Some nonsimple modules for centralizer algebras of the symmetric group (Q306221)

This article continues the work of the first two authors in [Commun. Algebra 44, No. 2, 837--850 (2016; Zbl 1348.20017)] on the centraliser algebra for the symmetric group. For \(l\leq n\), \(k\Sigma_n^{\Sigma_l}\) is the centraliser of \(\Sigma_l\) in \(k\Sigma_n\). In [loc. cit.], the authors develop the representation theory of these algebras in a parallel approach to James's theory of Specht modules. Given \(\mu\) a partition of \(l\) and \(\lambda\) a partition of \(n\), they define \(k\Sigma_n^{\Sigma_l}\)-modules \(\mathcal{M}^{(\lambda,\mu)} = \text{Hom}_{k\Sigma_l}(M^\mu, \text{res}^{\Sigma_n}_{\Sigma_l}M^\lambda)\), where \(M^\mu\) and \(M^\lambda\) are the usual permutation modules. Then, in the spirit of James's kernel intersection theorem, they defined modules \(\mathcal{S}^{(\lambda,\mu)}\), \(\mathcal{S}^{(\lambda,\mu)\bot}\) and \(\mathcal{D}^{(\lambda,\mu)} = \mathcal{S}^{(\lambda,\mu)}/ (\mathcal{S}^{(\lambda,\mu)} \cap \mathcal{S}^{(\lambda,\mu)\bot})\) and conjectured that the modules \(\mathcal{D}^{(\lambda,\mu)}\) are either simple or zero, and that \[ \mathcal{D}^{(\lambda,\mu)} \cong \text{Hom}_{k\Sigma_l}(D^\mu, \text{res}^{\Sigma_n}_{\Sigma_l}D^\lambda). \] They proved these conjectures under the condition that \(l<p\). The present paper provides counterexamples to these conjectures (in every characteristic for the former). Namely, the authors show that outside of characteristic 3, \[ \mathcal{D}^{((p+2,1),(p))} \cong\text{Hom}_{k\Sigma_l}(D^{(p)}, \text{res}^{\Sigma_n}_{\Sigma_l}D^{(p+2,1)}) \] is neither simple nor zero, while in characteristic 3, \(\mathcal{D}^{((5,1),(3))}\) is in fact a quotient of \(\text{Hom}_{k\Sigma_l}(D^{(3)}, \text{res}^{\Sigma_n}_{\Sigma_l}D^{(5,1)})\), and neither are simple or zero. The paper proceeds by direct computation, working through the structure of these modules quite thoroughly. At the end of the paper, the authors further demonstrate that these conjectures are not necessarily corrected by choosing a different partial order on partitions (used in the kernel intersection theorem statement, and thus the definitions of \(\mathcal{S}^{(\lambda,\mu)}\) and \(\mathcal{D}^{(\lambda,\mu)}\)), with a small counterexample in characteristic 2. The concluding remarks outline some questions left open by this work. Particularly interesting is that all modules \(\mathcal{D}^{(\lambda,\mu)}\) appearing in the paper have simple heads, and the authors speculate that all simple \(k\Sigma_n^{\Sigma_l}\)-modules may appear as simple heads of \(\mathcal{D}^{(\lambda,\mu)}\).