let $s,z$ be two complex parameters. Consider the integral : $$I(s,z)=s\int_{1}^{\infty}x^{z}\zeta^{'}(sx)dx$$ Computing the integral for $\Re(s)>1$ is easily done, using the Dirichlet series representation of the zeta function. Namely : $$I(s,z)=-s\int_{1}^{\infty}x^{z}\left(\sum_{n=2}^{\infty}\log (n)n^{-sx}\right)dx=\sum_{n=2}^{\infty}\left(s\log n\right)^{-z}\Gamma\left(1+z,s\log n\right)$$ Where $\Gamma(\cdot,\cdot)$ is the incomplete gamma function. It seems that $I(s,z)$ is entire in the $z$ plane, but not defined for $\Re(s)<1$ -as expected-. How to derive the analytic continuation for $I(s,z)$ in $s$ ?


For $s > 1$ with the change of variable $x= y/s$ $$I_z(s) = s \int_1^\infty x^z \zeta'(sx)dx=s^{-z} \int_s^\infty y^z \zeta'(y)dy$$ and by analytic continuation for every $s \in \mathbb{C} \setminus (-\infty,1]$.

$ I_z(s)$ has two branch points : at $s=0$ and another one at $s=1$ caused by the pole of $\zeta'(s)$.

  • $\begingroup$ I am sorry, but i don't understand this part of your answer "and by analytic continuation for every $s\in \mathbb{C}\setminus(-\infty,1]$". How did you arrive at this result ? say i want to compute $I_{z}\left(\frac{1}{2}+i\epsilon\right)$, what formula should i use ? $\endgroup$ – Mohammad Al Jamal Oct 31 '17 at 8:58
  • $\begingroup$ @MohammadAlJamal Don't you see the RHS is analytic on $ \mathbb{C}\setminus(-\infty,1]$ ? ($\int_s^\infty$ is a contour integral for $s$ complex) $\endgroup$ – reuns Oct 31 '17 at 19:34

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.