Complex Analysis

George Cain

(c)Copyright 1999 by George Cain. All rights reserved.

Table of Contents
Chapter One - Complex Numbers 1.1 Introduction 1.2 Geometry 1.3 Polar coordinates Chapter Two - Complex Functions 2.1 Functions of a real variable 2.2 Functions of a complex variable 2.3 Derivatives Chapter Three - Elementary Functions 3.1 Introduction 3.2 The exponential function 3.3 Trigonometric functions 3.4 Logarithms and complex exponents Chapter Four - Integration 4.1 Introduction 4.2 Evaluating integrals 4.3 Antiderivatives Chapter Five - Cauchy's Theorem 5.1 Homotopy 5.2 Cauchy's Theorem Chapter Six - More Integration 6.1 Cauchy's Integral Formula 6.2 Functions defined by integrals 6.3 Liouville's Theorem 6.4 Maximum moduli Chapter Seven - Harmonic Functions 7.1 The Laplace equation 7.2 Harmonic functions 7.3 Poisson's integral formula Chapter Eight - Series 8.1 Sequences 8.2 Series 8.3 Power series 8.4 Integration of power series 8.5 Differentiation of power series

Chapter Nine - Taylor and Laurent Series 9.1 Taylor series 9.2 Laurent series Chapter Ten - Poles, Residues, and All That 10.1 Residues 10.2 Poles and other singularities Chapter Eleven - Argument Principle 11.1 Argument principle 11.2 Rouche's Theorem

---------------------------------------------------------------------------George Cain School of Mathematics Georgia Institute of Technology Atlanta, Georgia 0332-0160 cain@math.gatech.edu

Chapter One

Complex Numbers
1.1 Introduction. Let us hark back to the first grade when the only numbers you knew were the ordinary everyday integers. You had no trouble solving problems in which you were, for instance, asked to find a number x such that 3x 6. You were quick to answer 2. Then, in the second grade, Miss Holt asked you to find a number x such that 3x 8. You were stumpedthere was no such number! You perhaps explained to Miss Holt that 32 6 and 3 3 9, and since 8 is between 6 and 9, you would somehow need a number between 2 and 3, but there isnt any such number. Thus were you introduced to fractions. These fractions, or rational numbers, were defined by Miss Holt to be ordered pairs of integersthus, for instance, 8, 3 is a rational number. Two rational numbers n, m and p, q were defined to be equal whenever nq pm. (More precisely, in other words, a rational number is an equivalence class of ordered pairs, etc.) Recall that the arithmetic of these pairs was then introduced: the sum of n, m and p, q was defined by n, m and the product by n, m p, q np, mq . p, q nq pm, mq ,

Subtraction and division were defined, as usual, simply as the inverses of the two operations. In the second grade, you probably felt at first like you had thrown away the familiar integers and were starting over. But no. You noticed that n, 1 p, 1 n p, 1 and also n, 1 p, 1 np, 1 . Thus the set of all rational numbers whose second coordinate is one behave just like the integers. If we simply abbreviate the rational number n, 1 by n, there is absolutely no danger of confusion: 2 3 5 stands for 2, 1 3, 1 5, 1 . The equation 3x 8 that started this all may then be interpreted as shorthand for the equation 3, 1 u, v 8, 1 , and one easily verifies that x u, v 8, 3 is a solution. Now, if someone runs at you in the night and hands you a note with 5 written on it, you do not know whether this is simply the integer 5 or whether it is shorthand for the rational number 5, 1 . What we see is that it really doesnt matter. What we have really done is expanded the collection of integers to the collection of rational numbers. In other words, we can think of the set of all rational numbers as including the integersthey are simply the rationals with second coordinate 1. One last observation about rational numbers. It is, as everyone must know, traditional to

1.1

write the ordered pair n, m as

n m

. Thus n stands simply for the rational number

n 1

, etc.

Now why have we spent this time on something everyone learned in the second grade? Because this is almost a paradigm for what we do in constructing or defining the so-called complex numbers. Watch. Euclid showed us there is no rational solution to the equation x 2 2. We were thus led to defining even more new numbers, the so-called real numbers, which, of course, include the rationals. This is hard, and you likely did not see it done in elementary school, but we shall assume you know all about it and move along to the equation x 2 1. Now we define complex numbers. These are simply ordered pairs x, y of real numbers, just as the rationals are ordered pairs of integers. Two complex numbers are equal only when there are actually the samethat is x, y u, v precisely when x u and y v. We define the sum and product of two complex numbers: x, y and x, y u, v xu yv, xv yu u, v x u, y v

As always, subtraction and division are the inverses of these operations. Now lets consider the arithmetic of the complex numbers with second coordinate 0: x, 0 and x, 0 u, 0 xu, 0 . u, 0 x u, 0 ,

Note that what happens is completely analogous to what happens with rationals with second coordinate 1. We simply use x as an abbreviation for x, 0 and there is no danger of confusion: x u is short-hand for x, 0 u, 0 x u, 0 and xu is short-hand for x, 0 u, 0 . We see that our new complex numbers include a copy of the real numbers, just as the rational numbers include a copy of the integers. Next, notice that x u, v z x, y may be written u, v x x, 0 u, v xu, xv . Now then, any complex number

1.2

x, y x y 0, 1

x, 0

0, y

When we let

0, 1 , then we have z x, y u, v y u xv x u v yu
2

y v. Then we have

Now, suppose z

x, y

y and w zw x xu

yv

0, 1 0, 1 1, 0 , and we have agreed that we can We need only see what 2 is: 2 2 safely abbreviate 1, 0 as 1. Thus, 1, and so zw xu yv xv yu

and we have reduced the fairly complicated definition of complex arithmetic simply to 1. ordinary real arithmetic together with the fact that 2 Lets take a look at divisionthe inverse of multiplication. Thus number you must multiply w by in order to get z . An example: z w x y x y u u v u v u xu yv yu xv u2 v2 xu yv yu xv 2 2 u v u2 v2 v v
z w

stands for that complex

Note this is just fine except when u 2 v 2 0; that is, when u by any complex number except 0 0, 0 .

0. We may thus divide

One final note in all this. Almost everyone in the world except an electrical engineer uses the letter i to denote the complex number we have called . We shall accordingly use i rather than to stand for the number 0, 1 . Exercises

1.3

1. Find the following complex numbers in the form x iy: a) 4 7i 2 3i b) 1 i 3 b) 51 2i c) 1 i i 2. Find all complex z x, y such that z2 3. Prove that if wz 0, then w 0 or z z 0. 1 0

1.2. Geometry. We now have this collection of all ordered pairs of real numbers, and so there is an uncontrollable urge to plot them on the usual coordinate axes. We see at once then there is a one-to-one correspondence between the complex numbers and the points in the plane. In the usual way, we can think of the sum of two complex numbers, the point in the plane corresponding to z w is the diagonal of the parallelogram having z and w as sides:

We shall postpone until the next section the geometric interpretation of the product of two complex numbers. The modulus of a complex number z x iy is defined to be the nonnegative real number x 2 y 2 , which is, of course, the length of the vector interpretation of z. This modulus is traditionally denoted |z|, and is sometimes called the length of z. Note that x2 | x, 0 | |x|, and so | | is an excellent choice of notation for the modulus. x iy. Thus |z| 2 The conjugate z of a complex number z x iy is defined by z Geometrically, the conjugate of z is simply the reflection of z in the horizontal axis: zz.

1.4

Observe that if z

iy and w z

u w

iv, then x x z u iy w. iy u v iv

In other words, the conjugate of the sum is the sum of the conjugates. It is also true that zw z w. If z x iy, then x is called the real part of z, and y is called the imaginary part of z. These are usually denoted Re z and Im z, respectively. Observe then that 2 Re z and z z 2 Im z. z z Now, for any two complex numbers z and w consider |z w| 2 z zz |z| 2 |z| 2 In other words, |z w| |z| |w| the so-called triangle inequality. (This inequality is an obvious geometric factcan you guess why it is called the triangle inequality?) Exercises 4. a)Prove that for any two complex numbers, zw z z b)Prove that w w . c)Prove that ||z| |w|| |z w|. 5. Prove that |zw|
z |z||w| and that | w | |z| |w|

w z wz

w wz |w| 2

z |w| 2

w z

ww |z| |w|
2

2 Re w z 2|z||w|

z w.

1.5

6. Sketch the set of points satisfying a) |z 2 3i| 2 c) Re z i 4 e)|z 1| |z 1| 4

b)|z 2i| 1 d) |z 1 2i| |z 3 f) |z 1| |z 1| 4

i|

1.3. Polar coordinates. Now lets look at polar coordinates r, of complex numbers. Then we may write z r cos i sin . In complex analysis, we do not allow r to be negative; thus r is simply the modulus of z. The number is called an argument of z, and there are, of course, many different possibilities for . Thus a complex numbers has an infinite number of arguments, any two of which differ by an integral multiple of 2 . We usually write arg z. The principal argument of z is the unique argument that lies on the interval , . Example. For 1 i, we have 1 i 2 cos 7 4 2 cos 4 399 2 cos 4
7 4

i sin 7 4 i sin 4 i sin 399 4

399 4

etc., etc., etc. Each of the numbers principal argument is 4 . Suppose z r cos zw i sin r cos rs cos and w

, and

is an argument of 1

i, but the

s cos sin sin i sin

i sin . Then i sin i sin cos sin cos

i sin s cos

rs cos cos

We have the nice result that the product of two complex numbers is the complex number whose modulus is the product of the moduli of the two factors and an argument is the sum of arguments of the factors. A picture:

1.6

We now define exp i , or e i by ei cos i sin

We shall see later as the drama of the term unfolds that this very suggestive notation is an excellent choice. Now, we have in polar form z where r re i ,

|z| and is any argument of z. Observe we have just shown that ei ei ei .

It follows from this that e i e

1. Thus 1 ei e
i

It is easy to see that z w Exercises 7. Write in polar form re i : a) i c) 2 e) 3 3i re i se i r cos s i sin

b) 1 i d) 3i

8. Write in rectangular formno decimal approximations, no trig functions: a) 2e i3 b) e i100 c) 10e i /6 d) 2 e i5 /4 9. a) Find a polar form of 1 i 1 i 3 . b) Use the result of a) to find cos 7 and sin 12 10. Find the rectangular form of 1 i
100

7 12

1.7

11. Find all z such that z 3 12. Find all z such that z 4

1. (Again, rectangular form, no trig functions.) 16i. (Rectangular form, etc.)

1.8

Chapter Two

Complex Functions
2.1. Functions of a real variable. A function : I C from a set I of reals into the complex numbers C is actually a familiar concept from elementary calculus. It is simply a function from a subset of the reals into the plane, what we sometimes call a vector-valued function. Assuming the function is nice, it provides a vector, or parametric, description of a curve. Thus, the set of all t : t e it cos t i sin t cos t, sin t , 0 t 2 is the circle of radius one, centered at the origin. We also already know about the derivatives of such functions. If t x t iy t , then x t iy t , interpreted as a vector in the plane, it is the derivative of is simply t tangent to the curve described by at the point t . Example. Let t part of the curve y t it 2 , 1 t x 2 between x 1. One easily sees that this function describes that 1 and x 1:

-1

-0.5

0.5 x

Another example. Suppose there is a body of mass M fixed at the originperhaps the sunand there is a body of mass m which is free to moveperhaps a planet. Let the location of this second body at time t be given by the complex-valued function z t . We assume the only force on this mass is the gravitational force of the fixed body. This force f is thus f GMm |z t | 2 zt |z t |

where G is the universal gravitational constant. Sir Isaac Newton tells us that mz t f GMm |z t | 2 zt |z t |

2.1

Hence, z Next, lets write this in polar form, z re i : k ei r2 GM z |z| 3

d 2 re i dt 2 where we have written GM

k. Now, lets see what we have. d re i dt r d ei dt dr e i dt

Now, d ei dt d cos dt sin i cos i d ei . dt (Additional evidence that our notation e i Thus, d re i dt cos i sin is reasonable.) i sin i cos i sin d dt d dt

dr e i r d ei dt dt d ei dr e i r i dt dt dr ir d e i . dt dt

Now,

2.2

d 2 re i dt 2

d 2 r i dr dt dt 2 dr ir d dt dt 2 d r r dt 2 re i
k r2

d dt

ir d 2 dt

ei

i d ei dt d 2 dt

i rd 2 dt

2 dr d dt dt

ei

Now, the equation

d2 dt 2

e i becomes
2

d2r dt 2

r d dt

i rd 2 dt

2 dr d dt dt

k . r2

This gives us the two equations d2r dt 2 and, rd 2 dt

2

r d dt

k , r2

2 dr d dt dt

0.

Multiply by r and this second equation becomes d r2 d dt dt This tells us that r2 d dt is a constant. (This constant is called the angular momentum.) This result allows us to get rid of d in the first of the two differential equations above: dt d2r dt 2 or, d2r dt 2 2.3
2

0.

r2

k r2

r3

k . r2

Although this now involves only the one unknown function r, as it stands it is tough to solve. Lets change variables and think of r as a function of . Lets also write things in 1 terms of the function s r . Then, d dt Hence, dr dt and so d2r dt 2 d dt ds d 2 2 2 d s s , d 2 s2 d d ds d dr r2 d ds , d d d dt d d . r2 d

and our differential equation looks like d2r dt 2 or, d2s d 2 s k . 2

2 2 2 2 s d s d 2 2 3

ks 2 ,

This one is easy. From high school differential equations class, we remember that s where A and 1 r A cos k , 2

are constants which depend on the initial conditions. At long last, r /k cos
2

where we have set eccentricity .

/k. The graph of this equation is, of course, a conic section of

Exercises 2.4

1. a)What curve is described by the function t 3t 4 i t 6 , 0 t b)Suppose z and w are complex numbers. What is the curve described by t 1 t w tz, 0 t 1 ? 2. Find a function x 10. 3. Find a function that describes that part of the curve y 4x 3 1 between x

1?

0 and

that describes the circle of radius 2 centered at z

2i .

4. Note that in the discussion of the motion of a body in a central gravitational force field, it was assumed that the angular momentum is nonzero. Explain what happens in case 0.

