On cocliques in commutative Schurian association schemes of the symmetric group

TL;DR

This paper investigates cocliques in association schemes derived from symmetric groups and specific subgroups, confirming their existence and eigenvalue properties in certain cases, and disproving in others.

Contribution

It provides new results on the existence of cocliques and their eigenvalues in association schemes related to symmetric groups and particular subgroup actions.

Findings

Confirmed coclique existence for certain group-subgroup pairs.

Disproved coclique existence for a specific symmetric group pair.

Analyzed eigenvalues related to the cocliques in these schemes.

Abstract

Given the symmetric group $G = \operatorname{Sym}(n)$ and a multiplicity-free subgroup $H\leq G$, the orbitals of the action of $G$ on $G/H$ by left multiplication induce a commutative association scheme. The irreducible constituents of the permutation character of $G$ acting on $G/H$ are indexed by partitions of $n$ and if $\lambda \vdash n$ is the second largest partition in dominance ordering among these, then the Young subgroup $\operatorname{Sym}(\lambda)$ admits two orbits in its action on $G/H$, which are $\mathcal{S}_\lambda$ and its complement. In their monograph [Erd\H{o}s-Ko-Rado theorems: Algebraic Approaches. {\it Cambridge University Press}, 2016] (Problem~16.13.1), Godsil and Meagher asked whether $\mathcal{S}_\lambda$ is a coclique of a graph in the commutative association scheme arising from the action of $G$ on $G/H$. If such a graph exists, then they also asked whether its smallest eigenvalue is afforded by the $\lambda$-module. In this paper, we initiate the study of this question by taking $\lambda = [n-1,1]$. We show that the answer to this question is affirmative for the pair of groups $\left(G,H\right)$, where $G = \operatorname{Sym}(2k+1)$ and $H = \operatorname{Sym}(2) \wr \operatorname{Sym}(k)$, or $G = \operatorname{Sym}(n)$ and $H$ is one of $\operatorname{Alt}(k) \times \operatorname{Sym}(n-k),\ \operatorname{Alt}(k) \times \operatorname{Alt}(n-k)$, or $\left(\operatorname{Alt}(k)\times \operatorname{Alt}(n-k)\right) \cap \operatorname{Alt}(n)$. For the pair $(G,H) = \left(\operatorname{Sym}(2k),\operatorname{Sym}(k)\wr \operatorname{Sym}(2)\right)$, we also prove that the answer to this question of Godsil and Meagher is negative.