30 20 10 2 1 0 1 2 x
25. P3( x ) 5 1 2 ( x 2 3 ) 1 ( x 2 3 ) 2 2 ( x 2 3 ) 3
y
5 4 3 2 1 0 1 2
4 x
Section 11.4 (page 777) 27. ( 2 1, 1 ) ex 1 1 29. 3 dx 5 ln Z x Z 1 x 1 x2 1 x3 x 4 18 1 4 ... 1 x 1 1C 96 CHAPTER 12
2 17. 0.18 5 11 1 n>15 19. an 5 871.1 ? a b 2
If we assume that the medicine is essentially gone from the bloodstream when the amount is less than 1 mg, it will take about 147 hours or a little over 6 days. Section 11.3 (page 777) "2 "2 p "2 p 2 21. P3( x ) 5 1 x2 B2 x2 B A A 2 2 4 4 4 "2 p 3 1 Ax 2 4 B 12
y
0.75 0.50 0.25 0 0 0.5 1.0 1.5 x
12.1 Algebra Flashback (page 791) 1. 5 14 3. 1 6 5. 1295 1296
Section 12.1 (page 792) 1. Since 3 A x 1 2 B dx 5 1, f ( x ) 5 x 1 2 is a

1
probability density function on 30, 14 .

2 1
3. Since 3
a probability density function on 31, 24 .

1
3 2 6 3 6 x 1 13 B dx 5 1, h( x ) 5 13x 2 1 13 is A 13 1 1 1 t 3 1 t 2 1 2 B dt 2 1, r( t ) 5 3t 3 1 t 2 1 2 A1 3
5. Since 3
0
is not a probability density function on 30, 14 .
A61
45
7. Since 3 r( t ) dt 2 1, r( t ) is not a probability density function on 30, 44 .

` 0
(d) 3 c ? B( x ) dx 5 0.389
29
9. Since 3 f ( x ) dx 5 1, f ( x ) is a probability density

2`
The probability that a mother chosen at random is at least 29 years old is about 38.9 percent.
25
function on ( 2` , ` ) . 11. c 5 3 15. c 5 19. c 5 10 3381 "p

24 16
(e) 3 c ? B( x ) dx 5 0.3973 13. c 5 1 20

16
17. c 5 2 < 1.772
The probability that a mother chosen at random is younger than 25 years old is 39.7 percent.
40
(f) 3 c ? B( x ) dx 5 0.3096
30
21. (a) 3 p( d ) dd 5 1 (b) 17.6 percent of high school dropouts are less than 18 years old but greater than 16 years old. (c) 82.4 percent of high school dropouts are 18 years old or older (up to age 24).
80
The probability that a mother chosen at random is in her thirties is 30.1 percent.
Number of deaths (in thousands)
27. (a)
12 10 8 6 4 2 0
23. (a) 3 p( w ) dw 5 1
16
0 10 20 30 40 50 60 70 80 90 Age
(b) About 9 percent of the female labor force is younger than 25 years old. (c) About 61.1 percent of the female labor force is at least 25 years old but less than 50 years old. (d) About 19.6 percent of the female labor force is 55 years old or older.
Number of births (in thousands)
25. (a)
1000 800 600 400 200 0
(b) D( x ) 5 20.1072 ? 0.972274x 1 (c) c 5 < 0.0024 415.966 (d) About 48.5 percent of deaths due to motor vehicle accidents involve people who are at least 15 years old but less than 35 years old. (e) About 85.2 percent of deaths due to motor vehicle accidents involve people who are at least 20 years old. (f ) About 14.8 percent of deaths due to motor vehicle accidents involve teenagers. 29. 97.6 percent of deaths attributed to cardiovascular disease involved people 44 years old or older. In Exercise 26(d), we found the percentage to be 98.4 percent. Because of modeling discrepancies, the results are not the same but are very similar. 31. Answers will vary. Responses should reflect the explanations given in the text. 33. Answers will vary. A sample student response follows: Computationally speaking, the area under the probability density function h( x ) 5 13 x 2 1 13 on 31, 24 is 0.52 on the interval
3 6
15 30 45 Age
(b) B( x ) 5 0.2069x3 2 22.396x2 1 731.47x 2 6400.5 1 (c) c 5 20,385.1989 < 0.0000491

45
3 c ? B( x ) dx 5 1
16
( 1, 1.6 ) . This means that with probability 0.52, the value of the random variable will be on the interval ( 1, 1.6 ) .
A62

c
35. The integral 3
3 2 6 x 1 13 B dx can be interpreted A 13
The probability of the patient surviving 10 or more years is about 95.9 percent. 27. Answers will vary. Student responses should reflect the explanations given in the text. 29.
b 1 1 3 b 2 a dx 5 b 2 a x ` a a 1 1 5 a ? bb 2 a ? ab b2a b2a 1 5 a b ? (b 2 a) b2a 51
37. F( C ) 5 0. The area under the probability density function on 3C, C4 is zero because there is no area. The probability that a random variable is at least C but less than C is zero.
62
as the probability that a random variable will take on values greater than 1 but less than c, where c is on 31, 24 .
39. F( 62 ) 5 3 f ( t ) dt represents the probability that

50
a randomly chosen student is at least 50 inches tall but less than 62 inches tall. 12.2 Algebra Flashback (page 801) 1. Answers will vary. 3. 2 2, 0, f ( 2 2 ) , f ( 0 ) , f ( 2 ) 5. Answers will vary. For example, f ( x ) 5 2 2e22x. Section 12.2 (page 802) 1. f ( x ) 5 5. f ( x ) 5 1 5 1 10 3. f ( x ) 5 1 6
31. If the stipulations are not true, then the total area under the probability density function on 30, ` ) will not be 1. ` b 2ax ` 2ax 3 be dx 5 2 a e ` 0 0 b b 5 lim a 2 e2a ? C b 2 e2a(0) a a CS` 5 b a
Note that the result will be 1 only if a 5 b and a 2 0, b 2 0. 33. F( x ) 5 2 e2ax 1 e2aC 35. There is a very small probability (0.055 percent) that you would have to wait more than 30 minutes.
7. P( x ) 5 0.06e20.06x 11. P( x ) 5 0.50e20.50x
9. P( x ) 5 0.12e20.12x
1
13. The probability of waiting 5 minutes or less for the bus is . 2 15. The probability is
1 16
that the spinner will land in
the first-quadrant green area. 17. Because the colored regions are of equal size, the area under the constant function will be the same for any color. If this scenario were not a uniform situation, then this would not be the case. 19. The probability is 50 percent that another phone call will come in within the next 3 minutes. 21. There is a 4.4 percent chance of failure during the warranty period. 23. There is a 90 percent chance that the patient will survive for more than 2 years. 25. P( x ) 5 0.0042 ? e20.0042 ? x
`
20.0042x dx < 0.959 3 0.0042e
12.3 Algebra Flashback (page 817) 1. a ( n2 2 1 ) 5 0 1 3 1 8 1 15 1 24 1 35

n51 6
5 85
3. a sin( np ) 5 0 1 0 1 0 1 0 1 0 1 0
n50
50 5. a 2n
n51 4 5
7. a n2
n51
9. a n2 1 2n
n51
10
A63
Section 12.3 (page 817) 1. (a) E( x ) 5 7 12 11 144
(b) Var( x ) 5
"11 (c) s( x ) 5 12 3. (a) E( x ) 5
17. (a) E( x ) 5 4 (b) Var( x ) 5 16 (c) s( x ) 5 4 19. (a) E( x ) 5 3 (b) Var( x ) 5 9 (c) s( x ) 5 3
25 3 5 5 !3 (d) s( x ) 5 or !3 3 (c) Var( x ) 5
59 < 1.31 45 (b) Var( x ) < 0.2665 (c) s( x ) < 0.5163
5. (a) E( x ) 5 1.8656 (b) Var( x ) 5 1.1867 (c) s( x ) 5 1.0894 7. (a) E( x ) 5 3 2
3 4 "3 (c) s( x ) 5 2 (b) Var( x ) 5 9. (a) E( x ) 5 6 5 "3 3 50
(b) Var( x ) 5 (c) s( x ) 5 11. (a) f ( x ) 5
21. (a) E( x ) 5 0.406. In the long run, $0.41 is won each time the lottery game is played. (b) Var( x ) 5 222,449.07, s( x ) 5 $471.65 The variance and standard deviation are very large when compared to the variation and standard deviation for the Fantasy 5 game. This tells us that winning amounts are much more spread out (deviate from the expected value or mean) than in the Fantasy 5 game. In fact, prizes vary from $2 to $100,000. In the Fantasy 5 game, prizes vary from $50 to $50,000. (c) It does not pay, in the long run, to play. While $0.406 is the expected amount won for each play, $1 is spent. The net loss is nearly $0.594 per play. 23. (a) E( d ) 5 19.51. The expected age of a high school dropout is about 19.5 years old. (b) Var( d ) 5 1.811 (c) s( d ) 5 1.346 years 25. (a) E( w ) 5 42.7. The expected age of a woman in the work force is 42.7 years old. (b) Var( w ) 5 163.9 (c) s( w ) 5 12.8 years 27. The expected wait time, or mean, is 5 minutes. 29. Since "3 < 1.73 minutes and < 2.31 "3 minutes, we see that the uniform random variable deviates from the mean (expected value) on 30, 84 more than it does on 30, 64 . 4
5"2
1 10 (b) E( x ) 5 5 (c) Var( x ) 5 (d) s( x ) 5 25 3 5
"3
13. (a) f ( x ) 5
1 15 25 (b) E( x ) 5 2 75 (c) Var( x ) 5 4 5"3 (d) s( x ) 5 2
31. Case 1, 21-month survival: Var( x ) 5 400, s( x ) 5 20
1 15. (a) f ( x ) 5 10 (b) E( x ) 5 0
Case 2, 14-month survival: Var( x ) 5 204.08, s( x ) 5 14.3
A64
The variance and standard deviation are smaller in Case 2. This tells us that the survival times in Case 2 deviate from the mean less than the survival times in Case 1. 33. Answers will vary. 35. Answers will vary. 37. Answers will vary.
14
deviations imply that the data are more spread out, thereby creating normal distributions that are relatively flatter. 13. E 17. A 21. 390, 1104 15. B 19. 0.3 percent 23. 46 percent
25. 27 percent 27. According to the normal distribution model, nearly 0 percent of NBA players are 7 feet tall or taller and nearly 0 percent of NBA players are under 6 feet tall. These results lead us to question the validity of the normal distribution model. Data found at www.nba.com show that in actuality about 11 percent of NBA players are 7 feet tall or taller and about 2 percent are under 6 feet tall. 29. The function P( x ) 5 s"2p 1 e2(x 2m) >(2s )
2 2
39. E( x ) 5 3 x ? f ( x ) dx
5
The mean represents the expected, or typical, shoe size of an adult male chosen at random. Knowing this value would help the bowling center owners be sure to have plenty of the correct shoe sizes available. 41. Refer to the textbook, Section 12.3, for details. Section 12.4 (page 832) 1.
is a probability density function. We know that the total area under a probability density function must be equal to 1; therefore, 3.
`
3 P( x ) dx 5 1.
2`
5.
7.
31. For the probability density function 1 P( x ) 5 e2x >2, we can say that "2p 1 1 1 2x >2 1 2x >2 2? 3 e dx 5 3 e dx because 0 "2p 21 "2p 1 2x >2 e of symmetry. That is, because P( x ) 5 "2p
2 2 2 2
9.
11. When the mean m is held constant and the standard deviation s is allowed to vary, the normal distribution gets flatter and flatter as the standard deviation increases. Since small standard deviations imply that the data are clustered around the mean, the normal distribution reaches a high maximum value at the mean. Larger standard
33. Answers will vary.
is symmetric with respect to the vertical axis, the area under the curve to the left of the vertical axis, 0 1 2x >2 e dx, is equal to the area under the 3 21 "2p curve to the right of the vertical axis, 1 1 2x >2 e dx. 3 0 "2p
2 2
35. The normal probability density function is maximized at x 5 m (the mean), where P 5
2.51 . s
37. The normal probability density function has inflection points at x 5 m 1 s and x 5 m 2 s.
A65
CHAPTER 12 REVIEW EXERCISES Section 12.1 (page 836)

1
Section 12.3 (page 838) 19. (a) E( x ) 5 4 (b) Var( x ) 5 (c) s( x ) 5 21. (a) E( x ) 5 16 3 4
1. Since 3 f ( x ) dx 5 1, f ( x ) is a probability density function. Also, f ( x ) . 0 on 30, 14 .

