Professional Documents
Culture Documents
Musielak
Lecture 22
Residue Theorem and Solution of Real Indefinite Integrals
In Lecture 21, I stated that we can solve several types of real definite integrals by
the application of Cauchy’s residue theorem. Those types of integrals are:
where 𝐹(sin 𝜃 , cos 𝜃) is a rational function of cos 𝜃 and sin 𝜃, which is finite on the
closed interval 0 ≤ 𝜃 ≤ 2𝜋. As explained in Lecture 21, we convert 𝐼 into a complex
contour integral 𝐽
𝐽 = ∮ 𝑓(𝑧)𝑑𝑧 ; 𝐶: |𝑧| = 1
𝐶
THEOREM 1. If 𝐹(sin 𝜃 , cos 𝜃) is a rational function of cos 𝜃 and sin 𝜃, and 𝐹 is finite
on the closed interval 0 ≤ 𝜃 ≤ 2𝜋, and if 𝑓(𝑧) is the complex function obtained from 𝐹
by the substitutions
𝑑𝑧 1 1
𝑑𝜃 = , cos 𝜃 = (𝑧 + 𝑧 −1 ), sin 𝜃 = (𝑧 − 𝑧 −1 ) (A)
𝑖𝑧 2 2𝑖
Then
2𝜋 𝑘
where the summation is taken over all singular points 𝑧𝑛 that lie within the unit circle
contour, |𝑧| = 1
2
We can use this method to evaluate the integral of any trigonometric rational function
between 0 and 2𝜋 , provided the denominator is never zero for any value of 𝜃.
We can also evaluate integrals with limits from 0 to 𝜋 if the integrand is even, since
the integral from 0 and 2𝜋 of an even periodic function is twice the integral from 0 to 𝜋
of the same function.
Example 1. Solve
2𝜋
𝑑𝜃
𝐼=∫
0 5 + 4 cos 𝜃
Solution: This is a definite real trigonometric integral which can be easily solved by
application of the residue theorem of complex analysis. As we did in the previous
lecture, by making the substitutions
𝑑𝑧 1
𝑑𝜃 = , cos 𝜃 = (𝑧 + 𝑧 −1 )
𝑖𝑧 2
1 𝑑𝑧 1 𝑑𝑧 1 𝑑𝑧
𝐽=∮ = ∮ = ∮
5 + 2(𝑧 + 𝑧 −1 ) 𝑖𝑧 𝑖 5𝑧 + 2𝑧 2 + 2 𝑖 (2𝑧 + 1)(𝑧 + 2)
𝐶 𝐶 𝐶
where 𝐶 is the unit circle, and we can solve 𝐽 by the residue theorem.
Note that the integrand has simple poles at 𝑧 = −1/2 and 𝑧 = −2, but the last pole is
outside 𝐶. Hence, we calculate the residue:
1 1 1
Res(@ − 1/2) = lim (𝑧 + 1/2) ∙ = | =
𝑧→−1/2 2(𝑧 + 1/2)(𝑧 + 2) 2(𝑧 + 2) 𝑧=−1/2 3
1 2𝜋𝑖
𝐽=
𝑖 3
Hence,
2𝜋
𝑑𝜃 2𝜋
𝐼=∫ =
0 5 + 4 cos 𝜃 3
Note: Verify your analysis by noting that the integral 𝐼 has a real value (not complex).
3
where 𝑓(𝑥) is continuous on (−∞, ∞). It is a rational function 𝑓(𝑥) = 𝑃(𝑥)/𝑄(𝑥) such
that the polynomials 𝑃 and 𝑄 have no common factors.
To evaluate an integral of the form (1) where the continuous rational function
𝑃(𝑥)
𝑓(𝑥) =
𝑄(𝑥)
by residue theorem, we replace 𝑥 by the complex variable 𝑧 and integrate the complex
function 𝑓(𝑧) over a closed contour 𝐶 that consists of 𝐶1 , which is the line segment
[−𝑅, 𝑅] on the real axis, and a semicircle 𝐶𝑅 of a radius large enough to enclose all
poles of 𝑓(𝑧) in the upper half-plane Re(𝑧) > 0. That is, 𝐶𝑅 is the semicircular contour
with radius 𝑅 → ∞. As shown in the sketch, the contour of integration is 𝐶 = 𝐶1 + 𝐶𝑅 .
