The density of rational points on a certain singular cubic surface (Q2503370): Difference between revisions

Let \(X\subset\mathbb{P}^3\) be the singular cubic surface given by \[ x_1x_2x_3= x_4(x_1+ x_2+ x_3)^2. \] This has exactly one singular point, which is of type \(D_4\). Moreover, it contains exactly 6 lines, all of which are defined over \(\mathbb{Q}\). Let \(U\) be the complement of the lines in \(X\), and let \(N_U(B)\) be the standard counting function for rational points on \(U\) with anticanonical height at most \(B\). In this case Manin's conjecture would predict that \(N_U(B)\sim cB(\log B)^6\), and the paper shows that \[ B(\log B)^6\ll N_U(B)\ll B(\log B)^6. \] The proof begins by passing to the universal torsor, given by \[ s_0 s_1 s_2 s_3 u_1 u_2 u_3= y_1 u_1 s_1+ y_2 u_2 s^2_2+ y_3 u_3 s^2_3, \] and one then has to count solutions of this equation. Producing the lower bound is relatively straightforward, but for the upper bound a variety of different techniques are used, depending on the relative sizes of the variables. To this extent the paper follows the reviewer's treatment [The density of rational points on Cayley's cubic surface. Proceedings of the Session in Analytic Number Theory and Diophantine Equations, 33 pp., Bonner Math. Schriften, 360, Univ. Bonn, Bonn (2003; Zbl 1060.11038)] of the Cayley cubic. However in addition to the use of the geometry of numbers as in the above work, one now needs the large sieve inequality and estimates for real character sums.