2 0
"3
< 2.31
3. Since 3 h( x ) dx 5 1, h( x ) is a probability density function. Also, h( x ) . 0 on 30, 14 . 5. c 5 9. (a) 3 124

Number of deaths 6000 5000 4000 3000 2000 0 1
7. c 5
5 32
7 10 (b) Var( x ) 5 0.033 (c) s( x ) 5 0.182
23. (a) E( x ) 5 1.176 (b) Var( x ) 5 1.384 (c) s( x ) 5 1.176 25. (a) E( x ) 5 15 2 (b) Var( x ) 5 18.75 (c) s( x ) 5 4.33 4 3 16 9
0 30 40 50 60 Age
(b) D( x ) 5 2 2.124x2 1 102.324x 1 5720.24 1 (c) c 5 252,604.87 (d) The probability is 23.9 percent that the age of a driver involved in a fatal accident will be between 30 and 39 years old. (e) The probability is 36.1 percent that the age of a driver involved in a fatal accident will be older than 50 years old. Section 12.2 (page 837) 11. 3 c dx 5 c ? x `
1 6
6 1
27. (a) E( x ) 5
(b) Var( x ) 5 (c) s( x ) 5 4 3
29. (a) E( x ) 5 25 (b) Var( x ) 5 208.3 (c) s( x ) 5 14.434 31. (a) E( x ) 5 200. The mean (or expected) number of miles traveled before breakdown is 200,000 miles. (b) Var( x ) 5 40,000 (c) s( x ) 5 200 33. The variance and standard deviation are smaller for the March 2006 data than for the March 2005 data. This tells us that the number of miles traveled before breakdown deviated from the mean less in March 2006 than in March 2005. Section 12.4 (page 839) 35. A. m 5 10 B. m 5 10 s 5 0.5 s52
5c?62c?1 55?c
Since we seek a function, f ( x ) , such that