𝑅 𝑘
∫ 𝑓(𝑧)𝑑𝑧 → 0, as 𝑅→∞
𝐶𝑅
so that the improper integral is easily evaluated by the residues of the poles of the
integrand that lie in the upper half plane.
(1) The complex function 𝑓(𝑧) is analytic in the upper half plane except at a finite
number of poles
(3) We have
|𝑓(𝑧)| ≤ 𝑀𝐿
Then the real improper integral (1) can be evaluated by a contour integral, and
𝑅 𝑘
Note, assumption (2) implies that 𝑓(𝑧) is analytic on 𝐶1 (the real axis)
∫ 𝑓(𝑧)𝑑𝑧 → 0, as 𝑅→∞
𝐶𝑅
𝑃(𝑥)
𝑓(𝑥) =
𝑄(𝑥)
where 𝑃(𝑥) and 𝑄(𝑥) are polynomials, 𝑄(𝑥) has no real roots, and deg 𝑄(𝑥) ≥
deg 𝑃(𝑥) + 2. (relative degree of 𝑓 is greater or equal to 2).
5
The following steps will facilitate the solution of this type of improper real integral:
Example 2. Evaluate
∞
1
𝐼=∫ 𝑑𝑥
−∞ 𝑥2 +1
Solution:
1
𝑓(𝑧) =
𝑧2 +1
(2) Check for relative degree of 𝑓(𝑧) = 𝑃(𝑧)/𝑄(𝑧). In this case, it is 2. Hence,
∫ 𝑓(𝑧)𝑑𝑧 → 0 as 𝑅 → ∞
𝐶𝑅
𝑎1 = 𝑖, 𝑎2 = −𝑖
(4) Identify poles which lie on the upper half plane. In this case, only 𝑎1 = 𝑖
𝑅
2𝜋𝑖 Res(@𝑎1 ) = ∮ 𝑓(𝑧)𝑑𝑧 = ∫ 𝑓(𝑥)𝑑𝑥
−𝑅
𝐶
6
Therefore
∞
1 (𝑧 − 𝑖) 1
𝐼=∫ 𝑑𝑥 = 2𝜋𝑖 Res(@𝑖) = 2𝜋𝑖 lim [ ] = 2𝜋𝑖 ( ) = 𝜋
−∞ 𝑥2 +1 𝑧→𝑖 (𝑧 + 𝑖)(𝑧 − 𝑖) 2𝑖
which we rewrite as
𝜋 𝜋
𝑖 𝑅𝑒 𝑖𝜃 𝑑𝜃 𝑖 𝑅𝑒 𝑖𝜃 𝑑𝜃 𝑖 𝜋 −3𝑖𝜃
∫ 2 2𝑖𝜃 + 1)2
=∫ 4 4𝑖𝜃
= 3∫ 𝑒 𝑑𝜃
0 (𝑅 𝑒 0 𝑅 𝑒 𝑅 0
∞ 𝑅 𝑘
Example 3. Evaluate
∞
1
𝐼=∫ 𝑑𝑥
0 𝑥2 +1
Solution: Observe, this is not an integral of the form (1), but we can make it by noting
that the integrand is even, and so we can write
7
1 ∞ 1
𝐼= ∫ 𝑑𝑥
2 −∞ 𝑥 2 + 1
Hence,
∞
1 𝜋
𝐼=∫ 𝑑𝑥 =
0 𝑥2 + 1 2
◼
This result should not be surprising since we know from calculus that
∞
1 −1 ∞
𝜋
∫ 𝑑𝑥 = tan 𝑥| 0 =
0 𝑥2 + 1 2
However, the method of solution studied here are easier to apply when such
elementary method become difficult or sometime impossible for some integrals.
