Let $\zeta$ be the zeta function of Riemann. Is the bound for

$$I_{T}=\int_{0}^{T} \Big|\log|\zeta(1/2 + it)| \Big| \mathrm{d}t$$ known ?

It seems to me that $I_{T} \ll T\log T$ since $\log|\zeta(1/2+it)|\ll \log t$ if $t$ is not an ordinate of a zero. But certainly this is too naive. Is there a better bound ?


We have $I_{T} \ll T\log T$ for $T\geq 2$. For this it suffices to verify that $$ \int_{T}^{T+1} \Big|\log|\zeta(1/2 + it)|\Big|\ dt\ll\log T,\qquad T\geq 2.\tag{$\ast$}$$ We can deduce this local bound from Theorem 9.6 (B) and surrounding material in Titchmarsh: The theory of the Riemann zeta-function. Indeed, this theorem implies (after taking the real part and applying the triangle inequality) that $$\Big|\log|\zeta(1/2 + it)|\Big|\leq\sum_{|t-\gamma|\leq 1} \Big|\log|1/2 + it-\rho)|\Big|+O(\log t),\qquad t\geq 2,$$ where $\rho=\beta+i\gamma$ runs through the zeros of $\zeta(s)$. It follows that $$\Big|\log|\zeta(1/2 + it)|\Big|\leq\sum_{|t-\gamma|\leq 1} \Big(1-\log|t-\gamma|\Big)+O(\log t),\qquad t\geq 2,$$ since $\log|t-\gamma|$ is nonpositive throughout the sum. The contribution of the term $1$ in the sum is $O(\log t)$ by Theorem 9.2, hence in fact $$\Big|\log|\zeta(1/2 + it)|\Big|\leq-\sum_{|t-\gamma|\leq 1} \log|t-\gamma|+O(\log t),\qquad t\geq 2.$$ On the other hand, from the proof of Theorem 9.7 we learn that there is a constant $A>0$ such that $$\int_T^{T+1}\sum_{|t-\gamma|\leq 1}\log|t-\gamma|\ dt\geq -A\log T,\qquad T\geq 2.$$ Combining the previous two inequalities, the bound $(\ast)$ follows.

  • 1
    $\begingroup$ Wow, that's really ingenious of you ! Thanks ! $\endgroup$ – ConsiderTheMathOverFlaw Dec 15 '18 at 16:56
  • $\begingroup$ @GHfroMO, done. $\endgroup$ – ConsiderTheMathOverFlaw Dec 15 '18 at 18:42
  • $\begingroup$ @ConsiderTheMathOverFlaw: At mathoverflow.net/questions/316668 I claimed that $\int_T^{2T}\zeta(\sigma+it)\,dt = T+O_\sigma(T^{1-\sigma})$ implies $\int_0^T\zeta(\sigma+it)\,dt = T+O_\sigma(T^{1-\sigma})$. This is can be seen as follows. Let $T\geq 2$ be fixed, and let $K$ be the largest integer for which $2^K\leq T$. Then, for any integer $1\leq k\leq K$, we have $\int_{T/2^k}^{2T/2^k}\zeta(\sigma+it)\,dt = T/2^k+O_\sigma(T^{1-\sigma}/2^{k(1-\sigma)})$. The sum of these yields $\int_{T_0}^T\zeta(\sigma+it)\,dt = T+O_\sigma(T^{1-\sigma})$, where $T_0:=T/2^K$ lies in $[1,2]$. Done. $\endgroup$ – GH from MO Dec 15 '18 at 18:43
  • $\begingroup$ Thanks, but where did $$\int_{T/2^k}^{2T/2^{k}} \zeta(\sigma+ it) \mathrm{d}t = T/2^{k} + O_{\sigma}(T^{1-\sigma}/2^{k(1-\sigma)})$$ come from ? $\endgroup$ – ConsiderTheMathOverFlaw Dec 15 '18 at 18:51
  • $\begingroup$ @ConsiderTheMathOverFlaw: It came from $\int_T^{2T}\zeta(\sigma+it)\,dt = T+O_\sigma(T^{1-\sigma})$ that I proved under that link. $T$ is arbitrary, so one can replace it by $T/2^k$. $\endgroup$ – GH from MO Dec 15 '18 at 20:47

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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