Whittaker modules for classical Lie superalgebras (Q2244089)

This article studies the category of Whittaker modules for classical Lie superalgebras. A Lie superalgebra \(\mathfrak{g}=\mathfrak{g}_{\overline{0}}\oplus\mathfrak{g}_{\overline{1}}\) is called classical if \(\mathfrak{g}_{\overline{0}}\) is reductive and acts semisimply on \(\mathfrak{g}\) under the adjoint action. For such a Lie superalgebra, one has a Cartan subalgebra, along with triangular decompositions \(\mathfrak{g}=\mathfrak{n}^-\oplus\mathfrak{h}\oplus\mathfrak{n}\), which unlike in the purely even case, are not all conjugate. However, up to conjugacy one may assume that \(\mathfrak{n}_{\overline{0}}\) and \(\mathfrak{n}^-_{\overline{0}}\) are fixed, which the author does. Further, for technical reasons, the author assumes that \(\mathfrak{h}=\mathfrak{h}_{\overline{0}}\). The category \(\widetilde{\mathcal{N}}\) of Whittaker modules consists of finitely generated \(\mathfrak{g}\)-modules which are finite over \(\mathfrak{n}_{\overline{0}}\) and \(Z(\mathcal{U}\mathfrak{g}_{\overline{0}})\). This in particular implies that the modules are finite over \(\mathfrak{n}\), as \(\mathcal{U}\mathfrak{n}\) is itself finitely generated over \(\mathcal{U}\mathfrak{n}_{\overline{0}}\). Further, these conditions have the pleasant property that modules from \(\widetilde{\mathcal{N}}\) lie in the category of Whittaker modules for \(\mathfrak{g}_{\overline{0}}\) upon restriction. The study of the category of Whittaker modules for reductive Lie algebras is well-developed (see [\textit{E. McDowell}, Proc. Am. Math. Soc. 118, No. 2, 349--354 (1993; Zbl 0774.17009); \textit{D. Miličić} and \textit{W. Soergel}, Comment. Math. Helv. 72, No. 4, 503--520 (1997; Zbl 0956.17004)]). Simple modules are classified by pairs \((\lambda,\zeta)\) where \(\lambda\in\mathfrak{h}^*\) and \(\zeta:\mathfrak{n}_{\overline{0}}\to\mathbb{C}\) is a Lie algebra homomorphism, modulo a certain relation. Further, there are standard Whittaker modules indexed by the same set which play an analogous role to Verma modules in category \(\mathcal{O}\). The composition factors of standard Whittaker modules may be computed in terms of Kazhdan-Lusztig combinatorics. One goal of this paper is to extend the above results to the super setting. This study was initiated in [\textit{I. Bagci} et al., Commun. Algebra 42, No. 11, 4932--4947 (2014; Zbl 1364.17008)], where the case of type I superalgebras was the main focus. A type I superalgebra \(\mathfrak{g}\) is one that admits a compatible \(\mathbb{Z}\)-grading \(\mathfrak{g}=\mathfrak{g}_{-1}\oplus\mathfrak{g}_0\oplus\mathfrak{g}_1\). In particular for type I superalgebras one has the Kac-induction functor \(\operatorname{Ind}_{\mathfrak{g}_0\oplus\mathfrak{g}_{1}}^{\mathfrak{g}}(-)\), which via inflation from \(\mathfrak{g}_0\) to \(\mathfrak{g}_0\oplus\mathfrak{g}_1\) defines a powerful functor from \(\mathfrak{g}_0\)-modules to \(\mathfrak{g}\)-modules. A main technical challenge for the study of Whittaker modules in the super setting is that \(\mathfrak{n}\) may have finite-dimensional irreducibles that are not one-dimensional; this happens whenever \([\mathfrak{n}_{\overline{1}},\mathfrak{n}_{\overline{1}}]\) does not lie in \([\mathfrak{n}_{\overline{0}},\mathfrak{n}_{\overline{0}}]\). For type I superalgebras, one may skirt around this issue by considering