Enumeration of curves via floor diagrams (Q2382298)

The authors give a process to compute the number of curves of degree \(d\) and genus \(g\) in real and complex projective spaces of dimension \(n\) (with \(n=2\) and arbitrary \(g\) or \(g=0\) and arbitrary \(n\)) passing through a given configuration \(\mathcal{P}\) of projective-linear subspaces of dimension \(l_j\). In the complex case these numbers do not depend on \(\mathcal{P}\). They are known as the Gromov-Witten numbers \(N_{d,\, g}^{(n)}(l_0,\ldots ,l_{n-2})\), and were computed by \textit{M. Kontsevich} [Commun. Math. Phys. 164, 525--562 (1994; Zbl 0853.14020)], and by \textit{L. Caporaso} and \textit{J. Harris} [Invent. Math. 131, 345--392 (1998; Zbl 0934.14040)]. However in the real case not so much is known. These numbers depend on the choice of the configuration \(\mathcal{P}\). \textit{J. Y. Welschinger} [Invent. Math 162, 195--234 (2005; Zbl 1082.14052); Duke Math. J. 127, 89--121 (2005; Zbl 1084.14056)] proved that for \(n=2, 3\) and \(g=0\) there is a compatible choice of signs \(\pm 1\) such that the number of the corresponding real curves counted with the sign, don't depend on the configuration \(\mathcal{P}\). These numbers are known as the Welschinger numbers and denoted by \(W_d^{(n)}\). The algorithm for computation of the numbers \(W_d^{(2)}\), as given by \textit{G. Mikhalkin} [J. Am. Math. Soc. 18, 313--377 (2005; Zbl 1092.14068)], \textit{I. Itenberg, V. Kharlamov} and \textit{E. Shustin} [Int. Math. Res. Not. 49, 2639--2653 (2003; Zbl 1083.14523)], produced several estimates for these numbers. In this note the authors introduce a new tool in tropical geometry called \textit{floor diagrams}, and give a technique for a simultaneous computation of \(N_{d,\, g}^{(n)}(l_0,\ldots ,l_{n-2})\) and \(W_d^{(n)}\) for \(n=2, 3\).