Remarks on the nef cone on symmetric products of curves (Q836906)

Consider \(C\) a smooth irreducible complex projective curve of genus \(g\) and assume that \(C\) has very general moduli (the point \([C]\) corresponding to \(C\) in the moduli space \({\mathcal M}_g\) is outside a countable collection of proper subvarieties of \({\mathcal M}_g\)). The paper under review studies the nef cone of numerically effective real divisors of the second symmetric product \(C^{(2)}\) of \(C\). There are two natural divisors on \(C^{(2)}\): the divisor \(X_p=\{p+q:q \in C\}\) associated to a point \(p \in C\) and the diagonal \(\Delta=\{q+q:q \in C\}\). Their clases in the Néron-Severi group are denoted respectively \(x\) (since it does not depend on \(p\)) and \(\delta\) and span a plane \(\Pi\) in the real vector space associated to the Néron-Severi group. The goal is to describe the cone \(N\) obtained by intersecting \(\Pi\) with the nef cone of \(C^{(2)}\). It is clear that the diagonal determines one boundary of \(N\) and one would like to understand the other boundary which is determined by the real number \(\tau(C)\) which is the infimum of the positive \(t\) such that \((t+1)x-{\delta \over 2}\) is ample. Hence the description of \(N\) becomes the computation of \(\tau(C)\). For \(g \leq 3\) the cone \(N\) is completely understood (see references in the paper) and for \(g \geq 4\) there is a conjecture saying that \(\tau(C)=\sqrt{g}\), which in particular means that the cone is as large as possible. The paper under review provides upper bounds for \(\tau(C)\) for very general curves of low genus. In fact, see Theorem 1, for \(5 \leq g \leq 8\) it follows that \(\tau(C) \leq \tau_g\), where \(\tau_5=9/4\), \(\tau_6=37/15\), \(\tau_7=189/71\) and \(\tau_8=54/19\), improving the previously known bounds (essentially \(\tau_g \leq g/[\sqrt{g}]\)).