Tropicalization of classical moduli spaces (Q475401)

This paper collects information about classical moduli spaces and their projective geometry constructions, in the spirit of \textit{I. Dolgachev} and \textit{D. Ortland} [Point sets in projective spaces and theta functions. Paris: Société Mathématique de France; Centre National de la Recherche Scientifique (1988; Zbl 0685.14029)], in order to produce explicit tropicalizations of these spaces. The basic aim, perhaps, is to explore the interaction between tropical geometry, classical projective geometry, and moduli spaces, through a series of concrete, low-dimensional examples. Some very nice constructions are considered, and very nice figures included, as well as certain combinatorial information about these tropicalized moduli spaces. The main tool, indeed used throughout the paper, is that many of the classical moduli spaces are the closure of the image of the composition of a linear embedding followed a monomial map. Reasons this perspective are so fruitful in the present context are, arguably, two-fold: (1) the tropicalization of a linear embedding is a very well-understood combinatorial construction, and the tropicalization of a monomial map is also manageable as, for instance, tropicalization is functorial with respect to toric equivariant morphisms, and (2) this composition of maps allows the authors to separate out the relevant combinatorics underlying these classical moduli spaces which is pertinent to their tropicalization. Elaborating on the second point: many moduli spaces naturally arise as quotients by group actions, but group actions and their quotients are not well-understood/behaved tropically, except it seems in certain cases such as permutations and diagonal torus actions. The authors employ the symplectic combinatorics of finite vector spaces in their study in a crucial way, building on earlier work of Sam, and use this to witness some intriguing ``shadows'' of group actions and related geometry in the tropical setting, in the convenient framework of moduli spaces where everything can, if luck permits, be interpreted in a modular way. In addition, the moduli spaces studied in this paper are all projective varieties, there is no stack theory, so the authors are able to bypass some technical aspects of tropical geometry which have not been developed yet. They are able to do this by rigidifying the objects they study, for instance by focusing on curves with level structure instead of just moduli of curves.