Mirror symmetry: from categories to curve counts

Abstract

We work in the setting of Calabi-Yau mirror symmetry. We establish conditions under which Kontsevich's homological mirror symmetry (which relates the derived Fukaya category to the derived category of coherent sheaves on the mirror) implies Hodge-theoretic mirror symmetry (which relates genus-zero Gromov-Witten invariants to period integrals on the mirror), following the work of Barannikov, Kontsevich and others. As an application, we explain in detail how to prove the classical mirror symmetry prediction for the number of rational curves in each degree on the quintic threefold, via the third-named author's proof of homological mirror symmetry in that case; we also explain how to determine the mirror map in that result, and also how to determine the holomorphic volume form on the mirror that corresponds to the canonical Calabi-Yau structure on the Fukaya category. The crucial tool is the `cyclic open-closed map' from the cyclic homology of the Fukaya category to quantum cohomology, defined by the first-named author in [Gan]. We give precise statements of the important properties of the cyclic open-closed map: it is a homomorphism of variations of semi-infinite Hodge structures; it respects polarizations; and it is an isomorphism when the Fukaya category is non-degenerate (i.e., when the open-closed map hits the unit in quantum cohomology). The main results are contingent on works-in-preparation [PS,GPS] on the symplectic side, which establish the important properties of the cyclic open-closed map in the setting of the `relative Fukaya category'; and they are also contingent on a conjecture on the algebraic geometry side, which says that the cyclic formality map respects certain algebraic structures.