Hyperelliptic curves in abelian varieties (Q1356320)

In this paper we prove the following main theorem: Let \(g\geq 3\), and let \(X\) be a generic abelian variety of dimension \(g\). Then \(X\) does not contain a hyperelliptic curve. To be more precise: Consider the moduli space \({\mathcal A}_g\) of polarized abelian varieties of dimension \(g\). Let \(K\) be a field, and let \({\mathcal A}\) be an irreducible component of \({\mathcal A}_g \otimes K\). Let \(\eta\) be the generic point of \({\mathcal A}\), and let \(X\) be the abelian variety over \(L:=K({\mathcal A})^a\) corresponding with \(\overline\eta \in {\mathcal A}(L)\) (here the superscript \(a\) represents taking the algebraic closure). Let \(C\subset X\) be an irreducible algebraic curve. If \(g\geq 3\), then the normalization \(C^\sim\) is not a hyperelliptic curve. This theorem in characteristic zero was proved by \textit{G. P. Pirola} [see Duke Math. J. 59, No. 3, 701-708 (1989; Zbl 0717.14021), section 2, theorem 2; see also: Abelian varieties, Proc. int. Conf., Egloffstein 1993, 237-249 (1995; Zbl 0837.14036), p. 238]. The proof by Pirola follows from the fact that hyperelliptic curves in an abelian variety in characteristic zero are ``rigid up to translation''. In positive characteristic this rigidity is no longer true, and we analyze the situation there [for surfaces, see \textit{C. Schoen}, J. Reine Angew. Math. 411, 196-204 (1990; Zbl 0702.14015), section 2]. These results by Pirola and by Schoen inspired us to give the proofs below. In Nieuw Arch. Wisk., IV. Ser. 13, No. 3, 427-434 (1995; Zbl 0859.14004), \textit{F. Oort} analyzed the proof by Pirola, and some details of that paper will be used and recorded here.