A Landau-Ginzburg/Calabi-Yau correspondence for the mirror quintic (Q332212)

A Landau-Ginzburg (LG) model is a pair \((W,G)\), often interpreted as a singularity, where \(W\) is a non-degenerate homogeneous polynomial on \(\mathbb{C}^N\) and \(G\) is a finite group of its automorphisms. In the early 1990s Vafa, Witten, etc., predicted existence of invariants of LG models that correspond to the Gromov-Witten invariants of a Calabi-Yau (CY) manifold. In 2013 Fan, Jarvis and Ruan defined the condidate invariants (now called the FJRW invariants) mathematically as intersection numbers on the moduli space of ``\(W\) curves''. Let \(W:=x_1^5+x_2^5+\dots +x_5^5\) and \(\jmath\) be the coordinatewise multiplication by the fifth primitive root of unity. For the Fermat quintic \(M:=\{W=0\}\) the LG/CY correspondence with \((W,\langle\jmath\rangle)\) follows from the recent work of \textit{A. Chiodo} and \textit{Y. Ruan} [Adv. Math. 227, No. 6, 2157--2188 (2011; Zbl 1245.14038)]. The story unfolds in two steps. First, let \(W_\psi:=W-\psi x_1\cdots x_5\) and define a family of mirror quintics \(\mathcal{W}_\psi:=\{W_\psi=0\}\subset\mathbb{P}^4/(\mathbb{Z}/5\mathbb{Z})^3\). By the Givental's mirror theorem, after a change of variables, components of the Givental's \(J\)-function of \(M\) give a basis of solutions to the Picard-Fuchs equations for \(\mathcal{W}_\psi\) around \(\psi=\infty\). Chiodo and Ruan showed that the \(J\)-function of the FJRW invariants of \((W,\langle\jmath\rangle)\) is similarly related to the Picard-Fuchs equations for \(\mathcal{W}_\psi\) around \(\psi=0\). The correspondence then follows via analytic continuation in \(\psi\) and a linear transformation. The authors prove a similar correspondence for the mirror quintic \(\mathcal{W}:=\mathcal{W}_0\). This is a first such example for a space which is not a complete intersection in a weighted projective space. The \(\psi=\infty\) part has already been proved by \textit{Y.-P. Lee} and \textit{M. Shoemaker} [Geom. Topol. 18, No. 3, 1437--1483 (2014; Zbl 1305.14025)]. Namely, after a change of variables, components of \(J^{\mathcal{W}}\) give solutions to the Picard-Fuchs equations of a holomorphic \((3,0)\) form on \(M_\psi:=\{W_\psi=0\}\subset\mathbb{P}^4\), the mirror family of \(\mathcal{W}\). The first derivatives of \(J^{\mathcal{W}}\) give solutions to the Picard-Fuchs equations of non-holomorphic families of 3-forms. The main result of the paper is that the FJRW \(J\)-function of \((W,(\mathbb{Z}/5\mathbb{Z})^4)\), and its first derivatives, after a change of variables give solutions to the Picard-Fuchs equations around \(\psi=0\). Analytic continuation combined with a linear symplectic transformation \(\mathbb{U}\) complete the correspondence, the first derivatives are needed to determine \(\mathbb{U}\) uniquely. In Givental's geometric formalism the correspondence can be rephrased as identifying analytic continuation of the ``small slices'' of Lagrangian cones determining the respective \(J\)-functions. It is conjectured that quantization of \(\mathbb{U}\) identifies the analytic continuation of the higher genus Gromov-Witten and FJRW theories. Analogous conjecture is still open even for the Fermat quintic.