Integral Chow rings of modular compactifications of $\mathcal{M}_{1,n\leq 6}$
Luca Battistella, Andrea Di Lorenzo
TL;DR
The paper delivers a complete integral Chow-ring computation for all modular compactifications of $\\mathcal{M}_{1,n}$ with $n\\le 6$ by exploiting the stack $\\mathcal{G}_{1,n}$ of log-canonically polarised Gorenstein genus-one curves and a two-tier stratification by tails and elliptic singularities. Through Vistoli-style patching and an explicit calculation of top Chern classes on strata, the authors produce presentations of $A^*(\\overline{\\mathcal{M}}_{1,n}(Q))$ generated by the Hodge class $\\lambda$, boundary divisors $\\tau_B$ (and $\\nu$ in codimension 2 for $n=6$), subject to Keel-type and bespoke relations. They establish Chow‑Künneth generation and cycle-map isomorphisms for all such compactifications, and they analyze Getzler’s relation integrality, showing a corrected integral form for $n=4$ while proving the integral Getzler relation fails in general. The results yield both explicit polynomial expressions for fundamental loci and a uniform framework for computing cohomology and point counts across a large family of modular compactifications, with potential applications in intersection theory and enumerative geometry of genus-one moduli spaces.
Abstract
For $n\leq 6$, we compute the integral Chow ring of every modular compactification of $\mathcal{M}_{1,n}$ parametrising only Gorenstein curves with smooth, distinct markings. These include the Deligne--Mumford, Schubert, and Smyth compactifications, and many more. They can all be excised from the stack of log-canonically polarised Gorenstein curves. The Chow ring of the latter admits a simple, combinatorial description, which we compute by patching along a natural stratification by core level. We further deduce that all these modular compactifications satisfy the Chow-Künneth generation property, that the cycle class map is an isomorphism, and for $n=4$ we study whether the Getzler's relation hold integrally.
