Kostant's problem for permutations of shape $(n-2,1,1)$ and $(n-3,2,1)$
Samuel Creedon, Volodymyr Mazorchuk
TL;DR
The paper delivers a complete combinatorial treatment of Kostant's problem for simple modules in the sl$_n$ principal block corresponding to permutations with RS-shapes $(n-2,1,1)$ and $(n-3,2,1)$. By relating Kostant positivity to left KL preorders, consecutive patterns, and Hecke-algebra data, it derives explicit criteria and reductions via Bruhat-walks and shift maps, including a full characterization for shape $(n-2,1,1)$: a permutation is Kostant negative precisely when it consecutively contains the pattern $14325$. For shape $(n-3,2,1)$, the involutions split into 13 types; indecomposability holds for all, and Kostant negativity is governed by the presence of four patterns: $2143$, $14325$, $1536247$, or $1462537$, with asymptotic shape results established for the associated two-row shape $(2,1)$. The Indecomposability Conjecture is verified in both families, while Kåhrström’s conjecture is resolved in the $(n-2,1,1)$ case and remains conditional (via CM25-2) in the $(n-3,2,1)$ case. Together, these results provide precise, shape-specific combinatorial criteria for Kostant positivity and deepen understanding of the interaction between Hecke-algebra combinatorics and category $ ilde{O}$ representations in type $A$.
Abstract
For a permutation $z$ in the symmetric group $\mathrm{S}_{n}$, denote by $L_{z}$ the corresponding simple highest weight module in the principal block of the BGG category $\mathcal{O}$ for the Lie algebra $\mathfrak{sl}_{n}(\mathbb{C})$. In this paper, we provide a combinatorial answer to Kostant's problem for the modules $L_{z}$ when $z$ has shape (associated Young diagram/integer partition via Robinson-Schensted correspondence) equal to $(n-2,1,1)$ or $(n-3,2,1)$. Moreover, we verify that certain closely related conjectures hold for such permutations, including the Indecomposability Conjecture, which states that applying any indecomposable projective functor to the corresponding simple highest weight module outputs either an indecomposable module or zero.
