On Balazard, Saias, and Yor's equivalence to the Riemann hypothesis (Q2258962): Difference between revisions

Let \(\zeta(s)\) be the Riemann zeta-function. One of the statements equivalent to the Riemann hypothesis (RH) is that the weighted integral of the logarithm of the function \(\zeta(s)\) along the critical line is equal to zero, i.e, \[ \int_{\text{Re}(s)=1/2}\frac{\log|\zeta(s)|}{|s|^2}|ds|=0. \] \textit{J.M. Borwein} et al. [J. Comput. Appl. Math. 121, No.1-2, 247--296 (2000; Zbl 0972.11077)] conjectured that under RH \[ I(T):=\int_{-T}^{T}\frac{\log|\zeta(1/2+it)|}{1/4+t^2}dt=O(T^{-2}) \] as \(T \to \infty\). The authors disprove this conjecture showing that, for any fixed positive \(\alpha<2\), \[ I(T)=O\bigg(\frac{1}{T^2}\frac{\log T}{(\log \log T)^2}\bigg) \] and \[ I(T)=\Omega\bigg(\frac{1}{T^2}\frac{\sqrt{\log T}}{(\log\log T)^{3/2}}\bigg) \] as \(T \to \infty\). In addition, the authors obtain unconditional formula for \(I(T)\) in terms of the nontrivial zeros \(\varrho=\beta+i\gamma\) of the Riemann zeta-function, i.e., \[ I(T)=2 \pi \sum_{-T \leq \gamma \leq T, \beta>1/2}\log\big|\frac{\varrho}{1-\varrho}\big|+O\bigg(\frac{1}{T^2}\log T\bigg). \] for \(T \geq 3\). They note that the Lindelöf hypothesis implies that the error term in last equality is \(o(T^{-2}\log T)\) as \(T \to \infty\). The authors give a new proof of a classical Omega theorem for the function \(S(t)\), which appears in the theory of the Riemann zeta-function.