2.2 Functions of a complex variable. The real excitement begins when we consider function f : D C in which the domain D is a subset of the complex numbers. In some sense, these too are familiar to us from elementary calculusthey are simply functions from a subset of the plane into the plane: fz f x, y u x, y iv x, y u x, y , v x, y

z2 x iy 2 x 2 y 2 2xyi. In other words, Thus f z z 2 looks like f z 2 2 2xy. The complex perspective, as we shall see, generally u x, y x y and v x, y provides richer and more profitable insights into these functions. The definition of the limit of a function f at a point z which we learned in elementary calculus: lim f z
z z0

z 0 is essentially the same as that

. As means that given an 0, there is a so that |f z L| whenever 0 |z z 0 | f z 0 . If f is you could guess, we say that f is continuous at z 0 if it is true that lim f z continuous at each point of its domain, we say simply that f is continuous. Suppose both lim f z and lim g z exist. Then the following properties are easy to establish:
z z0 z z0 z z0

2.5

lim f z
z z0

gz

lim f z
z z0

lim g z
z z0

lim f z g z
z z0

lim f z lim g z
z z0 z z0

and, lim
z z0

fz gz

lim f z
z z0

lim g z
z z0

provided, of course, that lim g z

z z0

0.

It now follows at once from these properties that the sum, difference, product, and quotient of two functions continuous at z 0 are also continuous at z 0 . (We must, as usual, except the dreaded 0 in the denominator.) It should not be too difficult to convince yourself that if z fz u x, y iv x, y , then lim f z
z z0

x, y , z 0

x 0 , y 0 , and

x,y

lim

x 0 ,y 0

u x, y

x,y

lim

x 0 ,y 0

v x, y

Thus f is continuous at z 0

x 0 , y 0 precisely when u and v are.

Our next step is the definition of the derivative of a complex function f. It is the obvious thing. Suppose f is a function and z 0 is an interior point of the domain of f . The derivative f z 0 of f is f z0 Example Suppose f z z 2 . Then, letting z z z 0 , we have lim
z z0

fz z

f z0 z0

2.6

lim
z z0

fz z

f z0 z0

lim
z 0

f z0 z0 2z 0 z

z z z 2 z z z z

f z0 z2 0
2

lim
z 0

lim
z 0 z 0

lim 2z 0 2z 0 No surprise herethe function f z Another Example Let f z zz. Then, lim
z 0

z 2 has a derivative at every z, and its simply 2z.

f z0

z z

f z0

lim
z 0

z0

z z0 z0 z z z

z z z

z0 z 0

lim z 0 z
z 0

lim
z 0

z0

z0 z z

Suppose this limit exists, and choose z lim

z 0

x, 0 . Then, lim
x 0

z0

z0 z z

z0 z0

z0 x x

z0 Now, choose z 0, y . Then, lim

z 0

z0

z0 z z

lim z0

y 0

z0 z0

i y

z0

i y i y

Thus, we must have z 0 z 0 z 0 z 0 , or z 0 0. In other words, there is no chance of this limits existing, except possibly at z 0 0. So, this function does not have a derivative at most places. Now, take another look at the first of these two examples. It looks exactly like what you 2.7

did in Mrs. Turners 3 rd grade calculus class for plain old real-valued functions. Meditate on this and you will be convinced that all the usual results for real-valued functions also hold for these new complex functions: the derivative of a constant is zero, the derivative of the sum of two functions is the sum of the derivatives, the product and quotient rules for derivatives are valid, the chain rule for the composition of functions holds, etc., etc. For proofs, you need only go back to your elementary calculus book and change xs to zs. A bit of jargon is in order. If f has a derivative at z 0 , we say that f is differentiable at z 0 . If f is differentiable at every point of a neighborhood of z 0 , we say that f is analytic at z 0 . (A z : |z z 0 | r, r 0 so that D S. set S is a neighborhood of z 0 if there is a disk D ) If f is analytic at every point of some set S, we say that f is analytic on S. A function that is analytic on the set of all complex numbers is said to be an entire function. Exercises 5. Suppose f z exist. 3xy ix y 2 . Find lim f z , or explain carefully why it does not
z 3 2i

6. Prove that if f has a derivative at z, then f is continuous at z. 7. Find all points at which the valued function f defined by f z 8. Find all points at which the valued function f defined by fz 2 i z 3 iz 2 4z 1 7i has a derivative. 9. Is the function f given by fz
z z
2

z has a derivative.

,z ,z

0 0

differentiable at z

0? Explain.

2.3. Derivatives. Suppose the function f given by f z u x, y iv x, y has a derivative at z z 0 x 0 , y 0 . We know this means there is a number f z 0 so that f z0 lim
z 0

f z0

z z

f z0

2.8

Choose z f z0

x, 0 lim
z 0

x. Then, f z0 u x0 u x0 z z f z0 iv x 0 x, y 0 x i u x0, y0 v x0 iv x 0 , y 0 v x0, y0

lim
x 0

x, y 0

lim
x 0

u x ,y 0 0 x Next, choose z f z0 0, y lim

z 0

x, y 0 u x 0 , y 0 x i v x0, y0 x

x, y 0 x

i y. Then, z z f z0 y y y iv x 0 , y 0 y i y i u x0, y0 u x0, y0 iv x 0 , y 0 y y u x0, y0

f z0 u x0, y0

lim lim

y 0

v x0, y0

v x0, y0

y 0

v x ,y 0 0 y

i u x0, y0 y

We have two different expressions for the derivative f z 0 , and so u x ,y 0 0 x or, u x0, y0 x u x0, y0 y v x ,y , 0 0 y v x ,y 0 0 x i v x0, y0 x v x ,y 0 0 y i u x0, y0 y

These equations are called the Cauchy-Riemann Equations. We have shown that if f has a derivative at a point z 0 , then its real and imaginary parts satisfy these equations. Even more exciting is the fact that if the real and imaginary parts of f satisfy these equations and if in addition, they have continuous first partial derivatives, then the function f has a derivative. Specifically, suppose u x, y and v x, y have partial x 0 , y 0 , suppose these derivatives are continuous at derivatives in a neighborhood of z 0 z 0 , and suppose 2.9

u x ,y 0 0 x u x0, y0 y We shall see that f is differentiable at z 0 . f z0 u x0 z f z0 z x, y 0 y

v x ,y , 0 0 y v x ,y . 0 0 x

u x0, y0 x

i v x0 i y

x, y 0

v x0, y0

Observe that u x0 x, y 0 y u x0, y0 u x0 u x0, y0 Thus, u x0 and, u x where, lim

z 0 1

x, y 0 y

u x0, y0

u x0, y0 .

x, y 0

u x0, y0

x u x

, y0

y,

, y0

u x ,y 0 0 x 0.

1,

Thus, u x0 x, y 0 y u x0, y0 y x u x0, y0 x

1

Proceeding similarly, we get

2.10

f z0 u x0 x
u x

z f z0 z x, y 0 y x0, y0
1

u x0, y0 x i
v x

i v x0 i y i
2

x, y 0 y i y
u y

y x0, y0

v x0, y0
3

x0, y0

v y

x0, y0

x where obtain,
i

,.

0 as z

0. Now, unleash the Cauchy-Riemann equations on this quotient and

f z0 x u x Here,
u x

z f z0 z i v i y u i x x x i y stuff . i v x i y x x
1

v x

stuff x i y

stuff Its easy to show that

lim stuff z z 0 and so, lim

z 0

0,

f z0

z z

f z0

u x

i v. x

In particular we have, as promised, shown that f is differentiable at z 0 .

Example Lets find all points at which the function f given by f z x 3 i 1 y 3 is differentiable. Here we have u x 3 and v 1 y 3 . The Cauchy-Riemann equations thus look like 3x 2 0 31 0. y 2 , and

2.11

The partial derivatives of u and v are nice and continuous everywhere, so f will be differentiable everywhere the C-R equations are satisfied. That is, everywhere x2 x 1 1 y 2 ; that is, where y, or x 1 y.

This is simply the set of all points on the cross formed by the two straight lines

4 3 2 1 -3 -2 -1 0 -1 -2 1 2 3

Exercises 10. At what points is the function f given by f z x3 i1 y

3

analytic? Explain.

11. Do the real and imaginary parts of the function f in Exercise 9 satisfy the Cauchy-Riemann equations at z 0? What do you make of your answer? 12. Find all points at which f z 2y ix is differentiable. 0 for all z D. Prove that f

13. Suppose f is analytic on a connected open set D, and f z is constant. 14. Find all points at which fz x
2

y x
2

y2

is differentiable. At what points is f analytic? Explain. 15. Suppose f is analytic on the set D, and suppose Re f is constant on D. Is f necessarily

2.12

constant on D? Explain. 16. Suppose f is analytic on the set D, and suppose |f z | is constant on D. Is f necessarily constant on D? Explain.

2.13

Chapter Three

Elementary Functions
3.1. Introduction. Complex functions are, of course, quite easy to come bythey are simply ordered pairs of real-valued functions of two variables. We have, however, already seen enough to realize that it is those complex functions that are differentiable that are the most interesting. It was important in our invention of the complex numbers that these new numbers in some sense included the old real numbersin other words, we extended the reals. We shall find it most useful and profitable to do a similar thing with many of the familiar real functions. That is, we seek complex functions such that when restricted to the reals are familiar real functions. As we have seen, the extension of polynomials and rational functions to complex functions is easy; we simply change xs to zs. Thus, for instance, the function f defined by fz z2 z 1

has a derivative at each point of its domain, and for z rational function fx x2 x 1.

0i, becomes a familiar real

What happens with the trigonometric functions, exponentials, logarithms, etc., is not so obvious. Let us begin. 3.2. The exponential function. Let the so-called exponential function exp be defined by exp z e x cos y i sin y ,

where, as usual, z x iy. From the Cauchy-Riemann equations, we see at once that this function has a derivative every whereit is an entire function. Moreover, d exp z dz Note next that if z x iy and w u iv, then exp z .

3.1

exp z

ex

cos y

i sin y

v i sin y cos v i sin v cos y sin v

e x e u cos y cos v e x e u cos y exp z exp w .

sin y sin v

i sin y cos v

We thus use the quite reasonable notation e z exp z and observe that we have extended the real exponential e x to the complex numbers. Example Recall from elementary circuit analysis that the relation between the voltage drop V and the current flow I through a resistor is V RI, where R is the resistance. For an inductor, the I, where C is relation is V L dI , where L is the inductance; and for a capacitor, C dV dt dt the capacitance. (The variable t is, of course, time.) Note that if V is sinusoidal with a frequency , then so also is I. Suppose then that V A sin t . We can write this as V Im Ae i e i t Im Be i t , where B is complex. We know the current I will have this same form: I Im Ce i t . The relations between the voltage and the current are linear, and so we can consider complex voltages and currents and use the fact that e i t cos t i sin t. We thus assume a more or less fictional complex voltage V , the imaginary part of which is the actual voltage, and then the actual current will be the imaginary part of the resulting complex current. What makes this a good idea is the fact that differentiation with respect to time t becomes d simply multiplication by i : dt Ae i t i Ae i t . If I be i t , the above relations between I current and voltage become V i LI for an inductor, and i VC I, or V for a i C capacitor. Calculus is thereby turned into algebra. To illustrate, suppose we have a simple RLC circuit with a voltage source V a sin t. We let E ae iwt .

Then the fact that the voltage drop around a closed circuit must be zero (one of Kirchoffs celebrated laws) looks like 3.2

i LI i Lb Thus, b In polar form, b

I i C b i C

RI Rb

ae i t , or a

a L

1 C

a R2 L
1 C 2

ei ,

where tan Hence, L R

1 C

. (R

0)

Im be i

Im R a
2

a L t
1 C 2

ei

1 C

sin

This result is well-known to all, but I hope you are convinced that this algebraic approach afforded us by the use of complex numbers is far easier than solving the differential equation. You should note that this method yields the steady state solutionthe transient solution is not necessarily sinusoidal. Exercises 1. Show that exp z 2. Show that
exp z exp w

2 i exp z

exp z . w. y 2k for any arg exp z and some

3. Show that |exp z |

e x , and arg exp z 3.3

integer k. 4. Find all z such that exp z 5. Find all z such that exp z 1 1, or explain why there are none. i, or explain why there are none. w have solutions? Explain.

6. For what complex numbers w does the equation exp z 7. Find the indicated mesh currents in the network:

3.3 Trigonometric functions. Define the functions cosine and sine as follows: cos z sin z where we are using e z exp z . e iz e iz e
iz iz

e 2i

First, lets verify that these are honest-to-goodness extensions of the familiar real functions, cosine and sineotherwise we have chosen very bad names for these complex functions. So, suppose z x 0i x. Then, e ix e Thus,
ix

cos x cos x

i sin x, and i sin x.

3.4

cos x sin x and everything is just fine.

e ix e ix

ix ix

, ,

e 2i

Next, observe that the sine and cosine functions are entirethey are simply linear combinations of the entire functions e iz and e iz . Moreover, we see that d sin z dz just as we would hope. It may not have been clear to you back in elementary calculus what the so-called hyperbolic sine and cosine functions had to do with the ordinary sine and cosine functions. Now perhaps it will be evident. Recall that for real t, sinh t Thus, sin it Similarly, cos it How nice! Most of the identities you learned in the 3 rd grade for the real sine and cosine functions are also valid in the general complex case. Lets look at some. sin 2 z cos 2 z 1 e iz e iz 2 e iz e iz 2 4 1 e 2iz 2e iz e iz e 2iz e 2iz 4 1 2 2 1 4 cosh t. e i it e 2i
i it

cos z, and d cos z dz

sin z,

et

e t , and cosh t

et

e t.

ie

i sinh t.

2e iz e

iz

2iz

3.5

It is also relative straight-forward and easy to show that: sin z cos z w w sin z cos w cos z cos w cos z sin w, and sin z sin w

Other familiar ones follow from these in the usual elementary school trigonometry fashion. Lets find the real and imaginary parts of these functions: sin z sin x iy sin x cos iy i cos x sinh y. i sin x sinh y. cos x sin iy

sin x cosh y In the same way, we get cos z

cos x cosh y

Exercises 8. Show that for all z, a)sin z 2 sin z; 9. Show that |sin z| 2 sin 2 x

b)cos z

cos z; cos 2 x

c)sin z sinh 2 y.

cos z.

sinh 2 y and |cos z| 2 0.

10. Find all z such that sin z 11. Find all z such that cos z

2, or explain why there are none.

