Computing Definite Integrals: Convergence and Cauchy's Principle Value, Study notes of Calculus

The convergence of definite integrals and introduces cauchy's principle value to avoid integrals that depend on how limits are taken. Topics include the definition of cauchy's principle value, the evaluation of integrals using contours, and the summation of infinite series.

Typology: Study notes

2010/2011

Uploaded on 10/04/2011

deville
deville 🇺🇸

4.7

(23)

390 documents

1 / 9

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
Computing Definite Integrals
Adrian Down
November 17, 2005
1 Convergence of real-valued functions
Suppose we have a continuous function f:RC. We would like to know
when the following integral exists.
lim
R2
R1
ZR2
R1
d(x)dx
As with infinite series, there is no single test. We can establish a sufficient
condition, however.
Fact. If C < such that
ZR2
R1
|f(x)|dx CR1, R2
then the limit (R1, R2 ) exists.
Example.
f(x) = cos(x)
1 + x2
|f(x)| 1
1 + x2
x1 |f(x)| x2
and Z
1
dx
x2<
1
pf3
pf4
pf5
pf8
pf9

Partial preview of the text

Download Computing Definite Integrals: Convergence and Cauchy's Principle Value and more Study notes Calculus in PDF only on Docsity!

Computing Definite Integrals

Adrian Down

November 17, 2005

1 Convergence of real-valued functions

Suppose we have a continuous function f : R → C. We would like to know

when the following integral exists.

lim

R 2 → ∞

R 1 → ∞

∫ R

2

−R 1

d(x)dx

As with infinite series, there is no single test. We can establish a sufficient

condition, however.

Fact. If ∃C < ∞ such that

∫ (^) R 2

−R 1

|f (x)|dx ≤ C ∀R 1 , R 2

then the limit (R 1 , R 2 → ∞) exists.

Example.

f (x) =

cos(x)

1 + x

2

|f (x)| ≤

1 + x

2

x ≥ 1 ⇒ |f (x)| ≤ x

− 2

and

1

dx

x

2

Similar arguments hold for x ≤ −1. Thus,

∫ (^) R 2

−R 1

|f (x)|dx < C < ∞

f satisfies the condition above, and so the integral exists.

Note. In the previous examples that we did, we proved directly that the

integral existed. The formulation above is good to use as a preliminary check

when approaching a problem to determine if the integral actually converges.

2

∫ ∞

e

ıx

x

dx

2.1 Limiting Behavior

e

ıx

x

|x|

lim N →∞

∫ N

1

dx

x

lim → 0

ln

As written, this integral is problematic for large positive or negative x, as

well as x = 0.

2.2 Related example

As a warm up, consider the simpler integral,

∫ (^1)

− 1

dx

x

We need to decide how to define this integral. Consider,

lim → 0

<|x|< 1

dx

x

Because the integrand is odd, the contributions on either side of the gap of

width  cancel exactly, leaving

lim → 0

Only the imaginary part of the integral remains,

P

−∞

e

ıx

x

dx = ıP

−∞

sin(x)

x

dx

We solve the first integral, although the second is generally more useful for

applications. The second follows immediately from the first.

2.4.1 Define a contour

Make a closed loop and use Cauchy’s Residue theorem. Traverse the real

axis in the usual direction. Take a rectangle with height R and length R.

Make a semi-circular indentation centered at 0 of radius . Call the whole

contour ΓR.

The function to integrate is

f (z) =

e

ız

z

2.4.2 Singularities

The only singularity is at z = 0. We have excluded this singularity from the

inside of the contour, so from Cauchy’s Residue Theorem,

Γ

f = 0 ∀, R

Note. Although the integral of f along the entire path Γ is 0, the integral

that we would like to calculate along the real axis is not 0.

2.4.3 Indentation in the contour

First consider the semi-circle about 0. It will be the most important integral.

On this path, expand e

ız to isolate the singular behavior of the function at

e

ız = 1 + (e

ız − 1)

Using this substitution,

C

e

ız

z

dz =

C

dz

z

C

e

ız − 1

z

dz

The first integral is related to one calculated previously.

C

dz

z

(2ıπ) = −ıπ

Note. The

1 2

appears because we previously integrated around an entire circle,

and it is negative because the circle is traversed in the opposite direction.

Using the Estimation Lemma,

C

e

ız − 1

z

dz

≤ π · max |z|=

e

ız − 1

z

The integrand can be be Taylor expanded about 0.

e

ız− 1

z

= ı +

ı

2

Since the function has an extension which is a convergent series at z = 0,

the singularity is removable. Since the function is differentiable on the rest

of the complex plane, it is bounded everywhere. Thus the integral tends to

0 as  → 0.

We will see that the only contribution to the integral comes from this

indentation in the contour, so

C

e

ız

z

dz → −ıπ

2.4.4 Other concours in ΓR

Now we show that the integral along all of the other paths is 0.

Horizontal path From the estimation lemma applied to the top of the

rectangle γ 2 ,

γ 2

e

ız

z

≤ L(γ 2 ) · max −R≤x≤R

|e

ıx − R

x + ıR

≤ (2R)

e

−R

R

R→∞

Vertical paths

limits,  → 0 and R → ∞.

0 = P

−∞

e

ıx

x

dx − ıπ

⇒ P

−∞

sin(x)

x

dx = π

Note. • This method of integrating around singular points holds for any

singularities on the real axis.

  • We could try to go around a path in the lower complex plane. However,

then we would have

e

ız = e

ıx e

−y y < 0

which is not bounded. Thus the desired limits cold not be taken.

3 Summation of infinite series

3.1 Motivating examples

Example. •

∞ ∑

n=

n^2

π

2

sin

2 (w)

∞ ∑

n=−∞

(w − nπ)^2

Note. – The poles of the two expression for the function are the

same, which is reassuring.

  • This series is unlike anything we have learned thus far. It is an

infinite series of functions. It is not a power series, nor is it a

Laurent series.

z

e

z − 1

z

∞ ∑

n=

2 z

2

z

2

2 n

2

Note. • We would like to see the relationship between these series that

allows their sums to be calculated.

  • The method that we develop considers a particular complex function,

from which a set of identities is developed. It cannot be used to generate

a formula given an arbitrary sum.

3.2 General method by example

g(z) =

cot(πz)

z

2

z

2

cos(πz)

sin(πz)

3.2.1 Singularities

sin(πz) vanishes at the real integers, which includes the singularity of

1 z^2

at

z = 0.

3.2.2 Construct a path

General method Consider a semi-circular path in the upper half plane of

radius r < 1. The path only encloses the singularity at z = 0. The integral

is, by Cauchy’s residue theorem,

γ

g = 2πı · res(g, 0)

If the radius of the path is increased to be 1 < r < 2,

γ

g = 2πı

n=− 1

res(g, n)

Our approach is to let the radius of the path of integration go to ∞. The

integral is then an infinite sum of the values of the residues of g(z) at the

singularities. We hope that the integral goes to 0 in the limit as R → ∞,

which will give an identity in terms of the sum of the residues.

From Cauchy’s Residue theorem,

γN

h = 2πı

+N ∑

n=−N

res(h, n)

We would like to show two things: that the integral on the left is 0, and the

explicit value for the residues.