The Riemann $\zeta$-function is here with analytical extension (e.g. with her functional equation).

Definition for $|x|<1$ and $-t\in\mathbb{R}\setminus\mathbb{N}$: \begin{align*} B_t(x+1):=&-\frac{2\Gamma(1+t)}{(2\pi)^t}\cos(\frac{\pi t}{2}) \sum\limits_{k=0}^\infty (-1)^k \frac{(2\pi x)^{2k}}{(2k)!}\zeta(t-2k) \\ &-\frac{2\Gamma(1+t)}{(2\pi)^t}\sin(\frac{\pi t}{2}) \sum\limits_{k=0}^\infty (-1)^k \frac{(2\pi x)^{2k+1}}{(2k+1)!}\zeta(t-1-2k) \end{align*}

Note: It’s possible to use complex variables but it’s not necessary here.

For $t\in\mathbb{N}_0$ and $x\in\mathbb{R}$ one gets the Bernoulli polynomials.

It’s not difficult to proof, that $$\frac{\partial}{\partial x}B_t(x+1)=t B_{t-1}(x+1)\,.$$

But how can one proof $$B_t(x+1)=B_t(x)+t x^{t-1}$$ for $t\geq 1$ and $x\in(-1;0)\,$?

(The consequence is a generalization of the Bernoulli polynomials $B_t(x)$ in relation to the index with $\,(1)\, t\geq 1\,$ and $x\in\mathbb{R}$ or $\,(2)\, -t\in\mathbb{R}\setminus\mathbb{N}$ and $x>0$.)

Application example for $B_t(x)\,$:

In combination with the Polylogarithm function ( https://en.wikipedia.org/wiki/Polylogarithm ) $$Li_s(z):=\sum\limits_{k=1}^\infty \frac{z^k}{k^s}$$ and the formula $$Li_s(e^\mu)=\Gamma(1-s)(-\mu)^{s-1}+\sum\limits_{k=0}^\infty \frac{\zeta(s-k)}{k!}{\mu}^k$$ for (complex $\mu$) $|\mu|<2\pi$ and complex $s\ne 1,2,3,...$ one gets (here with real $x$ and $t$) $$\Re(\sum\limits_{k=1}^\infty \frac{e^{i2\pi kx}}{(ik)^t})=\frac{(2\pi)^t}{2\Gamma(1+t)}B_t(x)$$ for $|x|<1$ and $t>0$ which is a generalization of the Fourier expansion of the Bernoulli polynomials. For derivations with respect to $t$ is $\ln(ik)=i\frac{\pi}{2}+\ln k$ (using the main branch of the logarithm).


1 Answer 1


Your definition seems somewhat complicated. Why don't you define your generalized Bernoulli polynomials via generaliation of the relation to the Hurwitz zeta function $B_n(x)=-n\zeta(1-n,x)?$ If you define $$B_t(x)=-t\zeta(1-t,x)$$ then you get from the well-know $\zeta$ property (http://dlmf.nist.gov/25.11.E4) $$\zeta(1-t,x)=\zeta(1-t,x+1)+\frac{1}{x^{1-t}}$$ $$\zeta(1-t,x+1)=\zeta(1-t,x)-\frac{1}{x^{1-t}}$$ $$-t\zeta(1-t,x+1)=-t\zeta(1-t,x)+t\frac{1}{x^{1-t}}$$ and finally $$B_t(x+1)=B_t(x)+tx^{t-1}$$ PS: Here is a reference to generalized Bernoulli polynomial of complex index: P.L. Butzer, M. Hauss, M. Leclerc, Bernoulli numbers and polynomials of arbitrary complex indices, Applied mathematics letters, 1992 available via ResearchGate

  • $\begingroup$ Thank you very much for your informations. With the definition of $B_t(x)$ you use the proof is possible (thanks). But it still remains to show, that the two different types of definitions are equivalent. The literatur researchgate.net/publication/… is an interesting list of theorems, but I don't see any proof or base for a proof. Theorem 1 and 7 are useful. $\endgroup$
    – user90369
    Sep 2, 2016 at 9:05
  • $\begingroup$ Maybe 25.11.9 of dlmf.nist.gov/25.11.E4 will help. Thanks again. $\endgroup$
    – user90369
    Sep 2, 2016 at 9:23

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.