Invariants of spectral curves and intersection theory of moduli spaces of complex curves (Q2514184)

In the early 90's the works of \textit{B. Dubrovin} [``Integrable systems in topological field theory'', Nuclear Physics B 379 (1992)] and \textit{M. Kontsevich} and \textit{Yu. Manin} [Commun. Math. Phys. 164, No. 3, 525--562 (1994; Zbl 0853.14020)] pointed a striking relationship between quantum cohomology and certain integrable systems, through the so called WDVV-equation. Solutions of an equation of the family of WDVV-equations (which are integrable systems) turned out to be generating functions of Gromov-Witten invariants of certain symplectic manifolds. In later works (quoted in the paper) further similar relations between integrable systems and cohomological invariants were founded. The question emerged if to any integrable system corresponds a set of cohomological invariants of some variety. A conjecture proposed by \textit{V. Bouchard} et al. [Commun. Math. Phys. 287, No. 1, 117--178 (2009; Zbl 1178.81214)] is that, if the spectral curve of an integrable system is mirror symmetric to a toric Calabi-Yau 3-fold \(\mathfrak{X}\), then the solutions to this system are generating functions of the Gromov-Witten invariants of \(\mathfrak{X}\). \smallskip The present paper is a step towards the setting of Bouchard-Klemm-Mariño-Pasquetti's conjecture. The main result of the paper is the correspondence between an integrable system and some intersection numbers in a space \(\mathcal{M}_{g,n}^{\mathfrak{b}}\) of Riemann surfaces with `colored' marked points. As explained by the author, such correspondence relates the spectral curve of an integrable system, and the Gromov-Witten invariants of a finite set of points. The author is able to recover with his result several previously known cases of the conjecture. To fully settle BKMP-conjecture it would remain to prove that the Gromov-Witten theory of a toric CY 3-fold reduces to the Gromov-Witten theory of a finite set of points.