Intertwining operators for real reductive groups (Q805764): Difference between revisions

The authors prove a difference equation for intertwining operators between (generalized) principal series representations of a real reductive Lie group G (possessing a maximal compact subgroup K). Let P be a parabolic subgroup of G with Langlands decomposition \(P=\circ MAN\). Let \(\sigma\), \(H_{\sigma}\) be a Hilbert representation of \(M=\circ MA\), and let \(H^{\infty}_{\sigma}\) denote the space of \(C^{\infty}\) vectors. It is assumed that this space coincides with the space of \(M\cap K\)- C\({}^{\infty}\) vectors (the assumption is fulfilled when dim \(\sigma\) \(<\infty\) or \(\sigma\) irreducible unitary). Let \(I^{\infty}_{\sigma}\) denote the space of smooth functions f: \(K\to H^{\infty}_{\sigma}\) transforming on the left according to \(\sigma| M\cap K\). Let \(\nu \in a^*_ c\). Then via restriction to K, the space of \(C^{\infty}\)- vectors for the principal series representation \(Ind^ G_ P(\sigma \otimes (\nu +\rho)\otimes 1)\) (here \(\rho =\rho_ P)\) may be identified with \(I^{\infty}_{\sigma}\), and one obtains the parabolically induced representation as a representation \(\pi_{P,\sigma,\nu}\) on \(I^{\infty}_{\sigma}\) (`the compact picture of the principal series'). Let \(\bar P\) be the opposite parabolic, and \(J_{\bar P| P}(\nu)\) the standard intertwining operator, viewed as an operator from \(I^{\infty}_{\sigma}\) into itself, intertwining \(\pi_{P,\sigma,\nu}\) with \(\pi_{\bar P,\sigma,\nu}\). For Re \(\nu\) sufficiently P-dominant, this operator is defined by an absolutely convergent integral, and depends holomorphically on \(\nu\). The difference equation now takes the form: \[ b_{\sigma}(\nu)J_{\bar P| P}(\nu)=J_{\bar P| P}(\nu +4\rho)\circ \pi_{P,\sigma,\nu +4\rho}(D_{\sigma}(\nu))\text{ on } I^{\infty}_{\sigma}. \] Here \(b_{\sigma}\) and \(D_{\sigma}\) are polynomial maps from \(a^*_ c\) into C and \(U(g_ c)^ K\) respectively. This equation allows one to extend the intertwining operators meromorphically in \(\nu\) as continuous linear endomorphisms of \(I^{\infty}_{\sigma}\). In the existing literature this meromorphic extension result was only established for the case that P is minimal. On the K-finite level it was proved for general P by using the subrepresentation theorem [see \textit{A. W. Knapp} and \textit{E. M. Stein}, Invent. Math. 60, 9-84 (1980; Zbl 0454.22010)]. The authors claim that the present more general result is important for applications to the theory of automorphic forms. The paper gives another application of the difference equation, namely a generalization of Cohn's determinant formula [cf. \textit{L. Cohn}, Analytic theory of the Harish-Chandra C-function (Lect. Notes Math. 429, 1974; Zbl 0342.33026)] for the restriction of the intertwining operators to a K-isotypic component (for cuspidal P this restriction is essentially a c- function). This determinant formula implies temperedness of Harish- Chandra's \(\mu\)-function (which for cuspidal P corresponds to the Plancherel density).