A Chabauty-Coleman bound for surfaces (Q6070678)

In this paper, the authors make a quantum leap towards an important problem which has seen substantial activity recently. The question is whether one can give a uniform bound for the number of rational points on an algebraic variety, when it is known that the number of such points must be finite. In the curve case, Mordell's conjecture, now Faltings' theorem, asserts that any algebraic curve of general type will have finitely many rational points. Faltings' original proof, and all subsequent proofs of this result, do in fact give numerical bounds for the number of rational points, but they are often poor and very far from the truth. Coleman's bounds, obtained by a careful refinement of Chabauty's methods, remains one of the best general bounds for the number of rational points on a curve of genus at least two provided that the conditions hold. The Chabauty-Coleman method has been employed fruitfully many times in the past decade, leading to a plethora of fantastic results, for example [\textit{J. Balakrishnan} et al., Ann. Math. (2) 189, No. 3, 885--944 (2019; Zbl 1469.14050); \textit{E. Katz} et al., Duke Math. J. 165, No. 16, 3189--3240 (2016; Zbl 1428.11121); \textit{M. Stoll}, J. Eur. Math. Soc. (JEMS) 21, No. 3, 923--956 (2019; Zbl 1428.11122)]. The higher dimensional case has seen far less activity, primarily due to its difficulty. It is often unknown, for a generic algebraic variety \(X\) of dimension \(d \geq 2\), what conditions must be applied to ensure that \(X(\mathbb{Q})\) will be finite. Being general type is not enough: for example, Fermat surfaces defined by equations of the shape \(x_1^d + x_2^d = x_3^d + x_4^d\) will have infinitely many projective rational points despite being of general type for \(d \geq 5\). Finiteness results are most abundant for subvarieties of abelian varieties, as supplied by work of Faltings. It is in this setting that the authors make their mark. The authors prove that for \(X\) a hyperbolic surface defined over \(\mathbb{Q}\) which can be embedded into an abelian variety \(A\) of rank at most \(1\) and for a suitable auxiliary prime \(p\), that the bound \[ \# X(\mathbb{Q}) \leq \# X^\prime(\mathbb{F}_p) + \frac{p-1}{p-2} \left(p + 4p^{\frac{1}{2}} + 5 \right) c_1^2(X), \] where \(X^\prime\) is the reduction of \(X\) modulo \(p\) and \(c_1^2(X)\) is the first Chern number of \(X\). This is a very nice result and a substantial generalization of Coleman's bounds for curves. Further, the methods introduced by the authors, using algebraic methods to replace certain analytic approaches used over \(\mathbb{C}\) that become unsuitable in the positive characteristic case, gives hope that similar approaches can prove far stronger results for surfaces, including towards a variant of Mazur's Conjecture B for surfaces.