NOTE: we could have closed the contour on the real axis with a semicircle in the lower
half-plane. Then, taking the contour so the real axis is traversed from −∞ to +∞ the
path would be clockwise. So we would need to take −2𝜋𝑖 times the residue of the pole
that is now encircled by 𝐶𝑅 in the lower half-plane. And as you can verify yourself, in
Example 3, you should get
1 1 𝜋
𝐼 = − 2𝜋𝑖 (− ) =
2 2𝑖 2
In calculus, the improper integral (1) is defined in terms of two distinct limits:
∞ 0 𝑅
∫ 𝑓(𝑥)𝑑𝑥 = lim ∫ 𝑓(𝑥)𝑑𝑥 + lim ∫ 𝑓(𝑥)𝑑𝑥
−∞ 𝑟→∞ −𝑟 𝑅→∞ 0
∞ 𝑅
∫ 𝑓(𝑥)𝑑𝑥 = lim ∫ 𝑓(𝑥)𝑑𝑥
−∞ 𝑅→∞ −𝑅
The limit may exist even though the improper integral is divergent.
For example, the integral
∞
∫ 𝑥 𝑑𝑥
−∞
is divergent since
8
𝑅
𝑅2
lim ∫ 𝑥 𝑑𝑥 = lim ( )=∞
𝑅→∞ 0 𝑅→∞ 2
However,
𝑅
𝑅 2 (−𝑅)2
lim ∫ 𝑥 𝑑𝑥 = lim ( − )=0
𝑅→∞ −𝑅 𝑅→∞ 2 2
This limit is called the Cauchy Principal Value or P.V. of the integral
∞ 𝑅
𝑃. 𝑉. ∫ 𝑓(𝑥)𝑑𝑥 = lim ∫ 𝑓(𝑥)𝑑𝑥 (4)
−∞ 𝑅→∞ −𝑅
When an integral of the form (1) converges, its Cauchy P.V. is the same as the value of
the integral. If the integral diverges, it may still possess a Cauchy P.V.
∞ 𝑅
𝑃. 𝑉. ∫ 𝑓(𝑥)𝑑𝑥 = lim ∫ 𝑓(𝑥)𝑑𝑥
−∞ 𝑅→∞ −𝑅
= 2𝜋𝑖 ∑ Res(𝑓(𝑧), 𝑎𝑘 )
𝑘=1
Solution: Let
9
1
𝑓(𝑧) =
(𝑧 2 + 1)(𝑧 2 + 9)
and since
let 𝐶 be the closed contour consisting of the interval [−𝑅, 𝑅] on the real axis, and a
semicircle 𝐶𝑅 of radius 𝑅 > 3 (to enclose all poles on the upper half-plane)
From (5),
𝑅
1 1 1
∮ 2 𝑑𝑧 = ∫ 𝑑𝑧 + ∫ 𝑑𝑥
(𝑧 + 1)(𝑧 2 + 9) (𝑧 2 + 1)(𝑧 2 + 9) 2 2
−𝑅 (𝑧 + 1)(𝑧 + 9)
𝐶 𝐶𝑅
= 𝐼1 + 𝐼2
where
1 1
Res @𝑖 = , Res @3𝑖 = −
16 𝑖 48 𝑖
so
1 1 𝜋
𝐼1 + 𝐼2 = 2𝜋𝑖 [Res @𝑖 + Res@3𝑖] = 2𝜋𝑖 ( − )=
16 𝑖 48 𝑖 12
Now let 𝑅 → ∞, and note that on 𝐶𝑅
1 𝜋𝑅
|𝐼2 | = | ∫ 𝑑𝑧 | ≤
(𝑧 2 + 1)(𝑧 2 + 9) (𝑅 2 − 1)(𝑅 2 − 9)
𝐶𝑅
|𝐼2 | → 0 as 𝑅 → ∞
