Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I want to compute the derivative of the Selberg $\zeta$-function:

$$ \mathcal{Z}(s)=\prod_{\gamma \; \text{primitive}} \prod_{n=0}^\infty (1-e^{-l(\gamma)(n+s)}); \qquad \Re(s)>1.$$

Where $\gamma$ are primitive closed geodesics on a given manifold and $l(\gamma)$ is their lenght. We can assume $s\in \mathbb{R}$. Observe that the given expression is justified by: $l(\gamma^n)= n l(\gamma)$.

A couple of reliable articles (Lou's "On the zeros of the derivative of the Selberg Zeta function" page 1143 for the statement and page 1141 for its definition of $\mathcal{Z}(s)$, and D'Hoker&Phong's "The geometry of String Perturbation theory, implicitly in the last passage of page 1005), and give, more or less explicitly, the following formula for the logarithmic derivative:

$$ \frac{\frac{d}{ds}\mathcal{Z}(s)}{\mathcal{Z}(s)} = \sum_{\gamma \; \text{primitive}} \sum_{n=1}^\infty\frac{l(\gamma) e^{-l(\gamma)ns}}{1-e^{-l(\gamma)n}}.$$

But my computation seems to give a slightly different result. Indeed, first we observe:

$$ \frac{d}{dx} \left(\prod_{n=0}^\infty f_n(x)\right)= \left(\prod_{n=0}^\infty f_n(x)\right)\left(\sum_{n=0}^\infty \frac{\frac{d}{dx}f_n(x)}{f_n(x)}\right).$$

Is it true? Wikipedia confirms it only in the finite case.

Then, applying it to $ \mathcal{Z}(s)$, I deduce:

$$ \frac{\frac{d}{ds}\mathcal{Z}(s)}{\mathcal{Z}(s)} = \sum_{\gamma \; \text{primitive}} \sum_{n=0}^\infty\frac{l(\gamma) e^{-l(\gamma)(n+s)}}{1-e^{-l(\gamma)(n+s)}}.$$

This expression seems to be close to the claimed one but I couldn't prove they are the same. So I guess I'm doing something wrong in the computation. Do you know how to do it?

Thank you very much!

share|cite|improve this question
I'm going to link them! – Giovanni De Gaetano Oct 9 '12 at 19:29
up vote 3 down vote accepted

Maybe something like that?

Write $$ \frac{1}{1-e^{-l(\gamma)n}} = \sum_{k=0}^\infty e^{-l(\gamma)kn} \qquad\text{and}\qquad \frac{e^{-l(\gamma)(k+s)}}{1-e^{-l(\gamma)(k+s)}} = \sum_{n=1}^\infty e^{-l(\gamma)(k+s)n}. $$ Then $$ \begin{align*} \sum_{n=1}^\infty\frac{e^{-l(\gamma)ns}}{1-e^{-l(\gamma)n}} & =\sum_{n=1}^\infty \sum_{k=0}^\infty e^{-l(\gamma)ns} e^{-l(\gamma)kn} \\ & = \sum_{k=0}^\infty \sum_{n=1}^\infty e^{-l(\gamma)(k+s)n} \\& = \sum_{k=0}^\infty \frac{e^{-l(\gamma)(k+s)}}{1-e^{-l(\gamma)(k+s)}} \end{align*} $$ multiply by $l(\gamma)$ and sum over the primitives and you seem to get what you want.

share|cite|improve this answer
Perfect! 'sorry for the delay in the answer and thank you very much! :) – Giovanni De Gaetano Oct 10 '12 at 7:52

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.