Gromov-Witten invariants of the Hilbert schemes of points of a \(K3\) surface (Q1676772)

Let \(S\) be a smooth projective \(K3\) surface, and let \(\text{Hilb}^d(S)\) be the Hilbert scheme of \(d\) points on \(S\). For \(\beta \in H^2(S;\mathbb Z)\), and the point class \(p\) in \(S\) the classes \(C(\beta):= \mathfrak p_{-1}(\beta)\mathfrak p_{-1}(p)^{d-1} 1\) and \(A:=\mathfrak p_{-2}(p)\mathfrak p_{-1}(p)^{d-2}1\) (when \(d\geq 2\)), defined by means of Nakajima creation operators, span \(H_2(\text{Hilb}^d(S);\mathbb Z)\). Now let \(\pi:S\to \mathbb P^1\) be an elliptically fibered \(K3\) surface and let \(\pi^{[d]}:\text{Hilb}^d(S)\to \text{Hilb}^d(\mathbb P^1)=\mathbb P^d\) be the induced Lagrangian fibration with generic fiber a smooth Lagrangian torus. Let \(L_z\) be the fiber of \(\pi^{[d]}\) over a point \(z\in \mathbb P^d\). Let \(\beta_h\) be a primitive effective curve class on \(S\) having intersection number 1 with the fiber class of \(\pi\) and self-intersection number \(2h-2\). For \(z_1, z_2\in \mathbb P^d\) and integer \(k\), let \(N_{d,h,k}:=\langle L_{z_1}, L_{z_2}\rangle_{C(\beta_h)+kA}\) be the 2-pointed Gromov-Witten invariant of \(\text{Hilb}^d(S)\) in class \(C(\beta_h)+kA\). One of the main results of the paper under review proves that \[ \sum_{h\geq 0}\sum_{k\in \mathbb Z}N_{d,h,k}y^k q^{h-1}=\frac{F(z,\tau)^{2d-2}}{\Delta(\tau)} \] where \(F(z,\tau)\) is the Jacobi theta function considered as a formal power series in the variables \(y=-e^{2\pi i z}\) and \(q=e^{2\pi i \tau}\) with \(|q|<1\). This formula specializes the famous Yau-Zaslow formula when \(d=1\). The paper under review also computes the genus 0 Gromov-Witten invariants of \(\text{Hilb}^d(S)\) for a few other natural incidence conditions in terms of Jacobi forms. In the case that \(\pi\) admits a section \(B\) and \(\beta_h=B+hF\) for the fiber class \(F\) the paper under review proves an explicit formula for the generating series of genus 0 arbitrary 2-pointed GW invariants of \(\text{Hilb}^2(S)\) in terms of a quasi-Jacobi form of index 1 and weight \(\leq\) 6. It is conjectured in this paper that the same formula holds for all \(d\) and this time the quasi-Jacobi form is of index \(d- 1\) and weight \(\leq 2 + 2d\). A couple of interesting applications are presented at the end. First, is a proof of a conjecture fir \(d=2\) relating genus 1 invariants of \(\text{Hilb}^d(S)\) to the Igusa cusp form. Second, a conjectural formula is given for the number of hyperelliptic curves on a \(K3\) surface passing through 2 general points.