GIT versus baily-borel compactification for quartic K3 surfaces

Abstract

Looijenga has introduced new compactifications of locally symmetric varieties that give a complete understanding of the period map from the GIT moduli space of plane sextics to the Baily-Borel compactification of the moduli space polarized K3’s of degree 2, and also of the period map of cubic fourfolds. On the other hand, the period map of the GIT moduli space of quartic surfaces is significantly more subtle. In our paper (Laza and O’Grady, Birational geometry of the moduli space of quartic K3 surfaces, 2016. ArXiv:1607.01324) we introduced a Hassett-Keel–Looijenga program for certain locally symmetric varieties of Type IV. As a consequence, we gave a complete conjectural decomposition into a product of elementary birational modifications of the period map for the GIT moduli spaces of quartic surfaces. The purpose of this note is to provide compelling evidence in favor of our program. Specifically, we propose a matching between the arithmetic strata in the period space and suitable strata of the GIT moduli spaces of quartic surfaces. We then partially verify that the proposed matching actually holds.