Derivative of Riemann zeta function

Click For Summary

Discussion Overview

The discussion revolves around evaluating the derivative of the Riemann zeta function at the origin, \(\zeta'(0)\), using its integral representation. Participants explore various mathematical approaches and techniques related to this evaluation, including differentiation under the integral sign and the behavior of integrals as \(s\) approaches zero.

Discussion Character

  • Technical explanation
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • One participant begins by stating the integral representation of the Riemann zeta function and expresses difficulty in evaluating its derivative at \(s=0\).
  • Another participant provides a differentiation approach involving the product rule and suggests a form for the derivative of the zeta function.
  • A subsequent post corrects an earlier misrepresentation of the integral's exponential term, clarifying the correct form of the integrand.
  • Participants discuss the implications of setting \(s=0\) on the convergence of integrals involved in the derivative evaluation.
  • One participant notes that while individual integrals diverge, combinations might converge, suggesting the use of limits and power series to handle the evaluation.
  • Another participant expresses frustration with the convergence of the final sum obtained from the power series approach, indicating a need for further assistance.

Areas of Agreement / Disagreement

There is no consensus on how to proceed with the evaluation of the integrals or the derivative. Participants express differing views on convergence and the methods to be used, indicating an unresolved discussion.

Contextual Notes

Participants mention potential divergences in integrals when \(s=0\) and the need for careful handling of limits and series expansions, but do not resolve these issues.

TriTertButoxy
Messages
190
Reaction score
0
I'm trying to evaluate the derivative of the Riemann zeta function at the origin, [itex]\zeta'(0)[/itex], starting from its integral representation
[tex]\zeta(s)=\frac{1}{\Gamma(s)}\int_0^\infty t^{s-1}\frac{1}{e^t-1}.[/tex]
I don't want to use a symbolic algebra system like Mathematica or Maple.

I am able to continue to [itex]s=0[/itex] and evaluate the zeta function there [itex]\zeta(0)=-1/2[/itex]. I'm just stuck on how to evaluate the derivative.

Can somebody show me how to do this starting from the integral representation? Thanks.
 
Physics news on Phys.org
d/ds{ts-1/Γ(s)} = {(s-1)ts-2Γ(s) - ts-1Γ'(s)}/Γ(s)2=f(s,t). The integrand is f(s,t)/(et-1)
 
\begin{array}\\\zeta (s)=\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{t^{s-1}}{e^{t-1}}dt\\
\frac{d}{ds}\zeta (s)=\frac{d}{ds}\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{t^{s-1}}{e^{t-1}}dt+\frac{1}{\Gamma (s)}\frac{d}{ds}\int_{0}^{\infty}\frac{t^{s-1}}{e^{t-1}}dt=\frac{-1}{\Gamma^2 (s)}\frac{d}{ds}\Gamma (s)\int_{0}^{\infty}\frac{t^{s-1}}{e^{t-1}}dt+\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{d}{ds}\frac{t^{s-1}}{e^{t-1}}dt\\

\zeta' (s)=\frac{-1}{\Gamma^2 (s)}\int_{0}^{\infty}t^{s-1}e^{-t}\ln{t}dt\int_{0}^{\infty}\frac{t^{s-1}}{e^{t-1}}dt+\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{t^{s-1}}{e^{t-1}}\ln{t}dt\end{array}

Now put s=0 and integrate!
 
Hi thanks, but the -1 is not in the exponential.
Its [itex]\frac{1}{e^t-1}[/itex] not [itex]\frac{1}{e^{t-1}}[/itex].
 
TriTertButoxy said:
Hi thanks, but the -1 is not in the exponential.
Its [itex]\frac{1}{e^t-1}[/itex] not [itex]\frac{1}{e^{t-1}}[/itex].

Oh yes you are right! My mistake :frown:


\begin{array}\\\zeta (s)=\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{t^{s-1}}{e^t-1}dt\\
\frac{d}{ds}\zeta (s)=\frac{d}{ds}\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{t^{s-1}}{e^t-1}dt+\frac{1}{\Gamma (s)}\frac{d}{ds}\int_{0}^{\infty}\frac{t^{s-1}}{e^t-1}dt=\frac{-1}{\Gamma^2 (s)}\frac{d}{ds}\Gamma (s)\int_{0}^{\infty}\frac{t^{s-1}}{e^t-1}dt+\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{d}{ds}\frac{t^{s-1}}{e^t-1}dt\\

\zeta' (s)=\frac{-1}{\Gamma^2 (s)}\int_{0}^{\infty}t^{s-1}e^{-t}\ln{t}dt\int_{0}^{\infty}\frac{t^{s-1}}{e^t-1}dt+\frac{1}{\Gamma (s)}\int_{0}^{\infty}\frac{t^{s-1}}{e^t-1}\ln{t}dt\end{array}

Now it is correct
 
Ok, that sounds good; but now what? How do I do those integrals? If I set s=0, the second integral doesn't converge...
 
TriTertButoxy said:
Ok, that sounds good; but now what? How do I do those integrals? If I set s=0, the second integral doesn't converge...

If you set s=0 (almost) all integrals(including Γ(s) and Γ'(s)) will diverge but a combination of them could actually converge.
My advice: calculate the integrals(you may need power series for some) and replace them with limits. Then, simplify everything and hope that the final limit won't be infinite
 
thanks. I've tried taking the power series. But in the end, the final sum doesn't converge. I'm afraid I'm going to need help till the bitter end on this one.
 

Similar threads

  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 5 ·
Replies
5
Views
4K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 10 ·
Replies
10
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 5 ·
Replies
5
Views
4K
  • · Replies 6 ·
Replies
6
Views
3K
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 8 ·
Replies
8
Views
4K