# Improper integral of $\frac{x}{e^x-1}$

This integral came up in an exercise on the estimation of the specific heat of a 1-D solid and is probably a standard integral, possibly one that can be solved by contour integration:

\begin{equation} \int_0^{+\infty} \frac{x}{e^x-1} dx \end{equation}

I have some rudimentary knowledge of contour integrals, but I can't come up with a proper path, also because of the many singularities along the imaginary axis. Any suggestion?

## 2 Answers

Make a change or variables, $e^{-x}=u$, and write the integral as

$$\int_0^\infty \frac{x e^{-x}}{1 - e^{-x}}dx$$

Now substituting gives

$$-\int_0^1 \frac{\log u}{1 - u}du$$

This is a known integral that evaluates to $\dfrac{\pi^2}{6}$.

If you want to prove it, you can use the dilogarithm. Let $1-u=x$, so that

$$-\int_0^1 \frac{\log(1 - x)}{x}dx$$

Now, since we're working on $(0,1)$ it is legitimate to use

$$\frac{{ - \log \left( {1 - x} \right)}}{x} = \sum_{n = 1}^\infty {\frac{{{x^{n - 1}}}}{n}}$$

Integrating termwise gives

$$\int_0^t \frac{-\log(1 - x)}{x} dx = \sum_{n = 1}^\infty \frac{t^n}{n^2} = \mathrm{Li}_2 (t)$$

Evaluating at $t=1$ gives

$$-\int_0^1 \frac{\log(1 - x)}{x} dx = \sum_{n = 1}^\infty \frac{1}{n^2} = \frac{\pi^2}{6}$$

Any text book which dicusses the Riemann Zeta function will likely have this.

We have for $\Re(z) \gt 1$, that

$$\zeta(z) \Gamma(z) = \int_{0}^{\infty} \frac{t^{z-1}}{e^t - 1} \text{d}t$$

Your integral is therefore

$$\zeta(2)\Gamma(2) = \frac{\pi^2}{6}$$

For an online discussion, see this: http://www.math.utah.edu/~milicic/zeta.pdf

• I pimped your Re symbol. – Pedro Tamaroff Apr 10 '12 at 20:18
• @PeterT.off: Thanks! – Aryabhata Apr 10 '12 at 20:19