Refined curve counting on complex surfaces (Q463151)

For a smooth projective complex curve \(C\) of genus \(g\), the ``MacDonald formula'' asserts that NEWLINE\[NEWLINE\sum_{n=0}^{\infty}q^{n-g+1} \chi(C^{[n]}) = \left( \frac{q}{(1-q)^2} \right)^{1-g}, NEWLINE\]NEWLINE where \(C^{[n]}\) is the Hilbert scheme of \(n\) points on \(C\). This formula has a relative version as follows. Let \(\mathcal{C}\to B\) be a family of reduced planar curves of arithmetic genus \(g\), and let \(\mathcal{C}^{[n]}\to B\) be the relative Hilbert scheme of \(n\) points. Motivated by string theory, consider the series NEWLINE\[NEWLINE\sum_{n=0}^{\infty}q^{n+1-g} \chi(\mathcal{C}^{[n]}) =\sum_{i=0}^{\infty} n_{\mathcal{C}/B}^i \cdot \left( \frac{q}{(1-q)^2} \right)^{i+1-g}, NEWLINE\]NEWLINE where the coefficients \(n^i\) on the right-hand side can be viewed as the ``virtual'' number of curves of cogenus \(i\) in the family \(\mathcal{C}\to B\). In good cases the \(n^i\) carry actual enumerative meaning [\textit{V. Shende}, Compos. Math. 148, No. 2, 531--547 (2012; Zbl 1312.14015)], [\textit{M. Kool} et al., Geom. Topol. 15, No. 1, 397--406 (2011; Zbl 1210.14011)]. In the paper under review, the authors pose the following question: does replacing the topological Euler characteristic on the left-hand side by more sophisticated invariants have an enumerative counterpart on the right?NEWLINENEWLINEA number of cases centering around the above question are explored in the paper. First for a single curve, replacing the Euler characteristic by the class in the Grothendieck ring of varieties, the authors speculate an enumerative meaning for the corresponding coefficients appearing on the right and relate it to geometry over \(\mathbb R\) (Conjecture 3 and Theorem 4). Next, the authors turn to the case of linear systems of curves on surfaces. Refining the Euler characteristic to Hirzebruch's \(\chi_{-y}\) genus, the authors conjecture that in good cases all the coefficients on the right vanish when the cogenus exceeds the dimension of the linear system (Conjecture 5), and they verify the conjecture for surfaces having numerically trivial canonical class. Assuming the conjecture in full generality, the authors further focus on the last nonvanishing coefficient and propose a generating function to describe it (Conjecture 7). Note that a special case of this conjecture (Conjecture 8) for irreducibly polarized \(K3\) surfaces and abelian surfaces has been established in the authors' subsequent work [``The chi-y genera of relative Hilbert schemes for linear systems on abelian and \(K3\) surfaces, \url{arXiv:1307.4316}]. Finally in Conjecture 12 the authors study a refinement of the Caporaso-Harris recursion for Severi degrees of plane curves [\textit{L. Caporaso} and \textit{J. Harris}, Invent. Math. 131, No. 2, 345--392 (1998; Zbl 0934.14040)], relating it to real enumerative geometry and to tropical geometry (Conjecture 13).