I have discovered (and found an elementary proof of) the following $$\zeta(2k)=(-1)^{k-1}\dfrac{\pi^{2k}}{2^{4k}-2^{2k}}\left[k\dfrac{2^{2k}}{(2k)!}+{\displaystyle \sum_{l=1}^{k-1}(-1)^{l}\dfrac{2^{2k-2l}}{(2k-2l)!}\dfrac{2^{4l}-2^{2l}}{\pi^{2l}}\zeta(2l)}\right]$$ ($k=1,2,3,\dots$)

My questions are:

  1. Is this formula new, or, how can I check if it is?

  2. Where should I publish such a result? (if the result or the proof is indeed new)


It's clearest to write the identities in terms of Bernoulli numbers $B_n$ using \begin{equation} \zeta (2 n) = \frac{(-1)^{n - 1} (2 \pi)^{2 n}}{2 (2 n)!} B_{2 n} . \end{equation} Remembering that $B_1 = \frac{1}{6}$ and that $B_n$ vanishes for all other odd $n$, the desired identity is \begin{equation} \sum_{k = 0}^{2 n} \binom{2 n}{k} 2^k (2^k - 1) B_k = 0 . \end{equation} In contrast, the identity of Song referred to by Carlo Beenakker is \begin{equation} \sum_{k = 0}^{2 n} \binom{2 n + 1}{k} 2^k B_k = 0 . \end{equation}

A useful reference is the book by N. Nielsen, Traité élémentaire des nombres de Bernoulli . It's necessary to convert notation from Nielsen's book using $B_n \rightarrow (-1)^{n - 1} B_{2 n}$, and use the formula for $T_n$ in equation (9) on page 56. The identity in equation (11) on page 132 is equivalent to \begin{equation} \sum_{k = 0}^{2 n} \binom{2 n}{k} 2^k B_k + 2 (2^{2 n - 1} - 1)B_{2 n} = 0 , \end{equation} which is a special case of a more general result attributed to previous authors. The identity in equation (12) on page 138 is equivalent to \begin{equation} \sum_{k = 0}^{2 n} \binom{2 n}{k} 2^{2 k} B_k + 2 (2^{2 n - 1} - 1)B_{2 n} = 0 . \end{equation} Taking the difference of these gives the desired identity.

The result is also given in equation VII on page 14 of the book by L. Saalschütz, Vorlesungen über die Bernoullischen Zahlen, which references three papers from 1843 by Stern, Schlömlich and Göpel.

One would expect a number of possible linear identities for the Bernoulli numbers. See, for example, M. W. Coffey, Bernoulli identities, zeta relations, determinant expressions, Mellin transforms, and representation of the Hurwitz numbers and references within.

  • $\begingroup$ Thank you so much! $\endgroup$ – Tom Moshaiov Jan 24 '19 at 20:54

I. Song, A recursive formula for even order harmonic series (1987):

with $C_k=\zeta(2k)$. If I'm not mistaken, this is the recursion formula in the OP.

  • $\begingroup$ Thank you very much. It is remarkably similar, but I have not been able to show one implies the other. $\endgroup$ – Tom Moshaiov Jan 20 '19 at 19:16

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.