Integral transforms and quantum correspondences (Q2006084)

Let \(w : \mathbb C^N \to \mathbb C\) be a quasi-homogeneous polynomial whose total degree \(d\) is equal to the sum of the weights of each variable. Let \(G \le SL_N (\mathbb C)\) be a diagonal subgroup of automorphisms of \(w\). In the article [\textit{Y.-P. Lee} et al., Ann. Sci. Éc. Norm. Supér. (4) 49, No. 6, 1403--1443 (2016; Zbl 1360.14133)] the commutativity of a diagram called LG/CY square is proven. The top row vertices of this square are \(GWT_0([\mathbb C^N /G])\), the genus zero Gromov-Witten theory of \([\mathbb C^N /G]\), and \(GWT_0(\mathrm{tot}(\mathcal O_{\mathbb P(G)}(-d)))\), the genus zero Gromov-Witten theory of a partial crepant resolution of \([\mathbb C^N /G]\). The bottom row vertices are \(FJRW_0(w, G)\), the genus zero FJRW theory of the Landau-Ginzburg model given by the pair \((w, G)\), and \(GWT_0(\mathcal Z)\), the genus zero Gromov-Witten theory of a hypersurface \(\mathcal Z\) defined as the vanishing locus of \(w\) in an appropriate finite quotient of weighted projective space \(\mathbb P(G)\). The arrows in the square are the crepant transformation conjecture, quantum Serre duality, the LG/CY correspondence, and the local GW/FJRW correspondence. The goal of the paper under review is to relate each of the above correspondences to an integral transform between appropriate derived categories, i.e. to lift the LG/CY square to the derived category to obtain a cube of relations. It is known that the crepant transformation conjecture (the top horizontal arrow of the LG/CY square) is compatible with a natural Fourier-Mukai transform. A similar result is known for the bottom horizontal arrow at least when \(G\) is cyclic. The paper under review shows that there are natural derived functors corresponding to both of the vertical arrows of LG/CY square as well, after restricting to subcategories of \(D([\mathbb C^N /G])\) and \(D(\mathrm{tot}(\mathcal O(-d)))\) with proper support. This requires a reformulation of both the local GW/FJRW correspondence and quantum Serre duality in terms of narrow quantum \(D\)-modules, which turns out to be a more natural way of describing these correspondences. Even though the corresponding square of derived functors does not commute in general the paper shows that the induced maps on \(K\)-theory commute.