6
3 f ( x ) dx 5 1.
1
5?c51 1 c5 5 1 f (x) 5 5 13. P( x ) 5 0.032e20.032x 15. The probability is that the wait time will be less 2 than 2 minutes. 17. The probability of a train breaking down in the first 100,000 miles of travel is about 39.3 percent.
1
Graph B is much flatter than Graph A because the standard deviation is so much greater.
A66
37. A. m 5 50 B. m 5 25
s55 s55
41. The probability is 7 percent that a randomly chosen player will have a batting average of 0.400 or better. 43. The probability is 31.6 percent that a randomly chosen player will have a batting average of 0.200 or less.
Graph A is the same shape as Graph B, but is shifted to the right because the value of m is greater. 39. About 55.3 percent of pregnancy lengths are greater than 182 days but less than 282 days.
45. 30.178, 0.3164
Index
D (Delta), 419 ! (Factorial), 751 m (Mu), 820, 821 p (Pi), 746 s (Sigma), see Standard deviation S (Sigma), see Summation notation Absolute extrema of continuous functions, 308, 315320 of cubic functions, 317, 319320, 366 in curve sketching, 368 definition of, 308, 386 at endpoints, 308 of exponential functions, 319 First Derivative Test for, 315317 graphs of, 308 of logarithmic functions, 369 maximum, see Maximum, absolute minimum, see Minimum, absolute point of diminishing returns at, 363 of profit function, 329, 333 of quadratic functions, 329, 333 of quartic functions, 315 of revenue function, 331 from second derivative, 362 of volume function, 336337 Absolute value function derivatives of, 241, 310311 extrema of, 310311 graphs of, 310, 767 in inequality solution, 767 Acceleration, 363, 387 Acceleration function, 363365 Accumulated change, 456458, 485489, 619 Addition Property of limits, 205, 250 Amplitude, 584585, 588, 631 Angle of impact, 617 Angle-to-radian conversion, 749 Annual growth rate, 127, 166 Antiderivatives, see also Integrals of constant functions, 396 of cosine functions, 618, 620 of cost function, 400 of cubic functions, 453454 of linear functions, 397, 453, 607 of position functions, 399 of power functions, 394, 395 of quadratic functions, 452 of rate of change function, 412 of sine functions, 623 of tangent functions, 626 units of, 399 Appreciation, 127 Arccosine functions, 592 Arcsine functions, 593 Area of circles, 384 of cross, 389 enclosed, 455, 482485 beneath function, see Area between function and x-axis in histograms, 782783 of a rectangle, 384 surface, see Surface area of trapezoids, 444 of triangles, 384, 389 between two functions, 481485, 510 Area between function and x-axis actual vs. signed, 454455 definite integral for, 438445, 452460, 465, 618 Fundamental Theorem of Calculus for, 453456 for normal probability density functions, 821, 828 for probability density functions, 786789, 821, 828 rate of change and, 411412 Riemann sums for, 412430, 451 summation notation for, 434437 Area function, 340, 342343, 377378 Arithmetic sequence, 725726, 775 Augmented matrix, 71 Average deviation, 807808 Average rate of change definition of, 250 difference quotient for, 177182, 188189, 196 inputs in, 178 instantaneous rate of change and, 188189, 196 in linear functions, 30, 177 outputs in, 178 in quadratic functions, 92 secant lines and, 217 from tabular data, 177180, 188189 from word problems, 180 Average of Riemann sums, 422 Axis of symmetry, 65 Bar graphs, 780781 Base, 116118, 136138, 141, 167 Binomials, 268 Break-even point, 46, 327 Calculator Technology Tips for aligning a data set, 125 for cubic regression, 98, 784785, 787 for definite integrals, 445446, 460, 620 for density functions, 784785 for derivatives, 270, 317318 for difference quotient, 191 on e, 137 using the Equation Solver, 382383 for Eulers Method, 653656 for expected value, 807 for exponential probability density functions, 801 for exponential regression, 126 for function graphs, 10 for instantaneous change, 191, 192 on intervals, 10, 16 on inverse normal feature, 826, 829 I1
I2
Index
Calculator Technology Tips (continued ) for linear regression, 3738 on list editing, 3738 on LnReg feature, 143 for logarithmic regression, 144 for logistic regression, 685 for normal probability density functions, 821829 for probability density function, 787791, 801, 821829 for quadratic regression, 68 for quartic regression, 98 for regression graphs, 3940 for Riemann sums, 424426 for scatter plots, 41 for sequence graphs, 727728 on ShadeNorm command, 824, 826, 828 for sine functions, 592593, 595, 597598, 612 for slope fields, 639, 643644 for standard deviation, 809810 for table of values, 358 on upper limit of x-value, 825 for variance, 809810 for x-intercept, 321 Capital, 523 Cartesian coordinate system 2-dimensional, 5, 747 3-dimensional, 594, 595 Chain Rule, 277283, 287, 290, 301 Change-of-base formula, 141, 167 Circles, 294295, 349, 377378, 384 Cobb-Douglas production function for Coca-Cola, 523, 576, 580 for Ford, 525, 543, 561, 566567, 575 purpose of, 516 Coefficient of determination, 39, 67 Common log, 137141, 166, 289 Common ratio, 725, 734, 735, 775 Composition of functions, 275278 Concavity of continuous functions, 354, 387 of cubic functions, 9395, 151, 353356, 366367, 747 in curve sketching, 368 definition of, 94, 166, 353 directionality of, 94, 353 of exponential functions, 118, 151
inflection point and, 93, 166, 354 of logarithmic functions, 137138, 151, 369370 of logistic functions, 151 of quadratic functions, 6365, 151, 747, 748 of quartic functions, 100, 151, 354355, 747 of quintic functions, 747 rhymes for, 94, 353, 354 second derivative and, 354, 748 of sine functions, 748 Cones, 384, 388, 527 Constant difference, 747 Constant functions antiderivatives of, 396 continuity in, 212 critical points/values of, 313 definition of, 62 degree of, 93 derivatives of, 258, 313 extrema of, 313 graphs of, 411 integration of, 396, 400 as probability density function, 796797 Constant Multiple Property of limits, 205, 250 Constant Multiple Rule for derivatives, 260261, 301 for integrals, 396, 408, 465 Constant of proportionality, 34, 640642 Constant Property of limits, 205, 250 Constant Rule for derivatives, 258, 301 for integrals, 396, 465 Constrained objective function, 563573, 578 Constraint function, 562563 Consumer surplus, 490492, 494, 497 Continuity, 9, 211214, 251, 307 Continuous exponential decay model, 799 Continuous functions concavity of, 354, 387 criteria for, 251 critical points/values of, 312319 definite integral for, 438445, 452460
definition of, 212, 307 extrema of, 307310, 312320, 353, 359362, 387 Fundamental Theorem of Calculus for, 451458, 463, 465, 485 graphs of, 9, 211, 307 inflection point of, 354 integration of, 438445, 452458, 465 limits of, 193194 testing for, 213214 Continuous random variables definition of, 780, 835 expected value of, 806807, 835 standard deviation of, 809810, 836 variance of, 809810, 813, 836 Convergence of sequences, 719, 775 Convergence of series, 739, 740, 767 Convergence of Taylor series, 765770, 775 Converging improper integrals, 500506, 510, 511 Correlation coefficient, 39 Cosine functions accumulated change in, 619 amplitude of, 584585, 588 antiderivatives of, 618, 620 applications of, 620623 derivatives of, 609610, 618, 623, 631 frequency of, 584, 586, 588 graphs of, see Graph of a cosine function hyperbolic, 774 integration of, 619, 630, 632 midline of, 584585 period of, 586 sine functions and, 584, 588589 solving of, 592 tangent functions and, 596 Cost function antiderivatives of, 400 derivatives of, 328 extrema of, 346 linear model for, 45 marginal cost and, 400, 466 marginal revenue and, 328 profit function and, 45, 329 revenue function and, 332
Index
I3
Critical points/values of constant functions, 313 of constrained objective functions, 563573 of continuous functions, 312319 of cubic functions, 312, 318 derivatives at, 312319 of exponential functions, 319 extrema and, 312317, 547557 hints for finding, 566 of Lagrange functions, 562573, 578 for multivariable functions, 547557, 562573, 577, 578 for one-variable functions, 312319, 331, 386 for power functions, 312313 for revenue function, 331 singular/stationary, 312313, 316, 386 Cross, area of, 389 Cross sections of a surface, 534536, 571 Cubic equations basic, 93, 151, 167 by calculator, 98, 784785, 787 coefficients of, 9495 from Excel, 104106 in piecewise functions, 158159 from tabular data, 9899 from word problems, 156 Cubic functions antiderivatives of, 453454 constant difference of, 747 critical points/values of, 312, 318 curve sketching of, 365368 definition of, 93 degree of, 93, 747 as density function, 784785 derivatives of, first, 235, 237238, 262, 308309, 393 derivatives of, second, 353, 749, 751 derivatives of, third, 749, 751752 differential of, 403 endpoints of, 452 equation for, see Cubic equations extrema of, 307309, 312, 315320, 353, 359, 366
Fundamental Theorem of Calculus for, 453456 graphs of, see Graph of a cubic function inflection points of, 93, 354, 355356, 366367, 747 instantaneous rate of change in, 190 integration of, 438439, 442443, 453456 intercepts of, 365366 piecewise, 158159 as probability density function, 787 Riemann sums for, 420422 secant line of, 182 second derivative of, 353, 749, 751 sine functions and, 749, 753 Taylor polynomials and, 749, 751752, 758760 Cubic regression, 98, 104106, 784785, 787 Cumulative distribution function, 795, 803804 Curve sketching, 365370 Cylinders, 299, 384 Definite integrals for area, 438445, 452460, 465, 618 by calculator, 445446, 460, 620 for continuous functions, 438445, 452460 definition of, 438, 452, 464 for exponential functions, 502 in Fundamental Theorem of Calculus, 452458 graphical meaning of, 439440, 618 improper integrals from, 499506 vs. indefinite integral, 438 properties of, 441, 465 for rate of change functions, 438445, 452460, 465 Riemann sums and, 438439 Degree of constant functions, 93 of cubic functions, 93, 747 definition of, 62, 166, 746 of linear functions, 63, 93, 747
of quadratic functions, 63, 93, 747 of quartic functions, 100, 747 of quintic functions, 747 Delta D , 419 Demand function for CDs, 204, 327 consumer surplus from, 492, 497 equation for, 327 price in, 490 profit function and, 332333 revenue function and, 328, 330331, 347 supply function and, 493495 for VHS tapes, 204 Density function, 783785, see also Probability density functions Dependent variable on Cartesian coordinate system, 5 definition of, 3 in differential equations, 637, 691 in multivariable functions, 516 range of, see Range from tabular data, 34 at x-intercept, 22 Depreciation, 127 Derivative at a point by calculator, 317318 definition of, 197 from Excel, 322 extrema from, 308320 formula for, 197198, 251 graphical meaning of, 221, 241 interpretation of, 200201 numerical estimate of, 226227 secant lines and, 241 slope and, 221, 241 tangent lines and, 221 Derivatives of absolute value function, 241, 310311 of area function, 377378 by calculator, 270, 317318 Chain Rule for, 277283, 287, 290, 301 of circles, 294295 of composition of functions, 277278 of constant functions, 258, 313 Constant Multiple Rule for, 260261, 301
I4
Index
Derivatives (continued ) Constant Rule for, 258, 301 of cosine functions, 609610, 618, 623, 631 of cost function, 328 at critical points/values, 312 319 of cubic functions, first, 235, 237238, 262, 308309, 393 of cubic functions, second, 353, 749, 751 of cubic functions, third, 749, 751752 definition of, 235, 251 derivative of, see Second derivative in differential equations, 637 of ex, 286288, 290 of endpoints, 226, 228 from Excel, 272, 322 of exponential functions, 200201, 238239, 285288, 290, 319 Exponential Rule for, 285287, 301 at extrema, 308320, 353, 387 fifth, 751, 752 First Derivative Test for, 313317, 360, 387 fourth, 750, 752 Generalized Exponential Rule for, 287, 301 Generalized Logarithmic Rule for, 289290, 301 Generalized Power Rule for, 278279, 301 graphical meaning of, 221, 607609 of horizontal lines, 258 implicit differentiation for, 293299, 301 interpretation of, 242245 Leibniz notation for, 257 of linear functions, 197198, 396, 748, 749 of logarithmic functions, 288291 Logarithmic Rule for, 288290, 301 mixed-second-order partial, 540, 577 notation for, 257, 300301, 395, 402403 numerical estimate of, 226227
partial, 528542, 562573, 577 at points, see Derivative at a point of position functions, 199200, 363365 of power functions, 233234, 258261, 266, 354 Power Rule for, 258261, 266, 301 Product Rule for, 266270, 301, 692 of profit function, 327 of quadratic functions, 236237, 243245, 261, 263, 607, 748 of quartic functions, 310, 355, 750 of quintic functions, 750, 751, 752 Quotient Rule for, 270271, 301 of revenue function, 328, 330, 362363 secant lines and, 241 Second Derivative Test, see Second Derivative Test second-order partial, 540542, 577 sign of, 313316 of sine functions, 606608, 611, 618, 631, 748755, 770 at singular values, 312 slope and, 198, 221, 241, 749 at stationary values, 312 Sum and Difference Rule for, 261262, 301 from tabular data, 226228 of tangent functions, 613, 632 tangent lines and, 221, 235, 308, 749 Taylor approximations for, 770 third, see Third derivative units of, 242 of volume function, 379381 Difference quotient for average change, 177182, 188189, 196 by calculator, 191 from Excel, 201202 formula for, 177, 190, 217, 251 graphs of, 180182 for instantaneous change, 189190, 196 limits of, 196 of profit function, 327 secant lines and, 180182, 217, 221 slope and, 177, 180
Differential equations definition of, 637, 691, 707, 708 dependent variable in, 637, 691 Eulers Method for, 651657 first-order, 691704, 707, 708 integrating factor method for, 692704, 708 in limited growth, 673681 linear, 691704, 708 for logistic functions, 683685 nonlinear, 692 notation for, 691 Reverse Product Rule for, 692703 separation of variables process for, 662670, 708 slope field for, 637644, 649 solution of, 637, 663664, 707 Differentials, 402403, 464 Differentiate, 257, 301 Differentiation definition of, 257, 301 implicit, 293299, 301 Directly proportional quantities, 3436 Direct Proportionality Test, 57 Discrete random variables definition of, 780, 835 expected value of, 804806 standard deviation of, 808, 836 variance of, 808809, 836 Distance function, see also Position functions for billiards, 532533 initial value of, 399 for minivan, 389, 399 for NYC subway trains, 838 for pistons, 623 projectile motion and, 617 from velocity function, 399, 623 velocity function from, 363365 Distribution cumulative, 795, 803804 exponential, 509, 798801, 813816 normal, 509, 820831 uniform, 509, 796797, 810813 Divergence of sequences, 719, 723724, 775 Divergence of series, 739740, 767 Divergence of Taylor series, 767, 775
Index
I5
Diverging improper integrals, 500, 503506, 510 Division Property of limits, 205, 250 Domain for constrained objective functions, 571 continuity and, 307 definition of, 10, 57 determination of, 336 of exponential functions, 116 of linear functions, 18 of logarithmic functions, 136, 149, 398 of multivariable functions, 516 of probability density functions, 786, 788 of random variables, 780 range and, 1011 restrictions on, 1112, 343344 of revenue function, 12, 331 D-Test, 548557, 563, 578 e in calculator, 137 as series, 746 Taylor approximations for, 764765, 771 value of, 137
ex
antiderivative of, 771 derivative of, 286288, 290 as integrating factor, 693694 integration of, 397, 406, 408409, 770771 properties of, 662 Endpoints of cubic functions, 452 estimating derivatives at, 226, 228 extrema at, 307, 308, 315 Equilibrium line/point, 491494 Error in difference of two functions, 483 line of best fit and, 3637 round-off, 35 Eulers Method, 651657, 707, 708 Even functions, 497 Event, 785 Excel Technology Tips for aligning a data set, 129 for cubic regression, 104106 for derivatives, 272, 322 for difference quotient, 201202
