Some remarks on the moments of \(| \zeta(1/2 + it)| \) in short intervals (Q949815): Difference between revisions

In this paper the author considers estimates for a Gaussian smoothing of powers of the Riemann Zeta--function \(\zeta(s)\). We set \[ J(G,T) = \frac 1{\sqrt \pi G} \int_{-\infty}^\infty | \zeta(\frac 12 +(T-G)\imath)| ^2 \exp(-u^2/G^2) \text{d} u \] which has ``width'' \(G\) around \(T\). We shall only be interested in the case where \(0 < G \leq T\). The Lindelöf Hypothesis would imply that \(J_k(G,T) = \mathcal O(T^\epsilon)\) for any \(\epsilon > 0\). The author relates the the function \(J(G,T)\) to some recent work of his on divisor problems to prove \[ \int_T^{2T} J(t,G)^m \text{d} t = \mathcal O(T^{1+\epsilon}) \] in the following cases where \(1 \leq m \leq 5\) and \(T^{c(m) + \epsilon} \leq G \leq T\) where \(c(1)=0, c(2)=0, c(3)= \frac 17, c(4)= \frac 7{36}\) and \(c(5) = \frac 29\).