Some evidence for the Coleman-Oort conjecture (Q2063002)

This paper is in the direction of the Coleman-Oort conjecture. Let \(\mathcal{M}_g\) (resp. \(\mathcal{A}_g\)) be the moduli space of smooth complex projective curves of genus \(g\) (resp. of principally polarized abelian varieties of dimension \(g\)) and let \(\tau: \mathcal{M}_g\to\mathcal{A}_g\) be the Torelli morphism which takes a curve to its jacobian. Let \(\mathcal{T}_g^\circ\) denote the image of \(\tau\) and \(\mathcal{T}_g\) its Zariski closure in \(\mathcal{A}_g\). The Coleman-Oort conjecture states that there are no positive dimensional Shimura subvarieties \(Z\) of \(\mathcal{A}_g\) that are generically contained \(\mathcal{T}_g^\circ\) for large \(g\), i.e., such that \(Z\subset\mathcal{T}_g\) and \(Z\cap\mathcal{T}_g^\circ\neq\emptyset\). Counterexamples are known for \(g\le 7\) and are constructed from families of (ramified finite) Galois coverings of curves satisfying a numerical condition (called \((*)\) in the paper under review) and can be classified (for example, see [\textit{P. Frediani} et al.; Int. Math. Res. Not. 2015, No. 20, 10595--10623 (2015; Zbl 1333.14023); Geom. Dedicata 181, 177--192 (2016; Zbl 1349.14104); Ann. Sc. Norm. Super. Pisa, Cl. Sci. (5) 22, No. 4, 1597--1619 (2021; Zbl 1494.14024)].) In this paper the authors classify all (positive dimensional) families of Galois coverings of curves satisfying the condition \((*)\) for \(g\le 100\). They show that such a family can only appear when \(g\le 7\), in which case the classification is known [loc. cit.]. The proof is Computer-Assisted: to enumerate all the families of Galois covers and then check condition \((*)\). This is a daunting task. In order to make the algorithm more efficient and economical, the authors use some theoretical arguments that reduce the input data to a great extent.