Group-valued equivariant localization (Q1568821)

Let \(G\) be a compact Lie group with Lie algebra \({\mathfrak g}\) and let \(M\) be a compact manifold with smooth \(G\)-action. Let \(\beta\) be a \(G\)-equivariant differential form on \(M\) with values in the polynomial algebra over \({\mathfrak g}\). Then \(\int_M\beta\) is an invariant polynomial on \({\mathfrak g}\) and the main result of the paper is a localization formula \[ \left\langle \int_M \beta, e^{2\pi i\langle., \xi\rangle} \right\rangle =\sum_F\int_F {i^*_F (\beta, e^{2\pi i\langle.,\xi\rangle}) \over\text{Eul}(\nu_F,2\pi i\xi)}, \] where the right hand side is the Fourier transform of \(\int_M\beta\), which via the symbol map is viewed as a distribution on \({\mathfrak g}\), supported at 0. The left hand side of the localization formula consists of a sum ranging over the connected components \(F\) of the zero set of the vector field \(\xi_M\) generated by \(\xi\) and \(\text{Eul} (\nu_F,2\pi i\xi)\) denotes the equivariant Euler form of the normal bundle. There also is a \(G\)-valued version in which \(\xi\) is replaced by the character of a finite dimensional representation of \(G\). As an application this localization formula is used to describe intersection pairings on reduced spaces. For moduli spaces of flat connections on 2-manifolds this leads to formulas previously conjectured by \textit{E. Witten} [in J. Geom. Phys. 9, No. 4, 303-368 (1992; Zbl 0768.53042)]. The proof uses the properties of the group valued equivariant deRham theory given in [\textit{A. Alekseev} and \textit{E. Meinrenken}, Invent. Math. 139, No. 1, 135-172 (2000; Zbl 0945.57017)].