triangular decompositions with \(\mathfrak{n}=\mathfrak{n}_{\overline{0}}\oplus\mathfrak{g}_1\); indeed here \(\mathfrak{n}_{\overline{1}}=\mathfrak{g}_1\) is supercommutative. In this paper, the author avoids the issue for an arbitrary Lie superalgebra by restricting consideration to the irreducibles of \(\mathfrak{n}\) that are one-dimensional, and allowing \(\mathfrak{n}\) to vary. To explain, the irreducible representations of \(\mathfrak{n}\) are indexed (up to parity) by characters \(\zeta:\mathfrak{n}_{\overline{0}}\to\mathbb{C}\). Let us write \(C_{\mathfrak{n}}(\zeta)\) for an irreducible representation corresponding to \(\zeta\) (it is unique up to parity). Note that \(C_{\mathfrak{n}}(\zeta)\) is one-dimensional if and only if \(\zeta([\mathfrak{n}_{\overline{1}},\mathfrak{n}_{\overline{1}}])=0\). By the same argument as in the even setting, one can show that we have a block decomposition \(\widetilde{\mathcal{N}}=\bigoplus\limits_{\zeta}\widetilde{\mathcal{N}}(\zeta)\). In order to ensure that a triangular decomposition exists for which \(C_{\mathfrak{n}}(\zeta)\) is one-dimensional, the author assumes that \(\mathfrak{l}_{\zeta}\), which is a Levi subalgebra of \(\mathfrak{g}_{\overline{0}}\) constructed from \(\zeta\), is also a Levi subalgebra of \(\mathfrak{g}\). We refer to the paper for the construction of \(\mathfrak{l}_{\zeta}\); it is constructed exactly as in the even setting. The author refers to such \(\zeta\) as being admissible. Then for admissible \(\zeta\) and for \(\lambda\in\mathfrak{h}^*\), the author constructs the standard Whittaker module \(\widetilde{M}(\lambda,\zeta)\). The author then proves that \(\widetilde{M}(\lambda,\zeta)\) has a simple top \(\widetilde{L}(\lambda,\zeta)\), that every simple module in \(\widetilde{\mathcal{N}}\) is isomorphic to \(\widetilde{L}(\lambda,\zeta)\) for some \(\lambda\), and gives an exact condition on when \(\widetilde{L}(\lambda,\zeta)\cong\widetilde{L}(\mu,\zeta)\). However, unlike in the type I case, this argument leaves unconsidered the blocks \(\widetilde{\mathcal{N}}(\zeta)\) for which \(\zeta\) is not admissible. The author goes through the list of basic classical Lie superalgebras and explains which \(\zeta\) are admissible in these cases. After the construction of standard Whittaker modules and the classification of simple modules for admissible \(\zeta\), the author goes on to prove a Miličić-Soergel type equivalence between blocks of \(\widetilde{\mathcal{N}}(\zeta)\) (again for \(\zeta\) admissible) and a category of Harish Chandra \((\mathcal{U}\mathfrak{g},\mathcal{U}\mathfrak{g}_{\overline{0}})\) bimodules. This gives rise to nontrivial equivalences of blocks of the category of Whittaker modules. Finally, the extension of the Backelin functor to the case of type I superalgebras is studied. It is shown to take Verma modules to standard Whittaker modules, simple modules to either simple Whittaker modules or 0, and Kac-inductions of simple modules to Kac-inductions of simple Whittaker modules or 0. Using this, one obtains that the calculation of composition factors in a standard Whittaker module for a type I superalgebra can be reduced to the computation of composition factors of Verma modules in category \(\mathcal{O}\) (see Theorem C for a precise result). For \(\mathfrak{gl}(m|n)\) and \(\mathfrak{osp}(2|2n)\) such computations are known.