Character formulæ and GKRS multiplets in equivariant \(K\)-theory (Q1941771)

Let \(G\), \(H\) be compact connected Lie groups and \(X\) a compact \(G\)-space. Let \(P \to X\) be a \(G\)-equivariant principal \(H\)-bundle over \(X\) and \(H^{(\tau)}\) denote a central extension of \(H\) by \(U(1)\). Then the degree zero part of the \(\mathbb Z/2\mathbb Z\)-graded \(K\)-group \(K^*_G(X, \tau)\) is defined as the Grothendieck group of \(G\times H^{(\tau)}\)-equivariant complex vector bundles over \(P\) such that \(U(1) \subset H^{(\tau)}\) acts by scalar multiplication on the fibers. This group has a multiplication morphism of the form \(\mu : K^*_G(X, \tau_1)\times K^*_G(X, \tau_2) \to K^*_G(X, \tau_1+\tau_2)\) which can be defined in the usual way. This paper deals with establishing two main theorems in the case where \(H\) is a closed subgroup of \(G\) with the same rank as \(G\). Let \(\omega_M\) denote the pullback of the central extension \(Spin^c(n)\) of \(SO(n)\) via the tangent representation \(H \to SO(n)\) of \(M=G/H\). We consider the composition \(\mathcal{P}_X\) of \(\mu : K^*_H(X, \tau) \times K^*_H(X, \omega_M-\tau) \to K^*_H(X, \omega_M)\) and \(i_D : K^*_H(X, \omega_M) \to K^*_G(X)\) where \(i_D\) denotes an induction map defined using a twisted version of a Dirac operator in ordinary \(K\)-theory. Then the first main theorem is that if \(\pi_1(G)\) is torsion-free, then this bilinear paring \(\mathcal{P}_X\) is nonsingular. This theorem is a generalization of the result for the case of a maximal torus in \textit{M. Harada}'s and the authors' earlier paper [Math. Res. Lett. 17, No. 3, 507--527 (2010; Zbl 1234.19002)]. Let \(T\) be a common maximal torus in \(H \subset G\) and \(W^H=W_G/W_H\) where \(W_G\), \(W_H\) are the Weyl groups of \(G\), \(H\) respectively. From the definition we know that there is a map \(\partial_H : K^*_T(X,\omega_{G/T}) \to K^*_T(X, \omega_{G/T}-\omega_{H/T})=K^*_T(X, \omega_M)\). For \(w \in W^H\) and \(a \in K^*_T(X, \omega_{G/T})\) put \(a_w=\partial_H(w^{-1}(a)) \in K^*_T(X, \omega_M)\). Then the second main theorem can be stated as follows: if \(H \neq G\), then \(\Sigma_{w \in W^H}\text{det}(w)a_w\) vanishes when viewed as an element of \(K^*(X, \omega_M)\). This theorem also includes the results of \textit{B. Gross} et al. on multiplets of representations [Proc. Natl. Acad. Sci. USA 95, No. 15, 8441--8442 (1998; Zbl 0918.17002)] as a special case where \(X\) is a point.