Regular strongly typical blocks of \(\mathcal O^{\mathfrak q}\) (Q1048101)

From the text: ``We use the technique of Harish-Chandra bimodules to prove that regular strongly typical blocks of the category \(\mathcal O\) for the queer Lie superalgebra \(\mathfrak q_n\) are equivalent to the corresponding blocks of the category \(\mathcal O\) for the Lie algebra \(\mathfrak{gl}_n\). In more detail: For \(n\in\mathbb N\) let \(\mathfrak q_n\) denote the queer Lie superalgebra and \(\mathcal O^{\mathfrak q}\) denote the category \(\mathcal O\) for \(\mathfrak q_n\). The category \(\mathcal O^{\mathfrak q}\) decomposes into a direct sum of blocks, which can be typical or atypical. Atypical blocks are very complicated and may contain infinitely many simple objects. Typical blocks are much easier and are always equivalent to module categories over finite-dimensional algebras. In [the first author, J. Algebra Appl. 6, No. 5, 731--778 (2007; Zbl 1236.17009)] it was shown that the finite-dimensional algebras describing typical blocks are always properly stratified in the sense of \textit{V. Dlab} [C. R. Acad. Sci., Paris, Sér. I, Math. 331, No. 3, 191--196 (2000; Zbl 0964.16009)]. Among all typical blocks one separates strongly typical ones, which are described [Frisk (loc. cit.)] by quasi-hereditary algebras in the sense of [\textit{E. Cline, B. Parshall} and \textit{L. Scott} [Mem. Am. Math. Soc. 591 (1996; Zbl 0888.16006)], [\textit{V. Dlab} and \textit{C. M. Ringel}, Ill. J. Math. 33, No. 2, 280--291 (1989; Zbl 0666.16014)]. A very general conjecture about the combinatorial structure of the category \(\mathcal O^{\mathfrak q}\) is given in [\textit{J. Brundan}, Adv. Math. 182, 28--77 (2004; Zbl 1048.17003)], 4.8]. In the special case of regular strongly typical blocks this conjecture says that multiplicities of simple highest weight supermodules in Verma supermodules for \(\mathfrak q_n\) are given by Kazhdan-Lusztig combinatorics. For this special case a much stronger conjecture was formulated in [Frisk (loc. cit.), 3.9], namely that strongly typical blocks of \(\mathcal O^{\mathfrak q}\) are equivalent to the corresponding blocks of the category \(\mathcal O\) for the Lie algebra \(\mathfrak{gl}_n\). A strong evidence for this conjecture, established in [Frisk (loc. cit.)], was a similarity between the quasi-hereditary structures in both cases. Moreover, [Frisk (loc. cit.), 3.9] contains also an explicit conjecture for the structure of all typical blocks. The aim of this paper is to prove the conjecture from [Frisk (loc. cit.), 3.9] (and hence the conjecture from [Brundan (loc. cit.), 4.8] as well) for regular strongly typical blocks. There is a natural restriction functor from \(\mathcal O^{\mathfrak q}\) to the category \(\mathcal O\) for the Lie algebra \(\mathfrak{gl}_n\). However, unlike the case of most of the other Lie superalgebras, this restriction functor does not induce an equivalence in a straightforward way. The problem is that the highest weights of Verma supermodules over \(\mathcal O^{\mathfrak q}\) are not one-dimensional (because the Cartan subsuperalgebra of \(\mathfrak q\) is not commutative). Subsequently, under restriction Verma supermodules are not mapped to the corresponding Verma modules but rather to direct sums of Verma modules (see Proposition 2). This suggests that the naive restriction functor is a direct sum of several copies of some ``smaller'' functor, which defines the desired equivalence of categories. This is exactly what we prove in this paper. The main idea of the proof is to realize the induction functor (the left adjoint to the restriction) as a tensor product with some Harish-Chandra bimodule. This requires several definitions and some technical work as we are forced to go beyond the original categories and instead work with the so-called thick version of the category \(\mathcal O\). Unfortunately, along the way we use some properties of Harish-Chandra bimodules, which require an additional assumption of regularity of the blocks we work with. The main result of the paper is the equivalence of blocks of categories \(\mathcal O\) for \(\mathfrak q_n\) and \(\mathfrak{gl}_n\), see Theorem 1. This extends earlier results of \textit{I. Penkov} and \textit{V. Serganova} [Indag. Math. New Ser. 3, 419--466 (1992; Zbl 0849.17030)] and \textit{M. Gorelik} [J. Am. Math. Soc. 15, No. 1, 167--184 (2002; Zbl 0985.17011)] to the case of the algebra \(\mathfrak q_n\) and verifies conjectures from [Frisk (loc. cit.), 3.9] and [Brundan (loc. cit.), 4.8] in our setup. At the same time Theorem~1 is a refinement (in the case of \(\mathfrak q_n\)) of the main result of [\textit{I. Penkov}, Monatsh. Math. 118, No. 3--4, 267--313 (1994; Zbl 0883.17028)].''