Lower bounds for the maximum of the Riemann zeta function along vertical lines (Q289854): Difference between revisions

Let \(\zeta(s)\) denote the Riemann zeta-function. A fundamental problem is the estimation of \(\zeta(\alpha+it)\) for \(1/2\leq\alpha\leq1\), and the author proves the lower bound \[ \max_{0\leq t \leq T}|\zeta(\alpha + it)| \geq \exp\left(\frac{c_\alpha(\log T)^{1-\alpha}} {(\log\log T)^\alpha}\right)\leqno(1) \] for fixed \(1/2 < \alpha < 1\), large \(T\), and \(c_\alpha = 0.18(2\alpha-1)^{1-\alpha}\). The bound (1) was proved long ago by \textit{H. L. Montgomery} [Comment. Math. Helv. 52, 511--518 (1977; Zbl 0373.10024)]. The present proof of (1) is new, and the constant \(c_\alpha\) better than Montgomery's. It is based on a modified version of \textit{K. Soundararajan} [Math. Ann. 342, No. 2, 467--486 (2008; Zbl 1186.11049)] and some ideas of \textit{T. Hilberdink} [Acta Arith. 139, No. 4, 341--367 (2009; Zbl 1252.11071)]. The method of proof enables the author to estimate from below the measure of \(t \in (0,T]\) for which \(|\zeta(\alpha + it)|\) is of the order given by (1). In a note at the end of the paper, the author remarks that recently \textit{A. Bondarenko} and \textit{K. Seip} [``Large GCD sums and extreme values of the Riemann zeta function'', Preprint, \url{arXiv:1507:05840}] use similar methods to obtain the new remarkable bound, for large \(T\), \[ \max_{0\leq t \leq T}|\zeta(1/2 + it)| \geq \exp\left(\frac{c\sqrt{\log T\log\log\log T}} {\sqrt{\log\log T}}\right)\qquad(c>0).\leqno(2) \] The bound (2) improves on previous bounds for \(\max\limits_{0\leq t \leq T}|\zeta(1/2 + it)|\) by a factor of \(\sqrt{\log\log\log T}\), and does not follow from (1) as the limiting case when \(\alpha\to1/2\).