4

How can I compute the following integral using residue calculus? I can't seem to compute it since the singularity point is on the contour boundary...

  • Example of this: http://math.stackexchange.com/questions/1110635/pole-on-a-contour-problem-with-integration – amcalde Apr 03 '17 at 23:50
  • The "bad" points are at $\pm i$. So taking the contour to be a half-circle of radius $2$ together with $[-1,1]$ will do it. Compute the whole contour with the residue theorem, and then subtract the contribution from just the half-circle itself. – Kaj Hansen Apr 03 '17 at 23:53
  • What contour are you trying to use? – Vik78 Apr 03 '17 at 23:57
  • @jamescho Hi James! It's been a while. I hope you're staying safe and healthy during the pandemic. I've reached out to contact you a few times, but am unsure whether you've received the notes? If you would, please let me know how I can improve my answer. I really want to give you the best answer I can. And feel free to up vote an answer as you see fit. ;-) – Mark Viola Feb 28 '21 at 19:41

1 Answers1

5

Let $I$ be the integral given by

$$I=\int_{-1}^1 \frac{\sqrt{1-x^2}}{1+x^2}\,dx \tag1$$


To evaluate $(1)$ using complex analysis, we will analyze the integral $J$

$$J=\int_C \frac{\sqrt{1-z^2}}{1+z^2}\,dz \tag2$$

where $C$ is the classical "dog-bone contour." (See other examples here and here for a primer with details).

We will cut the plane from $-1$ to $1$ such that

$$\sqrt{1-z^2}=-i\sqrt{z-1}\sqrt{z+1}$$

with $-\pi <\arg(z-1)\le \pi$ and $-\pi <\arg(z+1)\le \pi$.

Note that on $\mathbb{C}\setminus [-1,1]$, $f(z)$ is meromorphic with poles at $z=\pm i$.

Then, using the Residue Theorem and Cauchy's Integral Theorem, we can write $(2)$ as

$$\begin{align} J&=2\int_{-1}^{1}\frac{\sqrt{1-x^2}}{1+x^2}\,dx\\\\ &=2\pi i\text{Res}\left(\frac{\sqrt{1-z^2}}{1+z^2}, z=\pm i\right) -\int_0^{2\pi}\frac{\sqrt{1-R^2e^{i2\phi}}}{1+R^2e^{i2\phi}}\,iRe^{i\phi}\,d\phi\tag 3 \end{align}$$

where we the contributions from the integrals around the "small" circular contours centered at $\pm1$ vanish as their radii approach $0$.


The residues at $z=i$ and $z=-i$ are equal and given by

$$\text{Res}\left(\frac{\sqrt{1-z^2}}{1+z^2}, z=\pm i\right)=\frac{\sqrt{2}}{2i}$$


The integral over $\phi$ as $R\to \infty$ becomes (note that this is equivalent to $2\pi i$ times the residue at infinity)

$$\lim_{R\to \infty}\int_0^{2\pi}\frac{\sqrt{1-R^2e^{i2\phi}}}{1+R^2e^{i2\phi}}\,iRe^{i\phi}\,d\phi=-i2\pi $$


Putting it all together we see that

$$\begin{align} 2\int_{-1}^{1}\frac{\sqrt{1-x^2}}{1+x^2}\,dx&=2\pi i \frac{2\sqrt 2}{2i}-2\pi\\\\ &=2\pi (\sqrt 2-1) \end{align}$$

whereupon dividing by $2$ yields the coveted integral

$$I=\pi(\sqrt 2-1)$$

Mark Viola
  • 179,405
  • This was a long time ago, but maybe you could still clarify something? How do you get $\sqrt{1-z^2}=-i\sqrt{z-1}\sqrt{z+1}$? Because it seems you could just as well write $\sqrt{1-z^2}=i\sqrt{z-1}\sqrt{z+1}$ and the result would be different. How do you choose the sign here? – thedude Aug 23 '20 at 20:50
  • @thedude We need to define $f(z)=\sqrt{1-z^2}$ so that when $z\in(-1,1)$, $f(z)$ is real and positive. – Mark Viola Aug 23 '20 at 21:40
  • I still don't get it. I asked another question, if you want to give more details. It is here: https://math.stackexchange.com/questions/3801139/detail-on-the-the-choice-of-sign-when-computing-int-11-sqrt1-x2dx-by-r – thedude Aug 23 '20 at 21:53
  • @thedude Look at how the arguments were defined – Mark Viola Aug 23 '20 at 22:05
  • I am looking, but I don't see how they help me pick $-i$ over $i$. You can see from my question that I have made a fair effort, it is just this sign that still eludes me. I usually understand how to choose a branch of a multivalued funcion when moving continuously around a contour, but here I don't see any matter of continuity, just having to "pick" a sign. – thedude Aug 23 '20 at 22:22
  • In this problem herein, if we were to have written $\sqrt{1-z^2}=+i\sqrt{z-1}\sqrt{z-1}$, then for $z\in (-1,1)$ (e.g., $z=0$), we would have $\sqrt{1-z^2}<0$ (e.g., $\sqrt{1-0^2}=-1$) since $-\pi <\arg(z\pm 1)\le \pi$. Inasmuch as we should have instead $\sqrt{1-z^2}>0$ for $z\in (-1,1)$, the selection of $+i$ was not the correct choice. – Mark Viola Aug 23 '20 at 22:59
  • @jamescho Hi James! It's been a while. I hope you're staying safe and healthy during the pandemic. I've reached out to contact you a few times, but am unsure whether you've received the notes? If you would, please let me know how I can improve my answer. I really want to give you the best answer I can. And feel free to up vote an answer as you see fit. ;-) – Mark Viola Feb 28 '21 at 19:41