Counting hyperelliptic curves on an abelian surface with quasi-modular forms (Q481151)

Let \((A, L)\) be a polarized abelian surface with polarization of type \((1, h -1)\). Let \(N_{g,h}\) denote the number of hyperelliptic curves in the linear system \(|L|\) which are of geometric genus \(g\) and arithmetic genus \(h\), and define \[ F_g(u) = \sum_{h=g}^\infty N_{g,h}u^{h-1}. \] In the case \(g = 2\), all curves are hyperelliptic, and Göttsche proved that \(F_2(u)=E_2(u)+1/24\) where \(E_2\) is the Eisenstein series of weight 2. The paper under review, formulates a conjectural formula in terms of quasi modular forms for yet a refinement of the generating series above. This is defined by means of the number of Weierstrass points of the hyperelliptic curves that are lying over each point of \(A[2]\). The conjectural formula is based on the assumption of the validity of the Gromov-Witten crepant resolution conjecture for the resolution \(K\to A/\pm 1\) where \(K\) is the corresponding Kummer \(K3\) surface. To be able to use this conjecture, the following isomorphism of stacks of orbifold stable maps is proven: \[ \overline{M}([A/\pm 1];2g+2,\beta)\cong \overline{M}([\text{Sym}^2 A];2g+2,i_*\beta) \] where \(i:[A/\pm 1]\to [\text{Sym}^2 A]\) is induced from \(A\to A\times A\) sending \(a\mapsto (a,-a)\). The idea is then that counting hyperelliptic curves in \(A\) is equivalent to counting certain stacky rational curves in the orbifold \([A/ \pm1]\). Using the crepant resolution conjecture, this should be the same as counting certain rational curves in the smooth \(K3\) surface \(K\) which has been studied by Yau-Zaslow and Bryan-Leung. As an application of this formula, the paper under review provides a lower bound on the arithmetic genus of a hyperelliptic curve in \(A\). In particular it is proven that there are no smooth hyperelliptic curves on \(A\) of genus greater than \(5\).