7
$\begingroup$

What I'd like to find is an identity for $$\sum_{k=0}^{n-1}\csc\left(x+ k \frac{\pi}{n}\right)\csc\left(y+ k \frac{\pi}{n}\right)$$

here it can be shown that where $x=y$,

$$n^2 \csc^2(nx) = \sum_{k=0}^{n-1}\csc^2\left(x+ k \frac{\pi}{n}\right)$$

I also looked at possible residue methods involving csc

I thought I could use $\left( \sum^{n-1}_{j=0}Z_j\right)^2 = \sum^{n-1}_{j=0} Z_j^2 +\sum^{n-1}_{j=0}\sum^{n-1}_{i\neq j} Z_jZ_i$ but I have had no luck so far.

$\endgroup$

2 Answers 2

6
$\begingroup$

$$\bbox[5pt,border:2pt solid black]{\sum_{k=0}^{n-1}\csc\left(x+\frac{k\pi}n\right)\csc\left(y+\frac{k\pi}n\right)=\frac{n\sin n(x-y)}{\sin(x-y)\sin nx\sin ny}}$$

Steps to get it: $\cos nx$ is a polynomial in $\cos x$, then $\cos nx-\cos ny=F(\cos x)$ may be considered as a polynomial in $\cos x$ (with coefficients depending on $y$). Assume (temporarily) that $y$ is not a rational multiple of $\pi$; then the roots of $F(z)$ are given by $z_k=\cos(y+2k\pi/n)$ for $0\leqslant k\leqslant n-1$. Hence $F(z)=A\prod_{k=0}^{n-1}(z-z_k)$, with $A$ depending on $y$ only (in fact $A=2^{n-1}$ but we don't need it here; also, now the assumption on $y$ can be dropped by continuity). If we put $z=\cos x$ back, and take the logarithmic derivative w.r.t. $x$, we get $$\frac{n\sin nx}{\cos nx-\cos ny}=\sum_{k=0}^{n-1}\frac{\sin x}{\cos x-\cos(y+2k\pi/n)}.$$

To get the formula at the beginning, put $(x,y)\gets(x-y,x+y)$.

$\endgroup$
3
  • $\begingroup$ changing $x$ to $x-y$ and $y$ to $x+y$ in $\frac{n\sin nx}{\cos nx-\cos ny}$ is not resulting in ${\frac{n\sin n(x-y)}{\sin(x-y)\sin nx\sin ny}}$ is there an extra step of some kind needed here? $\endgroup$
    – onepound
    Commented May 17, 2023 at 21:17
  • $\begingroup$ @onepound: $\cos(A-B)-\cos(A+B)=2\sin A\sin B$. $\endgroup$
    – metamorphy
    Commented May 18, 2023 at 0:23
  • $\begingroup$ I see and then multiply by $\frac{1}{\sin(x-y)}$ to get ${\frac{n\sin n(x-y)}{\sin(x-y)\sin nx\sin ny}}$ all parts understood now thanks. $\endgroup$
    – onepound
    Commented May 18, 2023 at 8:50
2
$\begingroup$

Here is an alternate proof of the identity given by metamorphy:

\begin{equation} \sum_{k=0}^{n-1}\csc\left(x+\frac{k\pi}n\right)\csc\left(y+\frac{k\pi}n\right)=\frac{n\sin [n(x-y)]}{\sin(x-y)\sin nx\sin ny} \end{equation} Let $S(x,y)$ be the above sum. If $y$ is not of the form $y=\frac{k\pi}{n}$, note that $S$ is a $2\pi$-periodic meromorphic function of $x$ with simple poles at $x=\frac{k\pi}{n}$ for each integer $k$ with corresponding residues given by \begin{equation} \text{Res}\left(S,x=\frac{k}{n}\right)= \csc\left(y-\frac{k\pi}{n}\right) \end{equation} Now, note that the function \begin{equation} f(x,y)=\frac{n\sin [n(x-y)]}{\sin(x-y)\sin nx\sin ny} \end{equation} is also a $2\pi$-periodic meromorphic function of $x$ with simple poles at $x=\frac{k\pi}{n}$ for each integer $k$ with corresponding residues given by \begin{equation} \text{Res}\left(f,x=\frac{k}{n}\right)=(-1)^k\frac{n\sin [n(\frac{k\pi}{n}-y)]}{\sin(\frac{k\pi}{n}-y)\sin ny}= \csc\left(y-\frac{k\pi}{n}\right) \end{equation} This implies that $S-f$ is analytic and $2\pi$-periodic in $x$, and since $S-f$ vanishes as $\mathfrak{I}(x)\rightarrow\pm\infty$, then $S-f$ is bounded. Therefore by Louville's theorem, $S-f$ is constant with respect to $x$. By the same logic, $S-f$ is constant with respect to $y$, and thus, $S-f$ is constant. Finally, since $S-f$ vanishes as $\mathfrak{I}(x)\rightarrow\pm\infty$, then $S-f$ is precisely $0$, which completes the proof.

$\endgroup$
2
  • $\begingroup$ I took a quick attempt at your more recent question concerning the degree four version of this identity, and unfortunately could not find a way to apply this method there, but I might as well post it here in case someone finds it useful. $\endgroup$ Commented Jul 25, 2023 at 17:34
  • $\begingroup$ thank you for the alternative proof I had hoped another way might be shown and yes I agree somebody might find useful or at least interesting. $\endgroup$
    – onepound
    Commented Jul 25, 2023 at 18:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .