6
$\begingroup$

I proposed this question on MSE but some comments affirmed that is unsolved problem and no answer. I would like to see what MO say about it.

How do I evaluate this sum :$$\displaystyle\sum_{n=0}^{\infty} \frac{\sin(n!)}{\cos(n!)}$$

Note : I used many criterions of convergence to test whether it converges but i didn't succeed.

Thank you for any help .

$\endgroup$
12
  • 9
    $\begingroup$ I'd be surprised if it did converge! $\endgroup$ Jun 23, 2015 at 17:16
  • 4
    $\begingroup$ Presumably the set of values $\{ n! \bmod 2\pi\mathbb Z\}$ is dense in the interval $[0,2\pi]$, in which case $\tan(n!)$ will be arbitrarily large, infinitely often. So your series is highly unlikely to be absolutely convergent. (Whether one can prove this or not, I don't know.) Conditional convergence also seems unlikely. But it might be interesting to try to prove $\sum_{n\le X}\tan(n!)=o(X)$, or even $O(X^\epsilon)$ for some $\epsilon<1$. (I haven't thought about it, maybe this is easy, maybe not.) $\endgroup$ Jun 23, 2015 at 17:43
  • 3
    $\begingroup$ Actually, it's not clear that there's even a polynomial bound, i.e., $\sum_{n\le X}\tan(n!)=O(X^K)$ for some $K>0$. Rigorously, probably the best one can do is use transcendence estimates for $\pi$ to bound $n!$ away from multiples of $\pi/2$ where $\tan$ blows up, which is going to yield an effective, but horrible, upper bound in terms of $X$. $\endgroup$ Jun 23, 2015 at 17:53
  • 3
    $\begingroup$ On the other hand, $\sum_{n=0}^\infty \tan(\pi n! /e)$ will converge. $\endgroup$ Jun 23, 2015 at 19:16
  • 1
    $\begingroup$ May be add "x" and change $\sin(n!)$ to $\sin(n!x)$ ? In this form it is a so called lacunar Fourier series, there are a lot of results on convergence for them. May be some result will help? $\endgroup$
    – Sergei
    Jun 24, 2015 at 5:08

1 Answer 1

5
$\begingroup$

I computed $f(n)=\sum_{k=0}^n \tan(k!)$ for different $n$, and got the following plot. It does not seem to have a limit.

Sums for different n

Mathematica code:

Block[{$MaxExtraPrecision = 600},
 lst = Table[Tan[n!], {n, 0, 200}];
 ListPlot[N[Accumulate[lst], 200], PlotRange -> All]
 ]
$\endgroup$
9
  • $\begingroup$ i have the same plot with wolfram alpha but what about analytical proof ? $\endgroup$ Jun 23, 2015 at 17:46
  • 2
    $\begingroup$ It's probably more interesting to look at the values of something like $\sup_{n\le N} \frac{\log|f(n)|}{\log n}$ and see if it looks as if there might be a polynomial bound for $|f(n)|$. $\endgroup$ Jun 23, 2015 at 17:48
  • 3
    $\begingroup$ Joe Silverman gives a good argument. I guess to actually prove that $n!$ mod $2\pi \mathbb{Z}$ is dense is quite nontrivial, and perhaps even an open problem. $\endgroup$ Jun 23, 2015 at 17:48
  • $\begingroup$ @JoeSilverman: Ah, yes, it looks like $\log|f(n)|/\log(n) \leq 1.64$ for $n \leq 600$, and $g(N)=\sup_{n\leq N} \frac{\log|f(n)|}{\log(n)}$ is very close to being constant. $\endgroup$ Jun 23, 2015 at 17:57
  • 2
    $\begingroup$ @PerAlexandersson I upvoted to compensate for that. $\endgroup$ Jun 29, 2015 at 20:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.