1

I'm stucked with that integral because the solution I get is not correct according to WolframAlpha, what I have done is the following.

If we do the semicircle contour, we have the following,

$\int_{-r}^{r}\frac{dz}{(a+bz^2)^n}+\int_{\gamma}\frac{dz}{(a+bz^2)^n}=2\pi i \sum Res(\omega,z)$

Where the second integral tends to zero when r tends to infinity and the first one tends to the integral we want to solve.

We calculate the singularities (we want the one on the upper-half plane)

$a+bz^2=0\:; z_{0,1}=\pm \:i\sqrt{\frac{a}{b}}$

As it is a pole of order n, we calculate the residue as follows:

$\frac{1}{(n-1)!}g^{(n-1)}(z_0)\:\text{where}\:g(z)=(z-z_0)^n\frac{1}{(a+bz^2)^n}=\frac{1}{(z-z_1)^n}=(z+i\sqrt{\frac{a}{b}})^{-n}$

Which we can calculate recursively and results in

$\frac{-(2n-2)!}{(n-1)!^2}\frac{1}{(2i\sqrt{\frac{a}{b}})^{2n-1}}$

So

$I=2\pi i\:\frac{-(2n-2)!}{(n-1)!^2}\frac{1}{(2i\sqrt{\frac{a}{b}})^{2n-1}}=\frac{(2n-2)!}{(n-1)!^2}\frac{\pi}{2^{2n-2}\:(\sqrt{\frac{a}{b}})^{2n-1}}$

If a set values to the parameters, I don't get the same result as WolframAlpha, but I don't know what I'm doing wrong.

Thanks :)

1 Answers1

1

Your evalution of the residue was flawed. Note that

$$\begin{align} (z-z_0)^n \frac1{(a+bz^2)^n}&=\frac{(z-z_0)^n}{b^n (z-z_0)^n(z-z_1)^n}\\\\ &=\frac{1}{b^n(z-z_1)^n} \end{align}$$

Now, take $n-1$ derivatives with respect to $z$, then set $z=z_0$, and finally divide by $(n-1)!$.

Alternatively, we can simplify things by noting that

$$\frac{d^{n-1}}{da^{n-1}}\int_{-\infty}^\infty \frac{1}{a+bz^2}\,dz=(-1)^{n-1}(n-1)! \int_{-\infty}^\infty \frac{1}{(a+bz^2)^n}\,dz \tag1$$

Evaluting the integral on the left-hand side of $(1)$ using contour integration, we find that

$$\begin{align} \int_{-\infty}^\infty \frac1{a+bz^2}\,dz&=2\pi i \text{Res}\left(\frac1{a+bz^2}, z=i\sqrt{a/b}\right)\\\\ &=\pi/\sqrt{ab}\tag2 \end{align}$$

Now differentiate the right-hand side of $(2)$ $n-1$ times with respect to $a$ and divide by $(-1)^{n-1}(n-1)!$. Can you finish?

Mark Viola
  • 184,670
  • You're absolutely right, I made a pretty basic mistake when factoring the function. Your way of solving the integral seems extremely original to me, thank you so much for the contribution. If it's not too much to ask, how does your intuition lead you to solve it this way? – Pablo García García Jan 29 '25 at 11:47
  • 1
    @PabloGarcíaGarcía I suppose that the use of "Feynman's trick," which is an application to Leibniz's rule for differentiating under the integral, is one of the methodologies I saw years ago. – Mark Viola Jan 29 '25 at 14:58