Dirac cohomology and Dirac induction (Q424871)

Let \(G\) be a connected semisimple Lie group with maximal compact subgroup \(K\). Let \(\mathfrak{g}=\mathfrak{k}\oplus\mathfrak{p}\) be the corresponding Cartan decomposition of the complexified Lie algebra of \(G\). If \(M\) is a Harish-Chandra module for the pair \((\mathfrak{g},K)\), one can consider the action of Parthasarathy's Dirac operator \(D\) on \(M\otimes S\), where \(S\) is a spin module for the Clifford algebra of \(\mathfrak{p}\). In particular, Vogan defined the Dirac cohomology of \(M\) as the quotient of the kernel of \(D\) by the intersection of the image of \(D\) and the kernel of \(D\). The paper under review assumes that \(G\) and \(K\) have equal rank, so that \(G\) has discrete series representations. Then it is shown that the Dirac cohomology of \(L^2(G)\) gives a natural one-to-one correspondence between discrete series representations of \(G\) and irreducible representations of \(K\), with infinitesimal characters that are regular with respect to a root system of \(G\). This correspondence is closely related to the Dirac induction for the reduced group \(C^*\)-algebra of \(G\). The second part of the paper uses Kostant's cubic Dirac operator to define Dirac induction for a pair \((G,H)\), where \(G\) is a compact Lie group and \(H\) is a closed subgroup of \(G\). By definition, Dirac induction is a map from the character ring of the spin double cover of \(H\) into the character ring of \(G\). The definition is similar to an earlier definition of formal induction by Bott, but it is simpler and easier to compute.