0

I came across the following identity,

$$ \int_{-\infty} ^\infty dx \frac{e^{-i kx}}{e^{-ax} +1} = \frac{2\pi i}{a} \sum_{n=0}^{\infty} e^{-\frac{(2n+1)\pi k}{a}} = \frac{\pi i }{a\mathrm{sinh}\frac{k \pi}{a}} $$ I can to some extent see the first equality by doing the contour integration in the lower half plane which has poles at $\frac{i\pi(2 n +1)}{a}$. I however get the summation for negative integers as well, $$ \int_{-\infty} ^\infty dx \frac{e^{-i kx}}{e^{-ax} +1} = \frac{2\pi i}{a} \sum_{n=-\infty}^{\infty} e^{-\frac{(2n+1)\pi k}{a}} $$. I however have no idea how one can get the second equality. I am actually more interested in a slightly different integral, $\int_{-\infty} ^\infty dx \frac{e^{-i kx}}{e^{-ax} -1} $. Any help as to how one can obtain these identities is appreciated.

levitt
  • 339
  • 1
    $$\frac{2\pi i}{a}\sum_{n=0}^\infty e^{-(2n+1)\pi k/a}=\frac{2\pi ie^{-\pi k/a}}{a}\sum_{n=0}^\infty\big(e^{-2\pi k/a}\big)^n=\frac{2\pi ie^{-\pi k/a}}{a\big(1-e^{-2\pi k/a}\big)}$$ – mr_e_man Jul 16 '20 at 20:23
  • 1
    These integrals are very similar to the one you are trying to evaluate. – Zakhurf Jul 16 '20 at 21:23
  • Thanks, I have got it. I was mistaken to think that all the poles lie in the lower half plane. Half of them lie above which don't contribute. – levitt Jul 17 '20 at 09:41

1 Answers1

2

The second equality can be proven by writing the summation as a geometric progression.

\begin{align} \sum_{n=0}^\infty e^{-(2n+1)\frac{\pi k}{a}}&= e^{-\frac{\pi k}{a}}\sum_{n=0}^\infty\big(e^{\frac{-2\pi k}{a}}\big)^n \\&=\frac{e^\frac{-\pi k}{a}}{1-e^\frac{-2\pi k}{a}} \\ &= \frac{1}{e^{\frac{\pi k}{a}}-e^{-\frac{\pi k}{a}}} \\ &= \frac{1}{2\sinh{\frac{\pi k}{a}}} \end{align}

Note that the infinite summation converges only if $\frac{k}{a} > 0$.

Zakhurf
  • 878
  • 5
  • 13
  • Tank you. That answers part of my question. I am still looking for the first equality. I know it needs contour integration in the lower half plane. But I can't get it fully correctly. – levitt Jul 17 '20 at 05:01