On the distribution of positive and negative values of Hardy's \(Z\)-function (Q503706)

This paper investigates the distribution of positive and negative values of Hardy's function \[ Z(t):=\zeta\left(\tfrac{1}{2}+it\right)\chi\left(\tfrac{1}{2}+it\right)^{-1/2}, \] where \(\chi\) is the factor from the functional equation of the Riemann zeta-function, namely, \(\zeta(s)=\chi(s)\zeta(1-s)\). Note that \(Z(t)\) is real-valued and the non-trivial zeros of the Riemann zeta-function correspond to the zeros of Hardy's function, provided the Riemann Hypothesis holds true. Let \(2\leq H\leq T\) and set \[ I_{+}(T,H)=\left\{T<t\leq T+H : Z(t)>0\right\} \] and \[ I_{-}(T,H)=\left\{T<t\leq T+H : Z(t)>0\right\}. \] Numerical computations suggest that the Lebesgue measures of the above quantities are both close to \(H/2\) even if \(H\) is quite small compared to \(T\). So there should be no bias of \(Z(t)\) towards positive or negative values. The authors prove that \[ \mu(I_{+}(T,H))\gg T \quad \text{and} \quad \mu(I_{-}(T,H))\gg T, \] which gives support towards the above conjecture. To this end, they establish a lower bound for the integral \[ \int_T^{2T} |Z(t)|\left|B_X\left(\tfrac{1}{2}+it\right)\right|^2\mathrm{d}t \] and upper bounds for the integrals \[ \int_T^{2T} Z(t)\left|B_X\left(\tfrac{1}{2}+it\right)\right|^2\mathrm{d}t \] and \[ \int_T^{2T} Z(t)^2\left|B_X\left(\tfrac{1}{2}+it\right)\right|^4\mathrm{d}t, \] where \(\left|B_X\left(\tfrac{1}{2}+it\right)\right|^2\) is a mollifier depending on a positive parameter \(X\) which was used already by Hardy in his famous proof that a positive proportions of the zeta zeros are on the critical line.