New results on the value distribution of the Riemann zeta-function on the critical line (Q601780)

The authors present an interesting survey (with some graphical illustrations) concerning some aspects of the value distribution and the non-trivial zeros of the Riemann zeta function; they give results, and often the ideas of proofs. For example, section 2 mentions Bohr--Courant's result (1914) on the denseness of the set \(\{\zeta(\sigma + it),\;t\in\mathbb R\}\) in \(\mathbb C\) for any \(\sigma \in (\frac12, 1]\), and Voronin's universality theorem (1975). In section 3 the authors give a result of Garunkštis and Steuding, which states that the set \(\{(\zeta(\frac12 +it,\;\zeta^\prime(\frac12 +it));\, t\in \mathbb R\}\) is not dense in \(\mathbb C^2\), and also asymptotic formulae for the discrete moments \[ \sum_{{0<t\leq T}\atop {\zeta(\frac12+it)\in e^{i\Phi}\mathbb R}} \zeta\left(\frac12 +it\right) = 2e^{i\Phi} \cos \Phi \cdot \frac T{2\pi} \cdot \log\frac T{2\pi e} + \mathcal O\left(T^{\frac12 + \varepsilon}\right), \] and a similar result, where \(\zeta\left(\frac12 +it\right)\) is replaced with \(\left| \zeta\left(\frac12 +it\right)\right|^2\). Section 4 is concerned with Gram's law and a weak Gram's law (there is exactly one zero \(s=\frac12 +i\gamma\) of \(\zeta(s)\), for \(\gamma \in (g_n^{(\phi)}, g_{n+1}^{(\phi)}]\). The number \(g_n^{(\phi)}\) is defined by \((-1)^n Z(g_n^{(\phi)}) = e^{i\phi} \, \zeta\left(\frac12 + ig_n^{(\phi)}\right)\), where \(Z(t)\) denotes the function \(\zeta(\frac12+it) \cdot \Delta(\frac12 -it)^{\frac12}\), and \(\Delta(s) = 2^s \pi^{s-1} \Gamma(1-s) \sin\left(\frac{\pi s}2\right)\)). The authors sketch a proof of a result implying that even the weak Gram's law is violated infinitely often. Section 5 describes attempts, to use random matrices as a model for the behaviour of the zeta-function; Montgomery's pair correlation conjecture, a limit theorem due to Keating and Snaith (2000) are stated, and a limit theorem of Selberg (unpublished, a proof was given by Joyner 1986): \[ \lim_{T\to\infty} \frac1T \text{meas}\Biggl\{ t\in [T,2T]: \frac{\log \zeta(\frac12+it)}{\sqrt{\frac12 \log\log T\;}}\in\mathcal R\Biggr\} = \frac1{2\pi} \int_{\mathcal R} \exp\left(-\frac12 (x^2+y^2)\right) dxdy. \] [Here \(\mathcal R\) is a rectangle with sides parallel to the axes.] The last section describes a model for the behaviour of \(\zeta(s)\) due to Gonek (2007, preprint); this model uses a modified Euler product, \[ P_X(s) = \exp\left( \sum_{n\leq X^2} \frac{\Lambda_X(n)}{n^s \log n}\right). \] If Riemann's hypothesis is true, then (according to Gonek) \[ \zeta(s) = P_X(s) \;(1+o(1)), \quad \text{as } t\to\infty, \] in \(2\leq X \leq t^2\) and \( \frac12 + C \; \frac{\log\log 2t}{\log X} \leq \sigma \leq 1\), {where } \[ \Lambda_X(n) = \begin{cases} \Lambda(n) & \text{for }n\leq X,\\ \Lambda(n)\cdot\left(2 -\frac{\log n}{\log X}\right) & \text{for } X<n\leq X^2,\\ 0 & \text{for } n>X.\end{cases} \] Gonek's function \[ \zeta_X(s)= P_X(s) \cdot\left( 1+ \Delta(s) \;\frac{P_X(\bar s)}{P_X(s)}\right) \] does not have zeros for \(\sigma\not=\frac12\) and \(t\geq 10\).