3.4. Logarithms and complex exponents. In the case of real functions, the logarithm function was simply the inverse of the exponential function. Life is more complicated in the complex caseas we have seen, the complex exponential function is not invertible. There are many solutions to the equation e z w. If z 0, we define log z by log z ln|z| i arg z.

There are thus many log zs; one for each argument of z. The difference between any two of these is thus an integral multiple of 2 i. First, for any value of log z we have

3.6

e log z

e ln |z|

i arg z

e ln |z| e i arg z

z.

This is familiar. But next there is a slight complication: log e z ln e x z where k is an integer. We also have log zw ln |z||w| ln |z| log z for some integer k. There is defined a function, called the principal logarithm, or principal branch of the logarithm, function, given by Log z ln|z| iArg z, i arg zw ln |w| 2k i i arg w 2k i i arg e z x y 2k i

2k i,

i arg z log w

where Arg z is the principal argument of z. Observe that for any log z, it is true that log z Log z 2k i for some integer k which depends on z. This new function is an extension of the real logarithm function: Log x ln x iArg x ln x.

This function is analytic at a lot of places. First, note that it is not defined at z 0, and is not continuous anywhere on the negative real axis (z x 0i, where x 0.). So, lets suppose z 0 x 0 iy 0 , where z 0 is not zero or on the negative real axis, and see about a derivative of Log z : lim
z z0

Log z z

Log z 0 z0

lim
z z0

Log z e Log z

Log z 0 . e Log z 0 w 0 as z z 0 , this becomes

Now if we let w Log z and w 0 Log z 0 , and notice that w

3.7

lim
z z0

Log z z

Log z 0 z0

w w0

lim

w ew 1 z0 .

w0 e w0

1 e w0 Thus, Log is differentiable at z 0 , and its derivative is

1 z0

We are now ready to give meaning to z c , where c is a complex number. We do the obvious and define zc e c log z .

There are many values of log z, and so there can be many values of z c . As one might guess, e cLog z is called the principal value of z c . Note that we are faced with two different definitions of z c in case c is an integer. Lets see if we have anything to unlearn. Suppose c is simply an integer, c n. Then zn e n log z e n Log z
i 2k i

e nLog z e 2kn

e nLog z

There is thus just one value of z n , and it is exactly what it should be: e nLog z |z| n e in arg z . It is easy to verify that in case c is a rational number, z c is also exactly what it should be. Far more serious is the fact that we are faced with conflicting definitions of z c in case z e. In the above discussion, we have assumed that e z stands for exp z . Now we have a definition for e z that implies that e z can have many values. For instance, if someone runs at you in the night and hands you a note with e 1/2 written on it, how to you know whether this means exp 1/2 or the two values e and e ? Strictly speaking, you do not know. This ambiguity could be avoided, of course, by always using the notation exp z for e x e iy , but almost everybody in the world uses e z with the understanding that this is exp z , or equivalently, the principal value of e z . This will be our practice.

Exercises 12. Is the collection of all values of log i 1/2 the same as the collection of all values of 1 log i ? Explain. 2 13. Is the collection of all values of log i 2 the same as the collection of all values of 2 log i ? Explain.

3.8

14. Find all values of log z 1/2 . (in rectangular form) 15. At what points is the function given by Log z 2 16. Find the principal value of a) i i . 17. a)Find all values of |i i |. 1 analytic? Explain.

b) 1

4i

3.9

Chapter Four

Integration
4.1. Introduction. If integral integrals: :D C is simply a function on a real interval D , , then the

t dt is, of course, simply an ordered pair of everyday 3 rd grade calculus

t dt

x t dt

i y t dt,

where

xt

iy t . Thus, for example,

1

t2
0

it 3 dt

4 3

i. 4

Nothing really new here. The excitement begins when we consider the idea of an integral of an honest-to-goodness complex function f : D C, where D is a subset of the complex plane. Lets define the integral of such things; it is pretty much a straight-forward extension to two dimensions of what we did in one dimension back in Mrs. Turners class. Suppose f is a complex-valued function on a subset of the complex plane and suppose a and b are complex numbers in the domain of f. In one dimension, there is just one way to get from one number to the other; here we must also specify a path from a to b. Let C be a path from a to b, and we must also require that C be a subset of the domain of f.

4.1

Note we do not even require that a b; but in case a b, we must specify an orientation for the closed path C. We call a path, or curve, closed in case the initial and terminal points are the same, and a simple closed path is one in which no other points coincide. Next, let P be a partition of the curve; that is, P z 0 , z 1 , z 2 , , z n is a finite subset of C, such that a z 0 , b z n , and such that z j comes immediately after z j 1 as we travel along C from a to b.

A Riemann sum associated with the partition P is just what it is in the real case:
n

SP
j 1

f zj

zj,

where z j is a point on the arc between z j 1 and z j , and z j z j z j 1 . (Note that for a given partition P, there are many S P depending on how the points z j are chosen.) If there is a number L so that given any 0, there is a partition P of C such that |S P L|

whenever P P , then f is said to be integrable on C and the number L is called the integral of f on C. This number L is usually written f z dz.
C

Some properties of integrals are more or less evident from looking at Riemann sums: cf z dz
C

c f z dz
C

for any complex constant c.

4.2

fz
C

g z dz
C

f z dz
C

g z dz

4.2 Evaluating integrals. Now, how on Earth do we ever find such an integral? Let : , C be a complex description of the curve C. We partition C by partitioning the interval , in the usual way: t0 t1 t2 tn . Then a , t1 , t2 , , b is partition of C. (Recall we assume that t 0 for a complex description of a curve C.) A corresponding Riemann sum looks like
n

SP
j 1

tj

tj

tj

We have chosen the points z j sum by 1 in disguise:

n

t j , where t j

tj

t j . Next, multiply each term in the

SP
j 1

tj

tj tj

tj

tj
1

tj

tj

I hope it is now reasonably convincing that in the limit, we have

f z dz
C

t dt.

(We are, of course, assuming that the derivative

exists.)

Example We shall find the integral of f z x 2 y i xy from a different paths, or contours, as some call them. 0 to b 1 i along three

First, let C 1 be the part of the parabola y x 2 connecting the two points. A complex description of C 1 is 1 t t it 2 , 0 t 1:

4.3

1 0.8 0.6 0.4 0.2

0.2

0.4

0.6

0.8

Now,

2ti, and f

t2
1

t2

itt 2

2t 2

it 3 . Hence,

f z dz
C1 0 1

t dt

2t 2
0 1

it 3 1

2ti dt

2t 2
0

2t 4 5i 4

5t 3 i dt

4 15

Next, lets integrate along the straight line segment C 2 joining 0 and 1
1 0.8 0.6 0.4 0.2

i.

0.2

0.4

0.6

0.8

Here we have

it, 0

1. Thus,

i, and our integral looks like

4.4

f z dz
C2 0 1

t dt

t2
0 1

it 2 1

i dt

t
0

it 7i 6

2t 2 dt

1 2

Finally, lets integrate along C 3 , the path consisting of the line segment from 0 to 1 together with the segment from 1 to 1 i.
1 0.8 0.6 0.4 0.2

0.2

0.4

0.6

0.8

We shall do this in two parts: C 31 , the line from 0 to 1 ; and C 32 , the line from 1 to 1 Then we have f z dz
C3 C 31

i.

f z dz
C 32

f z dz.

For C 31 we have

t, 0

1. Hence,
1

f z dz
C 31 0

t 2 dt

1. 3

For C 32 we have

it, 0

t
1

1. Hence,

f z dz
C 32 0

it idt

1 2

3 i. 2

4.5

Thus, f z dz
C3 C 31

f z dz
C 32

f z dz

1 6 Suppose there is a number M so that |f z |

3 i. 2

M for all z C. Then

f z dz
C

t dt

|f

t |dt

M |

t |dt

ML,

where L Exercises

t |dt is the length of C.

1. Evaluate the integral

C 1 z C

z dz, where C is the parabola y

x 2 from 0 to 1

i.

2. Evaluate

dz, where C is the circle of radius 2 centered at 0 oriented

counterclockwise. 4. Evaluate f z dz, where C is the curve y

C

x 3 from 1

i to 1

i , and

fz

for y

0 0

4y for y

5. Let C be the part of the circle t small an upper bound as you can for

e it in the first quadrant from a z 2 z 4 5 dz .

1 to b

i. Find as

4.6

6. Evaluate f z dz where f z
C

2 z and C is the path from z

0 to z

2i 2i.

consisting of the line segment from 0 to 1 together with the segment from 1 to 1

4.3 Antiderivatives. Suppose D is a subset of the reals and : D C is differentiable at t. Suppose further that g is differentiable at t . Then lets see about the derivative of the composition g t . It is, in fact, exactly what one would guess. First, g where g z u x, y d g dt t u x t ,y t t xt iv x t , y t , iy t . Then, i v dx x dt v dy . y dt

iv x, y and t

u dx x dt

u dy y dt

The places at which the functions on the right-hand side of the equation are evaluated are obvious. Now, apply the Cauchy-Riemann equations: d g dt t u dx v dy i v dx x dt x dt x dt u i v dx i dy dt dt x x g t t. u dy x dt

The nicest result in the world! f z in D. Suppose moreover Now, back to integrals. Let F : D C and suppose F z that a and b are in D and that C D is a contour from a to b. Then

f z dz
C

t dt,

where : d F t dt

, F

C describes C. From our introductory discussion, we know that t t f t t . Hence,

4.7

f z dz
C

t dt

d F dt Fb

t dt Fa.

This is very pleasing. Note that integral depends only on the points a and b and not at all on the path C. We say the integral is path independent. Observe that this is equivalent to saying that the integral of f around any closed path is 0. We have thus shown that if in D the integrand f is the derivative of a function F, then any integral f z dz for C D is path independent. Example Let C be the curve y
1 x2 C

from the point z

1 z 2 dz.

i to the point z

i 9

. Lets find

This is easywe know that F z z 2 dz

C

z 2 , where F z 1 3 1 260 27 i
3

1 3

z 3 . Thus, i 9
3

728 i 2187

Now, instead of assuming f has an antiderivative, let us suppose that the integral of f between any two points in the domain is independent of path and that f is continuous. Assume also that every point in the domain D is an interior point of D and that D is connected. We shall see that in this case, f has an antiderivative. To do so, let z 0 be any point in D, and define the function F by Fz
Cz

f z dz,

where C z is any path in D from z 0 to z. Here is important that the integral is path independent, otherwise F z would not be well-defined. Note also we need the assumption that D is connected in order to be sure there always is at least one such path.

4.8

Now, for the computation of the derivative of F: Fz z Fz

L
z

f s ds,

where L

is the line segment from z to z

z.

Next, observe that

L
z

ds Fz

z. Thus, f z z z Fz

1 z

f z ds, and we have

L
z

fz

1 z

fs
L
z

f z ds.

Now then, 1 z fs
L
z

f z ds

1 | z| max |f s z max |f s

fz |:s L
z

fz |:s L fz |:s L

.
z

We know f is continuous at z, and so lim max |f s

z 0

0. Hence,

lim
z 0

Fz

z z

Fz

fz

lim
z 0

1 z

fs
L
z

f z ds

0.

4.9

In other words, F z f z , and so, just as promised, f has an antiderivative! Lets summarize what we have shown in this section: Suppose f : D C is continuous, where D is connected and every point of D is an interior point. Then f has an antiderivative if and only if the integral between any two points of D is path independent.

Exercises 7. Suppose C is any curve from 0 to 2i. Evaluate the integral cos z dz. 2

5 1 8. a)Let F z log z, 3 arg z . Show that the derivative F z z . 4 4 7 1 arg z . Show that the derivative G z b)Let G z log z, 4 z . 4 z : Re z 0 from i to i that does not c)Let C 1 be a curve in the right-half plane D 1 pass through the origin. Find the integral

1 dz. z
C1

d)Let C 2 be a curve in the left-half plane D 2 pass through the origin. Find the integral. 1 dz. z
C2

z : Re z

0 from i to i that does not

9. Let C be the circle of radius 1 centered at 0 with the clockwise orientation. Find 1 dz. z
C

10. a)Let H z zc, arg z . Find the derivative H z . 7 c b)Let K z z, 4 arg z . Find the derivative K z . 4 c)Let C be any path from 1 to 1 that lies completely in the upper half-plane and does not pass through the origin. (Upper half-plane z : Im z 0 .) Find 4.10

F z dz,
C

where F z

zi,

arg z

11. Suppose P is a polynomial and C is a closed curve. Explain how you know that P z dz 0.
C

4.11

Chapter Five

Cauchys Theorem
5.1. Homotopy. Suppose D is a connected subset of the plane such that every point of D is an interior pointwe call such a set a regionand let C 1 and C 2 be oriented closed curves in D. We say C 1 is homotopic to C 2 in D if there is a continuous function H : S D, where S is the square S t, s : 0 s, t 1 , such that H t, 0 describes C 1 and H t, 1 describes C 2 , and for each fixed s, the function H t, s describes a closed curve C s in D. The function H is called a homotopy between C 1 and C 2 . Note that if C 1 is homotopic to C 2 in D, then C 2 is homotopic to C 1 in D. Just observe that the function K t, s H t, 1 s is a homotopy. It is convenient to consider a point to be a closed curve. The point c is a described by a constant function t c. We thus speak of a closed curve C being homotopic to a constantwe sometimes say C is contractible to a point. Emotionally, the fact that two closed curves are homotopic in D means that one can be continuously deformed into the other in D.

Example Let D be the annular region D z : 1 |z| 5 . Suppose C 1 is the circle described by i2 t 2e , 0 t 1; and C 2 is the circle described by 2 t 4e i2 t , 0 t 1. Then 1 t i2 t H t, s 2 2s e is a homotopy in D between C 1 and C 2 . Suppose C 3 is the same circle as C 2 but with the opposite orientation; that is, a description is given by 4e i2 t , 0 t 1. A homotopy between C 1 and C 3 is not too easy to constructin 3 t fact, it is not possible! The moral: orientation counts. From now on, the term closed curve will mean an oriented closed curve.

5.1

Another Example Let D be the set obtained by removing the point z 0 from the plane. Take a look at the picture. Meditate on it and convince yourself that C and K are homotopic in D, but and are homotopic in D, while K and are not homotopic in D.

