40
$\begingroup$

I need to find a closed-form for the following integral. Please give me some ideas how to approach it: $$\int_0^{\infty}\frac{x^3}{(x^4+1)(e^x-1)}\mathrm dx$$

$\endgroup$
19
$\begingroup$

$$\int_0^{\infty}\frac{x^3}{(x^4+1)(e^x-1)}\mathrm dx=-\frac{\pi}{\sqrt8}-\log\sqrt{2\pi}-\frac{1}{2}\Re\ \psi\left(\frac{\sqrt[4]{-1}}{2\pi}\right),$$ where $\Re\ \psi(z)$ denotes the real part of the digamma function, $\psi(z)=\frac{\Gamma'(z)}{\Gamma(z)}$.


Solution: Use the approach from sos440's answer. To calculate the infinite sum and simplify the result we need the following: $$\frac{1}{n\left((2\,\pi\,n)^4+1\right)}=\frac{1}{n}-\frac{1}{4}\left(\frac{1}{n+\frac{(-1)^{1/4}}{2 \pi }}+\frac{1}{n+\frac{(-1)^{-1/4}}{2\pi}}+\frac{1}{n+\frac{(-1)^{3/4}}{2 \pi }}+\frac{1}{n+\frac{(-1)^{-3/4}}{2\pi}}\right),$$ $$\sum_{n=1}^\infty\left(\frac{1}{n}-\frac{1}{n+x}\right)=\gamma+\psi(1+x),$$ and the formulas (8), (9) from here.

$\endgroup$
  • 14
    $\begingroup$ A clean answer! $\endgroup$ – Sangchul Lee May 17 '13 at 6:18
  • $\begingroup$ Anyway, could you show me how you derived this neat formula? $\endgroup$ – Sangchul Lee May 17 '13 at 14:33
35
$\begingroup$

My calculation shows that

\begin{align*} \int_{0}^{\infty} \frac{x^3}{(x^4 + 1)(e^x - 1)} \, dx &= \frac{\gamma}{2} - \log\sqrt{2\pi} + \frac{\pi}{4} \frac{\sin\frac{1}{\sqrt{2}}}{\cosh\frac{1}{\sqrt{2}} - \cos\frac{1}{\sqrt{2}}} \\ &\quad - \frac{1}{2}\sum_{n=1}^{\infty} \frac{1}{n\left(1 + (2\pi n)^4\right)} \\ &\approx 0.389075976914101580976629 \cdots. \end{align*}

My solution is divided into several steps:

Step 1. Let us introduce the function

$$ I(s) = \int_{0}^{\infty} \frac{x^{s}}{(x^4+1)(e^{x}-1)} \, dx $$

It is easy to see that

$$ I(s) + I(s+4) = \Gamma(s+1)\zeta(s+1). \tag{1} $$

This allows us to extend $I(s)$ as a meromorphic function on $\Bbb{C}$.

Step 2. Consider a contour $C$ starting from $\infty + \epsilon i$, making a counter-clockwise turn around $z = 0$ and going back to $\infty - \epsilon i$ as follows:

enter image description here

If we use a logarithm function with the branch cut $[0, \infty)$, we see that

$$ (e^{2\pi i s} - 1)I(s) = \int_{C} \frac{z^{s}}{(z^4+1)(e^z - 1)} \, dz. \tag{2} $$

Also, for $ 1 < s < 2$ we can confirm that $(2)$ is rewritten as

$$ (e^{2\pi i s} - 1)I(s) = \lim_{n\to\infty} \int_{C_{n}} \frac{z^{s}}{(z^4+1)(e^z - 1)} \, dz, $$

where $C_n$ is the contour given by

enter image description here

Here, the condition $1 < s < 2$ is introduced in order to create an appropriate decay speed for the integral along the contour $C - C_{n}$. By applying the Cauchy integration formula, we have

$$ (e^{2\pi i s} - 1)I(s) = -2\pi i \sum_{\omega^4 = -1} \operatorname{Res}_{z=\omega} \frac{z^{s}}{(z^4+1)(e^z - 1)} - 2\pi i \sum_{n\neq 0} \operatorname{Res}_{z=2\pi i n} \frac{z^{s}}{(z^4+1)(e^z - 1)}. $$

Simplifying,

$$ I(s) = \frac{\pi}{\sin \pi s} \frac{e^{-i\pi s}}{4} \sum_{\omega^4 = -1} \frac{\omega^{s+1}}{e^{\omega} - 1} - \frac{2^{s}\pi^{s+1}}{\sin \frac{\pi s}{2}} \sum_{n=1}^{\infty} \frac{n^s}{1+(2\pi n)^4} \tag{3} $$

Since both sides define a meromorphic function for $\Re s < 3$, they coincide on this range.

Step 3. Combining $(1)$ and $(3)$, we have

\begin{align*} I(s+3) &= \Gamma(s)\zeta(s) - I(s-1) \\ &= \Gamma(s)\zeta(s) - \frac{\pi}{\sin \pi s} \frac{e^{-i\pi s}}{4} \sum_{\omega^4 = -1} \frac{\omega^{s}}{e^{\omega} - 1} - \frac{(2\pi)^{s}}{2\cos \frac{\pi s}{2}} \sum_{n=1}^{\infty} \frac{n^s}{n \left( 1+(2\pi n)^4 \right)} \end{align*}

Taking $s \to 0$, we obtain the desired result.

