I'm looking for a closed-form solution to this infinite series:


My attempt

All I've really been able to do is confirm this series converges for the specified values of $\alpha$ via the integral test:

$$\begin{align} S(\alpha)\leq I(\alpha)&=\int_{0}^{\infty}\frac{\log{(1+x)}}{(1+x)^{\alpha}-1}\mathrm{d}x\\ &=\int_{0}^{\infty}\frac{u\,e^u}{e^{\alpha u}-1}\mathrm{d}u\\ &=\int_{0}^{\infty}ue^u\sum_{n=1}^{\infty}e^{-n\alpha u}\mathrm{d}u\\ &=\sum_{n=1}^{\infty}\int_{0}^{\infty}ue^ue^{-n\alpha u}\mathrm{d}u\\ &=\sum_{n=1}^{\infty}\frac{1}{(n\alpha-1)^2},~~~\text{for }\Re(\alpha)>1\\ &=\frac{1}{\alpha^2}\psi^{(1)}{\left(1-\frac{1}{\alpha}\right)} \end{align}$$

where $\psi^{(1)}(x)$ is the first derivative of the digamma function.

I thought about expanding the logarithmic part in terms of its power series to write $S(\alpha)$ as a double-sum, and then try switching the order of summation, e.g.:

$$\begin{align} S(\alpha)&=\sum_{n=1}^{\infty}\frac{\log{(1+n)}}{(1+n)^{\alpha}-1}\\ &=-\sum_{n=1}^{\infty}\frac{1}{(1+n)^{\alpha}-1}\sum_{k=1}^{\infty}\frac{(-1)^kn^k}{k}\\ &=-\sum_{n=1}^{\infty}\sum_{k=1}^{\infty}\frac{(-1)^k}{k}\frac{n^k}{(1+n)^{\alpha}-1}\\ &=-\sum_{k=1}^{\infty}\frac{(-1)^k}{k}\sum_{n=1}^{\infty}\frac{n^k}{(1+n)^{\alpha}-1}. \end{align}$$

But I'm not sure how to proceed from there. Suggestions?

  • 4
    $\begingroup$ I would be astonished if this sum had a closed form. I'd be surprised if $S(2)$ had a closed form. $\endgroup$ – Gerry Myerson Jun 16 '14 at 9:32
  • $\begingroup$ @David H You can not expand $\log (1+n)$ as $\sum_{k=1}^{\infty}\frac{(-1)^{k-1}n^k}{k}$ as $n>1$. The radius of convergence of the power series for $\log (1+x)$ is equal to $1$. $\endgroup$ – Olivier Oloa Sep 5 '15 at 20:41

I'd be surprised if $S(2)$ had a closed form.

$S(2)$ admits a closed form in terms of the poly-Stieltjes constants unveiled here.

Moreover, $S(3),S(4),S(5),\cdots$, admit a closed form in terms of these special functions.

We have the following result.

Proposition. Let $\alpha=2,3,4,5,\ldots$.

Then $$ \bbox[15px,border:1px solid orange]{S(\alpha)=\frac1\alpha\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\:\gamma_1\!\!\left(1,1-e^{2k\pi i/\alpha}\right)} \tag1 $$ where $$ \gamma_1(a,b) = \lim_{N\to+\infty}\left(\sum_{n=1}^N \frac{\log (n+a)}{n+b}-\frac12 \log^2 \!N\right). \tag2 $$

Proof. For $\alpha=2,3,\ldots $, we have $$ \begin{align} S(\alpha):&=\lim_{N \to \infty}\sum_{n=1}^N\frac{\log{(1+n)}}{(1+n)^{\alpha}-1}\\ &=\frac1\alpha\lim_{N \to \infty}\left(\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\sum_{n=1}^N\frac{\log (n+1)}{n+1-e^{2k\pi i/\alpha}}\right)\\ &=\frac1\alpha\lim_{N \to \infty}\left(\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\left(\sum_{n=1}^N\frac{\log (n+1)}{n+1-e^{2k\pi i/\alpha}}-\frac12\log^2 N\right)+\frac12\log^2 N\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\right)\\ &=\frac1\alpha\lim_{N \to \infty}\left(\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\left(\sum_{n=1}^N\frac{\log (n+1)}{n+1-e^{2k\pi i/\alpha}}-\frac12\log^2 N\right)+0\right)\\ &=\frac1\alpha\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\lim_{N \to \infty}\left(\sum_{n=1}^N\frac{\log (n+1)}{n+1-e^{2k\pi i/\alpha}}-\frac12\log^2 N\right)\\ &=\frac1\alpha\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\gamma_1(1,1-e^{2k\pi i/\alpha}) \end{align} $$ where we have used a partial fraction decomposition over the complex numbers, $$ \frac{1}{X^\alpha-1}=\frac1\alpha\sum_{k=0}^{\alpha-1}\frac{e^{2k\pi i/\alpha}}{X-e^{2k\pi i/\alpha}},\quad \alpha=1,2,3,\ldots, \tag3 $$ giving $$ \frac{\log (n+1)}{(n+1)^\alpha-1}=\frac1\alpha\sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}\frac{\log (n+1)}{n+1-e^{2k\pi i/\alpha}},\quad \alpha=1,2,3,\ldots, $$ and we have used the standard fact that $\displaystyle \sum_{k=0}^{\alpha-1}e^{2k\pi i/\alpha}=0$.

  • $\begingroup$ Thank you very much for this response! I remember I had a very specific reason for asking it at the time (presumably it provided a way to solve some other difficult problem), but alas I've completely forgotten what I wanted to use it for. Out of curiosity, what sparked you to address this problem after all this time? $\endgroup$ – David H Sep 8 '15 at 20:28
  • $\begingroup$ You are welcome! I just stumbled upon your question. Thanks. @DavidH $\endgroup$ – Olivier Oloa Sep 9 '15 at 5:29

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.