Exercises 1. Suppose C 1 is homotopic to C 2 in D, and C 2 is homotopic to C 3 in D. Prove that C 1 is homotopic to C 3 in D. 2. Explain how you know that any two closed curves in the plane C are homotopic in C. 3. A region D is said to be simply connected if every closed curve in D is contractible to a point in D. Prove that any two closed curves in a simply connected region are homotopic in D.

5.2 Cauchys Theorem. Suppose C 1 and C 2 are closed curves in a region D that are homotopic in D, and suppose f is a function analytic on D. Let H t, s be a homotopy between C 1 and C 2 . For each s, the function s t describes a closed curve C s in D. Let I s be given by Is
Cs

f z dz.

Then,

5.2

Is
0

f H t, s

H t, s dt. t

Now lets look at the derivative of I s . We assume everything is nice enough to allow us to differentiate under the integral:
1

I s

d ds
1

f H t, s
0

H t, s dt t H t, s t H t, s t dt f H t, s
2

f H t, s
0 1

H t, s s H t, s s H t, s s

H t, s s t H t, s t s

dt

f H t, s
0 1

f H t, s

dt

f H t, s

H 1, s H 0, s f H 0, s . s s But we know each H t, s describes a closed curve, and so H 0, s f H 1, s I s f H 1, s H 1, s s f H 0, s I 1 , or f z dz.

C2

H 1, s for all s. Thus, 0,

H 0, s s

which means I s is constant! In particular, I 0 f z dz

C1

This is a big deal. We have shown that if C 1 and C 2 are closed curves in a region D that are homotopic in D, and f is analytic on D, then f z dz f z dz.
C1 C2

An easy corollary of this result is the celebrated Cauchys Theorem, which says that if f is analytic on a simply connected region D, then for any closed curve C in D,

5.3

f z dz
C

0.

In court testimony, one is admonished to tell the truth, the whole truth, and nothing but the truth. Well, so far in this chapter, we have told the truth and nothing but the truth, but we have not quite told the whole truth. We assumed all sorts of continuous derivatives in the preceding discussion. These are not always necessaryspecifically, the results can be proved true without all our smoothness assumptionsthink about approximation.

Example Look at the picture below and convince your self that the path C is homotopic to the closed path consisting of the two curves C 1 and C 2 together with the line L. We traverse the line twice, once from C 1 to C 2 and once from C 2 to C 1 .

Observe then that an integral over this closed path is simply the sum of the integrals over C 1 and C 2 , since the two integrals along L , being in opposite directions, would sum to zero. Thus, if f is analytic in the region bounded by these curves (the region with two holes in it), then we know that f z dz
C C1

f z dz
C2

f z dz.

Exercises 4. Prove Cauchys Theorem. 5. Let S be the square with sides x orientation. Find 5.4 100, and y 100 with the counterclockwise

1 dz. z
S

6. a)Find
C

1 z 1

dz, where C is any circle centered at z

1 with the usual counterclockwise 1 with the usual counterclockwise

orientation: t 1 Ae 2 it , 0 t 1. b)Find z 11 dz, where C is any circle centered at z orientation. c)Find

C 1 z2 1 C

dz, where C is the ellipse 4x 2

y2

100 with the counterclockwise

orientation. [Hint: partial fractions] d)Find orientation.

C 1 z2 1

dz, where C is the circle x 2

10x

y2

0 with the counterclockwise

8. Evaluate Log z
C 1 zn

3 dz, where C is the circle |z|

2 oriented counterclockwise. e 2 it , 0

9. Evaluate integer 1.
C

dz where C is the circle described by

1, and n is an

1 10. a)Does the function f z z have an antiderivative on the set of all z 1 , n an integer 1 ? b)How about f z zn

0? Explain.

11. Find as large a set D as you can so that the function f z on D.

e z have an antiderivative

12. Explain how you know that every function analytic in a simply connected (cf. Exercise 3) region D is the derivative of a function analytic in D.

5.5

Chapter Six

More Integration
6.1. Cauchys Integral Formula. Suppose f is analytic in a region containing a simple closed contour C with the usual positive orientation and its inside , and suppose z 0 is inside C. Then it turns out that f z0 1 2 i fz z 0 dz.

z
C

This is the famous Cauchy Integral Formula. Lets see why its true. Let 0 be any positive number. We know that f is continuous at z 0 and so there is a number such that |f z f z 0 | whenever |z z 0 | . Now let 0 be a number z : |z z 0 | is also inside C. Now, the function such that and the circle C 0 fz z z 0 is analytic in the region between C and C 0 ; thus z
C

fz z 0 dz
C0

fz z 0 dz.

We know that
C0

1 z z0

dz

2 i, so we can write

z
C0

fz z 0 dz

2 if z 0
C0

fz z 0 dz

f z0
C0

z 0 dz

fz z
C0

f z0 z 0 dz.

For z C 0 we have fz z f z0 z0 |f z |z . f z0 | z0 |

Thus,

6.1

z
C0

fz z 0 dz

2 if z 0
C0

fz z 2

f z0 z 0 dz 2 .

But is any positive number, and so fz z 0 dz

z
C0

2 if z 0

0,

or, f z0 1 2 i z
C0

fz z 0 dz

1 2 i

z
C

fz z 0 dz,

which is exactly what we set out to show. Meditate on this result. It says that if f is analytic on and inside a simple closed curve and we know the values f z for every z on the simple closed curve, then we know the value for the function at every point inside the curvequite remarkable indeed. Example Let C be the circle |z| the integral 4 traversed once in the counterclockwise direction. Lets evaluate

z2

cos z dz. 6z 5

We simply write the integrand as cos z z 6z 5

2

z fz

cos z 5 z 1

fz , z 1

where cos z . z 5 Observe that f is analytic on and inside C, and so,

6.2

cos z dz z 6z 5
2

fz dz z 1

2 if 1 i cos 1 2

2 i cos 1 1 5 Exercises

1. Suppose f and g are analytic on and inside the simple closed curve C, and suppose moreover that f z g z for all z on C. Prove that f z g z for all z inside C. 2. Let C be the ellipse 9x 2 Define the function g by 4y 2 36 traversed once in the counterclockwise direction.

gz
C

s2

1 ds.

Find 3. Find

a) g i

b) g 4i

e 2z dz, z 4
2

where C is the closed curve in the picture:

4. Find

e 2z z2 4

dz, where

is the contour in the picture:

6.3

6.2. Functions defined by integrals. Suppose C is a curve (not necessarily a simple closed curve, just a curve) and suppose the function g is continuous on C (not necessarily analytic, just continuous). Let the function G be defined by Gz
C

gs s z ds

for all z Consider,

C. We shall show that G is analytic. Here we go.

Gz

z z

Gz

1 z s

s z

1 z gs z s

z z

z g s ds

ds.

Next, Gz z z Gz
C

gs s z

ds
C

z s

g s ds

s z s z z
C

s z z z s z 2 z
2

g s ds ds.

gs z z s

Now we want to show that

6.4

lim
z 0

z
C

gs z z s

ds

0.

To that end, let M max |g s | : s C , and let d be the shortest distance from z to C. Thus, for s C, we have |s z| d 0 and also z| |s z| | z| d | z|. |s z Putting this all together, we can estimate the integrand above: gs z z s M | z| d 2

s for all s C. Finally,

z
C

gs z z s

ds

| z|

M length C , | z| d 2

and it is clear that gs z z s

lim
z 0

z
C

ds

0,

just as we set out to show. Hence G has a derivative at z, and G z

C

gs s z

ds.

Truly a miracle! Next we see that G has a derivative and it is just what you think it should be. Consider

6.5

G z

z z

G z

1 z 1 z 1 z

s s s

z z

1
2

s
2

1 z

z
2

g s ds g s ds

s z

z s

2s z z z 2 2 s z z s z 2 z z z
2

g s ds

2s s z

g s ds

Next, G z z z 2s z z z
2

G z z z s z
2

2
C

gs s z s 2

ds g s ds z
2

s 2s

zs z 2s z z z 2 s z 3 zs

g s ds z 2 z
2

2s
C

z 2s z 2 4 zs s z z 2 s z 3
2 3

g s ds

3 zs s z

2 z 2 z s z

g s ds

Hence, G z z z G z gs s z 3 zs s z |3m| d z 2 z 2 z s z
2 3

2
C

ds
C

g s ds

| z| where m max |s z| : s

2| z| M , z 2d3

C . It should be clear then that G z z z G z gs s z

lim
z 0

2
C

ds

0,

or in other words,

6.6

G z

2
C

gs s z

ds.

Suppose f is analytic in a region D and suppose C is a positively oriented simple closed curve in D. Suppose also the inside of C is in D. Then from the Cauchy Integral formula, we know that 2 if z
C

fs s z ds

and so with g

f in the formulas just derived, we have f z 1 2 i fs s z ds, and f z 2 2 i fs s z ds

for all z inside the closed curve C. Meditate on these results. They say that the derivative of an analytic function is also analytic. Now suppose f is continuous on a domain D in which every point of D is an interior point and suppose that f z dz 0 for every closed curve in D. Then we know that f has an antiderivative in Din other words f is the derivative of an analytic function. We now know this means that f is itself analytic. We thus have the celebrated Moreras Theorem: If f:D C is continuous and such that f z dz
C C

0 for every closed curve in D, then f is

analytic in D. Example Lets evaluate the integral

e z dz, z3

where C is any positively oriented closed curve around the origin. We simply use the equation f z 2 2 i fs s z ds

6.7

with z

0 and f s

e s . Thus, ie 0 i
C

e z dz. z3

Exercises 5. Evaluate
C

sin z dz z2

where C is a positively oriented closed curve around the origin. 6. Let C be the circle |z a)
C 1 z2 4

i|

2 with the positive orientation. Evaluate b)

C 1 z2 4
2

dz

dz

7. Suppose f is analytic inside and on the simple closed curve C. Show that f z z w dz
C C

fz z w

dz

for every w 8. a) Let

C. t e it , t . Evaluate

be a real constant, and let C be the circle e z dz. z

C

b) Use your answer in part a) to show that

e
0

cos t

cos

sin t dt

6.3. Liouvilles Theorem. Suppose f is entire and bounded; that is, f is analytic in the entire plane and there is a constant M such that |f z | M for all z. Then it must be true that f z 0 identically. To see this, suppose that f w 0 for some w. Choose R large M enough to insure that R |f w |. Now let C be a circle centered at 0 and with radius 6.8

max R, |w| . Then we have : M 1 2 i fs w

|f w | 1 M2 2 2

ds

M,

0; or, in other a contradiction. It must therefore be true that there is no w for which f w 0 for all z. This, of course, means that f is a constant function. What we words, f z have shown has a name, Liouvilles Theorem: The only bounded entire functions are the constant functions. Lets put this theorem to some good use. Let p z polynomial. Then pz an an 1 z an 2 z2 anzn an 1zn
1

a1z

a 0 be a

a0 zn

zn.
a

|a n | Now choose R large enough to insure that for each j 1, 2, , n, we have znj j 2n whenever |z| R. (We are assuming that a n 0. ) Hence, for |z| R, we know that

|p z |

|a n | |a n | |a n |

an 1 z an 1 z |a n | 2n

an 2 z2 an 2 z2 |a n | 2n

a0 zn

|z| n a0 zn |z| n

|a n | |z| n 2n

|a n | n |z| . 2 Hence, for |z| R, 1 |p z | 2 |a n ||z| n 2 . |a n |R n

Now suppose p z 0 for all z. Then p 1z is also bounded on the disk |z| R. Thus, p 1z is a bounded entire function, and hence, by Liouvilles Theorem, constant! Hence the polynomial is constant if it has no zeros. In other words, if p z is of degree at least one, there must be at least one z 0 for which p z 0 0. This is, of course, the celebrated 6.9

Fundamental Theorem of Algebra. Exercises 9. Suppose f is an entire function, and suppose there is an M such that Re f z z. Prove that f is a constant function. 10. Suppose w is a solution of 5z 4 z3 z2 7z 14 0. Prove that |w| 0, then p z 3. z aqz, M for all

11. Prove that if p is a polynomial of degree n, and if p a where q is a polynomial of degree n 1. 12. Prove that if p is a polynomial of degree n pz where k 1 , k 2 , cz z1
k1

1, then z2
k2

z k1

zj k2

kj

, kj.

, k j are positive integers such that n

13. Suppose p is a polynomial with real coefficients. Prove that p can be expressed as a product of linear and quadratic factors, each with real coefficients.

6.4. Maximum moduli. Suppose f is analytic on a closed domain D. Then, being continuous, |f z | must attain its maximum value somewhere in this domain. Suppose this happens at an interior point. That is, suppose |f z | M for all z D and suppose that |f z 0 | M for some z 0 in the interior of D. Now z 0 is an interior point of D, so there is a number R such that the disk centered at z 0 having radius R is included in D. Let C be a positively oriented circle of radius R centered at z 0 . From Cauchys formula, we know f z0 Hence,
2

1 2 i

s
C

fs z 0 ds.

f z0 and so,

1 2

f z0
0

e it dt,

6.10

M since |f z 0 e |
it

|f z 0 |

1 2

|f z 0
0

e it |dt

M.

M. This means
2

1 2

|f z 0
0

e it |dt.

Thus,
2 2

1 2

|f z 0
0

e |dt

it

1 2

M
0

|f z 0

e it | dt

0.

This integrand is continuous and non-negative, and so must be zero. In other words, R, and so |f z | M for all z C. There was nothing special about C except its radius we have shown that f must be constant on the disk . I hope it is easy to see that if D is a region ( connected and open), then the only way in which the modulus |f z | of the analytic function f can attain a maximum on D is for f to be constant. Exercises 14. Suppose f is analytic and not constant on a region D and suppose f z Explain why |f z | does not have a minimum in D. 0 for all z D.

15. Suppose f z u x, y iv x, y is analytic on a region D. Prove that if u x, y attains a maximum value in D, then u must be constant.

6.11

Chapter Seven

Harmonic Functions
7.1. The Laplace equation. The Fourier law of heat conduction says that the rate at which heat passes across a surface S is proportional to the flux, or surface integral, of the temperature gradient on the surface: k
S

T dA.

