\(K\)-theoretic Donaldson invariants via instanton counting (Q2390742)

The authors study the \(K\)-theoretic version of the Donaldson invariants from gauge theory, i.e., the holomorphic Euler characteristics of determinant line bundles on the moduli spaces of rank-\(2\) Gieseker semistable sheaves over an algebraic surface. For a smooth projective toric surface, these invariants and their wall-crossing formulas are determined in terms of the \(K\)-theoretic version of the Nekrasov partition function. An explicit generating function for the wall-crossing of these invariants is given by using elliptic functions and modular forms. Sect.~1 is devoted to background materials. Determinant line bundles, \(K\)-theoretic Donaldson invariants, Nekrasov partition function and the partition function with \(5D\) Chern-Simons term are recalled and defined. The \(K\)-theoretic Donaldson invariants for \(K3\)-surfaces are computed, and their relations with the Strange Duality are discussed. In Sect.~2, assuming that all the relevant walls are good, the authors express the wall-crossing components in terms of the holomorphic Euler characteristics of some virtual vector bundles over the Hilbert schemes \(X_2^{[n]}\) of \(n\) points on two copies of the smooth surface \(X\). In Sect.~3, the surface \(X\) is assumed to be toric. Results parallel to those in Sect.~2 hold for the equivariant version. Moreover, the equivariant wall-crossing components are expressed in terms of the \(K\)-theoretic Nekrasov partition function. In Sect.~4, after taking the nonequivariant limit, the authors obtain the wall-crossing components for a toric surface \(X\) in terms of modular forms and elliptic functions. Their main idea is to use the solution of the Nekrasov conjecture and its refinement. It turns out that these results hold for an arbitrary surface \(X\) as well. As an example, the Hilbert series of the determinant line bundles on the moduli spaces of rank-\(2\) Gieseker semistable sheaves over the projective plane \(X = \mathbb{P}^2\) are computed when the Chern classes are small. In Appendix A, the authors study Seiberg-Witten curves for the \(K\)-theoretic version with Chern-Simons terms. The corresponding results of Seiberg-Witten curves for the holomorphic version have been known. In particular, the authors show that the perturbation part of the Seiberg-Witten prepotential coincides with the genus-\(0\) part of the perturbation part introduced in Sect.~1, and that the Seiberg-Witten prepotential satisfies the contact term equation. These results complete the proof of Nekrasov's conjecture, started in an earlier work of the last two authors [Invent. Math. 162, No. 2, 313--355 (2005; Zbl 1100.14009); Transform. Groups 10, No. 3--4, 489--519 (2005; Zbl 1110.14015)], since the solution of the contact term equation is unique.