The quantum tropical vertex (Q2004526)

The quantum tropical vertex is an extension of the results of \textit{P. Bousseau} [Invent. Math. 215, No. 1, 1--79 (2019; Zbl 07015696)] to the setting of the classical tropical vertex group introduced by M. Gross, R. Pandharipande and B. Siebert. This proves the coincidence of two previously introduced refinements for scattering diagrams and some higher genera log Gromov-Witten invariants of some log Calabi-Yau surfaces. Additionally to this interpretation the author proves an integrality statement, known as Ooguri-Vafa/open BPS, about the generating series of the invariants and provide a refined conjecture about a coarser integrality. Two dimensional scattering diagrams are combinatorial objects. They consist in a family of half-lines \((D)\) in the plane, and to each half-line \(D\) is attached an element \(\theta_D\) in some group. In the setting of the Gross-Pandharipande-Siebert, this group is the tropical vertex group. A scattering diagram is called balanced if for each loop \(\gamma\) in the plane, the product of the elements \(g_D\) for the half-lines met by \(\gamma\) is equal to the identity. Given an unbalanced scattering diagram, there is a way to \textit{balance} it by adding some half-lines. As proven by Gross-Pandharipande-Siebert, the group elements attached to the half-lines added to balance the scattering diagram are related to log Gromov-Witten invariants. This relation passes through the intermediate step of tropical invariants and thus uses a correspondence theorem, proven by G. Mikhalkin for the planar case, and T. Nishinou and B. Siebert for the higher dimensional setting. Two refinements of the previous setting have been provided. The first one is due to M. Kontsevich and Y. Soibelman. It is obtained by considering a natural \(q\)-deformation of the tropical vertex group used in the scattering diagrams. As in the classical case, this deformation is related to a tropical count of rational curves using the refined multiplicity of Block and Göttsche, as worked out by S.A. Filippini and J. Stoppa. The second refinement is provided by Gross-Pandharipande-Siebert as they noticed that the log Gromov-Witten invariants appearing in the scattering diagram setting have a natural generalization to higher genera by the insertion of a top dimensional \(\lambda\)-class, which is in fact the setting of \textit{P. Bousseau} [Invent. Math. 215, No. 1, 1--79 (2019; Zbl 07015696)]. The values of the newly considered invariants are encoded in a generating series using a new indeterminate \(\hbar\) whose power is \(2g-2\), with \(g\) the genus. The content of the first theorem is then to prove using the results of [loc. cit.] that both refinement agree, using the change of variable \(q=e^{i\hbar}\). Concretely, this means that the refined \(q\)-invariant and the change \(q=e^{i\hbar}\) give the generating series of log Gromov-Witten invariants with the insertion of a top dimensional \(\lambda\)-class. The second result of the paper is a combination of the first theorem and an enhancement of results of Kontsevich-Soibelman. It consists in some integrality statement about the log Gromov-Witten invariants, known as Ooguri-Vafa/open BPS. Briefly, the generating series of these invariants become some specific rational function in \(q\) after the change of variable \(q=e^{i\hbar}\).