Gromov-Witten Invariants of Bielliptic Surfaces
Thomas Blomme
TL;DR
This work develops a concrete floor/pearl-diagram framework to compute Gromov-Witten invariants of bielliptic surfaces, leveraging Li degeneration and a decomposition into relative invariants of $ ilde E imesP^1$ to enumerate curves via pearl diagrams. It proves that generating series for enumerative GW-invariants are quasi-modular, with separate modular behavior in the elliptic directions, governed by $SL_2(Z)$ and congruence subgroups Γ_1(n). The paper further refines these invariants by inserting λ-classes, derives refined multiplicities for diagrams, and demonstrates how these refinements preserve modularity properties through multi-variable quasi-Jacobi form theory. Explicit genus-by-genus computations illustrate the structure, including the appearance of Eisenstein series and their twists, and the results point to broad future avenues in descendant and torsion-related refinements. Overall, the approach provides a rigorous, modularity-driven toolkit for exact GW-counts on bielliptic surfaces with potential extensions to other elliptic fibrations and refined enumerative theories.
Abstract
Bielliptic surfaces appear as quotient of a product of two elliptic curves and were classified by Bagnera-Franchis. We give a concrete way of computing their GW-invariants with point insertions using a floor diagram algorithm. Using the latter, we are able to prove the quasi-modularity of their generating series by relating them to generating series of graphs for which we also prove quasi-modularity results. We propose a refinement of these invariants by inserting a λ-class in the considered GW-invariants.
