The divisor problem for binary cubic forms (Q449712): Difference between revisions

Let \(C(x,y)\) be an integral cubic form which factors over the rationals into linearly independent linear forms, as \(L_1L_2L_3\). Then it is shown that \[ \sum_{x,y\leq X}\tau(C(x,y))=X^2P_C(\log X)+O_{\varepsilon,C}(X^{7/4+\varepsilon}) \] for some cubic polynomial \(P_C\). For comparison, the best known error term for the case in which \(C\) is irreducible has the exponent \(7/4\) replaced by \(15/8\) (see \textit{S. Daniel} [J. Reine Angew. Math. 507, 107--129 (1999; Zbl 0913.11041)]). The proof draws heavily from the work of \textit{R. de la Bretèche} and \textit{T. D. Browning} [Compos. Math. 144, No. 6, 1375--1402 (2008; Zbl 1234.11132), and J. Reine Angew. Math. 646, 1--44 (2010; Zbl 1204.11158)]. The author also considers the ternary additive divisor problem described by \[ T_h(X):=\sum_{h<n\leq X}\tau(n-h)\tau(n)\tau(n+h). \] It is conjectured that \(T_h(X)\sim c_hX(\log X)^3\) for a suitable explicitly given constant \(c_h>0\), but this seems far out of reach. In support of the conjecture it is shown that \[ \sum_{h\leq H}\left(T_h(X)-c_hX(\log X)^3\right)=o(HX(\log x)^3) \] for \(H\geq X^{3/4+\varepsilon}\), for any fixed \(\varepsilon>0\).