Orbifold quasimap theory (Q889937)

The theory of quasi-maps to GIT quotient varieties has been developed in [\textit{I. Ciocan-Fontanine} et al., J. Geom. Phys. 75, 17--47 (2014; Zbl 1282.14022)]. The current paper establishes and studies the theory of quasi-stable maps to GIT quotients that are Deligne-Mumford stacks. Let \(W\) be an irreducible affine variety equipped with a reductive linear algebraic group \(G\)-action. One considers the stack of quasi-stable maps to the Artin stack \([W/G]\). In general, this stack is not well-behaved -- it is not of finite type, and has infinitesimal automorphisms. The theory of quasi-maps introduces a sequence of conditions, called \textit{\(\epsilon\)-stability} for every small enough positive rational number \(\epsilon\). Each \(\epsilon\)-stability selects an open Deligne-Mumford substack of the stack of quasi-stable maps to \([W/G]\). When \(\epsilon\) is sufficiently large, the open substack coincides with the moduli of twisted stable maps to the GIT quotient stack as in [\textit{W. Chen} and \textit{Y. Ruan}, Contemp. Math. 310, 25--85 (2002; Zbl 1091.53058)], and [\textit{D. Abramovich} and \textit{A. Vistoli}, J. Am. Math. Soc. 15, No. 1, 27--75 (2002; Zbl 0991.14007)]. As \(\epsilon\) decreases, one obtains a sequence of moduli stacks by contracting the rational tails of the stable maps to the GIT quotient stacks determined by the \(\epsilon\)-stability conditions. The authors proved that these open substacks are proper over the affine quotient of \(W\). Furthermore, when \(W\) has at worst lci singularities, and the semi-stable locus \(W^{ss}\) is smooth, such an open substack carries a perfect obstruction theory, hence defines a Gromov-Witten type invariants, called the quasi-map invariants. Following [\textit{I. Ciocan-Fontanine} and \textit{B. Kim}, Adv. Math. 225, No. 6, 3022--3051 (2010; Zbl 1203.14014); Algebr. Geom. 1, No. 4, 400--448 (2014; Zbl 1322.14083)], the authors further investigate the wall-crossings of quasi-map invariants and their applications to mirror symmetry by introducing the \(J^{\epsilon}\)-function, and a generalization of Givental's small \(I\)-function. To define the \(J^{\epsilon}\)-function, the authors introduce a variantion of the moduli stack of quasi-maps, called the graph space. The graph space parameterizes morphisms from pre-stable orbifold curves to \([W/G]\times \mathbb{P}^1\) such that the projections to \([W/G]\) are quasi-maps, and the projections to \(\mathbb{P}^1\) are of degree one. The \(J^{\epsilon}\)-function is then defined using virtual localization of [\textit{T. Graber} and \textit{R. Pandharipande}, Invent. Math. 135, No. 2, 487--518 (1999; Zbl 0953.14035)] when there is a torus action on \(W\) commuting with \(G\). Similar to the case of GIT quotient varieties, conjectural properties of \(J^{\epsilon}\)-function have been provided which relate the \(J^{\epsilon}\)-function to the Lagrangian cone of the Gromov-Witten theory of the GIT quotient stack. These properties are verified when the torus action has certain good properties. To study the \(I\)-function, the authors further introduce a variation of the graph space by allowing one extra stacky marking on the source curve, which is called the stacky loop spaces. Similar to the case of GIT quotient varieties, the relation between the \(I\)-function and the Lagrangian cone of the Gromov-Witten theory of the GIT quotient stack is established with the presence of good torus actions. As an application, the authors study the case when the GIT quotient stack is a toric Deligne-Mumford stack. An explicit calculation of the \(I\)-function is provided. This leads to another proof of the mirror theorem for complete intersections of convex hypersurfaces in toric Deligne-Mumford stacks [\textit{T. Coates}, \textit{A. Corti}, \textit{H. Iritani} and \textit{H.-H. Tseng}, ``Some applications of the mirror theorem for toric stacks'', \url{arXiv:1401.2611}].