Computing node polynomials for plane curves (Q2392982)

Counting the number of projective plane curves of a given degree and number of nodes through the appropriate number of generic points is a wonderfully classical problem receiving modern attention and remarkable -- and beautiful -- results through the application of state-of-the-art enumerative techniques. The landmark result in this subject, called the Göttsche conjecture (which is based on a conjecture by Di Francesco and Itzykson, and which is now a theorem by work of Fomin and Mikhalkin) is that this number depends polynomially on the degree, once the number of nodes is fixed, and once the degree is large enough. See the introduction and references in the current paper for exact citations. The Fomin-Mikhalkin technique is to apply a ``Correspondence Theorem'' of Mikhalkin, which says that instead of counting algebraic curves, one can count tropical curves. This translates the very hard algebro-geometric problem into a very hard combinatorial problem, but Fomin-Mikhalkin succeed in establishing polynomiality through a clever enumerative apparatus called a floor diagram, first introduced (in a slightly less refined way) by Mikhalkin and Brugallé. Fomin-Mikhalkin essentially describe their count of floor diagrams (and hence tropical curves, and hence algebraic curves) in terms of a discrete integral of a more basic ``building block'' enumerative gadget. Block, in the present paper, notes that one can actually evaluate these sums symbolically by applying quite classical machinery (one summation result in particular is from 1631!). This leads rather directly to an explicit algorithm for computing these enumerative polynomials. However, there are certain bottlenecks in the algorithm which impede an efficient computation, and Block nicely addresses these and refines the algorithm to obtain a more feasible version. This is then literally fed into a computer to compute these polynomials explicitly for parameters significantly larger than were previously known, thereby verifying more cases of several conjectures about their shape/coefficients in the process (see Block's intro for details), and the polynomials are nicely listed explicitly in an appendix. Block also proves some remarkable results about the polynomiality threshold of these nodal curve counts: he extends the known cases of a conjecture of Göttsche on this polynomiality threshold, disproves a conjecture on it by Di Francesco and Itzykson, and prove a new general bound on the threshold (the latter, it appears, is the only explicit result in the paper valid for all choices of parameter and not dependent on the output of an algorithm that necessarily can only be applied for fixed values of the degree/node parameters -- and Block's proof of it beautifully uses the machinery developed in this paper). This paper is thus entirely combinatorial and computational -- algebraic geometry is non-existent here as that has been factored out by the work of Mikhalkin, Brugallé-Mikhalkin, and Fomin-Mikhalkin, yet remarkably within this combinatorial/tropical approach Block has managed to prove results and extend our knowledge of plane algebraic curves beyond what has been known for hundreds of years.