on dollar symbol in cell formula, 371 for Eulers Method, 656657 for exponential regression, 130131 formulas, pasting of, 12 for graphing a function, 1213 for linear regression, 51 for logarithmic regression, 145146 for multivariable functions, 384 for normal probability density functions, 829831 for probability density functions, 829831 for quadratic regression, 7980 for quartic regression, 104106 for Riemann sums, 428430 for scatter plots, 13, 4950 for sequence graphs, 728729 for table of values, 370371 for x-intercept, 322323 Expected value alternative formulas for, 819 by calculator, 807 definition of, 835 of normal distribution, 820 of probability density functions, 810816, 821, 836 of random variables, 804807, 813, 835 Exponential distribution, 509, 798801, 813816 Exponential equations aligning the data for, 123, 129 basic, 116, 151, 167 by calculator, 126 coefficients of, 116118, 285 from Excel, 130131 from points, 121122 from regression, 122124, 126, 130131 from tabular data, 119124, 724 from word problems, 127129, 156157 Exponential functions base of, 116118 common ratio in, 735 critical points/values of, 319 definition of, 116 derivatives of, 200201, 238239, 285288, 290, 319
differential of, 403 domain of, 116 equation for, see Exponential equations extrema of, 319 graphs of, see Graph of an exponential function growth factor of, 117, 127, 166, 167 growth rate of, 167, 208 initial value of, 116, 118, 166 instantaneous rate of change in, 238239 integration of, see Integration of exponential functions limits of, 200201 logarithmic functions and, 136, 140, 167 probability density, see Exponential probability density function sequences and, 724 Exponential probability density function, 509, 798801, 803, 813816, 835 Exponential proportionality, 293 Exponential regression, 122124, 126, 130131 Exponential Rule, see also Generalized Exponential Rule for derivatives, 285287, 301 for integrals, 397, 465 Exponents, properties of, 119, 167, 532, 662 Extrapolation, 68 Extrema absolute, see Absolute extrema of absolute value function, 310311 of area function, 340, 343 of constant functions, 313 of constrained objective functions, 563573 of continuous functions, 307310, 312320, 353, 359362, 387 of cost function, 346 critical points/values and, 312317, 547557 of cubic functions, 307309, 312, 315320, 353, 359, 366 in curve sketching, 367, 368
I6
Index
Extreme (continued ) definition of, 307, 308, 386 derivatives at, 308320, 353, 387 of discontinuous functions, 310311 at endpoints, 307, 308, 315 existence of, 387 of exponential functions, 319 First Derivative Test for, 313317, 387 graphs of, 308, 546 inflection points and, 357 interpretation of, 318320 of Lagrange functions, 562573 of logarithmic functions, 368369 of multivariable functions, 545557, 562573, 577 in parabolas, 329, 333 of piecewise functions, 310311 point of diminishing returns at, 363 of power functions, 312313 of profit function, 329, 333 of quadratic functions, 329, 333 of quartic functions, 310, 314315, 356357 relative, see Relative extrema of revenue function, 331 at saddle points, 546547, 577 from second derivative, 353, 356357, 359362 Second Derivative Test for, 359362, 387, 548557, 578 on surfaces, 545547 of volume function, 336337 Fibonacci sequence, 733 Finite geometric series, 734739, 741, 775 First Derivative Test, 313317, 360, 387 First-order differential equations, 691704, 707, 708 Fractions from repeating decimals, 742743 Frequency definition of, 584 of electricity, 602 notation for, 586 Function of two variables, 516, 577 Functions absolute value, see Absolute value function
amplitude of, 584585, 588 arccosine, 592 arcsine, 593 area, 340, 342343, 377378 composition of, 275278 constant, see Constant functions constrained objective, 563573, 578 constraint, 562563 continuous, see Continuous functions cost, see Cost function cubic, see Cubic functions cumulative distribution, 795, 803804 definition of, 2, 57 density, 783785 differentiable, 257, 301 discontinuous, 9, 212, 307 even, 497 exponential, see Exponential functions from graphs, 69 graphs of, on calculator, 10 graphs of, from Excel, 1213 hyperbolic, 774 increasing, 433 inputs to, 2 integrable, 394, 438 inverse, 140 Lagrange, 562, 578 limited growth, 673681 linear, see Linear functions logarithmic, see Logarithmic functions logistic, see Logistic functions magnitude of, 2122 midline of, 584585, 631 model selection strategies, 150160, 167 multivariable, see Multivariable functions normal probability density, 821831 notation for, 34 objective, 563573, 578 outputs of, 2 periodic, see Periodic functions piecewise, see Piecewise functions polynomials, see Polynomials position, see Position functions power, see Power functions
probability density, see Probability density functions profit, see Profit function quadratic, see Quadratic functions quartic, see Quartic functions quintic, see Quintic functions rate of change, see Rate of change functions revenue, see Revenue function sinusoidal, see Sinusoidal functions from tabular data, 23, 159160 tangent-line approximation for, 223226 uniform probability density, see Uniform probability density functions velocity, see Velocity function Vertical Line Test for, 79, 20, 57 volume, see Volume function Fundamental Theorem of Calculus, 451458, 463, 465, 485 Games, lottery, 804808, 817818 Generalized Exponential Rule, 287, 301 Generalized Logarithmic Rule, 289, 301 Generalized Power Rule, 278279, 301 General solution, 663664 Geometric sequence, 725, 734, 775 Graph of a cosine function horizontal shifts of, 584, 587588 rate of change on, 607609 slope of, 607 tangent lines to, 609 vertical shifts of, 584, 586588 Graph of a cubic function area between x-axis and, 420422, 439, 443, 454455 bends in, 95, 114 in Cartesian coordinate system, 747 concavity on, 9395, 151, 353356, 366367, 747 continuity in, 212 by curve sketching, 365368 from equation, 393 extrema in, 308 features of, 151
Index
I7
function values from, 6 from histogram, 783785 horizontal asymptote of, 151 inflection point of, 93 intercepts on, 365366 from tabular data, 56 Graph of an exponential function concavity of, 118, 151 continuity in, 212 directionality of, 798, 801 drawing of, 117119 horizontal asymptote of, 117, 151, 166 intercepts on, 116, 117 Graph of a linear function area between x-axis and, 444 in Cartesian coordinate system, 747 continuity in, 212 definition of, 18, 57 of horizontal lines, 21 intercepts of, 23 vs. nonlinear, 151 piecewise, 9 of secant line, 181182, 218221, 226 slope on, 2022, 181 from tabular data, 2829 of vertical lines, 20 Graph of a logarithmic function concavity in, 137138, 151, 369370 continuity in, 212 curve sketching for, 368369 drawing of, 137140 inflection points on, 369370 intercepts on, 137, 368 Taylor polynomial of, 757, 766 vertical asymptote on, 137, 138, 151 Graph of a logistic function concavity in, 151 continuity in, 212 growth pattern, 681683 horizontal asymptote of, 150, 151 vertical shifts of, 685 Graph of a power function continuity in, 212 inflection points of, 354 Riemann sums on, 417418 secant line to, 226 tangent lines to, 221222, 226
Graph of a quartic function bends in, 100, 114 concavity of, 100, 151, 354355, 747 continuity in, 212 inflection points on, 100, 354355, 747 tangent lines to, 310 Graph of a sine function in Cartesian coordinate system, 594, 595 concavity of, 748 horizontal shifts of, 584, 587588 intercepts on, 748 midline of, 584585 rate of change on, 607609 tangent lines to, 608, 748 vertical shifts of, 584, 586588 Graph of a tangent line to a circle, 294 to a cosine function, 609 to parabolas, 222223 to a power function, 221222, 226 to a quartic function, 310 to a sine function, 608, 748 to surfaces, 537538, 540 Graphs of absolute value function, 310, 767 of area between function and x-axis, 411, 439440, 443444, 454455, 457 of area between two curves, 482 bar, 780781 in Cartesian coordinate system, 5, 594, 595, 747 concavity of, see Concavity of constant functions, 411 constraints and, 571 continuity in, 9, 211, 307 of continuous functions, 9, 211, 307 of cosine functions, see Graph of a cosine function of cubic functions, see Graph of a cubic function of difference quotient, 180182 discontinuities in, 9, 307 of exponential functions, see Graph of an exponential function extrema in, 308, 546
of functions, on calculator, 10 of functions, in Excel, 1213 functions from, 69 histograms, 781785 horizontal shifts of, see Horizontal shifts on graphs of implicit differentiation, 297 of independent variable, 5, 607 inflection points on, see Inflection points of instantaneous rate of change, 607 intercepts on, see Intercepts of limited growth function, 676 limits from, 193194 linear equations from, 2526 of linear functions, see Graph of a linear function of logarithmic functions, see Graph of a logarithmic function of logistic functions, see Graph of a logistic function of maintenance level, 720, 721 for multivariable functions, see Surfaces of normal distribution, 821 of normal probability density function, 821828 origin in, see Origin of piecewise functions, 9, 307 of polynomials, 746747, 766 of probability density functions, 788, 821828 of quadratic functions, see Parabolas of quartic functions, see Graph of a quartic function of quintic functions, 747, 751, 753 range on, see Range of rate of change function, 411414, 417423, 437444 of regression, on calculator, 3940 of Riemann sums, 412414, 417421 scatter plots, see Scatter plots of secant lines, 181182, 218221, 226 of sequences, 720, 721, 727729 sigmoidal, 681 of sine functions, see Graph of a sine function of slope fields, 638
I8
Index
Graphs (continued ) slope on, see Slope supply and demand, 481, 491494 of tabular data, 57 of tangent functions, 597 variables in, 5 of velocity function, 423424 Vertical Line Test for, 79, 20, 57 vertical shifts on, see Vertical shifts on graphs of volume function, 517519 Growth factor, 117, 127, 166, 167 Growth rate, 167, 208 Half-life, 718719 Harmonic sequences, 733 Hemisphere, surface area of, 384 Histograms, 781785 Horizontal asymptote, 150, 151, 166 Horizontal lines area beneath, 411 asymptote, 117, 150, 151, 166 derivatives of, 258 exponential functions and, 116 functional notation for, 258 linear equation for, 23, 27, 58 midline, 584585, 631 slope of, 21, 308 Horizontal shifts on graphs general formula for, 124 of Taylor polynomials, 757758 Humidity, relative, 515, 526 Hyperbolic functions, 774 Implicit differentiation, 293299, 301 Improper integrals, 499506, 509511 Increasing functions, 433 Indefinite integrals vs. definite integral, 438 definition of, 394, 464 for exponential functions, 397398, 406, 408409, 472474, 477 notation for, 394395 Independent variable on Cartesian coordinate system, 5 definition of, 3 domain of, see Domain graphs of, 5, 607
in multivariable functions, 516, 577 in partial derivatives, 577 slope and, 21 from tabular data, 34 at y-intercept, 22 Indeterminate form of limits, 206 Index of summation, 434 Infinite geometric series, 739742, 775 Inflection points concavity and, 93, 166, 354 of continuous functions, 354 of cubic functions, 93, 354, 355356, 366367, 747 in curve sketching, 368 definition of, 166, 354, 386 extrema and, 357 location of, 387 of power functions, 354 of quartic functions, 100, 354355, 747 of quintic functions, 747 Initial position, 199 Initial value of distance function, 399 of exponential functions, 116, 118, 166 of investments, 134 Initial velocity, 199 Inputs, see also Independent variable in average rate of change, 178 extrapolation from, 68 to functions, 2 horizontal graphic shifts and, 124 in multivariable functions, 516, 577 notation for, 3 to sequence, 719 slope and, 1819 in tables, 2 Instantaneous rate of change average rate of change and, 188189, 196 by calculator, 191, 192 in cubic functions, 190 definition of, 250 difference quotient for, 189190, 196 in exponential functions, 238 239
graphs of, 607 in linear functions, 198 at multiple points, 236239 at point, 114, 221 in power functions, 196197, 233234 in quadratic functions, 192, 236237, 263 secant lines and, 221 slope and, 114, 198 Integral Rule, 398, 465 Integrals, see also Antiderivatives Constant Multiple Rule for, 396, 408, 465 Constant Rule for, 396, 465 definite, see Definite integrals differentials and, 402403 Exponential Rule for, 397, 465 improper, 499506, 509511 indefinite, see Indefinite integrals Integral Rule for 1> x, 398, 465 Power Rule for, 395, 465 Product Rule for, 472 Sum and Difference Rule for, 397, 465 Taylor approximations for, 770771 Integrand, 394, 438, 464 Integrating factor method, 692704, 708 Integration for accumulated change, 456458, 485489 by calculator, 445446, 460, 620 of constant functions, 396, 400 of continuous functions, 438445, 452460, 465 of cosine functions, 619, 630, 632 of cubic functions, 438439, 442443, 453456 definition of, 393, 394, 464 of e, 397, 406, 408409, 770771 of exponential functions, see Integration of exponential functions limits of, 438, 458460 of linear functions, 396397, 441442, 453 of logarithmic functions, 404, 459460, 474476, 478 of nonnegative functions, 438, 465 notation for, 394395, 464
Index
I9
by parts, 472479, 510511 of position functions, 399400 of power functions, 395396, 398399, 442, 500501 of quadratic functions, 439441 of rate of change function, 438445, 452460, 465 of sine functions, 623, 632 by substitution, 402, 404409, 458460, 464 of tangent functions, 626, 630, 632 of velocity function, 399 Integration of exponential functions with definite integrals, 502 with improper integrals, 499500, 502503 with indefinite integrals, 397398, 406, 408409, 472474, 477 Laplace transform and, 509 by parts, 477 Taylor approximation, 770771 Intercepts of cubic functions, 365366 in curve sketching, 367 of exponential functions, 116, 117 of linear functions, 2226, 2829, 49, 748 of logarithmic functions, 137, 368 of quadratic functions, 6466 of sine functions, 748 slope and, 23 from tangent function, 614 of Taylor polynomials, 748 x-axis, see x-intercept y-axis, see y-intercept Intervals in calculator, 10, 16 continuous functions on open, 212 of convergence for Taylor series, 765770, 775 enclosed area between, 455 of exponential probability density functions, 800 for frequency, 584 of normal probability density function, 821, 822 notation for, 177 of probability density functions, 786, 789, 797, 800, 821, 822 of random variables, 780
of uniform probability density functions, 797 Inverse functions, 140 l (Lambda), 562 Lagrange function, 562, 578 Lagrange multiplier, 562 Lagrange Multiplier Method, 562573, 578 Lambda l, 562 Laplace transform, 509 Left-hand limit, 193195 Left-hand sum for area estimation, 412420, 422423, 427 definite integral and, 438439 notation for, 436437 procedure for, 464 Leibniz, Gottfried, 257 Leibniz notation, 257, 280, 301 Limited growth, 673, 708 Limited growth function, 673681 Limit operations, 205, 250 Limits of continuous functions, 193194 definition of, 195 of difference quotient, 196 exist, 194, 250 from graphs, 193194 of improper integrals, 499506 indeterminate form of, 206 for infinite series, 767 at infinity, 206211, 251, 499506 of integration, 438, 458460 left-hand, 193195 of logistic functions, 209, 211 maintenance level, 719723, 774 notation for, 193 properties of, 205, 250 right-hand, 193195 of slope, 221 of summation, 434 summation notation and, 438 Linear equations algebraic construction of, 4448 for break-even point, 46 by calculator, 3738 differential, 691704, 708 for direct proportions, 3436 from Excel, 51
from graphs, 2526 for horizontal lines, 23, 27, 58 from points, 26, 613614 point-slope form of, 27, 58 from regression, 3739, 48 slope-intercept form of, 24, 58, 151 standard form of, 27, 2829, 56, 58 from tabular data, 2425, 3439 for tangent lines, 221223, 614 for vertical lines, 23, 27, 58 from word problems, 24, 27, 4448, 156 Linear functions antiderivatives of, 397, 453, 607 average rate of change in, 30, 177 coefficient of determination for, 39 constant difference of, 747 correlation coefficient of, 39 degree of, 63, 93, 747 derivatives of, 197198, 396, 748, 749 difference quotient and, 177 domain of, 18 equations for, see Linear equations Fundamental Theorem of Calculus for, 453 graphs of, see Graph of a linear function inflection points of, 747 instantaneous rate of change in, 198 integration of, 396397, 441442, 453 intercepts of, 2226, 2829, 49, 748 piecewise, 9, 4648, 57 range of, 18 rate of change in, 18, 30 slope of, see Slope of a line from tabular data, 18, 28, 4243 in three dimensions, 56 Vertical Line Test for, 20 from word problems, 4, 49 Linear regression, 3740, 48, 51 Lines average rate of change in, 30, 177 of best fit, 3637, 39, 4243 equilibrium, 491494 horizontal, see Horizontal lines
I10
Index
Lines (continued ) intercepts of, see Intercepts parallel, 32 perpendicular, 32 from points, 26, 613614 point-slope form of, 27, 58 secant, see Secant line slope-intercept form of, 2426, 58 slope of, see Slope of a line standard form of, 27, 2829, 56, 58 tangent, see Tangent lines vertical, see Vertical lines Vertical Line Test, 79, 20, 57 Logarithmic equations basic, 136, 151, 167 by calculator, 144 from Excel, 145146 from regression, 144146 from tabular data, 144 Logarithmic functions base of, 136138, 141 definition of, 136 derivatives of, 288291 differential of, 403 domain of, 136, 149, 398 equation for, see Logarithmic equations exponential functions and, 136, 140, 167 extrema of, 368369 graphs of, see Graph of a logarithmic function integration of, 404, 459460, 474476, 478 Taylor approximations for, 755758 Logarithmic regression, 144146 Logarithmic Rule, 288290, 301, see also Generalized Logarithmic Rule Logarithms calculation of, 136, 142143 common, see Common log natural, see Natural log rules for, 140143, 167, 662 Logistic equations basic, 150, 151, 167 by calculator, 685 differential equations and, 683685
in piecewise functions, 158159 Logistic functions equation of, see Logistic equations graphs of, see Graph of a logistic function limits of, 209, 211 from word problems, 156 Logistic growth, 681, 708 Logistic regression, 685 Magnitude, 2122 Maintenance level, 719723, 774 Matrix, augmented, 71 Maximum, absolute of continuous functions, 315320 of cubic functions, 317, 319320, 366 definition of, 308, 386 of exponential functions, 319 of logarithmic functions, 369 in parabolas, 329, 333 point of diminishing returns at, 363 of profit function, 329, 333 of quadratic functions, 329, 333 of quartic functions, 315 of revenue function, 331 from second derivative, 362 of volume function, 336337 Maximum, relative of absolute value function, 311 of constant functions, 313 of continuous functions, 307310, 312320, 353 of cubic functions, 309, 316320, 353, 366 definition of, 386 of discontinuous functions, 310311 First Derivative Test for, 313317, 387 of multivariable functions, 545557, 577 in parabolas, 329 point of diminishing returns at, 363 of profit function, 329, 333 of revenue function, 331 at saddle points, 546547, 577 Second Derivative Test for, 359362, 387, 548557, 578 on surfaces, 545547
of volume function, 336337 Mean Value Theorem, 205 Midline, 584585, 631 Midpoints, 783 Minimum, absolute of area function, 340 of continuous functions, 315320 of cubic functions, 317, 319320, 366 definition of, 308, 386 point of diminishing returns at, 363 of quartic functions, 315 of revenue function, 331 from second derivative, 362 Minimum, relative of area function, 343 of constant functions, 313 of continuous functions, 307310, 312320, 353 of cost function, 346 of cubic functions, 309, 315320, 353, 366 definition of, 386 of discontinuous functions, 310311 First Derivative Test for, 313317, 387 of multivariable functions, 545557, 577 in parabolas, 329 of quartic functions, 310, 315 of revenue function, 331 at saddle points, 546547, 577 Second Derivative Test for, 359362, 387, 548557, 578 on surfaces, 545547 Mixed second partial derivatives, 540, 577 Monotonic sequences, 732, 733 Multiplication Property of limits, 205, 250 Multivariable functions critical points/values for, 547, 547557, 562573, 577, 578 definition of, 516, 577 dependent variable in, 516 derivatives of, see Partial derivatives domain of, 516 D-Test for, 548557 from Excel, 384
Index
I11
extrema of, 545557, 562573, 577 graphs for, see Surfaces independent variable in, 516, 577 inputs in, 516, 577 Lagrange multiplier method for, 562573 outputs in, 516, 577 range of, 516 saddle points of, 546557 tables for, 516517, 519521 tangent line to, 537540 Mu m, 820 n, in Riemann sums, 417 Natural log change-of-base formula, 141 vs. common log, 139140, 289 curve sketching of, 368370 definition of, 137, 166 derivatives of, 289 differential of, 403 domain of, 398 of e, 287, 662 graphs of, 139140, 757, 766 infinite Taylor series for, 764, 765767 integration of, 404, 459460, 474476, 478 rules for, 141 Taylor approximations for, 755758, 764 Nonlinear differential equations, 692 Normal distribution, 509, 820831 Normal probability density function, 509, 821831, 835 Number lines, 314315 Objective function, 563573, 578 Ordered pair, 5 Origin 2-dimensional, 5 3-dimensional, 517 Outputs, see also Dependent variable in average rate of change, 178 extrapolation of, 68 of functions, 2 in multivariable functions, 516, 577 notation for, 3 of sequences, 719 slope and, 1819
in tables, 2 vertical shifts and, 124 Parabolas area between x-axis and, 419420, 423424, 436437, 440441, 451 axis of symmetry in, 65 bends in, 114 concavity of, 6365, 151, 747, 748 continuity in, 212 definition of, 166 drawing of, 64, 747 equation from, 66 extrema of, 329, 333 features of, 151 intercepts of, 6466 secant line of, 181 side steepness in, 6364 slope of, 607 tangent line to, 222223 vertex of, 6466, 167, 333 Parallel lines, 32 Partial derivatives, 528542, 562573, 577 Particular solution, 663664 Pascals Triangle, 236 Periodic functions cosine, see Cosine functions sine, see Sine functions tangent, see Tangent functions Periods, 584, 586, 631 Perpendicular lines, 32 Piecewise functions absolute value, see Absolute value function extrema of, 310311 graphs of, 9, 307 linear, 9, 4648, 57 from tabular data, 157159 from word problems, 4748 Points break-even, 46, 327 critical, see Critical points/values derivatives at, see Derivative at a point of diminishing returns, 362363 end, see Endpoints equilibrium, 491494 exponential equations from, 121122
inflection, see Inflection points instantaneous rate of change at, 114, 221 linear equations from, 26, 613614 midpoints, 783 saddle, 546557, 577, 578 slope at, 114, 221 in three-dimensional coordinate system, 518 Point-slope form of a line, 27, 58 Polynomials binomials, 268 coefficients of, 746 constant, see Constant functions cubic, see Cubic functions curve sketching of, 365368 definition of, 62, 746 degree of, see Degree even-powered, 750 graphs of, 746747, 766 growth rate of, 208 linear, see Linear functions multiplying, 268 odd-powered, 750 power, see Power functions quadratic functions, see Quadratic functions quartic, see Quartic functions quintic, see Quintic functions sine functions and, 748755, 769 sixth-degree, 754755, 766 Taylor, see Taylor polynomials terms of, 746, 754, 758 trinomials, 268 Position, 199, 387 Position functions, see also Distance function antiderivatives of, 399 derivatives of, 199200, 363365 integration of, 399400 from velocity function, 399 velocity function from, 363365 Power functions antiderivatives of, 394, 395 critical points/values for, 312313 definition of, 114 derivatives of, 233234, 258261, 266, 354 differential of, 403 equation for, 114, 151, 258 extrema of, 312313
I12
Index
Power functions (continued ) graphs of, see Graph of a power function instantaneous rate of change in, 196197, 233234 integration of, 395396, 398399, 442, 500501 limits of, 206207 Riemann sums for, 417418 second derivative of, 354 Power Property of limits, 205, 250 Power Rule, see also Generalized Power Rule for derivatives, 258261, 266, 301 for integrals, 395, 465 Probability density functions by calculator, 787791, 801, 821829 characteristics of, 821 creation of, 789791, 835 cumulative distribution functions and, 795 definition of, 509, 786, 835 domain of, 786, 788 from Excel, 829831 exponential, 509, 798801, 803, 813816, 835 graphs of, 788 as improper integral, 509 intervals of, 786, 789, 797, 800, 821, 822 normal, 509, 821831, 835 range of, 787 regression model for, 787791 from tabular data, 786787 uniform, 509, 797, 810813, 835, 836 Probability equation, 785 Producer surplus, 490493, 495 Product Rule for derivatives, 266270, 301, 692 related to integration by parts, 472 Profit function cost function and, 45, 329 demand function and, 332333 derivatives of, 327 difference quotient of, 327 extrema of, 329, 333 linear model for, 45 revenue function and, 45, 329 Pythagorean Theorem, 532
Quadratic equations aligning the data for, 67 basic, 63, 151, 167 by calculator, 68 coefficients of, 6364 concavity from, 65 discriminant of, 80 from Excel, 7980 in factored form, 92 from parabolas, 66 from regression, 68, 7980 from tabular data, 6978 vertex from, 65 y-intercept from, 65 Quadratic Formula, 77, 167 Quadratic functions antiderivatives of, 452 average rate of change in, 92 constant difference of, 747 definition of, 63 degree of, 63, 93, 747 derivatives of, 236237, 243245, 261, 263, 607, 748 equation for, see Quadratic equations extrema of, 329, 333 graphs of, see Parabolas instantaneous rate of change in, 192, 236237, 263 integration of, 439441 limits of, 199200 Riemann sums for, 419420, 423424, 426427, 436437, 451 second derivative of, 748 sine functions and, 748749 summation notation for, 436 437 tangent-line approximation for, 223226, 243244 Taylor polynomial, 748749 from word problems, 156 Quadratic regression, 68, 7980 Quartic equations basic, 100, 151, 167 by calculator, 98 coefficients of, 100 from Excel, 104106 from regression, 98, 104106, 790791 from tabular data, 101104
Quartic functions constant difference of, 747 cubic functions and, 151, 747 definition of, 100 degree of, 100, 747 derivatives of, 310, 355, 750 equation for, see Quartic equations extrema of, 310, 314315, 356357 fourth derivative of, 750 graphs of, see Graph of a quartic function as probability density function, 790791 second derivative of, 355, 750 sine functions and, 750 Taylor polynomial, 750 Quartic regression, 98, 104106, 790791 Quintic functions concavity of, 747 constant difference of, 747 degree of, 747 derivatives of, 750752 graphs of, 747, 751, 753 horizontal shifts of, 757758 inflection points on, 747 second derivative of, 750, 752 sine functions and, 750751, 753, 755 Taylor polynomial and, 750751, 752, 755, 757758 Quotient Rule, 270271, 301 Random variables continuous, see Continuous random variables definition of, 780 discrete, see Discrete random variables Range definition of, 10, 57 domain and, 1011 of linear functions, 18 midpoint of, 786 of multivariable functions, 516 of probability density functions, 787 Rate of change average, see Average rate of change
Index
I13
constant, see Slope differential equations for, see Differential equations in flow, 695697 instantaneous, see Instantaneous rate of change in linear functions, 18, 30 maximum/minimum, 356357 of second derivative, 749 second derivative for, 356357 slope field and, 637644, 649 of surfaces, 536539 total, over interval, 411416 units of, 178 Rate of change functions accumulated change in, 456458, 485489 antiderivatives of, 412 definite integral for, 438445, 452460, 465 Fundamental Theorem of Calculus for, 451458, 463, 465, 485 graphs of, 411414, 417423, 437444 Riemann sums for, see Riemann sums summation notation for, 434437 table of data for, 414417 Rational Exponent Property of limits, 205, 250 Rational expressions, rewriting of, 695 Ratio test, 767, 775 Rectangles, 345, 384, 417, 419 Recursive formulas, 724, 726, 775 Relative extrema of absolute value function, 310311, 311 of area function, 340, 343 of constant functions, 313 of constrained objective functions, 563 of continuous functions, 307310, 312320, 353 of cost function, 346 critical points/values and, 312317, 547557 of cubic functions, 309, 312, 315320, 316320, 353, 359, 366
in curve sketching, 367 definition of, 386 derivatives at, 308309, 353, 387 of discontinuous functions, 310311 at endpoints, 307, 315 existence of, 387 of exponential functions, 319 First Derivative Test for, 313317, 387 graphs of, 308, 546 inflection points and, 357 of logarithmic functions, 368369 maximum, see Maximum, relative minimum, see Minimum, relative of multivariable functions, 545557, 577 in parabolas, 329 of piecewise functions, 310311 point of diminishing returns at, 363 of power functions, 312313 of profit function, 329, 333 of quartic functions, 310, 314315, 356357 of revenue function, 331 at saddle points, 546547, 577 from second derivative, 353, 356357, 359362 Second Derivative Test for, 359362, 387, 548557, 578 on surfaces, 545547 of volume function, 336337 Repeating decimals, 742743 Revenue function cost function and, 332 critical points/values for, 331 demand function and, 328, 330331, 347 derivatives of, 328, 330, 362363 domain of, 12, 331 extrema of, 331 linear model for, 45 marginal revenue and, 328, 456 point of diminishing returns in, 362363 price and, 328, 330 profit function and, 45, 329 Reverse Product Rule, 692703 Riemann sums for area estimation, 412430, 451
average of, 422 by calculator, 424426 for cubic functions, 420422 definite integral and, 438439 definition of, 464 from Excel, 428430 graphs of, 412414, 417421 for power functions, 417418 for quadratic functions, 419420, 423424, 426427, 436437, 451 summation notation for, 436437 velocity function and, 423424 Right-hand limit, 193195 Right-hand sum for area estimation, 420422, 424, 427 definite integral and, 438439 procedure for, 465 summation notation for, 436 Round-off error, 35 Saddle points, 546557, 577, 578 Sample space, 785 Scatter plots on calculator, 41 definition of, 40 in Excel, 13, 4950 line of best fit on, 4243 Secant functions, 613 Secant line average rate of change and, 217 of cubic functions, 182 definition of, 250 derivatives and, 241 difference quotient and, 180182, 217, 221 graphs of, 181182, 218221, 226 instantaneous rate of change and, 221 of parabolas, 181 slope of, 180182, 217221, 227, 241 Second derivative concavity and, 354, 748 of cubic functions, 353, 749, 751 definition of, 353, 386 derivative of, see Third derivative extrema from, 353, 356357, 359362 notation for, 353 of power functions, 354
I14
Index
Second derivative (continued ) of quadratic functions, 748 of quartic functions, 355, 750 of quintic functions, 750, 752 for rate of change, 356357 rate of change of, 749 Second Derivative Test, see Second Derivative Test sign of, 355 of sine functions, 748755 Second Derivative Test for constrained objective functions, 563 for single-variable functions, 359362, 387 for two-variable functions, 548557, 578 Second-order partial derivatives, 540542, 577 Separation of variables process, 662670, 708 Sequences applications of, 718726 arithmetic, 725726, 775 by calculator, 727 convergence of, 719723 definition of, 719, 774 divergence of, 719, 723724, 775 from Excel, 728729 explicit formulas for, 724725, 726, 775 exponential functions and, 724 factorial, 751 Fibonacci, 733 geometric, 725, 734, 775 graphs of, 720, 721, 727729 harmonic, 733 inputs to, 719 maintenance level, 719723, 774 monotonic, 732, 733 notation for, 719720, 725 outputs of, 719 recursive formulas for, 724, 726, 775 strictly monotone, 732 terms of, 719720, 775 Series, 733743, 746, 763771, 775 Seventh-degree polynomials, 752, 766, 769 Sigma s, see Standard deviation Sigma S, see Summation notation Sigmoidal graphs, 681
Sign charts, 314316 Sine functions amplitude of, 584585, 588 antiderivatives of, 623 applications of, 587596 by calculator, 592593, 595, 597598, 612 cosine functions and, 584, 588589 derivatives of, 606608, 611, 618, 631, 748755, 770 frequency of, 584, 586, 588 graphs of, see Graph of a sine function hyperbolic, 774 as infinite sum, 763 integration of, 623, 632 period of, 586 polynomials and, 748755, 769, 770 rate of change in, 604608 second derivative of, 748755 slope of, 607, 748 table of data from, 604605 from tabular data, 583, 585589, 605606 tangent functions and, 596 Taylor approximations for, 748755, 763764 Taylor series for, 768770 Singular critical points/values, 312313, 316, 386 Sinusoidal functions cosine, see Cosine functions sine, see Sine functions Sixth-degree polynomials, 754755, 766 Slope common errors with, 19 correlation coefficient and, 39 of cosine functions, 607 of cross-section graph, 535536 of curve, at point, 221 definition of, 19 derivatives and, 198, 221, 241, 749 difference quotient and, 177, 180 directionality of, 536539 formula for, 58 independent variable and, 21 inputs and, 1819 instantaneous rate of change and, 114, 198
