Uniformity in Mordell-Lang for curves (Q2170807)

Let \(C\) be a smooth curve of genus \(g\geq 2\) with a model defined over a number field \(F\), a field extension of \(\mathbb{Q}\) of degree \(d\geq 1\) where we assume once and for all that all \(F\subset \overline{\mathbb{Q}}\) the algebraic closure of \(\mathbb{Q}\). By Faltings, \(C(F)\), the set of \(F\)-rational points of \(C\), is a finite set [\textit{G. Faltings}, Invent. Math. 73, 349--366 (1983; Zbl 0588.14026)]. Let \(\mathrm{Jac}(C)\) be the Jacobian of \(C\), and \(\mathrm{Jac}(C)(F)\) denotes the finitely generated abelian group of the \(F\)-points of the Jacobian of \(C\). \textit{B. Mazur} [Bull. Am. Math. Soc., New Ser. 14, 207--259 (1986; Zbl 0593.14021); Math. Sci. Res. Inst. Publ. 39, 199--227 (2000; Zbl 0990.14006)] formulated the following question: given \(g\geq 2\) and \(d\geq 1\) exists a constant \(c=c(g,d)\) depending only of \(g\) and \(d\) such that for any smooth curve \(C\) of genus \(g\) defined over a number field \(F\) with \([F:\mathbb{Q}]\leq d\) then \[ \# C(F)\leq c^{1+\rho} \] where \(\rho=\mathrm{rank}_{\mathbb{Z}}\mathrm{Jac}(C)(F)\)? The main result of the paper is to answer affirmatively this question. Inspired from the \textit{P. Vojta}'s proof [Ann. Math. (2) 133, No. 3, 509--548 (1991; Zbl 0774.14019)] of the Mordell Conjecture, the authors consider the Néron-Tate height: \[ \hat{h}:\mathrm{Jac}(C)(\overline{\mathbb{Q}})\rightarrow \mathbb{R} \] (where once and for all we fix \(P_0\in C(\overline{\mathbb{Q}})\), identify \(C\) with its image under Abel-Jacobi embedding \(C\hookrightarrow \mathrm{Jac}(C)\) via \(P_0\), and \(\Gamma\) is a subgroup of \(\mathrm{Jac}(C)(\overline{\mathbb{Q}})\) of finite rank \(\rho\)). Also consider the absolute logarithmic Weil height [\textit{E. Bombieri} and \textit{W. Gubler}, Heights in Diophantine geometry. Cambridge: Cambridge University Press (2006; Zbl 1115.11034)] \(h:\mathbb{P}_{\overline{\mathbb{Q}}}^m(\overline{\mathbb{Q}})\rightarrow \mathbb{R}\) with a fix immersion \(\iota:\mathcal{A}_{g,1}\rightarrow \mathbb{P}_{\overline{\mathbb{Q}}}^m\) where \(\mathcal{A}_{g,1}\) denotes the coarse moduli space of principally polarized varieties of dimension \(g\). Then, the question posed by Mazur in terms of height of the Jacobian instead by working over a number field of bounded degree can be formulated on the existence of two constants \(c_i(g,\iota)\) in the sense that if \(h(\iota([\mathrm{Jac}(C)]))\geq c_1(g,\iota)\) then \[ \# ((C(\overline{\mathbb{Q}})-P_0)\cap \Gamma) \leq c_2(g,\iota)^{1+\rho}. \] A usual strategy (since the work of Vojta) is divide \((C(\overline{\mathbb{Q}})-P_0)\cap \Gamma\) into two parts (where recall for abuse of notation \(C(\overline{\mathbb{Q}})\) by \(C(\overline{\mathbb{Q}})-P_0\subset \mathrm{Jac}(C)(\overline{\mathbb{Q}})\)): \begin{itemize} \item small points \(\{P\in C(\overline{\mathbb{Q}})\cap \Gamma:\hat{h}(P)\leq B(C)\}\), \item large points \(\{P\in C(\overline{\mathbb{Q}})\cap \Gamma:\hat{h}(P)\leq B(C)\}\), \end{itemize} where \(B(C)\) depends of certain height associated to \(C\). The choice of \(B(C)=constant\cdot max\{1,h(\iota([\mathrm{Jac}(C)]))\}\) acomodates Mumford and Vojta inequality to obtain an upper bound for large points in the form \(c_1(g)^{1+\rho}\) (see [\textit{P. Vojta}, Ann. Math. (2) 133, No. 3, 509--548 (1991; Zbl 0774.14019), Theorem 6.1] when \(\Gamma=\mathrm{Jac}(C)(F)\) or the works of \textit{S. David} and \textit{P. Philippon} [Comment. Math. Helv. 77, No. 4, 639--700 (2002; Zbl 1030.11026); IMRP, Int. Math. Res. Pap. 2007, Article ID rpm006, 113 p. (2007; Zbl 1163.11049)] and \textit{G. Rémond} [Invent. Math. 142, No. 3, 513--545 (2000; Zbl 0972.11054)]. The key point is to bound the number of small points. Previosly Rémond [loc. cit.] obtained explicit bounds on curves embedded in abelian varieties depending on a suitable