The asymptotic Schottky problem (Q2379370)

Let \(M_g\) be the coarse moduli space of smooth complex projective curves of genus \(g\geq 2\) and denote by \(A_g\) the moduli space of principally polarized complex abelian varieties of dimension \(g\). Torelli's theorem states that the map \(J: M_g\to A_g\) assigning to each curve its Jacobian variety is injective. If \(J_g:= J(M_g)\) denotes the so-called Jacobian locus in \(A_g\), then the classical, very long-standing Schottky problem (posed by \textit{F. Schottky} in 1888; JFM 20.0488.02; 1888; JFM 20.0489.01) asks for characterizing \(J_g\) inside \(A_g\) in algebraic, analytic, or purely geometric terms, i.e., for distinguishing Jacobians among principally polarized abelian varieties of dimension \(g\). During the past century, several approaches to solving the Schottky problem have been proposed and studied [cf. \textit{O. Debarre}, The Schottky problem: an update, in: Current topics in complex algebraic geometry. New York, NY: Cambridge University Press. Math. Sci. Res. Inst. Publ. 28, 57--64 (1995; Zbl 0847.14015)], with particularly spectacular progress having been achieved since the early 1980s. Recently, motivated by some earlier work of \textit{P. Buser} and \textit{P. Sarnak} [Invent. Math. 117, No. 1, 27--56 (1994; Zbl 0814.14033)], B. Farb suggested to study the Schottky problem from the viewpoint of ``large scale geometry'' [cf.: \textit{B. Farb}, in: Problems on mapping class groups and related topics. Providence, RI: American Mathematical Society (AMS). Proceedings of Symposia in Pure Mathematics 74, 11--55 (2006; Zbl 1191.57015)]. This approach is based on the fact that the moduli space \(A_g\) admits a so-called ``asymptotic cone'' denoted by \(\text{Cone}_\infty(A_g)\), which is defined as a certain Gromov-Hausdorff limit of rescaled pointed spaces [cf. \textit{M. Gromov}, in: Progr. Math. 152. Boston, MA: Birkhäuser (1999; Zbl 0953.53002)], and Farb's ``coarse Schottky problem'' can be stated as follows: Describe, as a subset of a \(g\)-dimensional Euclidean cone, the subset of \(\text{Cone}_\infty(A_g)\) determined by the Jacobian locus \(J_g\) in \(A_g\). In the paper under review, the authors provide a solution to the coarse Schottky problem by proving (Theorem 1.1) that \(J_g\) is coarsely dense in \(A_g\). This implies that the subset of \(\text{Cone}_\infty(A_g)\) determined by the Jacobian locus \(J_g\) actually coincides with \(\text{Cone}_\infty(A_g)\). Their proof also shows that the Jacobian locus of hyperelliptic curves is coarsely dense in \(A_g\) as well (Theorem 1.2.). Finally the authors study the boundary of the Jacobian locus \(J_g\) in the Baily-Borel compactification \(\overline A^{BB}_g\) and in the Borel-Serre compactification \(\overline A^{BS}_g\) of \(A_g\). Their precise description of these boundaries (Theorem 1.3. and Corollary 1.4.) is supplemented by a result identifying the boundary of the topological closure \(J_g\) in \(A_g\) itself as a product space defined by reducible Jacobians of dimension \(g\). In the course of these subtle investigations, the authors (re-)prove and use the following extension property (Proposition 1.5.) of the Torelli map: The map \(J: M_g\to A_g\) extends to an algebraic map from the Deligne-Mumford compactification \(\overline M^{DM}_g\) of \(M_g\) to the Baily-Borel compactification \(\overline A^{BB}_g\) of \(A_g\).