A genus-one FJRW invariant via two methods (Q2098203): Difference between revisions

FRJW theory, named after FAN, Jarvis, Ruan, and Witten, associates a cohomological field theory to each pair \((W,G)\) of a non-degenerate quasi-homogeneous polynomial \(W\) in \(n\)-variables and an Abelian group G of symmetries of \(W\). Given \(g,\ell\geq 0\) and a tuple of so-called narrow elements \(\gamma\in G^\ell\), a point of FRJW moduli space \(\overline{\mathcal{M}}_{g,\gamma}(G)\) (i.e., the moduli stack of \(G\)-spin \(\ell\)-pointed genus-\(g\) twisted nodal curves) is an \(\ell\)-marked genus \(g\) Riemann surface together with a tuple of \(n\) line bundles satisfying specific identities. Extending this definition, Chang-Li-Li-Liu introduced the moduli space with \(p\)-fields \(\overline{\mathcal{M}}_{g,\gamma}(G)^p\) whose elements include the additional datum of a choice of sections of the \(n\) line bundles. By construction, there is a forgetful map \(\overline{\mathcal{M}}_{g,\gamma}(G)^p_{g,\gamma}(G)\to \overline{\mathcal{M}}_{g,\ell}\) that can be used to pull back from/push forward to certain cohomology classes on the Deligne-Mumford space \(\overline{\mathcal{M}}_{g,\ell}\), and to stratify the moduli space. FRJW invariants are defined by integrating certain cohomology classes again the so-called Witten's top Chern class. The latter is a virtual fundamental class for \(\overline{\mathcal{M}}_{g,\gamma}(G)\). Via the so-called LG/CY correspondence, FJRW invariants are closely related to Gromov-Witten invariants. The polynomial equation \(W\) defines a complex projective variety \(X\) in some weighted projective space. When \(X\) is Calabi Yau, the LG/CY correspondence conjecture predicts that the FJRW theory of \((W,G)\) should agree in all genera with the Gromov-Witten theory of \(X\) (or more precisely \(X/G\)). The primary purpose of this paper is to calculate particular genus one 3-marked FJRW invariant of the Fermat cubic LG-pair \((W=x_1^3+x_2^3+x_3^3, G=\mathbb{Z}_3)\), where \(\mathbb{Z}_3\) acts by diagonal action of the 3-rd roots of unity on \(\mathbb{C}^3\). The calculation is done in two ways. One follows the original definition, and the other uses the Mixed Spin P-field moduli spaces and localization. The primary motivation for this calculation is the all genera LG/CY correspondence between the FRJW invariants of \((W=x_1^3+x_2^3+x_3^3, \mathbb{Z}_3)\) and GW invariants of the elliptic curve \(X=(x_1^3+x_2^3+x_3^3=0)\subset \mathbb{P}^2\). Proposition 2, borrowed from [\textit{J. Li}, ``Higher genus FJRW invariants of a Fermat cubic'', Preprint, \url{arXiv:2001.00343}], provides an explicit formula for Witten's top Chern class of the desired genus one 3-marked moduli space. Also, the forgetful map to \(\overline{\mathcal{M}}_{1,3}\) pushes forward certain calculations to the Deligne-Mumford space (e.g., see Corollary 1), where known formulas and computer programs (SAGE) can be used to do the calculations. Together, these result in the first proof of the main theorem in Section 2.5.1. The second proof, following [\textit{H.-L. Chang} et al., J. Differ. Geom. 120, No. 2, 251--306 (2022; Zbl 1490.14091)], uses mixed fields moduli space and localization calculations.