An effective universality theorem for the Riemann zeta function (Q1757305): Difference between revisions

The famous ``universality theorem'' for \(\zeta(s)\) of \textit{S. M. Voronin} [Sov. Math., Dokl. 16, 410 (1975; Zbl 0326.10035); translation from Dokl. Akad. Nauk SSSR 221, 771 (1975)] states the following. Let \(0<r<1/4\) and \(f\) be a non-vanishing, continuous function in \(|z| \leq r\) that is analytic in the interior. For any given \(\varepsilon>0\) one has \[ \liminf_{T\to\infty}\frac{1}{T}\mu\Bigl\{T \leq t \leq 2T:\max_{|z|\leq r} |\zeta(\frac{3}{4}+it+z)-f(z)| < \varepsilon\Bigr\} >0,\leqno(1) \] where \(\mu(\cdot)\) denotes Lebesgue measure. In this significant work, the authors obtain, for the first time, a rate of convergence in Voronin's universality theorem. They also give an explicit description for the limit in terms of the random model (\(p\) denotes primes) \[ \zeta(s,X) := \prod_p{\left(1- \frac{X(p)}{p^s}\right)}^{-1}, \] where \(\{X(p)\}_p\) is a sequence of independent random variables uniformly distributed on the unit circle. For example, they obtain from their Theorem 1.1 that the expression in liminf in (1) equals \[ {P}(\max_{|z|\leq r} |\zeta(\frac{3}{4}+z,X)-f(z)| < \varepsilon) + O((\log T)^{-(3/4-r)/11+o(1)}),\leqno(2) \] where \(P\) denotes probability. The proof uses many ideas from analysis and probability theory. It is stated that it seems difficult to obtain a rate of convergence better than the logarithmic one, as in (2) above. The authors also say that they have no idea as what should be the correct rate of convergence. The method can be generalized to other \(L\)-functions in the \(t\)-aspect, as well to families of \(L\)-functions in the conductor aspect.