0
$\begingroup$

I am reading The Riemann zeta function by H.M. Edwards , $1974$, pg. $52-53$. Here the author denotes by $\rho$ the non trivial zeros of Riemann zeta function

He proves that the following series in $\rho$ converges uniformly in any disc $|s|\leq K$ when the terms $\rho$ and $1-\rho$ are paired $$\sum_{\rho}\frac{1}{s-\rho}$$ For this he observes $$\left|\frac{1}{s-\rho}+\frac{1}{s-(1-\rho)}\right|=\left|\left[\frac{1}{\left(s-\frac{1}{2}\right)-\left(\rho-\frac{1}{2}\right)}\right]+\left[\frac{1}{\left(s-\frac{1}{2}\right)+\left(\rho-\frac{1}{2}\right)}\right]\right| $$ which gives $$ \left|\frac{1}{s-\rho}+\frac{1}{s-(1-\rho)}\right|=\left|\frac{2\left(s-\frac{1}{2}\right)}{\left(s-\frac{1}{2}\right)^2-\left(\rho-\frac{1}{2}\right)^2}\right|$$ Now we have $$\left|\frac{2\left(s-\frac{1}{2}\right)}{\left(s-\frac{1}{2}\right)^2-\left(\rho-\frac{1}{2}\right)^2}\right|\leq \frac{C}{\left|\rho-\frac{1}{2}\right|^2}$$ for all sufficiently large $\rho$ once $K$ is fixed and because $\sum_{\rho}\frac{1}{|\rho-\frac{1}{2}|^2}$ converges.

Question: How do we have the last inequality for all sufficiently large $\rho$? How does it ensure the uniform convergence?

Can someone please clear this doubt. Thank you.

$\endgroup$
5
  • $\begingroup$ @JyrkiLahtonen Sorry I could not follow. So we have an upper bound for the tail of the sum. How does it ensure uniform convergence? $\endgroup$
    – user1073119
    Jul 14, 2022 at 15:58
  • $\begingroup$ Wait a second. I only now realize that $\rho$ was not intended to be an integer variable. Scratch all that. $\endgroup$ Jul 14, 2022 at 16:22
  • $\begingroup$ @JyrkiLahtonen Yes $\rho$ is a complex number with real part between $0$ and $1$. $\endgroup$
    – user1073119
    Jul 14, 2022 at 16:23
  • $\begingroup$ Is it known that $\sum_\rho|\rho-\dfrac12|^{-2}$ converges? When I didn't read carefully, and mistakenly assumed $\rho$ to be an integer variable, the convergence was obvious. If the convergence of that sum is known for some other reason, then an argument like Weierstrass M-test should still work. $\endgroup$ Jul 15, 2022 at 12:41
  • $\begingroup$ @JyrkiLahtonen Yes that's okay. But I need an estimate valid for all $\rho$. Please see the answer below. But I am not able to understand why the second maximum exist or why $C_K$ should exist. $\endgroup$
    – user1073119
    Jul 15, 2022 at 12:44

1 Answer 1

1
$\begingroup$

When $|s|\le K$, we have $|s-\frac12|\le K+\frac12$. By the properties of $\rho$, we know that when $\Im(\rho)>\sqrt2(K+\frac12)$, there is $|\rho-\frac12|>\sqrt2(K+\frac12)$, so $|\rho-\frac12|>\sqrt2|s-\frac12|$. Consequently, we have $$ \left|{2\left(s-\frac12\right)\over\left(s-\frac12\right)^2-\left(\rho-\frac12\right)^2}\right|\le{2K+1\over|\rho-\frac12|^2-|s-\frac12|^2}<{4K+2\over|\rho-\frac12|^2}. $$

$\endgroup$
19
  • $\begingroup$ Thanks. But how do we know that $\Im (\rho)$ is large enough? We have the first $\Im (\rho)$ at $\Im(\rho)\approx 14.1$. Then do we have similar inequality to the above inequality for small $\rho$? $\endgroup$
    – user1073119
    Jul 15, 2022 at 9:09
  • $\begingroup$ Please prove an inequality valid for all non trivial zeros of Riemann zeta, $\rho$ $\endgroup$
    – user1073119
    Jul 15, 2022 at 10:15
  • $\begingroup$ As long as you can find a constant valid for large $\Im(\rho)$, you can automatically deduce that replacing $4K+2$ by a larger number will make the inequality work for all $\rho$. $\endgroup$
    – TravorLZH
    Jul 15, 2022 at 11:55
  • $\begingroup$ By large, I mean that there exists $T>0$ such that when $\Im(\rho)>T$, we have $|\rho-\frac12|>\sqrt2(K+\frac12)$, which is certainly possible because there are $\zeta(s)$ has infinitely many nontrivial zeros. $\endgroup$
    – TravorLZH
    Jul 15, 2022 at 11:56
  • $\begingroup$ I thank you for your guidance. Please show that how can we automatically deduce that replacing $4K+2$ by a larger number which will make the inequality work for all $\rho$. Please include this in your answer. $\endgroup$
    – user1073119
    Jul 15, 2022 at 12:03

You must log in to answer this question.