How to show the validity of the following integral inside the critical strip:

$$ \Gamma(s)\zeta(s)-\Gamma(s-1) = \int_{0}^{\infty}x^{s-2}\left(\frac{x}{e^x-1}-\frac{1}{e^x}\right)\,dx \qquad\colon\space Re\{s\}\gt0 \tag{1} $$

By definition: $$ \begin{align} \Gamma(s) &= \int_{0}^{\infty}\frac{x^{s-1}}{e^x}\,dx \space\colon\space Re\{s\}\gt0 \space\Rightarrow\space \Gamma(s-1) = \int_{0}^{\infty}x^{s-2}\left(\frac{1}{e^x}\right)\,dx \space\colon\space Re\{s\}\gt1 \\[4mm] \Gamma(s)\zeta(s) &= \int_{0}^{\infty}\frac{x^{s-1}}{e^x-1}\,dx = \int_{0}^{\infty}x^{s-2}\left(\frac{x}{e^x-1}\right)\,dx \quad\colon\space Re\{s\}\gt1 \end{align} $$ Thus: $$ \begin{align} \Gamma(s)\zeta(s)-\Gamma(s-1) &= \int_{0}^{\infty}x^{s-2}\left(\frac{x}{e^x-1}-\frac{1}{e^x}\right)\,dx \qquad\colon\space Re\{s\}\gt1 \\[4mm] &= \int_{0}^{\infty}x^{s-2}\left(\frac{x}{e^x-1}\color{red}{-1}-\frac{1}{e^x}\color{red}{+1}\right)\,dx \\[4mm] &= \int_{0}^{\infty}x^{s-2}\left(\frac{x}{e^x-1}-1\right)\,dx - \int_{0}^{\infty}x^{s-2}\left(\frac{1}{e^x}-1\right)\,dx \space\colon\space \color{red}{Re\{s\}\gt0} \end{align} $$ And, if I am not mistaken, this would imply:

$$ \begin{align} \Gamma(s)\zeta(s) &= \int_{0}^{\infty}x^{s-1}\left(\frac{1}{e^x-1}-\frac{1}{x}\right)\,dx \quad\colon\space 0\lt Re\{s\}\le1 \tag{2} \\[4mm] \Gamma(s) &= \int_{0}^{\infty}x^{s-1}\left(\frac{1}{e^x}-1\right)\,dx \space\quad\colon\space -1\lt Re\{s\}\le0 \tag{3} \end{align} $$ Is this correct? How can I show these results in another way? Appreciating.

  • $\begingroup$ anything unclear ? $\endgroup$ – reuns Jan 3 '17 at 12:54
  • $\begingroup$ @user1952009: Many thank for the answer. May I seek your patient, let us see other opinions/answers/votes. $\endgroup$ – Hazem Orabi Jan 3 '17 at 13:02
  • $\begingroup$ What do you want more ? If you are not sure of what is happening here, you need to study a complex analysis and a Fourier/Laplace transform course. The first thing is to see why $\frac{z}{e^z-1}$ is analytic at $z= 0$ $\endgroup$ – reuns Jan 3 '17 at 13:16
  • $\begingroup$ @user1952009: Thanks for been always interested in my posts, you definitely deserve a bounty for your skills. Regarding this question, appreciating if you could respond to my comment below, I think the answer is still uncompleted! $\endgroup$ – Hazem Orabi Jan 10 '17 at 20:40

Yes (1),(2),(3) are correct. A way to show (2) is with the Taylor expansion $\frac{x}{e^x-1}=1+\mathcal{O}(x)$ as $x \to 0$ so that $$\Gamma(s)\zeta(s) - \frac{1}{s-1} = \int_{0}^{\infty}x^{s-2}\left(\frac{x}{e^x-1}-1_{x < 1}\right)dx$$ (which is obvious for $Re(s)> 1$) converges and is analytic for $Re(s) > 0$.

Then, use $$\frac{1}{s} = \begin{cases} \ \int_0^1 x^{s-1}dx \ \ \text{ for } Re(s) > 0 \\ \ -\int_1^\infty x^{s-1}dx \text{ for } Re(s) < 0 \end{cases}$$ this is why the domain of convergence is so important when talking of a Laplace/Mellin transform.

Note the inverse Laplace/Mellin transforms $$\frac{1}{e^x-1} = \frac{1}{2i\pi}\int_{\sigma-i\infty}^{\sigma+i\infty} \zeta(s) \Gamma(s) x^{-s}dx, \quad \sigma > 1$$ $$ \quad \frac{1}{e^x-1}-\frac{1}{x} = \frac{1}{2i\pi}\int_{\sigma-i\infty}^{\sigma+i\infty} \zeta(s) \Gamma(s) x^{-s}dx, \quad \sigma \in (0,1)$$ the convergence being ensured by $\zeta(s) = \mathcal{O}(s)$ as $ Im(s) \to \infty$ while $\Gamma(s)$ decreases faster than any negative power of $s$ as $Im(s) \to \infty$, since $\Gamma(s) = \int_{-\infty}^\infty e^{-e^{-u}}e^{-su}du$ is the Laplace/Fourier transform of a Schwartz function for $Re(s) > 0$.

And see Riemann's proof of the functional equation, starting from $E(s) = \pi^{-s/2} \Gamma(s/2)\zeta(s) = \int_0^\infty x^{s/2-1}\sum_{n=1}^\infty e^{-\pi n^2 x}dx$ and using that $\theta(x) = 1+2\sum_{n=1}^\infty e^{-\pi n^2 x}$ is modular, that is $\theta(1/x) = x \theta(x)$ revealing that $E(s) = E(1-s)$.

  • $\begingroup$ I meant $\theta(1/x) = \sqrt{x}\theta(x)$ so it is modular but with a non-integral weight $\endgroup$ – reuns Jan 4 '17 at 23:52
  • $\begingroup$ Up to my knowledge, analytic continuation of $\,\zeta\,$ by Integral $(2)$ is known {i.e. books of E.C. Titchmarsh & Aleksandar Ivic}, while analytic continuation of $\,\Gamma\,$ by Integral $(3)$ is not. Would you please explain Integral $(3)$, at least in similar way of explaining Integral $(2)$. Thanks. $\endgroup$ – Hazem Orabi Jan 10 '17 at 20:41
  • $\begingroup$ @HazemOrabi For $(2)$ or $(3)$ it really works the same way. $e^{-x}-1_{x < 1} = \mathcal{O}(x)$ as $x \to 0$ so that $\Gamma(s) - \frac{1}{s} = \int_0^\infty x^{s-1} (e^{-x}-1_{x < 1})dx$ for $Re(s) > 0$ stays convergent and analytic for $Re(s) > -1$. Then add $\frac{1}{s} = -\int_0^\infty x^{s-1} 1_{x > 1} dx \ (Re(s) < 0)$ for obtaining $(2)$ $\endgroup$ – reuns Jan 10 '17 at 21:19
  • $\begingroup$ Meanwhile, for the benefit of public readers, whenever you have time, kindly review/revise your answer to include everything about the question on a bit more comprehensive/clear way. Do so, and I shall have no excuse not to accept it! $\endgroup$ – Hazem Orabi Jan 10 '17 at 22:05

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.