notion of height of \(C\) an artifact of the lower bounds for the Néron-Tate height, after David-Philippon and Nakayame proved stronger bounds in a power of an elliptic curve providing evidence of Mazur question in [\textit{S. David} and \textit{P. Philippon}, IMRP, Int. Math. Res. Pap. 2007, Article ID rpm006, 113 p. (2007; Zbl 1163.11049); \textit{S. David} et al., CRM Ser. (N.S.) 4, 143--164 (2007; Zbl 1230.14031)]. A main innovation of the paper under review is to prove a lower bound for the Néron-Tate height that is sufficiently strong to eliminate the dependency on the height of \(C\). Already the authors in [\textit{V. Dimitrov} et al., Int. Math. Res. Not. 2021, No. 2, 1138--1159 (2021; Zbl 1487.14056)] applied the earlier height lower bound in [\textit{Z. Gao} and \textit{P. Habegger}, Ann. Math. (2) 189, No. 2, 527--604 (2019; Zbl 1432.11060)] to answer positively Mazur question in a one-parameter family of smooth curves. In the paper, following the words of the authors, they find positive constants \(d_1,d_2,d_3\) and \(d_4\) that depends of \(g\) and \(\iota\) but not on \(C\) such that if \(h(\iota([\mathrm{Jac}(C)]))\geq d_1\), then for all \(P\in C(\overline{\mathbb{Q}})\) we have this dichotomy: \begin{itemize} \item \(P\) lies in a subset of \(C(\overline{\mathbb{Q}})\) of cardinality at most \(d_2\), \item or \(\{Q\in C(\overline{\mathbb{Q}}): \hat{h}(Q-P)\leq h(\iota([\mathrm{Jac}(C)])/d_3)\}<d_4\) \end{itemize} i.e. the small points in \(C(\overline{\mathbb{Q}})\cap\Gamma\) lie in a set of uniformly bounded cardinality or are contained in \((1+cte\cdot d_3)^{\rho}\) balls in the Néron-Tate metric where each ball contains at most \(d_4\)-points, and from such result the authors answer affirmative the above Mazur questions. A key result of the paper is a generalization of [\textit{Z. Gao} and \textit{P. Habegger}, Ann. Math. (2) 189, No. 2, 527--604 (2019; Zbl 1432.11060); Th.1.4]. More concretely, let \(S\) be a regular irreducible quasi-projective variety over \(\mathbb{Q}\) that is Zariski open in an irreducible projective variety \(\overline{S}\subseteq \mathbb{P}^m_{\overline{\mathbb{Q}}}\) and \(\pi:\mathcal{A}\rightarrow S\) be an abelian scheme of relative dimension \(g\geq 1\) such that carries a principal polarization and has level \(N\)-structure for some integer \(N\geq 3\). Let \(X\) be a closed irreducible subvariety of \(\mathcal{A}\) over \(\overline{\mathbb{Q}}\) that dominates \(S\), thus under the assumption ``non-degenerate'' exist constants \(l_1>0\) and \(l_2\geq 0\) and a Zariski open dense subset \(U\) of \(X\) with \[ \hat{h}_{\mathcal{A}}(P)\geq l_1 h(\pi(P))-l_2;\; \forall P\in U(\overline{\mathbb{Q}}). \] Such result for \(dim(S)=1\) is the previous cited [loc. cit.] and degeneracy behavior of \(X\) is fully studied in [\textit{Z. Gao}, Compos. Math. 156, No. 12, 2469--2509 (2020; Zbl 1481.11063)]. The authors apply the above key result for a concrete \(X\) that we will sketch next to finish this review. Denote by \(\mathcal{A}_{g,N}\) the moduli space of principally polarized \(g\)-dimensional abelian varieties with level \(N\)-structure and \(X_{g,N}\) the fine moduli space of smooth curves of genus \(g\) whose Jacobian is equipped with level \(N\)-structure, the authors take as geometrically irreducible varieties such moduli spaces, i.e., choosing and fixing a \(N\)-th root of unity once and for all. Take \(\mathcal{E}_{g,N}\rightarrow {X}_{g,N}\) the universal curve and \(\mathfrak{A}_{g,N}\rightarrow\mathcal{A}_{g,N}\) the universal abelian variety, then denoting by \(T_M\) the \(M\)-th Faltings-Zhang morphism the authors take \(X=T_M(\underbrace{\mathcal{E}_{g,N}\times_{X_{g,N}}\ldots \times_{X_{g,N}}\mathcal{E}_{g,N}}_{{(M+1)\text{-copies}}})\) in \(\underbrace{\mathfrak{A}_{g,N}\times_{X_{g,N}}\ldots\times_{X_{g,N}}\mathfrak{A}_{g,N}}_{M\text{-copies}}\) for a suitable \(M\).