Here k is the constant of proportionality, generally called the thermal conductivity of the substance (We assume uniform stuff. ). We further assume no heat sources or sinks, and we assume steady-state conditionsthe temperature does not depend on time. Now if we take S to be an arbitrary closed surface, then this rate of flow must be 0: k
S

T dA

0.

Otherwise there would be more heat entering the region B bounded by S than is coming out, or vice-versa. Now, apply the celebrated Divergence Theorem to conclude that T dV
B

0,

where B is the region bounded by the closed surface S. But since the region B is completely arbitrary, this means that T T x2
2

T y2

T z2

0.

This is the world-famous Laplace Equation. Now consider a slab of heat conducting material,

7.1

in which we assume there is no heat flow in the z-direction. Equivalently, we could assume we are looking at the cross-section of a long rod in which there is no longitudinal heat flow. In other words, we are looking at a two-dimensional problemthe temperature depends only on x and y, and satisfies the two-dimensional version of the Laplace equation: T x2
2

T y2

0.

Suppose now, for instance, the temperature is specified on the boundary of our region D, and we wish to find the temperature T x, y in region. We are simply looking for a solution of the Laplace equation that satisfies the specified boundary condition. Lets look at another physical problem which leads to Laplaces equation. Gausss Law of electrostatics tells us that the integral over a closed surface S of the electric field E is proportional to the charge included in the region B enclosed by S. Thus in the absence of any charge, we have E dA
S

0.

But in this case, we know the field E is conservative; let is, E Thus, E dA
S S

be the potential functionthat

dA.

Again, we call on the Divergence Theorem to conclude that must satisfy the Laplace equation. Mathematically, we cannot tell the problem of finding the electric potential in a 7.2

region D, given the potential on the boundary of D, from the previous problem of finding the temperature in the region, given the temperature on the boundary. These are but two of the many physical problems that lead to the Laplace equationYou probably already know of some others. Let D be a domain and let be a given function continuous on the boundary of D. The problem of finding a function harmonic on the interior of D and which agrees with on the boundary of D is called the Dirichlet problem. 7.2. Harmonic functions. If D is a region in the plane, a real-valued function u x, y having continuous second partial derivatives is said to be harmonic on D if it satisfies Laplaces equation on D : u x2
2

u y2

0.

There is an intimate relationship between harmonic functions and analytic functions. Suppose f is analytic on D, and let f z u x, y iv x, y . Now, from the Cauchy-Riemann equations, we know u x u y v , and y v. x

If we differentiate the first of these with respect to x and the second with respect to y, and then add the two results, we have u x2
2

u y2

v x y

v y x

0.

Thus the real part of any analytic function is harmonic! Next, if we differentiate the first of the Cauchy-Riemann equations with respect to y and the second with respect to x, and then subtract the second from the first, we have v x2
2

v y2

0,

and we see that the imaginary part of an analytic function is also harmonic. There is even more excitement. Suppose we are given a function harmonic in a simply connected region D. Then there is a function f analytic on D which is such that Re f . Lets see why this is so. First, define g by

7.3

gz

Well show that g is analytic by verifying that the real and imaginary parts satisfy the Cauchy-Riemann equations:
2 2 2

x since is harmonic. Next,

y2

y x

x y

Since g is analytic on the simply connected region D, we know that the integral of g around any closed curve is zero, and so it has an antiderivative G z u iv. This antiderivative is, of course, analytic on D, and we know that G z Thus, u x, y x, y u x i u y i .

h y . From this, u y h y, x, y c. In other

and so h y words, Re G Example

0, or h constant, from which it follows that u x, y u, as we promised to show.

The function x, y x 3 3xy 2 is harmonic everywhere. We shall find an analytic function G so that Re G . We know that G z x 3 3xy 2 iv, and so from the Cauchy-Riemann equations: v x u y 6xy

7.4

Hence, v x, y To find k y differentiate with respect to y : v y and so, k y ky 3y 2 , or y3 any constant. 3x 2 k y u x 3x 2 3y 2 , 3x 2 y ky.

If we choose the constant to be zero, this gives us v and finally, Gz u iv x3 3xy 2 i 3x 2 y y3 . 3x 2 y ky 3x 2 y y3,

Exercises 1. Suppose is harmonic on a simply connected region D. Prove that if maximum or its minimum value at some point in D, then is constant in D. assumes its

2. Suppose and are harmonic in a simply connected region D bounded by the curve C. Suppose moreover that x, y x, y for all x, y C. Explain how you know that everywhere in D. 3. Find an entire function f such that Re f function f. 4. Find an entire function f such that Re f function f. x2 3x y 2 , or explain why there is no such

x2

3x

y 2 , or explain why there is no such

7.5

7.3. Poissons integral formula. Let be the disk bounded by the circle . Suppose is harmonic on and let f be a function analytic on and z : |z| C , the function such that Re f . Now then, for fixed z with |z| gs fs z 2 sz

is analyic on . Thus from Cauchys Theorem g s ds

C C

fs z ds 2 sz

0.

We know also that fz 1 2 i fs s z ds.

C

Adding these two equations gives us fz 1 2 i 1 2 i 1 z f s ds

s
C

z
2

sz

|z| 2
2

sz

f s ds.

Next, let

e it , and our integral becomes

2 2

fz

1 2 i
2

|z| 2
2

e it
2

e it z

f e it i e it dt

|z| 2 2

e it

f e it z e

it

dt

|z| 2 2

f e it dt 2 | e it z|

Now, 7.6

x, y

Re f

|z| 2 2

e it dt. 2 | e it z|

Next, use polar coordinates: z

re i :
2

r,

r2

| e it

e it dt. 2 re i |

Now, | e it re i |
2

e it
2

re i r2

it

re

r2

r ei t

it

2r cos t

Substituting this in the integral, we have Poissons integral formula:

2

r,

r2

2 2 0

r2

e it 2r cos t

dt

This famous formula essentially solves the Dirichlet problem for a disk. Exercises
2

5. Evaluate
0

1 r 2 2r cos t

dt. [Hint: This is easy.]

6. Suppose is harmonic in a region D. If x 0 , y 0 D and if C D is a circle centered at x 0 , y 0 , the inside of which is also in D, then x 0 , y 0 is the average value of on the circle C. 7. Suppose is harmonic on the disk 0, 0 z : |z| 1 . Prove that dA.

7.7

Chapter Eight

Series
8.1. Sequences. The basic definitions for complex sequences and series are essentially the C from same as for the real case. A sequence of complex numbers is a function g : Z the positive integers into the complex numbers. It is traditional to use subscripts to indicate the values of the function. Thus we write g n z n and an explicit name for the sequence is seldom used; we write simply z n to stand for the sequence g which is such that i i gn z n . For example, n is the sequence g for which g n n . 0, there is an integer N such The number L is a limit of the sequence z n if given an that |z n L| for all n N . If L is a limit of z n , we sometimes say that z n L. It is relatively easy to see that if the converges to L. We frequently write lim z n u n iv n converges to L, then the two real sequences u n and complex sequence z n v n each have a limit: u n converges to Re L and v n converges to Im L. Conversely, if the two real sequences u n and v n each have a limit, then so also does the complex sequence u n iv n . All the usual nice properties of limits of sequences are thus true: lim z n wn lim z n lim w n ;

lim z n w n z lim wnn

lim z n lim w n ; and lim z n . lim w n

provided that lim z n and lim w n exist. (And in the last equation, we must, of course, insist that lim w n 0.) A necessary and sufficient condition for the convergence of a sequence a n is the celebrated Cauchy criterion: given 0, there is an integer N so that |a n a m | whenever n, m N . A sequence f n of functions on a domain D is the obvious thing: a function from the positive integers into the set of complex functions on D. Thus, for each z D, we have an ordinary sequence f n z . If each of the sequences f n z converges, then we say the sequence of functions f n converges to the function f defined by f z lim f n z . This pretty obvious stuff. The sequence f n is said to converge to f uniformly on a set S if for all n N and all z S. given an 0, there is an integer N so that |f n z f z | Note that it is possible for a sequence of continuous functions to have a limit function that is not continuous. This cannot happen if the convergence is uniform. To see this, suppose the sequence f n of continuous functions converges uniformly to f on a domain D, let z 0 D, and let 0. We need to show there is a so that |f z 0 f z | whenever 8.1

. Lets do it. First, choose N so that |f N z f z | . We can do this because |z 0 z| 3 of the uniform convergence of the sequence f n . Next, choose so that whenever |z 0 z| . This is possible because f N is continuous. |f N z 0 f N z | 3 Now then, when |z 0 z| , we have |f z 0 fz | |f z 0 |f z 0 3 and we have done it! Now suppose we have a sequence f n of continuous functions which converges uniformly on a contour C to the function f. Then the sequence is easy to see. Let length of C. Then, 0. Now let N be so that |f n z
C

fN z0 fN z0 | 3 3

fN z0 |f N z 0 ,

fN z fN z |

fN z |f N z

fz | fz |

f n z dz fz |
A

converges to f z dz. This for n N, where A is the

C

f n z dz
C C

f z dz
C

fn z A A

f z dz

whenever n

N.

Now suppose f n is a sequence of functions each analytic on some region D, and suppose the sequence converges uniformly on D to the function f. Then f is analytic. This result is in marked contrast to what happens with real functionsexamples of uniformly convergent sequences of differentiable functions with a nondifferentiable limit abound in the real case. To see that this uniform limit is analytic, let z 0 D, and let S z : |z z 0 | r D . Now consider any simple closed curve C S. Each f n is analytic, and so f n z dz 0 for every
C

n. From the uniform convergence of f n , we know that f z dz is the limit of the sequence
C

f n z dz , and so f z dz hence at z 0 . Truly a miracle.

C C

0. Moreras theorem now tells us that f is analytic on S, and

Exercises

8.2

1. Prove that a sequence cannot have more than one limit. (We thus speak of the limit of a sequence.) 2. Give an example of a sequence that does not have a limit, or explain carefully why there is no such sequence. 3. Give an example of a bounded sequence that does not have a limit, or explain carefully why there is no such sequence. 4. Give a sequence f n of functions continuous on a set D with a limit that is not continuous. 5. Give a sequence of real functions differentiable on an interval which converges uniformly to a nondifferentiable function.

8.2 Series. A series is simply a sequence s n in which s n a 1 a 2 a n . In other words, there is sequence a n so that s n s n 1 a n . The s n are usually called the partial sums. Recall from Mrs. Turners class that if the series true that lim a n
n n

aj

has a limit, then it must be

j 1

0.
n j 1

Consider a series

fj z

of functions. Chances are this series will converge for some

values of z and not converge for others. A useful result is the celebrated Weierstrass M-test: Suppose M j is a sequence of real numbers such that M j 0 for all j J, where J is some number., and suppose also that the series have |f j z | M j for all j J, then the series
n j 1 n

Mj

converges. If for all z D, we

j 1

fj z

converges uniformly on D. J so that

To prove this, begin by letting

0 and choosing N
n

Mj
j m

for all n, m that

N. (We can do this because of the famous Cauchy criterion.) Next, observe

8.3

fj z
j m j m

|f j z |
j m

Mj

This shows that

n j 1

fj z

converges. To see the uniform convergence, observe that

m 1

fj z
j m j 0

fj z
j 0

fj z

for all z D and n

N. Thus,
n m 1 m 1

lim
n j 0

fj z
j 0

fj z
j 0

fj z
j 0

fj z

for m

N.(The limit of a series

n j 0

aj

is almost always written as

j 0

a j .)

Exercises 6. Find the set D of all z for which the sequence 7. Prove that the series
n j 1 n j 1 n j 1 zn zn 3n

has a limit. Find the limit.

n j 1

aj

converges if and only if both the series

Re a j

and

Im a j

converge.

8. Explain how you know that the series |z| 5.

1 z

converges uniformly on the set

8.3 Power series. We are particularly interested in series of functions in which the partial sums are polynomials of increasing degree: sn z c0 c1 z z0 c2 z 8.4 z0
2

cn z

z0 n.

(We start with n

0 for esthetic reasons.) These are the so-called power series. Thus,
n j 0

a power series is a series of functions of the form

cj z

z0

Lets look first at a very special power series, the so-called Geometric series:
n

zj .
j 0

Here sn zs n 1 z z z2 z2 z3 z n , and zn 1.

Subtracting the second of these from the first gives us 1 z sn 1 zn 1.

If z 1, then we cant go any further with this, but I hope its clear that the series does not have a limit in case z 1. Suppose now z 1. Then we have sn Now if |z| 1 zn 1 . 1 z 0, and so 1 . z

z
1

1, it should be clear that lim z n

n

lim
j 0

zj

lim s n

Or, 1 , for |z| z

zj
j 0

1.

There is a bit more to the story. First, note that if |z| 1, then the Geometric series does not have a limit (why?). Next, note that if |z| 1, then the Geometric series converges 8.5

uniformly to

1 1 z

. To see this, note that

n j j 0

has a limit and appeal to the Weierstrass M-test. Clearly a power series will have a limit for some values of z and perhaps not for others. First, note that any power series has a limit when z z 0 . Lets see what else we can say. Consider a power series
n

cj z

z0

. Let lim sup

j

j 0

|c j | .

(Recall from 6 th grade that lim sup a k shall say R 0 if , and R if converges uniformly for all |z z 0 |

1 . (We lim sup a k : k n . ) Now let R 0. ) We are going to show that the series R and diverges for all |z z 0 | R.

First, lets show the series does not converge for |z 1 k 1 R

z0 | .

R. To begin, let k be so that

|z

z0 |

There are an infinite number of c j for which j |c j | each of these c j we have |c j z z0 j |

j

k, otherwise lim sup

|c j |

k. For

|c j | |z

z0 |

k|z

z0 |

1.

It is thus not possible for lim |c n z

n

z0 n |

0, and so the series does not converge. z0 | R. Let k be so

Next, we show that the series does converge uniformly for |z that 1 R Now, for j large enough, we have j |c j | k 1. z0 |

k. Thus for |z

, we have

8.6

|c j z
n j 0

z0 j |

|c j | |z

z0 |

k|z

z0 |

The geometric series of

n j 0

converges because k

1 and the uniform convergence

cj z

z0

follows from the M-test.

Example Consider the series

n j 0 1 j!

z j . Lets compute R

1/ lim sup

|c j |

lim sup

j! . Let
K 2K 2K !

