On the Riemann zeta-function and the divisor problem (Q1767434)

Let \(\Delta(T)\) and \(E(T)\) denote, respectively, the error terms in the Dirichlet divisor problem and in the mean square formula for \(\zeta(1/2 + it)\): \[ \Delta(T) = \sum_{n \leq T} d(n) - T(\log T + 2\gamma - 1) - 1/4, \] and \[ E(T) = \int_0^T | \zeta(1/2 + it)| ^2 \, dt - T\big( \log(T/(2\pi)) + 2\gamma - 1 \big), \] where \(d(n)\) is the number of positive divisors of \(n\), \(\zeta(s)\) is the Riemann zeta-function, and \(\gamma\) is Euler's constant. Some strong similarities in the behaviors of these two error terms have prompted the study of the possible relation between them. \textit{M.~Jutila} [Ark. Mat. 21, 75--96 (1983; Zbl 0513.10040)] showed that \(E(T)\) should actually be compared with \(2\pi\Delta^*(T/(2\pi))\), where \[ \Delta^*(T) = -\Delta(T) + 2\Delta(2T) - 0.5\Delta(4T). \] Let \(E^*(T) = E(T) - 2\pi\Delta^*(T/(2\pi))\). Jutila [op. cit.] proved the mean square bound \[ \int_0^T E^*(t)^2 \, dt \ll T^{4/3}(\log T)^3, \] which is consistent with the inequality \(E^*(T) \ll T^{1/6}(\log T)^{3/2}\). (Note that \(E(T)\) and \(\Delta(T)\) are conjectured to be of order \(T^{1/4}\).) In the paper under review, the author establishes the following bound for the fourth power moment of \(E^*(T)\): for any fixed \(\varepsilon > 0\), one has \[ \int_0^T E^*(t)^4 \, dt \ll_{\varepsilon} T^{16/9 + \varepsilon}. \] Furthermore, he shows how the method of proof of the above bound can be used to obtain the inequality \[ \sum_{r = 1}^R \bigg( \int_{t_r - G}^{t_r + G} | \zeta(1/2 + it)| ^2 \, dt \bigg)^4 \ll_{\varepsilon} T^{2 + \varepsilon}G^{-2} + RG^4T^{\varepsilon}, \] where \(T^{1/5 + \varepsilon} \leq G \ll T\), \(T < t_1 < t_2 < \cdots < t_r \leq 2T\), \(t_{r + 1} - t_r \geq 5G \, (1 \leq r < R)\).