On the bielliptic and bihyperelliptic loci (Q2225029)

Let \(\mathcal {A}_g\) be the moduli space of \(g\)-dimensional principally polarized Abelian varieties and \(j: \mathcal {M}_g\to \mathcal {A}_g\) the Torelli map. \(\mathcal{A}_g\) has a metric, the Siegel metric, and a conjecture of Coleman and Oort is related to totally geodesic subvarieties for the Siegel metric. \(\mathcal {M}_g\) get the metric and the second Gaussian map may be used to detect its totally geodesic subvarieties [\textit{F. Colombo} and \textit{P. Frediani}, Mich. Math. J. 58, No. 3, 745--758 (2009; Zbl 1191.14030); Trans. Am. Math. Soc. 362, No. 3, 1231--1246 (2010; Zbl 1196.14025); \textit{E. Colombo} et al., Int. J. Math. 26, No. 1, Article ID 15500005, 21 p. (2015; Zbl 1312.14076)]. In this paper the authors prove that for \(g\ge 4\) the locus of bielliptic curves is not totally geodesic in \(\mathcal {A}_f\), contrary to the case \(g=3\), and that it is not totally geodesic the locus \(\mathcal {BH}_{g,g'}\) of double coverings of genus \(g\) of hyperelliptic curves of genus \(g'\) (double hyperelliptic curves) for \(g \ge 3g'\ge 6\).