2

How to prove: \begin{equation} 11=10^{12}+10^{7}-45\sum_{n=1}^{999}\csc^4\frac{n\pi}{1000}\;. \end{equation}

This identity appears on my clock: enter image description here

Zacky
  • 27,674
  • 4
    Where did you come across this? – bsbb4 Aug 02 '19 at 23:12
  • 2
  • 1
    Why on earth is this question closed for missing details? It simply appears on a clock as OP said: https://www.cafepress.com/+math_clock_blue_green_wall_clock,755880297 Isn't that more than enough? – Zacky Aug 03 '19 at 10:02
  • 1
    @ㄴㄱ : Did you read the explaining text? "context include: background ... , possible strategies, your current progress, ..." – Lutz Lehmann Aug 03 '19 at 12:21
  • @LutzL Did you saw that J. Tilgner is a new contributor to this site? (Take care in asking for clarification, commenting, and answering).

    The fact that he collaborated and gave the source (that was the only thing required in the comments by others), looks enough imo.

    – Zacky Aug 03 '19 at 12:23
  • 2
    @ㄴㄱ : There is one complete answer, carrying out the calculations at the end really gives the desired result. So the formula is verified and one can get an impression on how heavy the machinery behind it is. Until J. Tilgner indicates whether he understood it and if not, what his general mathematical level is, nothing much more can be done. – Lutz Lehmann Aug 03 '19 at 15:50
  • 1
    @LutzL I agree too on that. Also this should be closed as a duplicate afterwards: https://math.stackexchange.com/questions/1877014/a-series-1012-107-45-sum-k-1999-csc4-frack-pi1000 – Zacky Aug 03 '19 at 16:03
  • it should not have been closed. – onepound Feb 28 '24 at 22:36

1 Answers1

2

Write $\cot\dfrac{n\pi}{1000}=c_n$

Now from Sum of tangent functions where arguments are in specific arithmetic series and Question regarding $f(n)=\cot^2\left(\frac\pi n\right)+\cot^2\left(\frac{2\pi}n\right)+\cdots+\cot^2\left(\frac{(n-1)\pi}n\right)$

$c_n,1\le n\le1000-1$ are the roots of $$\binom{1000}1c^{999}-\binom{1000}3c^{997}+\binom{1000}5c^{995}+\cdots-\binom{1000}{999}c=0$$

$\cot\dfrac{n\pi}{1000}=0,\dfrac{n\pi}{1000}=\dfrac{(2m+1)\pi}2\iff n=(2m+1)500$

$\implies c_n,1\le n\le1000-1,n\ne500$ are the roots of $$\binom{1000}1c^{998}-\binom{1000}3c^{998}+\binom{1000}5c^{994}+\cdots-\binom{1000}{999}=0$$

So, if $d_n=c^2_n$

$d_n,1\le n\le499$ are the roots of $$\binom{1000}1d^{499}-\binom{1000}3d^{498}+\binom{1000}5c^{497}+\cdots-\binom{1000}{999}=0$$

Again $\csc^4\dfrac{n\pi}{1000}=(d_n+1)^2=d_n^2+2d_n+1$

$$\sum_{n=1}^{999}\csc^4\dfrac{n\pi}{1000}=\csc^4\dfrac{500\cdot\pi}{1000}+2\sum_{n=1}^{499}\csc^4\dfrac{n\pi}{1000}=1+2\sum_{n=1}^{499}(d_n^2+2d_n+1)$$

Now $$\sum_{n=1}^{499}d_n=\dfrac{\binom{1000}3}{\binom{1000}1}$$

$$\sum_{n=1}^{499}d^2_n=\left(\sum_{n=1}^{499}d_n\right)^2-2\cdot\dfrac{{\binom{1000}5}}{\binom{1000}1}$$

  • There are no degree gaps in the polynomial for $d$. Thus would you not get $$\sum_{n=1}^{499}d_n=\dfrac{\binom{1000}3}{\binom{1000}1}, ~\text{ and }~ \sum_{n=1}^{499}d^2_n=\left(\dfrac{\binom{1000}3}{\binom{1000}1}\right)^2-2\cdot\dfrac{{\binom{1000}5}}{\binom{1000}1}=\dfrac{(999\cdot998)^2}{6^2}-2\dfrac{999\cdot998\cdot997\cdot996}{120}?$$ – Lutz Lehmann Aug 03 '19 at 07:04
  • @LutzL, Thanks for your observation – lab bhattacharjee Aug 03 '19 at 09:50