intercepts and, 23 limits of, 221 of linear functions, see Slope of a line magnitude of, 2122 of parabolas, 607 at point, 114, 221 of sine functions, 607, 748 step size for, 651 Slope field, 637644, 649, 707 Slope-intercept form of a line, 2426, 58 Slope of a line applications of, 1920 definition of, 18, 57 derivatives and, 198, 749 difference quotient and, 177, 180 errors in calculating, 19 formula for, 58 from graphs, 2122, 23 horizontal line, 21, 308 line of best fit, 39 magnitude of, 2122 perpendicular line, 32 rate of change and, 18, 198 secant line, 180182, 217221, 227, 241 sign of, 2122 tangent line, 221223, 608, 609, 614, 748 from two points, 2021, 613 vertical line, 20 in word problems, 49 Solution of differential equations, 637, 663664, 707 Speed, 198, 389 Spheres, 384 Squares, surface area of, 349 Standard deviation by calculator, 809810 of continuous random variables, 809810, 836 definition of, 835 of discrete random variables, 808, 836 of exponential probability density functions, 815816 of normal distribution, 821 of uniform probability density functions, 812813, 836 Standard form of a line, 27, 2829, 56, 58
Index
I15
Stationary critical points/values, 312313, 316, 386 Step size, 651, 707 Strictly monotone sequences, 732 Subtraction Property of limits, 205, 250 Sum and Difference Rule for derivatives, 261262, 301 for integrals, 397, 465 Sum of a finite geometric series, 737739, 741 Sum of an infinite geometric series, 739743, 775 Summation notation, 434438, 735736, 763, 806 Sum of the signed areas, 438 Sum of squares of the error, 3637 Supply function, 490, 493495, 497 Surface area of boxes, 339, 384, 524 of cans, 341 of circles, 349 of cones, 388 of a cylinder, 384 of hemisphere, 384 minimizing of, 339340 of rectangles, 345 of spheres, 384 of square, 349 Surfaces constraints and, 571 cross section of, 534536, 571 definition of, 516, 577 drawing of, 517518 extrema on, 545547 rate of change of, 536539 rotation of, 519521 saddle points on, 546557 from tabular data, 519522 tangent lines to, 537540 Tables average rate of change from, 177180, 188189 cubic model from, 9899 dependent variable from, 34 derivatives from, 226228 directly proportional data in, 3435 exponential models from, 119124, 724
of functional values, 358 functions from, 23, 159160 graphs of, 57 independent variable from, 34 inputs in, 2 linear models from, 18, 2425, 28, 3439, 4243 logarithmic models from, 144 logistic models from, 151153 model selection strategies, 150155, 157160 multivariable models from, 516517, 519521 outputs in, 2 piecewise models from, 157159 probability density functions from, 786787 quadratic models from, 6978 quartic models from, 101104 for Riemann sums, 414423 sine functions from, 583, 585589, 605606 surfaces from, 519522 for tangent lines, 222 Tangent functions antiderivatives of, 626 cosine functions and, 596 derivatives of, 613, 632 graphs of, 597 hyperbolic, 774 integration of, 626, 630, 632 sine functions and, 596 vertical asymptote of, 597 y-intercept from, 614 Tangent lines approximations from, 223226, 242244 to circles, 294295 to cosine functions, 609 definition of, 221, 250 derivatives and, 221, 235, 308, 749 equations for, 221223, 614 graphs of, see Graph of a tangent line of multivariable functions, 537540 to power functions, 221222 to quadratic functions, 222223 to sine functions, 608, 748 slope of, 221223, 608, 609, 614, 748
to surfaces, 537540 tables for, 222 Taylor polynomials centered at (0 = x), 748755 centered at (1 = x), 755760 definition of, 775 derivative approximation by, 770 as infinite sum, 763765 integral approximation by, 770771 Taylor series, 763771, 775 Technology Tips for calculators, see Calculator Technology Tips for Excel, see Excel Technology Tips Terms of a polynomial basic, 746 Taylor, 754, 758, 775 Terms of the sequence, 719720, 775 Third derivative, 749755 Torricellis Law, 710 Trapezoids, 444 Triangles, 384, 389, 532533 Trigonometric equations, 592 Trinomials, 268 Uniform distribution, 509, 796797, 810813 Uniform probability density functions, 509, 797, 810813, 835, 836 Upper limit of integration, 438 Variables changing of, 405 correlation coefficient of, 39 dependent, see Dependent variable in equation solution, 11 independent, see Independent variable index of summation, 434 in partial derivatives, 528529 random, see Random variables Variance alternative formulas for, 819 by calculator, 809810 of continuous random variables, 809810, 813, 836
I16
Index
Variance (continued ) definition of, 835 of discrete random variables, 808809, 836 of exponential probability density functions, 814816 of normal distribution, 821 of uniform probability density functions, 810813, 836 Velocity acceleration and, 387 data in, 198 definition of, 363 of free-falling objects, 198200 initial, 199 position and, 387 sign of, 198 of wind, 527 Velocity function area between x-axis and, 423424, 467 from distance function, 363365 distance function from, 399, 623 for free-falling objects, 466 graphs of, 423424 integration of, 399 from position functions, 363365 position functions from, 399 Riemann sums and, 423424 Verbal descriptions, see Word problems Vertex of a parabola, 6466, 167, 333 Vertical asymptote of integrands, 505 of logarithmic functions, 137, 138, 151
of tangent functions, 597 Vertical lines asymptote, see Vertical asymptote linear equation for, 23, 27, 58 slope of, 20 Vertical Line Test, 79, 20, 57 Vertical shifts on graphs definition of, 586, 631 general formula for, 124 of logistic functions, 685 of periodic functions, 584, 586588 Volume of boxes, 336, 542, 564565, 567571 of cans, 341 of cones, 384, 388 of cylinders, 299 of liquid in cans, 379 of pools, 381 of spheres, 384 Volume function, 336337, 379381, 517519 Von Bertalanffy Limited Growth model, 679681 Word problems average rate of change from, 80 cubic model from, 156 domain and, 11, 12 exponential model from, 127129, 156157 linear model from, 4, 24, 27, 4449, 49, 156 logistic model from, 156
model selection strategies, 155157 piecewise model from, 4748 quadratic model from, 156 x-intercept by calculator, 321 in Cartesian coordinate system, 5 of cubic functions, 365366 in curve sketching, 367 definition of, 22, 57 from Excel, 322323 of exponential functions, 117 of linear functions, 2223, 2526, 2829 of logarithmic functions, 137, 368 of quadratic functions, 6466 y-intercept in Cartesian coordinate system, 5 of cubic functions, 365366 in curve sketching, 367 definition of, 22, 57 of exponential functions, 116, 117 of linear functions, 2226, 2829, 49, 748 of logarithmic functions, 368 of quadratic functions, 6466 of sine functions, 748 from tangent function, 614 of Taylor polynomials, 748
Index of Applications
Accidents, aviation, 163, 253 Accidents, vehicle age and, 794, 819, 837 alcohol-related, 819, 837 cubic model for, 264 by decade, 253 non-alcohol-related, 173, 291 pedestrian-involved, 837 in South Carolina, 478479 Acid reflux medicine, 732 Advertising on billboards, 91, 246, 265, 303 on cable TV , 168, 303 by direct mail, 92, 242 Got milk? campaign, 468 by Kellogg Company, 522523, 543 in magazines, 168, 303 in newspapers, 169, 362363 on radio, 169, 303 revenue from, 480 on television, 168, 170, 265 in Yellow Pages, 169 Age of labor force, 792793, 818 Agriculture and farming industry certified organic cropland in, 87, 240241, 264 land for, 87, 185 net farm income in, 185 number of farms in, 161 AIDS adult and adolescent cases of, 512, 687 deaths from, 157159, 249, 512, 687 pediatric cases of, 163, 508, 647, 659, 687 Alcohol, 133, 148, 819, 837 Allergy medicine, 732 Alternating current, 602 Anti-inflammatory medicine, 776 AP Calculus exam number of participants, 62 quadratic model for, 67 rate of change in participants, 243, 450, 462 Apogee, 602603, 629 Apparel and textile industry, 91, 203, 273, 512 Apples cross section of, 534 dietary fiber in, 24, 32 yield per acre, 274, 349 Arthritis, 776 Assets, 579 Asthma medicine, 731 Average fixed interest rates, 101102 Avian flu, 498 Aviation industry, 55, 159160, 163, 253 Balls, 204 Bananas, 32 Banks, 109, 713 Baseball, 232, 254, 617, 839 Basketball Final Four tournament, 745 games, attendance at, 81, 254 height of players, 833 salaries of players, 115, 116117, 135136, 293 Beef, 443 Billboards, 91, 246, 265, 303 Billiards, 532533 Bioterrorist attacks, 676679 Births, 793, 839 Blood pressure medication, 721723 Body Mass Index, 526, 543 Books inventory costs for, 333335 marginal printing costs of, 466 production cost function for, 400, 401 sales of, 712 self-publishing of, 333335 Bottled water, 238239, 290291, 319, 449, 462 Bottles, sports, 497 Boxes design of, 338340 girth of, 336 Parcel Post, 335338, 564565 surface area of, 339, 384, 524 volume of, 336, 542, 564565, 567571 Boys, 244, 264, 265, 284 Bread, cost of, 231 Breast cancer, 803 Cable TV advertising on, 168, 303 homes with, 113, 690 subscriptions to, 184 Cakes, 672 Call centers, 802803, 818819 Candy, 128129, 497 Cans, 204, 341343, 379380 Cardiovascular disease, 718, 794 Carrying capacity, 642643 Cars, see Vehicles Cassette tapes demand model for, 332333 market share, 689 price of, 332 sales of, 174, 211 shipments of, 230, 246 Cell phones, plans for, 54 Cereal consumption of, 7576, 161162, 318319, 376, 440441 dietary fiber in, 24, 32 Certificates of deposit, 552554 Checking account balance, 2526 Cheddar cheese, 493495 Cheese, 450, 462, 493495 Chicken, 87, 266, 268, 442443 Cholesterol, 718, 720721, 823824 Church membership, 254 Cirrhosis, 216
Clean rooms, 705706 Clothing sales tax, 2425 Coffee, 649652, 660, 670 Colleges/Universities enrollment in private, 87, 230, 241, 496 enrollment in public, 53, 162 male freshmen in, 83 tuition, see Tuition and fees Colon cancer, 803 Commissions, 4445, 5354 Compact disks (CDs) case dimensions, 335 demand function for, 204, 327 demand model for, 204, 327, 330331 packing for shipment, 335338 sales vs. VHS tape, 82, 204 shipments of, 246247 weight of, 335 Compound interest annual, 736, 738739 continuous, 665666, 708 differential equations for, 665666 integrating factor method for, 700702 multivariable model for, 521522 partial derivatives for, 531532 standard formula for, 665 Computer industry, 111, 705706, see also Software industry Computers, 554556, 653654, 803, 819 Concert admission prices, 133 Construction, 660 Consumer energy consumption, 582, 629 Cost accumulated change in, 486488 at break-even point, 327 definition of, 327 fixed, 45, 55, 327 forecasting, 128129 of goods sold, 247 marginal, 327, 400, 401, 466 minimizing, 333335, 344346 operating expenses, 480 production and, 55 profit and, 45, 327, 347, 348 rate of change in, 481, 486
sales volume and, 45 variable cost, 45, 55, 327 Counterfeit money, 745 Credit Unions, Federal, 169 Daylight, hours of in Juneau, 601, 612, 617, 625, 629 in Miami, 600601, 616, 628 in Phoenix, 601, 616, 628 in Sydney, 601602, 616617, 628 Debit cards, 110 Debt/Asset Ratio, 579 Dental care, 47, 132, 147 Dice, 791 Direct mail advertising, 92, 242 Diving rings, 204 Dogs, 776 DVD players distribution of, 681 introduction of, 249, 681 price of, 216 rate of change in sales of, 683684 sales of, 681683 shipments of, 82, 224225 DVD shipments, 247, 326 DVD videos, 247 Earnings, see also Salaries/wages in apparel and textile industry, 273 employee, see Employee earnings in lumber industry, 273, 275, 282283 in paper industry, 273 price and, 543, 578579 in printing and publishing industry, 273 in rubber and plastics industry, 273 in software industry, 352 in vehicle industry, 248 Economics Cobb-Douglas, see CobbDouglas production function information technology sector in, 688 surplus in, 490495, 497
Elections age of voters, 780788, 807, 809810 funds for, 163, 252 Electricity, 582, 602, 629 Employee earnings in lumber and wood products industry, 275, 282283, 326 in software industry, 356357 in vehicle industry, 248 Employees, number of, 273, 388, 522, 543 Employer labor costs, 273274 Equity, 579 Fabricated metals shipments, 689 Fall equinox, 612 Ferris wheels, 589596, 611 Fetal biparietal diameter, 839 Firearms, 159160 Fish, 450, 679681, 686687, 712 Fixed cost, 45, 55, 327 Flight segment tax, 55 Flow rates, 380381, 695697 Fruit apples, see Apples banana, 32 melons, 304305 Games, board, 802 Games, dice, 791 Games, electronic annual sales of, 376 Grand Theft Auto 3 sales, 203 hardware sales of, 161, 257, 262, 326 Harry Potter revenue, 481 quarterly sales of, 203 rate of change in sales of, 712 revenue from, 9899, 481 software sales for, 240, 263, 326 U.S. sales of, 83 Warcraft sales, 688 worldwide sales of, 712 Garden fences, 343344 Girls height of, 247, 264, 284 weight of, 245, 253, 264265, 284 Golf courses, 209
Growth funds, 552554 Harry Potter and the Sorcerers Stone game, 481 Health care industry dentistry, 47, 132, 147 insurance payments to, 82 nursing homes, 7273, 203 premium payments to, 36 prescription drugs, see Pharmaceuticals Heartworm medicine, 776 Heat index, 516, 526527 Height in BMI, 526 of boxes, 524 of boys, 264, 284 of cans, 341343, 516 of Ferris wheel cars, 590596, 611 of girls, 247, 264, 284 of liquid in cans, 379 of men, 823 of pool water, 381 projectile motion and, 617 of rice in cans, 379380 in Torricellis Law, 710 High School dropouts from, 792, 818 enrollment in public, 107 principals salaries, 108, 229230 Homes with cable TV , 113, 690 central air conditioning in, 85 garages in, 85 mortgages for, 84, 101102 owner-occupied, 184 price of, 85, 86, 164, 227, 240 size of, 84, 230, 240, 244245 steel- vs. wood-frame, 526 with VCRs, 113, 149, 293 Homicide from arguments over money/property, 508, 688 rate in U.S., 247, 375, 713 from romantic triangles, 233, 236237, 375 Hotel/motel rates, 151153, 688 Ice cream cones for, 388, 527
cost of, 82, 192, 217221 Ice formation, 641 Income, per capita, 245246, 251, 471, 488489 Income, personal average, 31 in California, 88, 248 in Colorado, 88, 248 commissions and, 4445, 5354 in Connecticut, 83, 248 disposable, 184 in Florida, 163, 469 in Louisiana, 88, 469 in Minnesota, 89 in Mississippi, 83 in New Mexico, 468 in U.S., 31 in Utah, 53 Income ranking, 53 Income taxes, 164165 Inflation, 128129 Influenza, 498500, 508 Information technology sector, 688 Interest average fixed rates of, 101102 compound, see Compound interest definition of, 700 Internet access to, at home/work, 31, 183 access to, by public schools, 109, 647, 659, 687 use of, cubic model for, 375 use of, per person, 113 use of, rate of change in, 647648, 659, 687, 709710, 711 Inventory costs, 333335 Investments certificates of deposit, 552554 with compound interest, see Compound interest finite geometric series for, 736 forecasting value of, 521522, 552554 growth funds, 552554 initial value of, 134 mutual funds, 102104 rate of change in, 665666 for retirement, 53, 134, 700702 shares, see Shares stocks, see Stocks
IQ tests, 820, 821822, 824826, 832833 Juggling pins, 204 Labor force, age of, 792793, 818 Lakes, 697700 Landscaping, 343346 Leather industry, 170, 375 Light bulbs, 798801, 815816 Liver disease, 216 Logistic equations for DVD player sales, 682 Lollipops, 497 Lottery, 804808, 817818 Lumber and wood products industry, 273, 275, 282283, 326 Lung cancer, 803, 819 Magazines, 168, 303 Magnetic wall calendars, 497 Malaria, 216 Manufacturing industry, 171 Margarine, 326, 449450 Marginal cost, 327, 400, 401, 466 Marginal profit, 327329, 332333 Marginal revenue, 327329, 433434, 443445, 456457 Medicine, see Pharmaceuticals Melons, 304305 Military, 88, 293, 302, 690 Milk advertising for, 468 consumption of, 84, 285, 288, 302, 468 dietary fiber in, 32 linear model for consumption of, 27 Moons orbit, 602 603, 629 Mortgages, 84, 101102 Movie tickets for A Beautiful Mind, 513 model selection strategies, 154155 for My Big Fat Greek Wedding, 376, 689 for The Passion of the Christ, 509 price of, 232 purchase model for, 320 revenue from, 320
(Continued on page I17 following the Subject Index)
Index of Applications (Continued)

