1
$\begingroup$

Denote by $\zeta$ the Riemann zeta function. I just learnt from this question $L_2$ bounds for tails of $\zeta(s)$ on a vertical line

that $\int_{T}^{\infty} \frac{\zeta(1/2 + it)}{1/4 + t^2}\mathrm{d}t=O(T^{-1}\log T)$ as $T\rightarrow +\infty$. But is the numerical value for the related integral

$$\int_{-\infty}^{\infty} \frac{\zeta(1/2 + it)}{1/4 + t^2} \mathrm{d}t $$ known?

$\endgroup$
6
  • 2
    $\begingroup$ Why should this integral converge? $\endgroup$ – Stanley Yao Xiao Oct 27 '18 at 19:44
  • $\begingroup$ $\frac{\zeta(s)}{s} = \frac{1}{s-1}-\int_1^\infty (x-\lfloor x \rfloor) x^{-s-1}dx$ and $\frac{\zeta(s+1)}{s} = \frac{1}{s^2}-\frac{\gamma}{s} -\int_0^\infty (\sum_{n \le x}\frac{1}{n} - \log(x)+\gamma) x^{-s-1}dx$ thus $F(t)=\zeta(1/2+it)(\frac{1}{1/2+it}+\frac{1}{1-(1/2+it)})-\frac{1}{1/2+it-1}+\frac{1}{(1/2+it-1)^2}-\frac{\gamma}{1/2+it-1}$ is the Fourier transform of $f(u)=(e^u-\lfloor e^u \rfloor) e^{-u/2}-(\sum_{n \le e^u} \frac{1}{n}+u-\gamma)e^{u/2}$ whence $\lim_{T \to \infty}\int_{-T}^T F(t)dt = \lim_{u \to 0} \frac{f(u)+f(-u)}{2}$ $\endgroup$ – reuns Oct 27 '18 at 21:08
  • $\begingroup$ @reuns, thanks. So it seems to me that $\lim_{u\rightarrow 0} \frac{f(u)+f(-u)}{2}=0$, isn't it ? $\endgroup$ – acc10 Oct 27 '18 at 21:21
  • 2
    $\begingroup$ @StanleyYaoXiao, i think the integral should converge since $\zeta(1/2 + it)=o(t)$. $\endgroup$ – acc10 Oct 27 '18 at 21:23
  • $\begingroup$ No because $f(u)$ is supported on $u > 0$ (the $u < 0$ part is in the poles I substracted from $\frac{\zeta(1/2+it)}{t^2+1/4}$) $\endgroup$ – reuns Oct 27 '18 at 21:24
2
$\begingroup$

For $\Re(s) > 1$, $$\frac{\zeta(s)}{s} = \int_1^\infty (\sum_{1 \le n \le x} 1) x^{-s-1}dx, \qquad \frac{\zeta(s)}{s-1} = \int_1^\infty (x\sum_{1 \le n \le x} \frac1n) x^{-s-1}dx$$ Thus for $\Re(s) \in (0,1)$ $$\frac{\zeta(s)}{s}+\frac{\zeta(s)}{1-s}= \int_0^\infty ((\sum_{1 \le n \le x} 1-x)-x(\sum_{1 \le n \le x} \frac1n-\log x-\gamma)) x^{-s-1}dx$$

So that $\frac{\zeta(1/2+it)}{1/4+t^2}$ is the Fourier transform of $$g(u) = ((\sum_{1 \le n \le e^u} 1)-e^u-e^u(\sum_{1 \le n \le e^u} \frac1n) + e^u u+e^u\gamma)e^{-u/2}$$ and $$\int_{-\infty}^\infty \frac{\zeta(1/2+it)}{1/4+t^2} dt = 2\pi\lim_{u \to 0} \frac{g(u)+g(-u)}{2}=2\pi( \gamma-1)$$

The integral converges absolutely because $\zeta(1/2+it) = O(t^c),c < 1$, and for the same reason you can shift the contour to $1/2+i(-\infty,+\infty)$ to $+\infty+i(-\infty,+\infty)$) and evaluate the integral as $-2 \pi \text{Res}(\frac{\zeta(s)}{s(1-s)},1)$.

$\endgroup$
1

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.