Mean square discrepancy bounds for the number of lattice points in large convex bodies (Q1809913)

Let \(D\) be a convex domain in \(\mathbb{R}^d\) containing the origin in its interior. Suppose that \(D\) is sufficiently smooth and that the Gaussian curvature of the boundary is nonzero throughout. Let \(N_D(t)\) denote the number of lattice points in the dilated domain \(tD\). Further let \[ \Delta_D(t)= \frac {N_D(t)- t^d \operatorname {vol}(D)} {t^d \operatorname {vol}(D)} \] denote the discrepancy function. The main purpose of the paper is to consider the mean square discrepancy \[ G_D(R)= \Biggl( \frac{1}{R} \int_R^{2R} |\Delta_D(t) |^2 dt \Biggr)^{1/2} \] of the lattice remainder. \textit{W. G. Nowak} [Math. Proc. Camb. Philos. Soc. 98, 1-4 (1985; Zbl 0552.10032)] proved that \(G_D(R)= O(R^{-3/2})\) for planar domains and \textit{W. Müller} [Acta Arith. 80, 89-100 (1997; Zbl 0871.11070)] \(G_D(R)= O(R^{-2+\varepsilon})\) for \(d\geq 4\). The main results of the paper are the estimations \(G_D(R)= O(R^{-2})\) for \(d\geq 4\) and \(G_D(R)= O(R^{-2}\log R)\) for \(d=3\). Furthermore, it is proved that \[ \Biggl(\frac{1}{h} \int_R^{R+h} |\Delta_D(t) |^2 dt \Biggr)^{1/2}\leq C(D) R^{-3/2} \] holds for \(\log R\leq h\leq R\) when \(d=2\).