The homomorphisms between scalar generalized Verma modules associated to maximal parabolic subalgebras (Q2493469)

Let \(\mathfrak g\) be a simple finite dimensional Lie algebra over \(\mathbb C\), fix a Borel subalgebra \(\mathfrak b=\mathfrak h + \mathfrak u\), and let \(\mathfrak p=\mathfrak l + \mathfrak n\) be a parabolic subalgebra containing \(\mathfrak b\), with Levi component \(\mathfrak l\) and nilpotent part \(\mathfrak n\). If \(E(\lambda)\) is a one-dimensional \(\mathfrak l\)-module with weight \(\lambda\), extend it to \(\mathfrak p\) by letting \(\mathfrak n\) act trivially, and consider the generalized Verma module \(M_{\mathfrak p}(\mu)=\mathcal U(\mathfrak g)\otimes_{\mathcal U(\mathfrak p)}E(\lambda)\), where \(\mu=\lambda+\rho\) with \(\rho\) the half sum of positive roots determined by \(\mathfrak b\). In this paper, the author considers the question of the existence of nontrivial \(\mathfrak g\)-module homomorphisms \(\phi:M_{\mathfrak p}(\mu)\longrightarrow M_{\mathfrak p}(\nu)\) for weights \(\mu\) and \(\nu\). The main result is a complete classification of such homomorphisms for the case where \(\mathfrak p\) is a maximal parabolic subalgebra. Then he indicates how such homomorphisms can be used to construct homomorphisms between generalized Verma modules induced from more general parabolics. The classification is quite specific, on a case by case basis for each Lie type. The argument uses a combination of Vogan's translation principle, Jantzen's irreducibility criterion, and Kazhdan-Lusztig theory.