Integration--Cauchy Principal Value, Residue, line integration, poles

Click For Summary

Discussion Overview

The discussion revolves around the evaluation of the integral $$\int_0^{\infty}\frac{(\log x)^2}{1 + x^2}dx$$ using techniques from complex analysis, specifically the Cauchy Principal Value, residue theory, and contour integration. Participants explore the implications of the multivalued nature of the logarithm and the appropriate contour to use for the integration.

Discussion Character

  • Technical explanation
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • One participant expresses uncertainty about how to approach the integral, noting the difficulty with the logarithm's definition.
  • Another participant points out that the logarithm is multivalued and suggests the use of a branch cut, proposing the keyhole contour as a potential method.
  • A later reply clarifies that the logarithm should be considered as the real logarithm for this integral, referencing a specific book for further guidance.
  • One participant proposes defining the function $$f(z) = \frac{\log^{2} z}{1+z^{2}}$$ and discusses the implications of the branch cut along the negative imaginary axis.
  • There is a detailed exploration of the contour integral, including the evaluation of residues and the limits of integration as they approach zero and infinity.
  • Another participant confirms the need for a small semi-circle around the origin to avoid the logarithmic singularity, indicating that a keyhole contour is not necessary.

Areas of Agreement / Disagreement

Participants generally agree on the need to consider the multivalued nature of the logarithm and the use of contour integration techniques. However, there are differing views on the specific contours to use and the interpretation of the logarithm in this context, indicating that the discussion remains unresolved.

Contextual Notes

Participants mention the potential for different interpretations of the logarithm and the necessity of defining branch cuts, which could affect the evaluation of the integral. The discussion also highlights the complexity of contour integration in the presence of singularities.

Dustinsfl
Messages
2,217
Reaction score
5
$$
\int_0^{\infty}\frac{(\log x)^2}{1 + x^2}dx = \frac{\pi^3}{8}
$$

I have no idea what to do with this integral. I can't see it is even and do 1/2 the integral from -infinity to infinity since log -x doesn't make sense.
 
Physics news on Phys.org
$\log x$ is a multivalued function, so there needs to be a branch cut somewhere. Have you used the keyhole contour before?
 
Random Variable said:
$\log x$ is a multivalued function, so there needs to be a branch cut somewhere. Have you used the keyhole contour before?

No I think log is supposed to be viewed as the real log in this integral.
If you have Lang's book, I can tell you what page it is on.
 
Let me correct myself. You're going to let $\displaystyle f(z) = \frac{\log^{2} z}{1+z^{2}}$ and $\log z$ is a multivalued function.

---------- Post added at 11:42 ---------- Previous post was at 10:29 ----------

Define the cut along the negative imaginary axis (including the origin). Then the contour is exactly the same as the contour for $\displaystyle \int_{0}^{\infty} \frac{\sin^{2} x}{x^{2}} \ dx $.

let $\displaystyle f(z) = \frac{(\log z)^{2}}{1+z^{2}}$ and recall that $\displaystyle \log z = \ln z + i \arg z$

$\displaystyle \int_{-R}^{-r} f(x) \ dx + \int_{C_{r}} f(z) \ dz + \int_{r}^{R} f(x) \ dx + \int_{C_{R}} f(z) \ dz = 2 \pi i \ \text{Res} [f,i] $

$\displaystyle = 2 \pi i \lim_{z \to i} \ (z-i) \frac{(\log z)^2}{1+z^{2}} = 2 \pi i \frac{(\log i)^{2}}{2i} = \pi \left( \ln |i| + i \arg{i} \right)^{2} = \pi \left(\frac{i \pi}{2} \right)^{2} = \frac{-\pi^{3}}{4}$so we have $\displaystyle \int_{-R}^{-r} \frac{(\ln |x| + i \pi)^{2}}{1+x^{2}} \ dx + \int_{C_{r}} f(z) \ dz + \int_{r}^{R} \frac{(\ln|x| + i0)^{2}}{1+x^{2}} \ dx + \int_{C_{R}} f(z) \ dz = \frac{-\pi^{3}}{4}$If you let $r$ to go to zero and $R$ go to infinity, the second the fourth integrals will go to zero. You can show that using the ML inequality.$\displaystyle \text{PV} \left( \int_{-\infty}^{0} \frac{(\ln |x|)^{2} + 2 \pi i \ln |x| - \pi^{2}} {1+x^{2}} + \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx \right)$

$ \displaystyle =\text{PV} \left( \int_{0}^{\infty} \frac{(\ln x)^{2} + 2 \pi i \ln x - \pi^{2}} {1+x^{2}} + \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx \right) = \frac{-\pi^{3}}{4} $now equate the real parts on both sides

$ \displaystyle 2 \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx - \pi^{2} \int_{0}^{\infty} \frac{1}{1+x^{2}} \ dx = \frac{-\pi^{3}}{4} $ (I dropped the PV label since both integrals converge.)

so $\displaystyle 2 \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx - \frac{-\pi^{3}}{2} = \frac{-\pi^{3}}{4} \implies \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx = \frac{\pi^{3}}{8}$ (Whew)
 
Random Variable said:
Let me correct myself. You're going to let $\displaystyle f(z) = \frac{\log^{2} z}{1+z^{2}}$ and $\log z$ is a multivalued function.

---------- Post added at 11:42 ---------- Previous post was at 10:29 ----------

Define the cut along the negative imaginary axis (including the origin). Then the contour is exactly the same as the contour for $\displaystyle \int_{0}^{\infty} \frac{\sin^{2} x}{x^{2}} \ dx $.

let $\displaystyle f(z) = \frac{(\log z)^{2}}{1+z^{2}}$ and recall that $\displaystyle \log z = \ln z + i \arg z$

$\displaystyle \int_{-R}^{-r} f(x) \ dx + \int_{C_{r}} f(z) \ dz + \int_{r}^{R} f(x) \ dx + \int_{C_{R}} f(z) \ dz = 2 \pi i \ \text{Res} [f,i] $

$\displaystyle = 2 \pi i \lim_{z \to i} \ (z-i) \frac{(\log z)^2}{1+z^{2}} = 2 \pi i \frac{(\log i)^{2}}{2i} = \pi \left( \ln |i| + i \arg{i} \right)^{2} = \pi \left(\frac{i \pi}{2} \right)^{2} = \frac{-\pi^{3}}{4}$so we have $\displaystyle \int_{-R}^{-r} \frac{(\ln |x| + i \pi)^{2}}{1+x^{2}} \ dx + \int_{C_{r}} f(z) \ dz + \int_{r}^{R} \frac{(\ln|x| + i0)^{2}}{1+x^{2}} \ dx + \int_{C_{R}} f(z) \ dz = \frac{-\pi^{3}}{4}$If you let $r$ to go to zero and $R$ go to infinity, the second the fourth integrals will go to zero. You can show that using the ML inequality.$\displaystyle \text{PV} \left( \int_{-\infty}^{0} \frac{(\ln |x|)^{2} + 2 \pi i \ln |x| - \pi^{2}} {1+x^{2}} + \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx \right)$

$ \displaystyle =\text{PV} \left( \int_{0}^{\infty} \frac{(\ln x)^{2} + 2 \pi i \ln x - \pi^{2}} {1+x^{2}} + \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx \right) = \frac{-\pi^{3}}{4} $now equate the real parts on both sides

$ \displaystyle 2 \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx - \pi^{2} \int_{0}^{\infty} \frac{1}{1+x^{2}} \ dx = \frac{-\pi^{3}}{4} $ (I dropped the PV label since both integrals converge.)

so $\displaystyle 2 \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx - \frac{-\pi^{3}}{2} = \frac{-\pi^{3}}{4} \implies \int_{0}^{\infty} \frac{(\ln x)^{2}}{1+x^{2}} \ dx = \frac{\pi^{3}}{8}$ (Whew)

So the contour is the semi-circle in the upper half plane?
 
Yes, but with a small semi-circle about the origin to avoid the logarithmic singularity there. A keyhole contour was not needed.
 

Similar threads

  • · Replies 2 ·
Replies
2
Views
1K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 29 ·
Replies
29
Views
3K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 11 ·
Replies
11
Views
3K