Mortgages (continued ) sales model for, 513 for Spider-Man, 513 Mozzarella cheese, 450, 462 Museum admission fees, 31, 4748 Music market, 174 Mutual fund availability, 102104 Nail fungus medicine, 731 National Health and Nutrition Examination Survey (NHANES), 823, 833 Natural gas storage, 633 Nets, insecticide-treated, 216 News, spread of, 673675 Newspapers, 169, 184, 231, 362363 Newtons Law of Heating and Cooling, 660, 666670, 708, 710712 Nursing home care, 7273, 203 Nutrition calories in snacks, 1920 fat in diet, 5455, 59 fiber in diet, 24, 32 Pain medicine, 731, 732 Pallets, revenue from, 348 Paper industry, 273 Payroll, 90, 203 Pennies, 204 Perigee, 602, 629 Pharmaceuticals half-life of, 718719 income from, 89 maintenance level of, 719723 prescriptions for, 216 profit from, 89 sales of, 6972 sequences for patient monitoring, 718723, 730733, 776 series for patient monitoring, 733736, 738, 741, 745, 776 spending on, 203, 241 Pistons, 620623 Pollution of lakes, 691, 697700 from oil spills, 377378, 388, 389 Pools, 380381, 467, 705 Population of Africa, 216 of Akron, 122124 of Alaska, 32 of Arizona, 164, 187 carrying capacity and, 642643 of Hawaii, 144, 303 of Indonesia, 405 of Louisiana, 186 of Massachusetts, 186 of Missouri, 186 of Montana, 185 of North Dakota, 186 of Pennsylvania, 186 of Queen Creek, 164 rate of change in, 637639, 642643 of U.S., 247, 302, 304305, 647, 659 of Utah, 186 of Washington, 180, 200201 of West Virginia, 186 of world, 3233 of Wyoming, 186 Position Functions for Ferris wheel cars, 590, 596, 611 of free falling objects, 198, 199 for minivan, 399, 401, 402 for roller coaster trains, 352, 376 Poverty level, 184 Pregnancy, 839 Price-to-Earnings ratio (P/E), 543, 578579 Principals salaries, 108, 229230 Printing and publishing industry, 273274, 388 Prison rate, 87, 241, 302, 496 Private schools, 90, 107, 230, 512 Profit accumulated change in, 487488 at break-even point, 46 cost and, 45, 327, 347, 348 definition of, 327 demand and, 332333 gross, 347, 480 marginal, 327329, 332333 maximization of, 328330, 332333, 348 model selection strategies, 156 operating, 89, 247, 348, 349, 388, 470 rate of change in, 487 revenue and, 45, 327, 347349 sales volume and, 4546 Projectile motion, 617 Public schools enrollment in, 107, 512 enrollment rate of change in, 512 faculty salaries in, 173, 203, 292 with Internet access, 109, 647, 659, 687 teacher rate of change in, 512 teachers in, 90, 108, 173 Radio advertising on, 169, 303 wages in, 112, 230, 246 Rape, 508 Reading scores, 184 Rectal cancer, 803 Restaurants fat in food from, 5455, 59 number of, 172 sales by, see Restaurant sales wait time in, 789791 Restaurant sales full-service, 170, 240 McDonalds, 4243, 91, 173, 254, 350, 462 rate of change in, 281 U.S., 246 I17
I18
Retirement investments for, 53, 134, 700702 savings for, 723725 Revenue accumulated change in, 456457, 486488 from advertising, 480 at break-even point, 327 definition of, 327 demand and, 328, 330331, 347 forecasting, 9899 gross, 247 marginal, 327329, 433434, 443445, 456457 maximization of, 320, 330331, 348, 554556 model selection strategies, 156 from net sales, 388 price and, 320, 328, 330 profit and, 45, 327, 347349 quarterly, 15 rate of change in, 481, 486 sales volume and, 45, 74, 320, 328, 330, 348 Rice, 379380 River flow rate, 248, 411412, 426427, 432, 467 Roller coaster, 352, 376 Rubber and plastics industry, 273 Rumors, spread of, 639 Russet potatoes, 497 Salaries/wages, see also Employee earnings in apparel and textile industry, 91, 512 for basketball players, 115, 116117, 135136, 293 commissions and, 4445, 5354 in computer industry, 111 of electronics store employees, 4 hourly, earnings from, 4, 34 in manufacturing industry, 171 model selection strategies, 156 of principals, 108, 229230 in radio, 112, 230, 246 rate of change in, 512 of sailors, 93, 9597, 375 in software industry, 111, 112
for teachers, 3536, 173, 190, 203, 292 in television, 112, 231 variables in, 528 in vehicle industry, 349, 360362 withholding of savings from, 723725 Sales tax, 2425, 55 Sales volume cost and, 45 forecasting of, 156157 model selection strategies, 156 profit and, 4546 revenue and, 45, 74, 320, 328, 330, 348 Salt water, 695697 Savings accounts, 723725, 736, 738739 Schools college/university, see Colleges/Universities enrollment in, 156, 629 faculty, see Teachers high, see High School principals salaries, 108, 229230 private, 90, 107, 230, 512 public, see Public schools student-to-teacher ratio in, 107, 230 third-grade reading scores, 184 Self-publishing, 333335 Semiconductor manufacturing, 705706, 731 Shares, see also Stocks earnings per, 15, 543, 579 price of, 53, 579 trading volume, 110 Shellfish harvests, 633634 Shuttle bus wait time, 796797, 802, 813, 818 Smallpox, 676679 Social Security, 3536, 243244, 253 Software forecasting sales of, 554556 for games, 240, 263, 326 for PDAs, 526 Software industry earnings in, 352 employee earnings in, 356357 licensing fees, 480
wages in, 111, 112 Solid waste disposal, 31, 54, 59 Sports baseball, 232, 254, 617, 839 basketball, see Basketball running, 188, 252 Spring equinox, 612 Squirt guns, 328330 Stocks, see also Shares limit orders for, 180 on NYSE list, 111 P/E for, see Price-to-Earnings ratio (P/E) prices of, 15, 58, 178180 Summer solstice, 612 Sunstroke, 526 SUVs, see Vehicles Take-home pay, 31 Taxes flight segment, 55 income, 164165 sales, 2425, 55 on solid waste disposal, 54 on teachers salaries, 3536 Teachers contracts for, 35 in private schools, 90, 512 in public schools, 90, 512 rate of change in, 512 salaries for, 3536, 173, 190, 203, 292 taxes paid by, 3536 to-student ratio, 107, 230 Telephone, cellular plans for, 54 Television advertising on, 168, 170, 265 cable, see Cable TV sets, prices of, 133, 148 wages in, 112, 231 Temperature, air accumulated change in, 619620 actual vs. apparent, 515 average daily, 629 average monthly high, 583, 585589, 604606, 611, 624 in heat index, 516, 526527 in Newtons Law of Heating and Cooling, 666669 object temperature and, 640641
I19
Temperature, air (continued ) rate of change in, 604606, 611, 618 in wind chill, see Wind chill Temperature, liquid, 649652, 660, 668670, 671672 Temperature, of an object, 640641, 666669 Toxic chemical protective suits, 731 Trucks, see Vehicles Tuition and fees at Green River, 177178, 203, 231, 292 at Maricopa, 161 for Washington state, 162, 307 Unemployment rate, 184 Vans, see Vehicles Variable cost, 45, 55, 327 Vegetables, 304305 Vehicles accidents in, see Accidents, vehicle color of, 23 depreciation of, 127, 731 distance traveled by, 363365, 389, 399
fleet sales of, 350351 fuel consumption of, 7678, 86, 229 fuel costs for, 3435 leasing of, 86, 228 payments for, 726 position function for, 399, 401402 shipments of, 23 speed of, 198 speedometer in, 393 used, value of, 54, 59 velocity function for, 399, 401402, 423424, 432, 462 velocity of, 198 Vehicle industry, 248, 349, 360362 Velocity function for minivan, 399, 401402, 423424, 432, 462 for pistons, 620, 623 VHS tapes, 82, 204 Video games, see Games, electronic Videotape recorders (VCRs) homes with, 113, 149, 293 price of, 248, 684 rate of change in price of, 684685
VHS tapes for, 82, 204 Voltage, 602 Wages, see Salaries/wages Warranties, 803 Water balloons, 199200 Weight in BMI, 526 of boys, 244, 265, 284 of CDs, 335 of girls, 245, 253, 264265, 284 of men, 833 of Parcel Post packages, 335 Wildcat populations, 634 Wind chill by calculator, 383 forecasting changes to, 533534 model for, 381, 527, 543 rate of change in, 381382, 383, 385 variables in, 516, 528 Wine prices, 133, 148 Winter solstice, 612 Work hours, 793
Index of Businesses, Products, and Associations
Advanced Placement (AP) Program Calculus exam, 62, 67, 243, 450, 462 courses offered by, 62 Africa, 216 Akron, 122124 Alaska Barrow, 629 Exxon Valdex oil spill, 377378, 388, 389 Juneau, 601, 612, 617, 625, 629 population of, 32 Albacore tuna, 712 Apple Computers debt/asset ratio of, 579 earnings per share, 579 iPods, 347, 348, 481, 490493 sales of, 216 share prices, 579 stock prices, 15, 58, 179180 Arizona Arizona State University, 802, 818, 833 Chandler, 705 Fantasy 5 lottery, 804808 Glendale, 717 Maricopa Community College, 161, 660 Phoenix, see Phoenix The Pick lottery, 817818 population of, 164, 187 Queen Creek, 164 Arizona State University, 802, 818, 833 Army active-duty personnel, 293, 690 Aspirin, 730 ATACAND, 721723 Australia, 601602, 616617, 628, 712 Azithromycin, 733736, 738, 745 BeautiControl Cosmetics, 4546 Beautiful Mind, A, 513 Black Sea fish, 679681 Blizzard Entertainment, 688 Buick, 351 Cadillac, 351 California, 88, 248, 741742 Canada, Ontario, 582 Candesartan cilexetil, 721723 Centers for Disease Control and Prevention on AIDS, see AIDS on BMI, 526 on Hong Kong flu, 498500 on mens height, 823 on mens weight, 833 on smallpox bioterrorist attack, 676677 Centre for Population Biology, 633, 634
Chandler, 705 Cheerios, 24 Chevrolet, 351 Church of Jesus Christ of Latter-Day Saints, 254 Claritin, 732 Click2Learn Corporation, 58 Coca-Cola Cobb-Douglas model for, 523, 576, 580 cost rate of change model for, 496497 marginal revenue model for, 433434, 443445, 456457 production volume of, 156157, 348 revenue model for, 348 revenue rate of change model for, 496 Code Red computer virus, 653654 College Board, The AP Calculus exam, 62, 67, 243, 450, 462 courses offered by, 62 Colorado, 88, 248, 629 Colorado Springs, 629 Columbia River, 432, 467 Connecticut, 83, 248, 582 Consumer Price Index, 132, 175 Cooperative Association for Internet Data Analysis, 653 Copper shark, 686687 Cowcod, 687 Crabs, Dungeness, 633634 Dave Matthews Band, 133 Digital Think Corporation, 15 Disneyland, 796797, 813 Don Millar and Associates, 676677 Dow Jones Industrial Average, 183 Ducks, mallard, 633 Electronic Arts cost model for, 481, 486488 earnings per share, 543, 579 profit model for, 480, 487488 revenue model for, 9899, 481, 486 488 share prices, 579 stock prices, 543 Enumclaw Transfer Station, 31, 54, 59 Experience Music Project, 31, 4748 Exxon Valdez, 377378, 388, 389 Fantasy 5 lottery, 804808 Federal Bureau of Investigation, 508 Federal Reserve Bank of Boston, 745 Ferris, George, 589
FLASH bus service, 802, 818 Florida manufacturing wages in, 171 Miami, 600601, 616, 628 personal income in, 163, 469 Flounder, 450, 686 Ford Motor Company Cobb-Douglas model for, 525, 543, 561, 566567, 575 employee earnings at, 248 wages in, 349, 360362 Frito-Lay, 247, 348349 Fruit Smiles, 1920 Fruity Pebbles, 32 Gatorade, 388, 496 GE-500 Semi-Automatic cone maker, 527 General Electric, 579 General Mills cereals, 32 Giant Drop, 466, 467 Glendale, 717 Golden Pillar Publishing, 400, 401, 466 Grand Theft Auto 3 video game, 203 Green River Community College, 177178, 203, 231, 292 Hall Kinion, 5354 Harbor Capital Funds, 53 Harvard University, 833 Hawaii, 144, 303 Help Ministries, 802803, 818819 Hong Kong flu, 498500, 508 Idaho, Snake River, 432 IFC Films, My Big Fat Greek Wedding, 376, 689 Indiana, 745 Indianapolis, 745 Indonesia, 405 Intamin, 466, 467 Intel Corporation, 705 Interceptor Flavor Tabs, 776 iPods, 347, 348, 481, 490493 Johnson & Johnson cost model for, 247, 496 cost rate of change model for, 512 net income of, 89 profit model for, 470 revenue model for, 247, 496 revenue rate of change model for, 512 Jordan, Michael, 115 Joyner, Florence Griffith, 188 Juneau, 601, 612, 617, 625, 629
Kellogg Company advertising by, 522523, 543 cost of goods sold by, 347 employees in, 522, 543 sales of, 74, 347, 522523, 543 King County, 54 Lamisil, 731 Lipitor, 718, 720721 Louisiana, 88, 186, 469 Lutherans, 303 Lynx, Canadian, 634 Macintosh OS, 554556 Major League Baseball, 232, 254, 839 Maricopa Community College, 161, 660 Massachusetts, 186, 582 Maxtor MaXLine Pro 500 disk drives, 803, 819 McDonalds coffee temperature, 649, 670 fat in food from, 5455, 59 number of, 172 sales of, 4243, 91, 173, 254, 350, 462 Mercedes-Benz, 59 Miami, 600601, 616, 628 Michigan, 171, 582 Microsoft, 554556, 579 Minnesota Canadian lynx in, 634 deer in, 715716 Department of Natural Resources, 715 Northstar Cash lottery, 818 personal income in, 89 Trout Lake, 691 Mississippi, 83 Mississippi River, 691 Missouri, 186 Montana, 185 Muskego, 731 My Big Fat Greek Wedding, 376, 689 Naproxen, 776 NASA, 633 NASDAQ, 110, 246 NASDAQ Composite Index, 184 NBA height of players, 833 salaries of players, 115, 116117, 135136, 293 NCAA Basketball, 81, 745 New Jersey, 582 New Mexico, 468 Newton, Issac, 257, 666 New York City, 498500, 508, 673674, 838 New York State New York City, 498500, 508, 673674, 838 power outage in, 582 New York Stock Exchange (NYSE), 110, 111, 246 North Carolina, 53
North Dakota, 186, 633 Northstar Cash lottery, 818 Ohio, 122124, 582 Olympics, 247, 252 Ontario, 582 Oregon, Rogue River, 426427 Palm, 526 Passenger Facility Charge, 55 Passion of the Christ, The, 509 Pennsylvania, 186, 582 People for the Ethical Treatment of Animals, 375 Pepsi, 349, 496 Percocet, 732 Phoenix air temperature in, 583, 585589, 604606, 611, 618620, 624 daylight, hours of, 601, 616, 628 electricity cost in, 629 Pick, The lottery, 817818 Pickerel fish, 686 Picture People studio, 55 Pontiac, 351 Post cereal, 32 Prilosec, 732 Quantitative Environmental Learning Project, 633634 Queen Creek, 164 Qwest, 54 Rainbow trout, 712 Real Networks, Inc., 480 Red mullet, 686 Rogue River, 426427 San Diego, 741742 SAT Reasoning Test, 833 Saturn Corporation, 127 Seattle cellular phone plans in, 54 Experience Music Project, 31, 4748 sales tax in, 2425 Space Needle, 204 weather forecast for, 581 September 11, 2001 terrorist attacks, 673674 Singulair, 731, 776 Snake River, 432 Snoqualmie River, 248, 432 South Carolina, 478479, 526 Southwest Airlines, 55 Space Needle, 204 Space Shuttle, 633 Special K Red Berries cereal, 347 Spider-Man, 513 SplashData, 526 Sprint PCS, 54
Starbucks accumulated change in sales of, 458 rate of change in sales of, 451 sales income of, 86, 249, 350 Subway Restaurants, 172 Super Bowl, 717, 741742 Swan River, 691 Sydney, 601602, 616617, 628 Tacoma, 55 TIAA-CREF, 53, 134 Toyota, 54, 731 Transportation Statistics Bureau, 160 Tropicana, 388, 496 Trout Lake, 691 Turbot, 686 Twister game, 802 Uline Shipping Supply Specialists, 348 United Parcel Service, 335 U.S. Centers for Medicare & Medicaid Services, 7273 U.S. Department of Housing and Urban Development (HUD), 526 U.S. National Center for Education Statistics, 512 U.S. Postal Service, 335, 564565 Utah, 53, 186 Vermont, 582 Warcraft game, 688 Washington college tuition/fees in, 162, 307 Columbia River, 432, 467 Enumclaw Transfer Station, 31, 54, 59 Green River Community College, 177178, 203, 231, 292 King County, 54 per capita income in, 245246, 488489 population of, 180, 200201 Seattle, see Seattle Snoqualmie River, 248, 432 Tacoma, 55 Washington State University, 53, 162 Wenatchee River, 432 Washington State University, 53, 162 Wenatchee River, 432 West Virginia, 186 Wheaties, 32 Whiting fish, 679681 Windows OS, 554556 Wisconsin, Muskego, 731 World Health Organization, 216, 498 Wyoming, 186 Yahtzee, 791 Yellow Pages, 169
Derivative Rules
For differentiable functions f ( x ) , g( x ) , and u 5 h( x ) and constants n and c, the following rules apply: Constant Rule d (c) 5 0 dx Power Rule d n ( x ) 5 nx n21 dx Constant Multiple Rule d 3c # f ( x ) 4 5 c # f r( x ) dx Sum and Difference Rule d 3 f ( x ) 6 g( x ) 4 5 f r( x ) 6 gr( x ) dx Product Rule d 3 f ( x ) # g( x ) 4 5 f r( x ) g( x ) 1 gr( x ) f ( x ) dx Quotient Rule d f (x) f r( x ) g( x ) 2 gr( x ) f ( x ) c d 5 dx g( x ) 3g( x ) 4 2 Exponential Rule d x 3b 4 5 ln( b ) # bx dx Logarithmic Rule d 1 3logb x4 5 dx ln( b ) # x Chain Rule d 5 f 3g( x ) 46 5 f r 3g( x ) 4 gr( x ) dx Generalized Exponential Rule d u 3b 4 5 ln( b ) # buur dx Generalized Logarithmic Rule 1 d 3logb u4 5 ur dx ln( b ) # u Chain Rule (Alternate Form) dy dy # dt 5 dx dt dx Generalized Power Rule d n ( u ) 5 nu n21ur dx
Integral Rules
For integrable functions f ( x ) , g( x ) , u( x ) , and v( x ) and constants n, b, c, and C, the following rules apply: Power Rule xn 1 1 n 3x dx 5 n 1 1 1 C for n 2 2 1 Rule for 1 x
1 3 x dx 5 ln 0 x 0 1 C Constant Multiple Rule # # 3 3c f ( x ) 4 dx 5 c 3 f ( x ) dx Sum and Difference Rule 3 3 f ( x ) 6 g( x ) 4 dx 5 3 f ( x ) dx 6 3g( x ) dx Exponential Rule bx x 3b dx 5 ln( b ) 1 C Integration by Parts 3u dv 5 uv 2 3v du
Common Functions
Constant Linear
Quadratic Cubic Quartic Power Exponential Logarithmic
Equation
y5c y 5 mx 1 b y 5 ax 2 1 bx 1 c y 5 ax 3 1 bx 2 1 cx 1 d y 5 ax 4 1 bx 3 1 cx 2 1 dx 1 k y 5 xn y 5 ab x y 5 logb( x )