K be any positive integer and choose an integer m large enough to insure that 2 m n! Now consider K n , where n 2K m: n! Kn 2K m ! 2K 2K m K 2K ! 2 m 2K 1 K m 2K m 1 K m K 2K 2K 1 2K !

K. Reflect on what we have just shown: given any number K, there is a Thus n n! , and so the number n such that n n! is bigger than it. In other words, R lim sup j j! series
n j 0 1 j!

zj

converges for all z.

n j 0

Lets summarize what we have. For any power series R

1 lim sup j |c j |

cj z

z0

, there is a number R and does not

such that the series converges uniformly for |z

z0 |

.) The number R is converge for |z z 0 | R. (Note that we may have R 0 or R called the radius of convergence of the series, and the set |z z 0 | R is called the circle of convergence. Observe also that the limit of a power series is a function analytic inside the circle of convergence (why?). Exercises 9. Suppose the sequence of real numbers
j

has a limit. Prove that

8.7

lim sup

lim

For each of the following, find the set D of points at which the series converges: 10.
n j 0 n j 0 n j 0 n j 0

j!z j .

11.

jz j .

12.

j2 3j

zj .

13.

1 j 2 2j j! 2

z 2j

8.4 Integration of power series. Inside the circle of convergence, the limit

Sz
j 0

cj z

z0

is an analytic function. We shall show that this series may be integrated term-by-termthat is, the integral of the limit is the limit of the integrals. Specifically, if C is any contour inside the circle of convergence, and the function g is continuous on C, then

g z S z dz
C j 0

cj g z z
C

z 0 j dz.

Lets see why this. First, let 0. Let M be the maximum of |g z | on C and let L be the length of C. Then there is an integer N so that

cj z
j n

z0

ML

8.8

for all n

N. Thus,

gz
C j n

cj z

z0

dz

ML

ML

Hence,
n 1

g z S z dz
C j 0

cj g z z
C

z 0 j dz
C

gz
j n

cj z

z0

dz

, and we have shown what we promised.

8.5 Differentiation of power series. Again, let

Sz
j 0

cj z

z0 j.

Now we are ready to show that inside the circle of convergence,

S z
j 1

jc j z

z0

j 1

Let z be a point inside the circle of convergence and let C be a positive oriented circle centered at z and inside the circle of convergence. Define gs 1 2 is z ,

and apply the result of the previous section to conclude that

8.9

g s S s ds
C j 0

cj g s s
C

z 0 j ds, or s z0 j ds. Thus s z 2

j 1

1 2 i

Ss s z

ds
j 0

cj 1 2 i jc j z

S z
j 0

z0

as promised! Exercises 14. Find the limit of

n

j
j 0

1 zj .

For what values of z does the series converge? 15. Find the limit of
n j 1

zj j

For what values of z does the series converge? 16. Find a power series
n j 0

cj z

such that

1 z

cj z
j 0

1 j , for |z

1|

1.

17. Find a power series

n j 0

cj z

such that

8.10

Log z
j 0

cj z

1 j , for |z

1|

1.

8.11

Chapter Nine

Taylor and Laurent Series

9.1. Taylor series. Suppose f is analytic on the open disk |z z 0 | r. Let z be any point in this disk and choose C to be the positively oriented circle of radius , where r. Then for s C we have |z z 0 | 1 1 1 1 z z0 j s z0 j 1

s since | z s
z0 z0

z0

z0

z0

z z0 s z0

j 0

1. The convergence is uniform, so we may integrate fs s z ds

C

j 0

fs s z0 1 2 i

j 1

ds fs s z0

z 0 j , or

fz

1 2 i

fs s z ds
C

j 0

j 1

ds

z0 j.

We have thus produced a power series having the given analytic function as a limit:

fz
j 0

cj z

z 0 j , |z

z0 |

r,

where cj 1 2 i fs z0 z0. ds.

j 1

This is the celebrated Taylor Series for f at z We know we may differentiate the series to get

f z
j 1

jc j z

z0

j 1

9.1

and this one converges uniformly where the series for f does. We can thus differentiate again and again to obtain

z
j n

jj

1 j

1 cj z

z0

j n

Hence, f
n

z0 cn

n!c n , or f
n

z0 . n!

But we also know that cn This gives us f

n

1 2 i

fs s z0

n 1

ds.

z0

n! 2 i

fs z0

n 1

ds, for n

0, 1, 2,

This is the famous Generalized Cauchy Integral Formula. Recall that we previously derived this formula for n 0 and 1. What does all this tell us about the radius of convergence of a power series? Suppose we have

fz
j 0

cj z

z0 j,

and the radius of convergence is R. Then we know, of course, that the limit function f is analytic for |z z 0 | R. We showed that if f is analytic in |z z 0 | r, then the series converges for |z z 0 | r. Thus r R, and so f cannot be analytic at any point z for which |z z 0 | R. In other words, the circle of convergence is the largest circle centered at z 0 inside of which the limit f is analytic.

9.2

Example Let f z exp z at z 0 0 is e z . Then f 0 f 0 f

n

1, and the Taylor series for f

ez
j 0

1 zj j!

and this is valid for all values of z since f is entire. (We also showed earlier that this particular series has an infinite radius of convergence.)

Exercises 1. Show that for all z, 1 z j!

ez

e
j 0

1 j.

2. What is the radius of convergence of the Taylor series 3. Show that 1 z i j 1 i j1

n j 0

cjzj

for tanh z ?

1 for |z i| 2.
1 1 z

j 0

4. If f z

, what is f

10

i ?

5. Suppose f is analytic at z 0 and f 0 f 0 f 0 0. Prove there is a function g 3 analytic at 0 such that f z z g z in a neighborhood of 0. 6. Find the Taylor series for f z sin z at z 0 0.

7. Show that the function f defined by 9.3

fz

sin z z

for z for z

0 0

is analytic at z

0, and find f 0 .

9.2. Laurent series. Suppose f is analytic in the region R 1 |z z 0 | R 2 , and let C be a positively oriented simple closed curve around z 0 in this region. (Note: we include the .) We shall show that for z C in this region possiblites that R 1 can be 0, and R 2 bj z z0

fz
j 0

aj z

z0

j j 1

where aj 1 2 i
C

fs z0

j 1

ds, for j

0, 1, 2,

and bj 1 2 i fs s z0 ds, for j 1, 2, .

j 1

The sum of the limits of these two series is frequently written

fz
j

cj z

z0 j,

where cj 1 2 i fs s z0 ds, j 0, 1, 2, .

j 1

This recipe for f z is called a Laurent series, although it is important to keep in mind that it is really two series.

9.4

Okay, now lets derive the above formula. First, let r 1 and r 2 be so that R 1 r 1 |z z 0 | r 2 R 2 and so that the point z and the curve C are included in the region r 1 |z z 0 | r 2 . Also, let be a circle centered at z and such that is included in this region.

Then fs sz is an analytic function (of s) on the region bounded by C 1 , C 2 , and , where C 1 is the circle |z| r 1 and C 2 is the circle |z| r 2 . Thus, fs s z ds
C2 C1

fs s z ds

fs s z ds.

(All three circles are positively oriented, of course.) But

fs s z

ds

2 if z , and so we have

2 if z
C2

fs s z ds
C1

fs s z ds.

Look at the first of the two integrals on the right-hand side of this equation. For s C 2 , we have |z z 0 | |s z 0 |, and so s 1 z 1

s s s 1 1

z0 z0 z0

z 1

z0
z z0 s z0

1 z s

j 0

z0 z0 z

j 0

1 z0

j 1

z0 j.

9.5

Hence, fs s z ds
C2

j 0

C2

fs s z0 fs s z0

j 1

ds

z0

.
j 0 C

j 1

ds

z0

For the second of these two integrals, note that for s C 1 we have |s 1 1 1 z0 1 1 s
j 0

z0 |

|z

z 0 |, and so

z z

z0 1 z0 s

s s z
j 1

z0 z0 z0
j

s z0 z z0 j

z0

j 0

z s 1 z0

1 z0
j 1

j 1

z0

j 1

1 z0

j 1

1 z0

As before, fs s z ds
C1

j 1

C1

fs s z0 fs s z0

j 1

ds

1 z0

j 1

j 1

ds

1 z z0

Putting this altogether, we have the Laurent series: fz 1 2 i fs s z ds

C2

1 2 i

fs s z ds
C1

j 0

1 2 i

fs s z0

j 1

ds

z0

j j 1

1 2 i

fs s z0

j 1

ds

1 z0

Example

9.6

Let f be defined by fz 1 . 1. Lets find the Laurent series for f

zz

First, observe that f is analytic in the region 0 valid in this region. First, fz 1

|z|

zz

1 z

1 . 1

From our vast knowledge of the Geometric series, we have 1 z

fz

zj.
j 0

Now lets find another Laurent series for f, the one valid for the region 1 First, z Now since | 1 | z 1, we have 1 1 z 1 1 1 z 1 1 1 z 1 1
1 z

|z|

z and so

1 z

z
j 0

j j 1

z j,

fz fz

1 z z j.
j 2

1 z

z
j 1

Exercises 8. Find two Laurent series in powers of z for the function f defined by

9.7

fz

1 z2 1 z

and specify the regions in which the series converge to f z .

9. Find two Laurent series in powers of z for the function f defined by fz 1

z1

z2

and specify the regions in which the series converge to f z . 10. Find the Laurent series in powers of z 1 for f z
1 z

in the region 1

|z

1|

9.8

Chapter Ten

Poles, Residues, and All That

10.1. Residues. A point z 0 is a singular point of a function f if f not analytic at z 0 , but is analytic at some point of each neighborhood of z 0 . A singular point z 0 of f is said to be isolated if there is a neighborhood of z 0 which contains no singular points of f save z 0 . In . other words, f is analytic on some region 0 |z z 0 | Examples The function f given by fz has isolated singular points at z 0, z 1

zz

4 2i.

2i, and z

Every point on the negative real axis and the origin is a singular point of Log z , but there are no isolated singular points. Suppose now that z 0 is an isolated singular point of f . Then there is a Laurent series

fz
j

cj z

z0

valid for 0 |z z 0 | R, for some positive R. The coefficient c residue of f at z 0 , and is frequently written Res f.
z z0

of z

z0

is called the

Now, why do we care enough about c 1 to give it a special name? Well, observe that if C is any positively oriented simple closed curve in 0 |z z 0 | R and which contains z 0 inside, then c 1 2 i f z dz.
C

10.1

This provides the key to evaluating many complex integrals. Example We shall evaluate the integral e 1/z dz
C

where C is the circle |z| 1 with the usual positive orientation. Observe that the integrand has an isolated singularity at z 0. We know then that the value of the integral is simply 2 i times the residue of e 1/z at 0. Lets find the Laurent series about 0. We already know that 1 zj j!

ez
j 0

for all z . Thus, 1z j! 1 z 1 1 2! z 2

e 1/z
j 0

The residue c

1, and so the value of the integral is simply 2 i.

Now suppose we have a function f which is analytic everywhere except for isolated singularities, and let C be a simple closed curve (positively oriented) on which f is analytic. Then there will be only a finite number of singularities of f inside C (why?). Call them z 1 , z 2 , , z n . For each k 1, 2, , n, let C k be a positively oriented circle centered at z k and with radius small enough to insure that it is inside C and has no other singular points inside it.

10.2

Then, f z dz
C C1

f z dz
C2 z z1 n

f z dz
Cn z z2

f z dz 2 i Res f
z zn

2 i Res f 2 i
k 1

2 i Res f

Res f.
z zk

This is the celebrated Residue Theorem. It says that the integral of f is simply 2 i times the sum of the residues at the singular points enclosed by the contour C. Exercises Evaluate the integrals. In each case, C is the positively oriented circle |z| 1.
C

2.

e 1/z dz.
1 z

2.
C

sin

dz.

3.
C

cos
1 z C

1 z

dz.
1 z

4.

sin

dz.

5.
C

1 z

cos

1 z

dz.

10.3

10.2. Poles and other singularities. In order for the Residue Theorem to be of much help in evaluating integrals, there needs to be some better way of computing the residuefinding the Laurent expansion about each isolated singular point is a chore. We shall now see that in the case of a special but commonly occurring type of singularity the residue is easy to find. Suppose z 0 is an isolated singularity of f and suppose that the Laurent series of f at z 0 contains only a finite number of terms involving negative powers of z z 0 . Thus, fz c z
n n

c z z0

z0

n 1

c
n 1

z0
n

z0

c0

c1 z

z0

Multiply this expression by z z z z0 nf z

:
n

n 1

z0

z z

z0

n 1

What we see is the Taylor series at z 0 for the function z of z z 0 n 1 is what we seek, and we know that this is
n 1

z 0 n f z . The coefficient

n . The sought after residue c

1

z0 1!

is thus c Res f
z z0 n 1

z0 , 1!

where

z0 nf z .

Example We shall find all the residues of the function fz ez . z z2 1

2

First, observe that f has isolated singularities at 0, and i. Lets see about the residue at 0. Here we have

10.4

z The residue is simply 0 : z Hence,

z2f z

ez

z2

1 e z 2ze z . z2 1 2

Res f
z 0

1.

Next, lets see what we have at z z and so

i: z ifz ez , z z i
2

Res f z
z i

ei . 2i

In the same way, we see that Res f

z ez z2 1 i

e i. 2i

Lets find the integral

C

z2

dz , where C is the contour pictured:

10.5

This is now easy. The contour is positive oriented and encloses two singularities of f; viz, i and i. Hence, ez dz z2 z2 1 2 i Res f Res f
z i z i

2 i

ei 2i 2 i sin 1.

e i 2i

Miraculously easy! There is some jargon that goes with all this. An isolated singular point z 0 of f such that the Laurent series at z 0 includes only a finite number of terms involving negative powers of z z0 nf z z z 0 is called a pole. Thus, if z 0 is a pole, there is an integer n so that z 0. The number n is called the order of the pole. Thus, in the is analytic at z 0 , and f z 0 preceding example, 0 is a pole of order 2, while i and i are poles of order 1. (A pole of order 1 is frequently called a simple pole.) We must hedge just a bit here. If z 0 is an isolated singularity of f and there are no Laurent series terms involving negative powers of z z 0 , then we say z 0 is a removable singularity. Example Let then the singularity z sin z ; z 0 is a removable singularity: fz fz 1 sin z z 2 1 z 3! 1 z z z4 5! z3 3! z5 5!

and we see that in some sense f is really analytic at z the right thing there.

