On Frobenius-destabilized rank-2 vector bundles over curves (Q2427584)

Let \(X\) be a smooth projective curve of genus \(g\geq 2\) defined over an algebraically closed field \(k\) of positive characteristic \(p\). In the first section, the authors prove stability of Frobenius push-forwards of line bundles on \(X\). Recently this problem received quite a lot of attention (see \textit{X. Sun}'s paper [Invent. Math. 173, No. 2, 427--447 (2008; Zbl 1195.14060)] for similar results in higher dimension and for history of the problem). Then in Section 2 they use Hirschowitz's generalization of the Nagata-Segre theorem to prove that on \(X\) there exist stable rank \(2\) vector bundles with trivial determinant, whose Frobenius pull-backs are unstable. The first example of such a bundle was constructed by \textit{D. Gieseker} [Ann. Sci. Éc. Norm. Supér. (4) 6, 95--101 (1973; Zbl 0281.14013)] and it was used by \textit{Y. Laszlo} and \textit{C. Pauly} [Adv. Math. 185, No. 2, 246--269 (2004; Zbl 1055.14038)] to give another proof of the fact that such examples exist on every curve of genus at least \(2\). The main result of the paper under review is computation of the number of Frobenius destabilized rank \(2\) stable vector bundles on an ordinary curve of genus \(2\). This is used to compute the degree of the map induced by the Frobenius morphism on the moduli space of rank \(2\) stable vector bundles with trivial determinant. Independently, \textit{B. Osserman} [Trans. Am. Math. Soc. 360, No. 1, 273--305 (2008; Zbl 1140.14030)] computed the number of Frobenius destabilized rank \(2\) stable vector bundles with trivial determinant for a general genus \(2\) curve. He also computed the degree of the map induced by the Frobenius morphism on the moduli space of rank \(2\) stable vector bundles with trivial determinant on a general genus \(2\) curve [see, Math. Ann. 336, No. 4, 963--986 (2006; Zbl 1111.14031]). The methods used by the authors are very different to those used by \textit{B. Osserman} and more complicated, but their result is much more precise.