$\endgroup$
  • 4
    $\begingroup$ Nice technique! (+1) Apparently Mathematica can express the remaining infinite sum as a sum of digamma functions over the roots of some 4th order polynomial. $\endgroup$ – Start wearing purple May 16 '13 at 8:58
  • 1
    $\begingroup$ @O.L.: what version Mathematica? V. 8.0.4 could't do anything with this. $\endgroup$ – Ron Gordon May 16 '13 at 21:21
  • 1
    $\begingroup$ @Jonathan, I drew them using Mathematica. $\endgroup$ – Sangchul Lee May 18 '13 at 0:22
  • 1
    $\begingroup$ @sos440 Could you please tell what Mathematica functions do you use to draw contour diagrams like this? mathematica.stackexchange.com/questions/25626/… $\endgroup$ – Vladimir Reshetnikov May 22 '13 at 0:16
  • 1
    $\begingroup$ It's available since 6.0.x where x is some x > 0. Don't remember exactly. $\endgroup$ – Szabolcs May 22 '13 at 3:48
8
$\begingroup$

$\newcommand{\+}{^{\dagger}} \newcommand{\angles}[1]{\left\langle\, #1 \,\right\rangle} \newcommand{\braces}[1]{\left\lbrace\, #1 \,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\, #1 \,\right\rbrack} \newcommand{\ceil}[1]{\,\left\lceil\, #1 \,\right\rceil\,} \newcommand{\dd}{{\rm d}} \newcommand{\down}{\downarrow} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,{\rm e}^{#1}\,} \newcommand{\fermi}{\,{\rm f}} \newcommand{\floor}[1]{\,\left\lfloor #1 \right\rfloor\,} \newcommand{\half}{{1 \over 2}} \newcommand{\ic}{{\rm i}} \newcommand{\iff}{\Longleftrightarrow} \newcommand{\imp}{\Longrightarrow} \newcommand{\isdiv}{\,\left.\right\vert\,} \newcommand{\ket}[1]{\left\vert #1\right\rangle} \newcommand{\ol}[1]{\overline{#1}} \newcommand{\pars}[1]{\left(\, #1 \,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\pp}{{\cal P}} \newcommand{\root}[2][]{\,\sqrt[#1]{\vphantom{\large A}\,#2\,}\,} \newcommand{\sech}{\,{\rm sech}} \newcommand{\sgn}{\,{\rm sgn}} \newcommand{\totald}[3][]{\frac{{\rm d}^{#1} #2}{{\rm d} #3^{#1}}} \newcommand{\ul}[1]{\underline{#1}} \newcommand{\verts}[1]{\left\vert\, #1 \,\right\vert} \newcommand{\wt}[1]{\widetilde{#1}}$ $\ds{\int_{0}^{\infty}{x^{3} \over \pars{x^{4} + 1}\pars{\expo{x} - 1}}\,\dd x: \ {\large ?}}$

\begin{align}&\color{#c00000}{% \int_{0}^{\infty}{x^{3} \over \pars{x^{4} + 1}\pars{\expo{x} - 1}}\,\dd x} =\half\int_{0}^{\infty}\pars{{1 \over x^{2} + \ic} + {1 \over x^{2} - \ic}}\, {x \over \expo{x} - 1}\,\dd x \\[3mm]&=\Re\int_{0}^{\infty}{x\,\dd x \over \pars{x^{2} + \ic}\pars{\expo{x} - 1}} =\Re\int_{0}^{\infty}{x\,\dd x \over \bracks{x^{2} + \ic\,/\pars{4\pi^{2}}}\pars{\expo{2\pi x} - 1}}\tag{1} \end{align}

The last integral in $\pars{1}$ is related to the Digamma function $\ds{\Psi\pars{z}}$ by means of the identity ${\bf\mbox{6.3.21}}$: $$ \Psi\pars{z} = \ln\pars{z} - {1 \over 2z} -2\int_{0}^{\infty} {t\,\dd t \over \pars{t^{2} + z^{2}}\pars{\expo{2\pi t} - 1}}\,,\qquad \verts{{\rm arg}\pars{z}} < {\pi \over 2} $$

Then, $\pars{1}$ is reduced to $\ds{\pars{~\mbox{with}\ \root{\ic} = \expo{\ic\pi/4} ={1 \over \root{2}}\pars{1 + \ic}~}}$: \begin{align} &\color{#44f}{\large% \!\!\!\!\!\int_{0}^{\infty}{x^{3} \over \pars{x^{4} + 1}\pars{\expo{x} - 1}}\,\dd x} =\Re\bracks{-\,\half\,\Psi\pars{\root{\ic} \over 2\pi} +\half\,\ln\pars{\root{\ic} \over 2\pi} - {1 \over 4}\,{2\pi \over \root{\ic}}} \\[3mm]&=\color{#44f}{\large% -\,\half\,\Re\Psi\pars{\root{\ic} \over 2\pi} -\half\,\ln\pars{2\pi} -{\root{2} \over 4}\,\pi} \approx 0.3891 \end{align}

$\endgroup$
  • $\begingroup$ Is there a way to simplify the $\Re\Psi(\cdot)$? $\endgroup$ – Akiva Weinberger Aug 27 '14 at 14:12
  • $\begingroup$ @columbus8myhw I don't think so as far as I know. I was checking this page but it seems there isn't any thing simpler than that. Thanks. $\endgroup$ – Felix Marin Aug 27 '14 at 17:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.