On the piecewise quasipolynomiality of double tropical Welschinger invariants
Vincenzo Reda
TL;DR
The paper addresses the problem of counting real curves on toric surfaces via double tropical Welschinger invariants associated with $h$-transverse polygons, proving that these counts are piecewise quasipolynomial across chamber structures. It develops a robust framework based on real floor diagrams, a correspondence theorem to tropical curves, and weighted Ehrhart theory to convert weighted lattice sums into higher-dimensional unweighted lattice counts, enabling precise chamberwise descriptions. The main contributions include proving quasipolynomiality for the counting maps $G_{(\mathbf{d}^r;\mathbf{d}^l),\mathbf{c},g}^{n_1,n_2}$ (and its $\mathbf{c}$-dependent extension) with degree $g$, introducing $s$-real multiplicities and new combinatorial Welschinger-type numbers, and providing explicit examples to illustrate the piecewise structure. The results enhance understanding of tropical enumerative invariants, their chamber structure, and potential deformation behavior, with implications for real curve counts on a broad class of toric surfaces, beyond Hirzebruch surfaces.
Abstract
Ardila and Brugallé conjectured that double tropical Welschinger invariants of Hirzebruch surfaces are piecewise quasipolynomial. In this work, we prove the conjecture holds in full generality, i.e. for toric surfaces corresponding to h-transverse polygons. Furthermore, we define new combinatorial Welschinger-type numbers for h-transverse polygons and show that they are likewise piecewise quasipolynomial.
