The theory of \(N\)-mixed-spin-\(P\) fields (Q2024746): Difference between revisions

This work is the first of a series of three papers building on previous work [\textit{H.-L. Chang} et al., Camb. J. Math. 7, No. 3, 319--364 (2019; Zbl 1430.14104); ``An effective theory of GW and FJRW invariants of quintics Calabi-Yau manifolds'', \url{arXiv:1603.06184}, to appear in J. Differ. Geom.] on mixed-spin-P (MSP) fields by three of the authors (Huai-Liang Chang, Jun Li, Wei-Ping Li) along with Chiu-Chu Melissa Liu. Recall that expanding the B-model topological string free energy of the mirror quintic in the vicinity of the large complex structure limit yields the Gromov-Witten invariants of the quintic while expanding it in the vicinity of the orbifold point yields the Fan-Jarvis-Ruan-Witten (FJRW) invariants of the quintic. Both these limits can regarded as Landau-Ginzburg models associated to two different GIT quotients of \([\mathbb C^6/\mathbb G_m]\). So, by constructing an intermediary master space and considering a suitable Landau-Ginzburg model on this master space, one can hope to create a bridge between Gromov-Witten and FJRW theory. This is the idea underlying the notion of MSP fields. Concretely, an MSP field associated to the Fermat quintic consists of the following data: \begin{itemize} \item a twisted curve \(\Sigma\subset \mathscr C\) (i.e.\ a nodal curve \(\Sigma\) in a family \(\mathscr C\) with orbifold structure at its nodes, roughly speaking) of fixed genus \(g\) and \(\ell\) marked points away from the nodes; \item two invertible sheaves (i.e. line bundles, roughly speaking) \(\mathscr L\) and \(\mathscr N\) defined on the complement of the marked points, with \(\mathscr L\) having prescribed monodromies \(\gamma=(\gamma_i)_{i=1}^{\ell}\) around the marked points and the pair \((\mathscr L\otimes \mathscr N, \mathscr N)\) having fixed degrees \(\boldsymbol d = (d_0,d_\infty)\); \item a collection of various sections of the sheaves involved: five sections \((\varphi_j)_{j=1}^5\) of \(\mathscr L\), a section \(\rho\) of the fifth power of the dual sheaf \(\mathscr L^\vee\) twisted with the log-canonical sheaf of the curve, and a section \(\mu\) of the product \(\mathscr L\otimes \mathscr N\). \end{itemize} In this paper, the above definition is upgraded to that of an N-mixed-spin-P (NMSP) field. This has an additional positive integer parameter \(N\) and replaces the single section \(\mu\) of \(\mathscr L\otimes \mathscr N\) in the definition of an MSP field by a tuple of \(N\) sections \((\mu_k)_{k=1}^N\). The point of this replacement is that the resulting \((\mathbb C^\times)^N\)-equivariant theory is easier to work with than the \(\mathbb C^\times\)-equivariant theory in the MSP case. The localisation formula, for instance, simplifies greatly in the large \(N\) limit. As a consequence of this simplication, the authors are able to prove several structural results concerning the Gromov-Witten potential function of quintic Calabi-Yau threefolds in subsequent instalments of this series. The main result of the current paper is an NMSP analogue of previous results regarding MSP fields, namely that the moduli space of NMSP fields of fixed numerical data \((g,\gamma,\boldsymbol d)\) associated to the Fermat quintic is a separated Deligne-Mumford stack that is locally of finite type and has a \((\mathbb C^\times)^N\)-equivariant perfect obstruction theory. In particular, the moduli space has a \((\mathbb C^\times)^N\)-equivariant virtual fundamental class localised on a cosection of the obstruction sheaf which can be used to obtain integral invariants. The authors furthermore show that the generating functions of these NMSP invariants admit a certain polynomial structure and a decomposition into a part involving Gromov-Witten invariants and a part involving FJRW invariants.