0 if we would just define it to be

A singularity that is neither a pole or removable is called an essential singularity. Lets look at one more labor-saving trickor technique, if you prefer. Suppose f is a function:

10.6

fz

pz , qz 0, while q z 0 0, and p z 0 0.

where p and q are analytic at z 0 , and we have q z 0 Then fz pz qz p z0 q z0 z p z0 z z0

q z0 2

z0 z

z0

and so z z z0 f z p z0 q z0 p z0 z z0 q z0 z z0 2 .

Thus z 0 is a simple pole and Res f

z z0

z0

p z0 . q z0

Example Find the integral cos z dz, ez 1

where C is the rectangle with sides x 1, y , and y 3 . The singularities of the integrand are all the places at which e z 1, or in other words, the points z 0, 2 i, 4 i, . The singularities enclosed by C are 0 and 2 i. Thus, cos z dz ez 1 2 i Res f Res f ,
z 0 z 2 i

where fz cos z . ez 1

10.7

Observe this is precisely the situation just discussed: f z analytic, etc.,etc. Now, pz q z Thus, Res f
z 0

pz qz

, where p and q are

cos z . ez

cos 0 1 e
2

1, and 2 e2 cosh 2 .

Res f
z 2 i

cos 2 i e2 i

Finally, cos z dz ez 1 2 i Res f Res f

z 0 z 2 i

2 i1 Exercises

cosh 2

6. Suppose f has an isolated singularity at z 0 . Then, of course, the derivative f also has an isolated singularity at z 0 . Find the residue Res f .
z z0

7. Given an example of a function f with a simple pole at z 0 such that Res f carefully why there is no such function.
z z0

0, or explain

8. Given an example of a function f with a pole of order 2 at z 0 such that Res f explain carefully why there is no such function. 9. Suppose g is analytic and has a zero of order n at z 0 (That is, g z 0.). Show that the function f given by h z0 fz 1 gz z
z z0

0, or

z 0 n h z , where

10.8

has a pole of order n at z 0 . What is Res f ?

z z0

10. Suppose g is analytic and has a zero of order n at z 0 . Show that the function f given by fz has a simple pole at z 0 , and Res f
z z0

g z gz

n.

11. Find cos z dz, z2 4 6.

where C is the positively oriented circle |z| 12. Find

tan zdz,
C

where C is the positively oriented circle |z| 13. Find

2 .

z2

1 z

dz,

where C is the positively oriented circle |z|

10.

10.9

Some Applications of the Residue Theorem Supplementary Lecture Notes MATH 322, Complex Analysis Winter 2005
Pawel Hitczenko Department of Mathematics Drexel University Philadelphia, PA 19104, U.S.A. email: phitczenko@math.drexel.edu

would like to thank Frederick Akalin for pointing out a couple of typos.

Introduction

These notes supplement a freely downloadable book Complex Analysis by George Cain (henceforth referred to as Cains notes), that I served as a primary text for an undergraduate level course in complex analysis. Throughout these notes I will make occasional references to results stated in these notes. The aim of my notes is to provide a few examples of applications of the residue theorem. The main goal is to illustrate how this theorem can be used to evaluate various types of integrals of real valued functions of real variable. Following Sec. 10.1 of Cains notes, let us recall that if C is a simple, closed contour and f is analytic within the region bounded by C except for nitely many points z0 , z1 , . . . , zk then Z f (z)dz = 2i
k X j=0

Resz=zj f (z),

where Resz=a f (z) is the residue of f at a.

2
2.1

Evaluation of Real-Valued Integrals.

Denite integrals involving trigonometric functions

We begin by briey discussing integrals of the form Z 2 F (sin at, cos bt)dt.
0

(1)

Our method is easily adaptable for integrals over a dierent range, for example between 0 and or between . Given the form of an integrand in (1) one can reasonably hope that the integral results from the usual parameterization of the unit circle z = eit , 0 t 2. So, lets try z = eit . Then (see Sec. 3.3 of Cains notes), cos bt = eibt + e 2
ibt

z b + 1/z b , 2 dt =

sin at =

eiat

e 2i

iat

za

1/z a . 2i

Moreover, dz = ieit dt, so that

dz . iz 1/z a z b + 1/z b , 2i 2

Putting all of this into (1) yields Z

2

F (sin at, cos bt)dt =

za

dz , iz

where C is the unit circle. This integral is well within what contour integrals are about and we might be able to evaluate it with the aid of the residue theorem. 2

It is a good moment to look at an example. We will show that Z 2 cos 3t dt = . 5 4 cos t 12 0

(2)

Following our program, upon making all these substitutions, the integral in (1) becomes Z Z (z 3 + 1/z 3 )/2 dz 1 z6 + 1 = dz 4(z + 1/z)/2 iz i C z 3 (10z 4z 2 4) C 5 Z 1 z6 + 1 = dz 2i C z 3 (2z 2 5z + 2) Z 1 z6 + 1 = dz. 2i C z 3 (2z 1)(z 2) The integrand has singularities at z0 = 0, z1 = 1/2, and z2 = 2, but since the last one is outside the unit circle we only need to worry about the rst two. Furthermore, it is clear that z0 = 0 is a pole of order 3 and that z1 = 1/2 is a simple pole. One way of seeing it, is to notice that within a small circle around z0 = 0 (say with radius 1/4) the function z6 + 1 (2z 1)(z 2)

is analytic and so its Laurent series will have all coe cients corresponding to the negative powers of z zero. Moreover, since its value at z0 = 0 is 06 + 1 (2 0 1)(0 2) = 1 , 2

the Laurent expansion of our integrand is 1 z6 + 1 1 1 1 1 a1 = 3 + a1 z + . . . = + 2 + ..., 3 (2z 3 z 1)(z 2) z 2 2z z which implies that z0 = 0 is a pole of order 3. By a similar argument (using a small circle centered at z1 = 1/2) we see that z1 = 1/2 is a simple pole. Hence, the value of integral in (2) is z6 + 1 z6 + 1 2i Resz=0 + Resz=1/2 . z 3 (2z 1)(z 2) z 3 (2z 1)(z 2) The residue at a simple pole z1 = 1/2 is easy to compute by following a discussion preceding the second example in Sec. 10.2 in Cains notes: z6 + 1 z 3 (2z 1)(z 2) = z6 + 1 z6 + 1 = 5 z 3 (2z 2 5z + 2) 2z 5z 4 + 2z 3

is of the form p(z)/q(z) with p(1/2) = 2 6 + 1 6= 0 and q(1/2) = 0. Now, q 0 (z) = 10z 4 20z 3 + 6z 2 , so that q 0 (1/2) = 10/24 20/23 + 6/22 = 3/23 . Hence, the residue at z1 = 1/2 is p(1/2) = q 0 (1/2) (26 + 1) 23 = 26 3 65 . 24

The residue at a pole of degree 3, z0 = 0, can be obtained in various ways. First, we can take a one step further a method we used to determine the degree of that pole: since on a small circle around 0, z6 + 1 (2z 1)(z 2) = (2z z6 1)(z 2) + (2z 1 1)(z 2) . (3)

is analytic, the residue of our function will be the coe cient corresponding to z 2 of Taylor expansion of the function given in (3). The rst term will not contribute as its smallest non-zero coe cient is in front of z 6 so we need to worry about the second term only. Expand each of the terms 1/(2z 1) and 1/(z 2) into its Taylor series, and multiply out. As long as |2z| < 1 we get (2z 1 1)(z 2) = = = = = 1 1 1 2 1 2z 1 z/2 1 z z2 2 2 1 + 2z + 2 z + . . . 1 + + 2 + . . . 2 2 2 1 1 2 1 2 2 1 + A1 z + 2 z + z + 4z + . . . 2 2 4 1 1 2 1 + A1 z + 5 + z + ... 2 4 1 a1 21 + z + z2 + . . . 2 2 8

so that the residue at z0 = 0 is 21/8 (from the calculations we see that A1 = 2 + 1/2, but since our interest is the coe cient in front of z 2 we chose to leave A1 unspecied). The same result can be obtained by computing the second derivative (see Sec. 10.2 of Cains notes) of 1 3 z6 + 1 z 3 2! z (2z 1)(z 2) ,

and evaluating at z = 0. Alternatively, one can open Maple session and type: residue((z^6+1)/z^3/(2*z-1)/(z-2),z=0); to get the same answer again. Combining all of this we get that the integral in (2) is Z 1 z6 + 1 1 21 65 65 63 dz = (2i) = = , 2i C z 3 (2z 1)(z 2) 2i 8 24 24 12 as required.

2.2

Evaluation of improper integrals involving rational functions

Z
1

Recall that improper integral f (x)dx

R

is dened as a limit

R!1

lim

provided that this limit exists. When the function f (x) is even (i.e. f (x) = f ( x), for x 2 R) one has Z R Z 1 R f (x)dx = f (x)dx, 2 R 0 and the above integral can be thought of as an integral over a part of a contour CR consisting of a line segment along the real axis between R and R. The general idea is to closethe contour (often by using one of the semi-circles with radius R centered at the origin), evaluate the resulting integral by means of residue theorem, and show that the integral over the addedpart of CR asymptotically vanishes as R ! 0. As an example we will show that Z 1 dx = . (4) (x2 + 1)2 4 0 Consider a function f (z) = 1/(z 2 + 1)2 . This function is not analytic at z0 = i (and that is the only singularity of f (z)), so its integral over any contour encircling i can be evaluated by residue theorem. Consider CR consisting of the line segment along the real axis between R x R and the upper semi-circle AR := {z = Reit , 0 t }. By the residue theorem Z dz 1 = 2iResz=i . 2 2 (z 2 + 1)2 CR (z + 1) The integral on the left can be written as Z R Z dz dz + . (z 2 + 1)2 (z 2 + 1)2 R AR Parameterization of the line segment is (t) = t + i 0, so that the rst integral is just Z R Z R dx dx =2 . 2 + 1)2 2 + 1)2 R (x 0 (x Hence, Z
R

f (x)dx,

dx = iResz=i (x2 + 1)2

1 (z 2 + 1)2 5

1 2

AR

dz . (z 2 + 1)2

(5)

Since

1 = (z 2 + 1)2 (z

1 , i)2 (z + i)2

and 1/(z + i)2 is analytic on the upper half-plane, z = i is a pole of order 2. Thus (see Sec. 10.2 of Cains notes), the residue is d 1 2 2 1 1 2 (z i) = = = = 2 (z + i)2 3 3 3 dz (z i) (z + i) z=i (2i) 4i 4i z=i which implies that the rst term on the right-hand side of (5) is i = . 4i 4 Thus the evaluation of (4) will be complete once we show that Z dz lim = 0. R!1 A (z 2 + 1)2 R But this is straightforward; for z 2 AR we have |z 2 + 1| so that for R > 2 |z|2 1 = R2 1,

(6)

1 1 . (z 2 + 1)2 (R2 1)2

Using our favorite inequality Z

C

g(z)dz M length(C),

(7)

where |g(z)| M for z 2 C, and observing that length(AR ) = R we obtain Z dz R ! 0, 2 2 (R2 1)2 AR (z + 1) as R ! 1. This proves (6) and thus also (4).

2.3

Improper integrals involving trigonometric and rational functions.

Integrals like one we just considered may be spiced upto allow us to handle an apparently more complicated integrals with very little extra eort. We will illustrate it by showing that Z 1 cos 3x 2 dx = 3 . 2 + 1)2 e 1 (x 6

We keep the same function 1/(x2 + 1)2 , just to illustrate the main dierence. This time we consider the function e3iz f (z), where f (z) is, as before 1/(z 2 +1)2 . By following the same route we are led to Z R Z Z e3xi 2 3zi 3zi dx = 2iResz=i (f (z)e f (z)e dz = 3 f (z)e3zi dz. 2 2 e R (x + 1) AR AR At this point it only remains to use the fact that the real parts of both sides must the same, write e3xi = cos 3x + i sin 3x, and observe that the real part of the left-hand side is exactly the integral we are seeking. All we need to do now is to show that the real part of the integral over AR vanishes as R ! 1. But, since for a complex number w, |Re(w)| |w| we have Z Z 3iz Re f (z)e dz f (z)e3iz dz ,
AR AR

and the same bound as in (7) can be established since on AR we have since AR is in the upper half-plane so that y Remark. (i) This approach generally works for many integrals of the form Z 1 Z 1 f (z) cos azdz and f (z) sin azdz,
1 1

|e3iz | = |e3i(x+iy) | = |e

3y 3xi

0.

|=e

3y

1,

where f (z) is a rational function (ratio of two polynomials), where degree of a polynomial in the denominator is larger than that of a polynomial in the numerator (although in some cases working out the bound on the integral over AR may be more tricky than in the above example). The following inequality, known as Jordans inequality, is often helpful (see Sec. 2.4 for an illustration as well as a proof) Z e R sin t dt < . (8) R 0

(ii) The integrals involving sine rather than cosine are generally handled by comparing the imaginary parts. The example we considered would give Z 1 sin 3x dx = 0, 2 2 1 (x + 1)

but that is trivially true since the integrand is an odd function, and, clearly, the improper integral Z 1 sin 3x dx (x2 + 1)2 0 converges. For a more meaningful examples see Exercises 4-6 at the end. 7

2.4

One more example of the same type.

