$(1).$ Show that: $$ \lim_{s\to0^+}\,\left[\sum_{n=1}^{\infty}\cos\left(\pi\frac{n}{m}\right)\frac{1}{n^s}\right]=\color{red}{-\frac{1}{2}} \quad\colon\space\forall\,m\in\mathbb{N}^{+}\tag{1} $$ $(2).$ Find a closed-form for: $$ \lim_{s\to0^+}\,\left[\sum_{n=1}^{\infty}\sin\left(\pi\frac{n}{m}\right)\frac{1}{n^s}\right]=\color{red}{\,\,\,\,?\,\,\,\,} \quad\colon\space\,\,\,\,m\in\mathbb{N}^{+}\tag{2} $$

Both series converge by Dirichlet's test for $\mathrm{Re}(s)>0$.
I could not find a good reason way the first series shall converge to the same constant!!
Thanks for you help.

  • $\begingroup$ The limit in $(2)$ equals $\dfrac12 \cot\left(\dfrac \pi{2m}\right)$. $\endgroup$ – Tianlalu Oct 10 '18 at 0:36
  • $\begingroup$ @Tianlalu: Thanks, but how to show it (if it is the correct answer)? $\endgroup$ – Hazem Orabi Oct 10 '18 at 0:39
  • $\begingroup$ If we can prove $\lim_{s\to0^+}\operatorname{Li}_s(z)= \operatorname{Li}_0(z)$, then the question is trivial. $\endgroup$ – Kemono Chen Dec 14 '18 at 0:54

$$f(z,s)=\sum_{n=1}^{\infty}\frac{z^n}{n^s}$$ can be analytically continue to all complex $z$ except $z\in \mathbf R \bigwedge z\ge1$ and for complex $s$ in the disk $|s|<1$. So we need only to evaluate $f(z,s=0)$ for $|z|<1$. That is $$f(z,s=0)=\sum_{n=1}^{\infty}z^n=\frac z{1-z},\, \forall |z|<1.$$ Analytically continue this fraction to $|z|=1\bigwedge z\ne 1$, we have $\forall\theta\in\mathbf R\bigwedge\theta\ne0$ $$\lim_{s\rightarrow 0}f(e^{i\theta},s)= \lim_{z\rightarrow e^{i\theta}\atop s\rightarrow 0} f(z,s)=\lim_{z\rightarrow e^{i\theta}} f(z,s=0)=\frac {e^{i\theta}}{1- e^{i\theta}}=-\frac12+\frac i2\cot\frac \theta 2.$$ Separating out the real and imaginary parts of the above equation, we obtain the desired results.

  • $\begingroup$ ... can be analytically continue to the whole complex plane except at $z\in \mathbf R \bigwedge z\ge1 \bigwedge s=1$ No, consider $s=2$, it is not even meromorphic in the complex plane. And the imaginary part is missing a factor $\frac12$. Just to be rigorous, the general idea of your answer is correct. $\endgroup$ – Kemono Chen Dec 14 '18 at 3:37
  • $\begingroup$ @KemonoChen: You are right. Thank you. I have made the correction. I will add even more details later. $\endgroup$ – Hans Dec 14 '18 at 9:20
  • $\begingroup$ I wish if I can split the bounty, many thanks. $\endgroup$ – Hazem Orabi Dec 18 '18 at 9:37
  • $\begingroup$ @HazemOrabi: Haha. No worries. $\endgroup$ – Hans Dec 18 '18 at 21:12

Denote $\frac\pi m=x$, rewrite $$\lim_{s\to0^+}\left[\sum_{n=1}^{\infty}\cos\left(\pi\frac{n}{m}\right)\frac{1}{n^s}\right]=\lim_{s\to0^+}\sum_{n=1}^\infty\frac{e^{ixn}+e^{-ixn}}{2n^s}\\ =\lim_{s\to0^+}\sum_{n=1}^\infty\frac{e^{ixn}+e^{-ixn}}{2n^s}\\ =\lim_{s\to0^+}\frac12\left(\operatorname{Li}_s(e^{ix})+\operatorname{Li}_s(e^{-ix})\right)$$where $\operatorname{Li}_s(z)$ is the analytic continuation of $\sum_{n=1}^\infty\frac{z^n}{n^s}$ and is named polylogarithm function.
Notice that $\operatorname{Li}_s(z)$ is holomorphic in $s$ in a neighborhood of the origin whatever $z$ is, hence continuous in $s\in(0,\epsilon)$. We have $$\lim_{s\to0^+}\frac12\left(\operatorname{Li}_s(e^{ix})+\operatorname{Li}_s(e^{-ix})\right)=\frac12\left(\operatorname{Li}_0(e^{ix})+\operatorname{Li}_0(e^{-ix})\right)=\frac{1}{2} \left(\frac{e^{i x}}{1-e^{i x}}+\frac{e^{-i x}}{1-e^{-i x}}\right)=\frac12$$ Similarly, $$\lim_{s\to0^+}\,\left[\sum_{n=1}^{\infty}\sin\left(\pi\frac{n}{m}\right)\frac{1}{n^s}\right]=\frac1{2i}\left(\operatorname{Li}_0(e^{ix})-\operatorname{Li}_0(e^{-ix})\right)=\frac{1}{2} i \left(\frac{e^{-i x}}{1-e^{-i x}}-\frac{e^{i x}}{1-e^{i x}}\right)=\frac12\cot\frac x2$$ Evaluation of $\operatorname{Li}_0(z)$
$$\sum_{n=1}^\infty\frac{z^n}{n^0}=z\sum_{n=0}^\infty z^n=\frac{z}{1-z}(|z|<1)$$ Since $\operatorname{Li}_0(z)$ is the analytic continuation of it, $\operatorname{Li}_0(z)=\frac{z}{1-z}$.

  • $\begingroup$ According to this development, $m$ need not be restricted to the natural numbers. $\endgroup$ – Mark Viola Dec 14 '18 at 2:08
  • $\begingroup$ It is slightly surreal to see a solution jumping to fine properties of polylogarithms to "help" an OP whose background is obviously lacking these. $\endgroup$ – Did Dec 15 '18 at 6:51

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.