1

How would you evaluate $$\sum_{n=1}^{\infty}\left(\frac{\sin(n)}{n}\right)^2$$ Wolfram|Alpha says it equals $\;\pi/2-1/2.\;$ If the solution is complicated, ... I can handle complicated.

Angelo
  • 12,328
Royce Martin
  • 101
  • 6
  • Some context would help. Do you know about Fourier series and Parseval's theorem? –  Jun 14 '21 at 07:48
  • 5
  • 2
    I disagree that the proposed duplicate https://math.stackexchange.com/questions/3611785/evaluating-sum-k-1-infty-left-frac-sintkk-right2 is in fact a duplicate. In that question, the OP specifically requested a solution using the Poisson summation formula, so the answers use that method. But there are other approaches, including my answer below which uses Parserval's theorem (and which would therefore be inappropriate as an answer to the proposed duplicate) but which is in fact easier because computing Fourier coefficients for a rectangular pulse is easier than for a triangular pulse. –  Jun 14 '21 at 08:31

2 Answers2

5

$\displaystyle\sum_{n=1}^∞ \dfrac{\sin(nx)}{n} =\frac{1}{2i} \sum_{n=1}^∞ \dfrac{(e^{ix})^n}{n}-\dfrac{(e^{-ix})^n}{n}=-\frac{1}{2i}({\log(1-e^{ix})-\log(1-e^{-ix})})$

$= \displaystyle \dfrac{1}{2i} \log\left(\frac{(1-e^{-ix})}{(1-e^{ix})}\right)= \dfrac{1}{2i} \log(-e^{-ix})= \dfrac{\log(e^{iπ})-\log(e^{ix})}{2i} =\dfrac{π-x}{2}. $

Now $\displaystyle\int_0^x \sum_{n=1}^∞\dfrac{ \sin(nx)}{n}dx = \int_0^x \dfrac{π-x}{2}dx $ $\implies \displaystyle -\sum_{n=1}^∞ \left(\dfrac{\cos(nx)}{n^2} -\dfrac{\cos(0)}{n^2}\right)= \dfrac{2πx-x^2}{4}$ $\implies \displaystyle\sum_{n=1}^∞ \dfrac{2\sin^2(\frac{nx}{2})}{n^2} = \dfrac{2πx-x^2}{4} .$

Put $x=2$ the answer will be $\displaystyle\sum_{n=1}^∞ \dfrac{\sin^2(n)}{n^2} =\dfrac{π-1}{2} $.

Edit

In this way we can also solve Basel problem.

$\displaystyle\sum_{n=1}^∞ \dfrac{\sin^2(\frac{nπ}{2})}{n^2} = \dfrac{\frac{π}{2}(π-\frac{π}{2})}{2} = \dfrac{π^2}{8} $

Which means $\displaystyle\dfrac{1}{1^2} +\dfrac{1}{3^2} +\dfrac{1}{5^2} +...= \dfrac{π^2}{8} \implies \dfrac{3\zeta(2)}{4} =\dfrac{π^2}{8} \implies \zeta(2)=\dfrac{π^2}{6}$

  • If your answer is different from those found in the duplicate, it is advised to answer that question instead. If a rather similar anwer is given for the duplicate, do not answer at all. – Gary Jun 14 '21 at 08:27
  • @Gary I disagree that the proposed duplicate https://math.stackexchange.com/questions/3611785/evaluating-sum-k-1-infty-left-frac-sintkk-right2 is in fact a duplicate. In that question, the OP specifically requested a solution using the Poisson summation formula, so the existing answers use that method and any future answers to that question should do the same. –  Jun 14 '21 at 08:29
2

The proposed duplicate Evaluating $\sum_{k=1}^{\infty}\left(\frac{\sin(tk)}{k}\right)^2$ does this the hard way, using the Poisson summation formula, which requires finding the Fourier series for a triangular function. (The OP of that question explicitly requested a Poisson summation formula solution.)

It's easier to compute the Fourier series for a rectangular function and use Parseval's theorem, so my solution uses that approach.

Let $f$ be the $2\pi$-periodic function defined as follows: $$f(x) = \begin{cases} 1 & \text{if }|x| < 1 \\ 0 & \text{if }1 \leq |x| \leq \pi \\ \end{cases}$$ Then $f$ has a Fourier series representation $$f(x) = \sum_{n=-\infty}^{\infty}a_n e^{inx}$$ where $$a_n = \frac{1}{2\pi}\int_{-\pi}^{\pi}f(x)e^{-inx}$$ for each $n \in \mathbb Z$. When $n=0$, this reduces to $$\begin{aligned} a_n &= \frac{1}{2\pi}\int_{-\pi}^{\pi}f(x)\ dx \\ &= \frac{1}{2\pi}\int_{-1}^{1}\ dx \\ &= \frac{1}{\pi} \end{aligned}$$ For $n \neq 0$ we have $$\begin{aligned} a_n &= \frac{1}{2\pi}\int_{-\pi}^{\pi}f(x)e^{-inx}\ dx \\ &= \frac{1}{2\pi}\int_{-1}^{1}e^{-inx}\ dx \\ &= \frac{1}{-i2\pi n}(e^{-in} - e^{in}) \\ &= \frac{\sin(n)}{\pi n} \end{aligned}$$ Now we can apply Parseval's theorem, which says that if $a_n$ are the Fourier coefficients of $f$ and $b_n$ are the Fourier coefficients of $g$, then $$\frac{1}{2\pi}\int_{-\pi}^{\pi}f(x)\overline g(x)\ dx = \sum_{n=-\infty}^{\infty}a_n \overline b_n,$$ where the bar denotes complex conjugation. (In words, the inner product of $f$ and $g$ equals the inner product of their Fourier coefficient sequences.) In particular, for the special case $f=g$, Parseval's theorem gives us $$\frac{1}{2\pi}\int_{-\pi}^{\pi}|f(x)|^2\ dx = \sum_{n=-\infty}^{\infty}|a_n|^2$$ For our function $f(x)$, we have $|f(x)|^2 = 1$ for $|x| < 1$ and $|f(x)|^2 = 0$ otherwise, so the integral on the LHS becomes $$\frac{1}{2\pi}\int_{-1}^{1}\ dx = \frac{1}{\pi}$$ And, since our Fourier coefficients are real and even ($a_n = a_{-n}$), the sum on the RHS becomes $$a_0^2 + 2\sum_{n=1}^{\infty}a_n^2 = \frac{1}{\pi^2} + \frac{2}{\pi^2}\sum_{n=1}^{\infty}\left(\frac{\sin(n)}{n}\right)^2 = \frac{1}{\pi^2} + \frac{2}{\pi^2} S$$ where $S$ is your desired sum. Summarizing: $$\frac{1}{\pi} = \frac{1}{\pi^2} + \frac{2}{\pi^2} S$$ Multiplying by $\pi^2$ gives us $$\pi = 1 + 2S$$ and therefore $\displaystyle S = \frac{\pi - 1}{2}$ as desired.