Equivariant cohomology and wall crossing formulas in Seiberg-Witten theory (Q1574681)

Seiberg-Witten invariants for a four manifold \(X\) are defined in terms of the algebraic topology of a moduli space of solutions to the (perturbed) Seiberg-Witten equations. Since the equations require a choice of certain geometric data on \(X\), the moduli spaces, and thus the invariants, can depend on these choices. When \(b_+^2(X)=1\), the invariants are affected in a non-trivial way by the choices. The possible choices lie in a parameter space which is divided by walls into chambers; variations within a chamber have no affect on the Seiberg-Witten invariants, but crossing a wall between two chambers results in a change. It is these changes which are described by the formulas in the title of this paper. To derive the formulas, a path between two points in the parameter space is used to construct a cobordism between the Seiberg-Witten moduli spaces corresponding to the end points. Topological differences between the moduli spaces, and thus differences in the Seiberg-Witten invariants, can be attributed to singularities in the cobordism. The authors show how to modify this construction in such a way that the singularities are replaced by the fixed point set of an \(S^1\)-action on a smooth manifold. The key is to replace the full Seiberg-Witten gauge group by a subgroup of based gauge transformations. Since the moduli spaces are quotients by the gauge group, this leads to enlarged moduli spaces, but with residual \(S^1\)-actions. It then follows almost immediately from the definitions of the Seiberg-Witten invariants that the wall-crossing formulae become special cases of localization formulae by \textit{J. Kalkman} [J. Reine Angew. Math. 458, 37-52 (1995; Zbl 0851.57029)] for equivariant cohomology classes. The main results of the paper use this approach to derive the wall crossing formula of \textit{T. J. Li} and \textit{A. Liu} [Math. Res. Lett. 2, No. 6, 797-810 (1995; Zbl 0871.57017)] for ordinary Seiberg-Witten invariants, and also the formulas of \textit{C. Okonek} and \textit{A. Teleman} [Int. J. Math. 7, No. 6, 811-832 (1996; Zbl 0959.57029)] for a generalization of the invariants.