0618611045

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

0618611045

Uploaded by

Copyright:

Available Formats

T FOR

HOUGHTON MIFFLIN HARCOURT PUBLISHING COMPANY

Boston New York

1 Functions and Linear Models

Chapter 1 Study Sheet 57 Chapter 1 Review Exercises 58 Make It Real Project 1 60

7 Advanced Integration Techniques and Applications 471

8 Multivariable Functions and Partial Derivatives 515

11 Sequences and Series

12 Probability and Calculus

Answers to Odd-Numbered Exercises Index I1

This page intentionally left blank

Real Applications! Real Data! Real Learning!

Average Rate of Change

Calculate the average rate of change of a function over an interval

Limits and Instantaneous Rates of Change

Limits and Continuity

The Derivative as a Slope: Graphical Methods

The Derivative as a Function: Algebraic Method

3.1 Average Rate of Change

Interpreting the Derivative

3.1 Average Rate of Change

Calculate the average rate of change of a function over an interval

Calculating an Average Rate of Change

Years (since 19941995) (t) 0 1 2 3 4 5 6 7

Change in Tuition Cost from Prior Year (dollars)

Source: Green River Community College.

Real Applications! Real Data! Real Learning!

Using Technology to Graph a Sequence

Excel Technology Tips

Real Applications! Real Data! Real Learning!

Analyzing Shuttle Bus Wait Times

P( x ) 5 15 2 0 5 15 to compute the expected value. 1 E( x ) 5 ( 0 1 15 ) 2 5 15 2

3.6 Interpreting the Derivative

(Source: Modeled from Recording Industry Association of America data.)

deaths, where t is the number of years since 1990.

Interpret the meaning of H( 10 ) 5 185 and Hr( 10 ) 5 2 14 . 20.

Association of America data.)

Modeled from www.babybag.com data.)

Interpret the meaning of H( 5 ) 5 43.6 and Hr( 5 ) 5 2.6 . 21.

(Source: Modeled from www.olympicwomen.co.uk data.)

women, where t is the number of years since 1900.

Interpret the meaning of W( 92 ) 5 2662 and Wr( 92 ) 5 150. 16.

Interpret the meaning of R( 2 ) 5 32,317 and Rr( 2 ) 5 3810 . 22.

(Source: Modeled from U.S. Bureau of the Census data.)

million people, where t is the calendar year.

million dollars, where t is the number of years

Interpret the meaning of P( 2 ) 5 2056 and Pr( 2 ) 5 93.5 .

Real Applications! Real Data! Real Learning!

Show You Know Conceptual Exercises

6.4 The Definite Integral

6.3 Using Sums to Approximate Area

Algebra Flashback Exercises

6.3 Algebra Flashback

t12 ; 33, 114 t22

Real Applications! Real Data! Real Learning!

Vertical and Horizontal Shifts

Just In Time Algebra Review

Chapter Study Sheet

Real Applications! Real Data! Real Learning!

dy 5 y2 dx dy 5 sin y dx dy 5 cos y dx dy 5 2y2 dx

Make It Real Projects

Where to find data

Additional ResourcesGet the Most out of Your Textbook!

Supplements for the Student