Z
1

Here we will show that

sin x dx = . (9) x 2 0 This integral is quite useful (e.g. in Fourier analysis and probability) and has an interesting twist, namely a choice of a contour (that aspect is, by the way, one more thing to keep in mind; clever choice of a contour may make wonders). Based on our knowledge form the previous section we should consider f (z) = eiz /z and try to integrate along our usual contour CR considered in Sections 2.2 and 2.3. The only problem is that our function f (z) has a singularity on the contour, namely at z = 0. To avoid that problem we will make a small detour around z = 0. Specically, consider pick a (small) > 0 and consider a contour consisting of the arc AR that we considered in the last two sections followed by a line segment along a real axis between R and , followed by an upper semi-circle centered at 0 with radius and, nally, a line segment along the positive part of the real exit from to R (draw a picture to see what happens). We will call this contour BR, and we denote the line segments by L and L+ , respectively and the small semi-circle by A . Since our function is analytic with BR, its integral along this contour is 0. That is Z Z ix Z Z R ix eiz e eiz e dz = 0 + dx + dz + dx, z x z x AR R A Since Z
R

eix dx = x

by combining the second and the fourth integral we obtain Z R ix Z Z e e ix eiz eiz dx + dz + dz = 0. x AR z A z Since the integrand in the leftmost integral is 2i we obtain 2i Z
R

ix

dx,

eix

e 2ix Z

ix

= 2i eiz dz z

sin x , x Z eiz dz = 0, z !

sin x dx = x

AR

or upon dividing by 2i Z R sin x i dx = x 2

AR

eiz dz + z

eiz dz z

The plan now is to show that the integral over AR vanishes as R ! 1 and that Z eiz lim dz = i. (10) !0 A z 8

To justify this last statement use the usual parameterization of A : z = eit , 0 t . Then dz = ieit dt and notice (look at your picture) that the integral over A is supposed to be clockwise (i.e. in the negative direction. Hence, Z eiz dz = z Z

eie ieit dt = i eit

it

eie dt.

it

Thus to show (10) it su ces to show that Z it lim eie dt = .

!0 0

To this end let us look at Z Z it eie dt =

0

eie dt

it

dt =

We will want to use once again inequality (7). Since the length of the curve is we only need to show that |eie But we have |eie
it it

eie

it

1 dt .

1| ! 0,

as

! 0.

(11)

1| = =

|ei(cos t+i sin t) 1| = |e sin t ei cos t 1| |e sin t ei cos t e sin t + e sin t 1| e sin t |ei cos t 1| + |e sin t 1| |ei cos t 1| + |e sin t 1|.

For 0 t , sin t 0 so that e sin t 1 and thus the second absolute value is actually equal to 1 e sin t which is less than sin t, since for any real u, 1 u e u (just draw the graphs of these two functions). We will now bound |ei cos t 1|. For any real number v we have |eiv 1|2 = (cos v 1)2 + sin2 v = cos2 v + sin2 v 2 cos v + 1 = 2(1 cos v). We will show that

v2 2 If we knew that, then (11) would follow since we would have 1 cos v |eie
it

(12)

1| |ei cos t

1| + |e

sin t

1| (| cos t| + | sin t|) 2 ! 0.

But the proof of (12) is an easy exercise in calculus: let h(v) := v2 + cos v 2 9 1.

Than (12) is equivalent to h(v) 0 for v 2 R.

Since h(0) = 0, it su ces to show that h(v) has a global minimum at v = 0. But h0 (v) = v sin v so that h0 (0) = 0, and h00 (v) = 1 cos v 0.

That means that v = 0 is a critical point and h(v) is convex. Thus, it has to be the mininimum of h(v). All that remains to show is that Z eiz lim dz = 0. (13) R!1 A z R This is the place where Jordans inequality (8) comes into picture. Going once again through the routine of changing variables to z = Reit , 0 t , we obtain Z Z Z eiz iReit dz = i e dt eiR(cos t+i sin t) dt AR z 0 0 Z Z = eiR cos t e R sin t dt e R sin t dt , R 0 0

where the last step follows from Jordans inequality (8). It is now clear that (13) is true. It remains to prove (8). To this end, rst note that Z Z /2 e R sin t dt = 2 e R sin t dt.
0 0

Thats because the graph of sin t and (thus also of e R sin t ) is symmetric about the vertical line at /2. Now, since the function sin t is concave between 0 t , its graph is above the graph of a line segment joining (0, 0) and (/2, 1); in other words 2t . sin t Hence Z /2 Z /2 2R e R sin t dt e t dt = 1 e R 2R 2R 0 0 which proves Jordans inequality.

Summation of series.
1 X 1 2 = , n2 6 n=1

As an example we will show that (14)

10

but as we will see the approach is fairly universal. Let for a natural number N CN be a positively oriented square with vertices at (1 i)(N + 1/2) and consider the integral Z cos z dz. (15) 2 CN z sin z Since sin z = 0 for z = 0, 1, 2, . . . the integrand has simple poles at 1, 2, . . . and a pole of degree three at 0. The residues at the simple poles are
z!k

lim

(z

k) cos z cos k (z k) 1 = lim = . z 2 sin z k 2 z!k sin((z k)) k 2 /3. To see that you can either ask

The residue at the triple pole z = 0 is Maple to nd it for you by typing:

residue(cos(Pi*z)/sin(Pi*z)/z^2,z=0); in your Maple session or compute the second derivative (see Sec. 10.2 of Cains notes again) of 1 3 cos z 1 z cos z z = , 2! z 2 sin z 2 sin z and evaluate at z = 0, or else use the series expansions for the sine and cosine, and gure out from there, what s the coe cient in front of 1/z in the Laurent series for cos z z 2 sin z around z0 = 0. Applying residue theorem, and collecting the poles within the contour CN we get ! ! Z N N N X 1 X 1 X 1 cos z dz = 2i + + = 2i +2 . 2 3 k 2 k 2 3 k 2 CN z sin z
k=1 k= 1 k=1

The next step is to show that as N ! 1 the integral on the left converges to 0. Once this is accomplished, we could pass to the limit on the right hand side as well, obtaining ! N 2X 1 lim = , 2 N !1 k 3
k=1

which is equivalent to (14). In order to show that the integral in (15) converges to 0 as N ! 1, we will bound Z cos z dz z 2 sin z CN using our indispensable inequality (7).

11

First, observe that each side of CN has length 2N + 1 so that the length of the contour CN is bounded by 8N + 4 = O(N ). On the other hand, for z 2 CN , |z| so that N+ 1 2 N,

1 1 2, 2 z N

so that it would be (more than) enough to show that cos z/ sin z is bounded on CN by a constant independent of N . By Exercise 9 from Cains notes | sin z|2 = sin2 x + sinh2 y, where z = x + iy. Hence cos z sin z
2

| cos z|2 = cos2 x + sinh2 y,

cos2 x + sinh2 y cos2 x sinh2 y = + sin2 x + sinh2 y sin2 x + sinh2 y sin2 x + sinh2 y cos2 x + 1. 2 sin x + sinh2 y

On the vertical lines of the contour CN x = (N + 1/2) so that cos(x) = 0 and sin(x) = 1. Hence the rst term above is 0. We will show that on the horizontal lines the absolute value of sinh y is exponentially large, thus making the rst term exponentially small since sine and cosine are bounded functions. For t > 0 we have et e t et 1 sinh t = . 2 2 Similarly, for t < 0 sinh t = so that, in either case et 2 e
t

e 2 e|t|

1 . 2 Since on the upper horizontal line |y| = N + 1/2 we obtain | sinh t| | sinh y| e(N +1/2) 2 1 ! 1,

as N ! 1. All of this implies that there exists a positive constant K such that for all N 1 and all for z 2 CN cos z | | K. sin z Hence, we conclude that Z cos z 8N + 4 dz K , z 2 sin z N2 12

CN

which converges to 0 as N ! 1. This establishes (14). The above argument is fairly universal and applies generally to the sums of the form 1 X f (n).
n= 1

Sums from 0 (or 1) to innity are then often handled by using a symmetry of f (n) or similarly simple observations. The crux of the argument is the following observation: Proposition 3.1 Under mild assumptions on f (z),
1 X

f (n) =

n= 1

cos z Res f (z) , sin z

where the sum extends over the poles of f (z). In our example we just took f (z) = 1/z 2 . Variations of the above argument allow us to handle other sums. For example, we have Proposition 3.2 Under mild assumptions on f (z),
1 X

( 1)n f (n) =

n= 1

Res

f (z) sin z

where the sum extends over the poles of f (z).

13

Exercises.
1. Evaluate 2. Evaluate 3. Evaluate 4. Evaluate 5. Evaluate 6. Evaluate 7. Evaluate 8. Evaluate Z Z Z Z Z Z Z Z
0 0 1 1 1 1 1 1 2 1 0 1 1 1

2x2 1 dx. x4 + 5x2 + 4 x dx. (x2 + 1)(x2 + 2x + 2) x4 dx . +1 sin x dx. + 4x + 5

Answer: /4. Answer: Answer: Answer: Answer: Answer: /5. p . 2 2

x2

sin 2. e sin 1. 2e e p
2

x sin x dx. x4 + 4 x sin x dx. x2 + 2x + 5 dt . 1 + sin2 t cos2 3x dx. 5 4 cos 2x

Answer:

2.

Answer:

3 . 8

9. Show that

1 X 1 4 = . n4 90 n=1 1 X ( 1)n+1 2 = . 2 n 12 n=1

10. Use a suggestion of Proposition 3.2 to show that

14

Chapter Eleven

Argument Principle
11.1. Argument principle. Let C be a simple closed curve, and suppose f is analytic on C. Suppose moreover that the only singularities of f inside C are poles. If f z 0 for all z C, then f C is a closed curve which does not pass through the origin. If t, is a complex description of C, then t f t , t t

is a complex description of . Now, lets compute

f z dz fz

f f

t t

t dt.

But notice that

t . Hence, f f 1 dz z t t t dt t

f z dz fz

t dt n2 i,

where |n| is the number of times winds around the origin. The integer n is positive in case is traversed in the positive direction, and negative in case the traversal is in the negative direction. Next, we shall use the Residue Theorem to evaluate the integral
C f z fz

dz. The singularities

of the integrand ff zz are the poles of f together with the zeros of f. Lets find the residues at these points. First, let Z z 1 , z 2 , , z K be set of all zeros of f. Suppose the order of the 0. Thus, z z j n j h z and h z j zero z j is n j . Then f z 11.1

f z fz

z h z hz

zj

nj

h z nj z zj z z j nj h z nj . z zj

nj 1

hz

Then z and Res f z zj f The sum of all these residues is thus N n1 n2 nK. nj. z zj f z fz z zj h z hz n j,

Next, we go after the residues at the poles of f. Let the set of poles of f be P p 1 , p 2 , , p J . Suppose p j is a pole of order m j . Then hz is analytic at p j . In other words, fz hz z pj . z pj
mj

fz

mj

Hence, f z fz z h z hz Now then, pj

mj

h z mj z pj z p j 2m j mj . z p j mj

mj 1

hz

pj hz

mj

11.2

z and so

pj

mj

f z fz

pj

mj

h z hz

mj,

Res f z pj f The sum of all these residues is P Then, f z dz fz m1

pj

mj.

m2

mJ

2 iN

P ;

and we already found that f z dz fz n2 i,

where n is the winding number, or the number of times winds around the originn 0 means winds in the positive sense, and n negative means it winds in the negative sense. Finally, we have n N P,

n K is the number of zeros inside C, counting multiplicity, or the where N n 1 n 2 m J is the number of poles, counting the order. order of the zeros, and P m 1 m 2 This result is the celebrated argument principle. Exercises 1. Let C be the unit circle |z| 1 positively oriented, and let f be given by

11.3

fz

z3.

How many times does the curve f C wind around the origin? Explain. 2. Let C be the unit circle |z| 1 positively oriented, and let f be given by fz z2 2.

z3

How many times does the curve f C wind around the origin? Explain.

a 1 z a 0 , with a n 3. Let p z anzn an 1zn 1 C is the circle |z| R positively oriented, then p z dz pz 2n i.

0. Prove there is an R

0 so that if

4. How many solutions of 3e z

0 are in the disk |z|

1? Explain.

5. Suppose f is entire and f z is real if and only if z is real. Explain how you know that f has at most one zero.

11.2 Rouches Theorem. Suppose f and g are analytic on and inside a simple closed contour C. Suppose moreover that |f z | |g z | for all z C. Then we shall see that f and f g have the same number of zeros inside C. This result is Rouches Theorem. To see why it is so, start by defining the function t on the interval 0 t 1 : t 1 2 i f z fz tg t dz. tg z

Observe that this is okaythat is, the denominator of the integrand is never zero: |f z Observe that tg z | ||f t | t|g t || ||f t | |g t || 0. t is the

is continuous on the interval 0, 1 and is integer-valued 11.4

number of zeros of f tg inside C. Being continuous and integer-valued on the connected set 0, 1 , it must be constant. In particular, 0 1 . This does the job! 0 1 2 i f z dz fz

is the number of zeros of f inside C, and 1 1 2 i f z fz g z dz gz

is the number of zeros of f

g inside C.

Example How many solutions of the equation z 6 5z 5 z 3 2 0 are inside the circle |z| 1? Rouches Theorem makes it quite easy to answer this. Simply let f z 5z 5 and let 6 3 6 3 gz z z 2. Then |f z | 5 and |g z | |z| |z| 2 4 for all |z| 1. Hence |f z | |g z | on the unit circle. From Rouches Theorem we know then that f and f g have the same number of zeros inside |z| 1. Thus, there are 5 such solutions. The following nice result follows easily from Rouches Theorem. Suppose U is an open set (i.e., every point of U is an interior point) and suppose that a sequence f n of functions analytic on U converges uniformly to the function f. Suppose further that f is not zero on the circle C z : |z z 0 | R U. Then there is an integer N so that for all n N, the functions f n and f have the same number of zeros inside C. This result, called Hurwitzs Theorem, is an easy consequence of Rouches Theorem. 0 for some . Now let N be large enough Simply observe that for z C, we have |f z | on C. It follows from Rouches Theorem that f and to insure that |f n z f z | f fn f f n have the same number of zeros inside C.

Example
z 1 z z2 , converges On any bounded set, the sequence f n , where f n z n! z uniformly to f z e , and f z 0 for all z. Thus for any R, there is an N so that for 2 zn has modulus R. Or to put it another way, given n N, every zero of 1 z z2 n! 2 zn an R there is an N so that for n N no polynomial 1 z z2 has a zero inside the n!
2 n

11.5

circle of radius R.

Exercises 6. Show that the polynomial z 6 7. How many solutions of 2z 4 4z 2 2z 3 1 has exactly two zeros inside the circle |z| 2z 2 2z 9 0 lie inside the circle |z| 1? 1.

8. Use Rouches Theorem to prove that every polynomial of degree n has exactly n zeros (counting multiplicity, of course). 9. Let C be the closed unit disk |z| 1. Suppose the function f analytic on C maps C into the open unit disk |z| 1that is, |f z | 1 for all z C. Prove there is exactly one w C such that f w w. (The point w is called a fixed point of f .)

11.6