𝑅
1 𝜋
lim ∫ 𝑑𝑥 =
𝑅→∞ −𝑅 (𝑧 2 2
+ 1)(𝑧 + 9) 12
or
∞ 𝑅
1 1 𝜋
𝑃. 𝑉. ∫ 2 2
𝑑𝑥 = lim ∫ 𝑑𝑥 =
−∞ (𝑥 + 1)(𝑥 + 9) 𝑅→∞ −𝑅 (𝑧 2 + 1)(𝑧 2 + 9) 12
Suppose 𝑓(𝑧) = 𝑃(𝑧)/𝑄(𝑧), where the degree of 𝑃(𝑧) is 𝑛 and the degree of 𝑄(𝑧) is 𝑚 ≥
𝑛 + 2. If 𝐶𝑅 is a semicircular contour 𝑧 = 𝑅𝑒 𝑖𝜃 , 0 ≤ 𝜃 ≤ 𝜋, then
∫ 𝑓(𝑧)𝑑𝑧 → 0 as 𝑅 → ∞ (7)
𝐶𝑅
This means that the integral along 𝐶𝑅 tends to zero, as 𝑅 becomes very large and tends
to ∞ when the denominator of 𝑓(𝑧) is of a power at least 2 more than its numerator. In
the previous example, the conditions of this theorem are satisfied since the degree of
𝑃(𝑧) = 1 is 0, and the degree of 𝑄(𝑧) = (𝑧 2 + 1)(𝑧 2 + 9) is 4, since
Solution: Let
1 1
𝑓(𝑧) = =
𝑧4 + 1 (𝑧 − 𝑎1 )(𝑧 − 𝑎2 )(𝑧 − 𝑎3 )(𝑧 − 𝑎4 )
which has 2 simple poles in the upper half-plane, at 𝑎1 = 𝑒 𝜋𝑖/4 and at 𝑎2 = 𝑒 3𝜋𝑖/4.
Hence,
11
1 1 1 1
Res(𝑓(𝑧), 𝑎1 ) = − − 𝑖, Res(𝑓(𝑧), 𝑎2 ) = − 𝑖
4√2 4√2 4√2 4√2
Therefore,
∞ 𝑛
1 𝜋
𝑃. 𝑉. ∫ 4
𝑑𝑥 = 2𝜋𝑖 ∑ Res(𝑓(𝑧), 𝑎𝑛 ) =
−∞ 𝑥 + 1 1
√2
■
Example 7. Evaluate
∞
𝑥2
𝐼=∫ 2 2 2 2
𝑑𝑥 , 𝑎 > 0, 𝑏>0
−∞ (𝑥 + 𝑎 )(𝑥 + 𝑏 )
𝑧2
𝑓(𝑧) =
(𝑧 2 + 𝑎2 )(𝑧 2 + 𝑏 2 )
∫ 𝑓(𝑧)𝑑𝑧 → 0 as 𝑅 → ∞
𝐶𝑅
We also note that 𝑓(𝑧) has poles at 𝑧 = ±𝑖𝑎 and 𝑧 = ±𝑖𝑏, but only the poles 𝑖𝑎 and 𝑖𝑏 lie
on the upper half plane.
𝑧2
∮ 𝑑𝑧 = 2𝜋𝑖[Res(𝑓(𝑧), 𝑧 = 𝑖𝑎) + Res(𝑓(𝑧), 𝑧 = 𝑖𝑏)]
(𝑧 2 + 𝑎2 )(𝑧 2 + 𝑏 2 )
𝐶
that is
∞
𝑥2 𝑎 𝑏 𝜋
𝐼=∫ 2 2 2 2
𝑑𝑥 = 2𝜋𝑖 [ 2 2
+ 2 2
]=
−∞ (𝑥 + 𝑎 )(𝑥 + 𝑏 ) 𝑖 2(𝑎 − 𝑏 ) 𝑖 2(𝑏 − 𝑎 ) 𝑎+𝑏
Final Comment: The residue theorem requires 𝑓(𝑧) to be analytic inside and on 𝐶.
Thus, the residue theorem does not apply if there are singular points on 𝐶.
12
We will evaluate this type of integral by considering a contour integral that can be
easily solved with the residue theorem. We leave this analysis for next lecture.