You are on page 1of 2

From the Ch.

10 Notes:

Integral Test

Given

t
n =1

If

t dn converges then series converges.


n 1

Integral Test: If f is a continuous, positive, decreasing function on [1, ) with f (n) = an , then the series

an converges if and only if the improper integral


n =1

f ( x)dx converges.
1

So essentially, if the area under a curve from 1 to infinity is finite (the integral converges/exists), then the sum of the series also exists. This should make sense because an integral is the exact area under the curveit is the summation of infinitely many tiny rectangles (of width dx). (Think back to the Riemann sums; they were rectangles under the curve and then to find the exact area, we allowed the number of rectangles to approach infinity.) In a sum however, we only have a rectangle every integer numbereach rectangle is only 1 wide. This is because the sigma symbol just goes by integersie 1, 2, 3, etc. It never has 1.1 or 1.5. So because we have fewer rectangles, it would follow that this sum would be slightly less than the integraltherefore, if the integral exists, than the sum will also exist.

To determine if the integral converges:


This goes back to section 8.8improper integrals As we learned,

xp

dx converges for values of p > 1 and diverges for values of p 1 .


1

1 dx , p=1 x x =1 (because x is to the first power). 11 so the integral diverges and therefore the sum also diverges.

So if we were given

x , we could test this by seeing if the integral converges. In

Sometimes you need to combine powers from the numerator and the denominator to find what p is for 2x3 1 the equation. If you are given 2 , you would rewrite with an integral (you can 2 x =1 ( x 1) ( x 4 ) ( x 2) essentially skip the step of rewriting for a polynomial expression like this), add the degrees of x in the numerator and in the denominator independently then add/subtract. In this case, the numerators degree is 3, and the denominators degree is 5 (because you have an x2 * x2 * x). Subtract 5-3, and you get p=2. x3 1 You could also look at this as 5 which would simplify to 2 , also giving you p=2. 2>1, so the series x x would converge.

Some more involved problems:

1 k ln k k =2

1 2 x ln x u = ln x

(k changes to x just because thats convention; you could leave it as k if you want) (2 to infinity because our initial term was k=2)

du = 1 dx x
1 1 du ln 2 xu 1x = ln 2 u

Note that limits of integration change because of the u-sub

p = 1 and 1 1; diverges
Similar problem: n(ln n )
n= 2

2 2 x(ln x )3 dx u = ln x 1 du = dx x 2 2 du ln 2 xu 3 1x = ln 2 u 3 p = 3 and 3 > 1; converges

Note that evaluating the integral would NOT give you the answer to the sum; it would be an over approximation. (We havent learned any way so far to find the sum of this question)