Visible lattice points in the sphere (Q2466032)

Let \(R\) be a large real and let \(N^*(R)\) denote the number of primitive lattice points in \(\mathbb Z^3\) that lie inside the ball of radius \(R\) centered at the origin: \[ N^*(R)= \#\left\{(n_1,n_2,n_3)\in \mathbb Z^3: \gcd(n_1,n_2,n_3)= 1, \;n_1^2+ n_2^2+ n_3^2\leq R^2\right\}. \] We define \(E^*(R)= N^*(R)- \frac 43\pi R^3/\zeta(3)\), where \(\zeta(s)\) the Riemann zeta-function. The remainder \(E^*(R)\) is related to the remainder in the sphere problem. In particular, \textit{D. R. Heath-Brown}'s estimate for the latter remainder [Lattice points in the sphere. Number theory in progress. Volume 2: Elementary and analytic number theory. Berlin: de Gruyter. 883--892 (1999; Zbl 0929.11040)], implies that \(E^*(R)\ll_{\varepsilon} R^{21/16+ \varepsilon}\) for any fixed \(\varepsilon> 0\). In the paper under review, the authors study the oscillations of \(E^*(R)\). Their main result states that \(E^*(R)= \Omega(R(\log R)^{1/2})\). The standard approach to such \(\Omega\)-results is to estimate the second power moment \(M_2(R)= \int| E^*(t)|^2\, d\nu_R(t)\) for a suitably chosen measure \(\nu_R\) supported in \((R, 2R]\). In contrast to other similar problems, an asymptotic formula for \(M_2(R)\) appears to be out of reach at present, so the authors derive a lower bound for \(M_2(R)\) by calculating the correlation of \(E^*(t)\) and an auxiliary trigonometric polynomial \(g(t)\).