The McKay correspondence for isolated singularities via Floer theory (Q6113403)

The authors reprove by novel symplectic methods the generalized McKay correspondence for isolated singularities \(\mathbb{C}^n/G\) with finite subgroups \(G\subset \mathrm{SL}(n,\mathbb{C})\). \textit{V. V. Batyrev}'s original proof in [J. Eur. Math. Soc. (JEMS) 1, No. 1, 5--33 (1999; Zbl 0943.14004)] used algebraic geometry and motivic integration. The correspondence relates the cohomology of a crepant resolution \(\pi:Y\to\mathbb{C}^n/G\) to the representation theory of \(G\). In particular, the Euler characteristic of \(Y\) is equal to the number of conjugacy classes in \(G\) and \(\dim H^{2k}(Y,\mathbb{F})\) (odd cohomology vanishes) is equal to the number of age \(k\) (under Reid's age grading) conjugacy classes, assuming that the characteristic of \(\mathbb{F}\) is \(0\) or coprime to all positive integers \(\leq|G|\). The proof is based on Floer theory. A Hamiltonian that is standard away from the singular fiber \(\pi^{-1}(0)\) is used to define Floer cohomology groups and their direct limit, the symplectic cohomology \(SH^*(Y)\). The positive complex \(SH^*_+(Y)\) is defined, roughly, as the quotient of \(SH^*(Y)\) by the Morse subcomplex of constant \(1\)-periodic orbits. This construction is usually done for exact convex symplectic manifolds, where Floer's action functional provides the necessary filtration on the symplectic chain complex. The authors construct a novel filtration that works to the same effect without the exactness condition. There is a long exact sequence computing \(H^*(Y)\) from \(SH^*(Y)\) and \(SH^*_+(Y)\), and it is proved that \(SH^*(Y)\) vanishes and \(SH^{*\,-1}(Y)\simeq H^*(Y)\) with the grading by the Conley-Zehnder index. The McKay correspondence then follows by the Morse-Bott spectral sequence argument that constructs orbits generating the cohomology classes based on eigenvalues of elements of conjugacy classes of \(G\). To get to the limit of the spectral sequence, the authors use \(S^1\)-equivariant theory as an intermediary. The authors announce an extension of their work to non-isolated singularities that was also supposed to provide a natural basis for \(H^*(Y)\) labeled by conjugacy classes of \(G\), a still open problem originally posed by \textit{M. Reid} [in: Séminaire Bourbaki. Volume 1999/2000. Exposés 865--879. Paris: Société Mathématique de France. 53--72, Exp. No. 867 (2002; Zbl 0996.14006)], but it does not seem to have appeared since. They also express hope that their approach can provide new insights into the ``crepant resolution conjecture''.