Since Hjortnaes (and later Apéry), we know that

$$ \zeta(3)=\frac{5}{2} \sum_{k=1}^{\infty}{\frac{ (-1)^{k-1}} {k^{3}\binom {2k}{k}}}. $$

I read somewhere that there might be a similar expression for $\zeta(5)$:

$$ \zeta(5)=\frac{a}{b} \sum_{k=1}^{\infty}{\frac{ (-1)^{k-1}} {k^{5}\binom {2k}{k}}}, $$

where $a$ and $b$ are positive nonzero integers. I know that no such $a$ and $b$ have been found yet, otherwise Apéry's proof of the irrationality of $\zeta(3)$ could be extended to $\zeta(5)$. Are there any results suggesting that there exist similar sums for $\zeta(2n+1)$, $n>1$:

$$ \zeta(2n+1)=\frac{a}{b} \sum_{k=1}^{\infty}{\frac{ (-1)^{k-1}} {k^{2n+1}\binom {2k}{k}}}? $$

Conversely, are there any results that rule out the existence of such sums for certain odd positive integers?

Finally, what are the current known bounds on $a$ and $b$, if such sums can exist?

  • $\begingroup$ The paper by Nan-Yue and Williams [1] is worth consulting; equation (4.20) onwards involving the golden ratio $\Phi$. ( [1] Nan-Yue, Zhang; Williams, Kenneth S. Values of the Riemann zeta function and integrals involving $\log(2\,{\rm sinh}(\theta/2))$ and $\log(2\sin(\theta/2))$. Pacific J. Math. 168 (1995), no. 2, 271--289. projecteuclid.org/euclid.pjm/1102620561 ) $\endgroup$ – James Arathoon Jun 27 '18 at 15:45

The first terms of the continued fraction of $$ \frac{1}{\zeta(5)}\sum_{k\geq 1}\frac{(-1)^{k+1}}{k^5 \binom{2k}{k}} $$ are given by $$[0; 2, 10, 1, 1, 5, 1, 1, 1, 1, 2, 2, 1, 3, 3, 1, 66, 4, 135, 3, 1, \ 6, 1, 29,\ldots]$$ so if the previous number is $\frac{a}{b}\in\mathbb{Q}$, we have $\min(a,b)>3\cdot 10^{12}$.
In other terms, it looks pretty unlikely that the series $\sum_{k\geq 1}\frac{(-1)^{k+1}}{k^5 \binom{2k}{k}}$ provides a rational multiple of $\zeta(5)$.
The classical results $$ \sum_{n\geq 1}\frac{1}{n^2\binom{2n}{n}}=\frac{1}{3}\zeta(2),\quad \sum_{n\geq 1}\frac{(-1)^{n+1}}{n^3\binom{2n}{n}}=\frac{2}{5}\zeta(3),\quad \sum_{n\geq 1}\frac{1}{n^4\binom{2n}{n}}=\frac{17}{36}\zeta(4)$$ (stating that in some cases the hypergeometric $\phantom{}_{p+1} F_p\left(1,1,1\ldots;\tfrac{3}{2},2,2,\ldots;\pm\tfrac{1}{4}\right)$ has a nice closed form in terms of the $\zeta$ function) can be seen as consequences of creative telescoping, or as consequences of symmetry relations for polylogarithms (see [1]). About $\frac{1}{k^5}$, creative telescoping produces spurious terms and the functional relations for $\text{Li}_4$ and $\text{Li}_5$ are pretty messy, so there are substantial obstructions in generalizing the classical approaches for producing a simple irrationality proof for $\zeta(5)$.

  • $\begingroup$ Thank you. Could you please point me to an article that describes the hypergeometric function as closed forms in terms of $\zeta$? That would be an interesting read. $\endgroup$ – Klangen Jun 27 '18 at 14:57
  • 1
    $\begingroup$ @PreservedFruit: here it is a good one - arxiv.org/abs/0801.1591 $\endgroup$ – Jack D'Aurizio Jun 27 '18 at 15:02
  • $\begingroup$ Thank you for the link! $\endgroup$ – Klangen Jun 27 '18 at 16:12

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.