I recently, out of interest, tried to prove Euler's formula $\zeta{(2n)}=(-1)^{n-1}\frac{(2\pi)^{2n}}{2(2n)!}B_{2n}$ for all $n\in\mathbb{N}$. I adapted Euler's original proof for $\zeta(2)=\frac{\pi^2}{6}$:

We have the well known formulas $\sin(x)=\sum_{k=0}^{\infty}\frac{(-1)^kx^{2k+1}}{(2k+1)!}$ and $\sin(x)=x\prod_{k=1}^{\infty}\left(1-\frac{x^2}{k^2\pi^2}\right)$. The product formula can be transformed into a series expansion as well but in order to keep it neat, some definitions first:

For the real sequence $\left(a_i\right)_{i\in\mathbb N}$ with $\sum a_i$ convergent, define $$ A_n:=\sum_{i=1}^{\infty} a_i^n $$ $$ \alpha_n:=\sum_{(x_1,\dots,x_n)\in M_n}\prod_{r=1}^{n} a_{x_r} $$ Where $M_n:=\{(x_1,\dots,x_n)\in\mathbb N\space|\space x_1<\dots<x_n\}$. Now we write $c_i=-\frac{1}{i^2\pi^2}$. When we expand the product, we get only odd exponents of $x$ and the coefficient of $x^{2k+1}$ for $k≥1$ is $\gamma_k$. By comparison of the coefficients, we obtain $\gamma_k=\frac{(-1)^k}{(2k+1)!}$. Now, through telescoping, it can be shown algebraically that for every sequence $\left(a_i\right)_{i\in\mathbb N}$ with $\sum a_i$ convergent we have: $$\sum_{i=1}^{n}{(-1)^{n-i}\alpha_{n+1-i}\space A_{i}}=A_{n+1}+(-1)^{n+1}(n+1)\alpha_{n+1}$$ For all $n\in\mathbb N_0$. Since $C_n=\sum_{i=1}^{\infty} \left(-\frac{1}{i^2\pi^2}\right)^n=\frac{(-1)^n}{\pi^{2n}}\zeta{(2n)}$ we have: $$ \sum_{i=1}^{n}{(-1)^{n-i}\gamma_{n+1-i}\space C_{i}}=\sum_{i=1}^{n}{(-1)^{n-i}\frac{(-1)^{n+1-i}}{(2n-2i+3)!}\frac{(-1)^i}{\pi^{2i}}\zeta{(2i)}}=\sum_{i=1}^{n}{(-1)^{i-1}\frac{\zeta{(2i)}}{(2n-2i+3)!\pi^{2i}}}=\frac{(-1)^{n+1}}{\pi^{2n+2}}\zeta{(2n+2)}+(-1)^{n+1}(n+1)\frac{(-1)^{n+1}}{(2n+3)!}=\frac{(-1)^{n+1}}{\pi^{2n+2}}\zeta{(2n+2)}+\frac{(n+1)}{(2n+3)!} \iff \sum_{i=1}^{n+1}{(-1)^{i-1}\frac{\zeta{(2i)}}{(2n-2i+3)!\pi^{2i}}}=\frac{(n+1)}{(2n+3)!} $$ For all $n\in\mathbb N_0$. Now we are ready to prove the formula $\zeta{(2n)}=(-1)^{n-1}\frac{(2\pi)^{2n}}{2(2n)!}B_{2n}$ with induction. With the equation above we can easily verify that it is true for $n=1$; therefore we assume that it holds for all $1≤i≤n$. The assertion $\zeta{(2n+2)}=(-1)^{n}\frac{(2\pi)^{2n+2}}{2(2n+2)!}B_{2n+2}$ is then equivalent to: $$ \sum_{i=1}^{n+1}{(-1)^{i-1}\frac{(-1)^{i-1}\frac{(2\pi)^{2i}}{2(2i)!}B_{2i}}{(2n-2i+3)!\pi^{2i}}}=\frac{(n+1)}{(2n+3)!}\iff \sum_{i=1}^{n+1}{\binom{2n+3}{2i}2^{2i-1}B_{2i}}=n+1\iff \sum_{i=0}^{2n+2} \binom{2n+3}{i}2^iB_i=0 $$ This identity can be proven by comparing the coefficients in the equation $\frac{e^x-1}{x}\frac{2x}{e^{2x}-1}=\frac{2}{e^x+1}$.

Is this proof valid? Nonetheless, if it is, it isn't very elegant. Are there faster ways to prove it?

  • 2
    $\begingroup$ I saw a nice and short proof somewhere using the Fourier series of even Bernoulli polynomials once you took some basics proprieties as granted. $\endgroup$ – yultan Jun 12 '15 at 13:06

I usually prove it in the following way. Since: $$ \frac{t}{e^t-1} = \sum_{n\geq 0}\frac{B_n}{n!}t^n \tag{1}$$ it follows that: $$ \coth z-\frac{1}{z} = \sum_{n\geq 1}\frac{4^n\,B_{2n}}{(2n)!}z^{2n-1}.\tag{2}$$ On the other hand, by taking the logarithmic derivative of the Weierstrass product for the $\sinh $ function it follows that: $$\begin{eqnarray*} \coth z -\frac{1}{z} &=& \sum_{n\geq 1}\frac{d}{dz}\log\left(1+\frac{z^2}{n^2 \pi^2}\right)\\&=&\sum_{n\geq 1}\frac{2z}{\pi^2 n^2+z^2}\\&=&\sum_{n\geq 1}\sum_{m\geq 1}\frac{2z^{2m-1}}{\pi^{2m}n^{2m}} \\&=&\sum_{m\geq 1}\frac{2\,\zeta(2m)}{\pi^{2m}}z^{2m-1}\tag{3}\end{eqnarray*}$$ and we have the claim by comparing the coefficients in the RHSs of $(2)$ and $(3)$.

The intermediate identities are often very useful, too.


In order to compute the values $\zeta(2n),$ for $n\in\mathbb{N}^+,$ use equation Riemann's functional equation for the Zeta function and the double angle sine formula to obtain

\begin{equation}\tag{1} \zeta(2n)=\frac{(2\pi)^{2n}\zeta(1-2n)}{2\Gamma(2n)\cos(n\pi)}=\frac{(-1)^n(2\pi)^{2n}}{2(2n-1)!}\zeta(1-2n). \end{equation}

Using the fact that, for $n\in\mathbb{N},$ \begin{equation}\tag{2} B_n=(-1)^{n+1}n\zeta(1-n). \end{equation}

We can use (2) to rewrite (1) as \begin{equation}\tag{3} \zeta(2n)=\frac{(-1)^n(2\pi)^{2n}}{2(2n-1)!}\zeta(1-2n)=\frac{(-1)^{n+1}(2\pi)^{2n}}{2(2n)!}B_{2n}. \end{equation}

  • $\begingroup$ how to deduce from this result that $\zeta(2k) $ -->1 as k --> $\infty $? $\endgroup$ – Dxdxdada Jan 4 at 10:46

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.