A theory of semiprimitive groups (Q1643094)

One of the most fundamental theorems in the theory of finite permutation groups is the O'Nan-Scott theorem classifying finite primitive permutation groups in types. This was generalized by \textit{C. E. Praeger} [J. Lond. Math. Soc., II. Ser. 47, No. 2, 227--239 (1992; Zbl 0738.05046)] who gave a similar description for quasiprimitive permutation groups. A permutation group is called quasiprimitive if all of its non-trivial normal subgroups are transitive. The subject of this paper is a broader class of permutation groups, so-called semiprimitive groups, groups all of whose normal subgroups are transitive or semiregular. Such groups were first investigated by \textit{Á. Bereczky} and \textit{A. Maróti} [J. Algebra 319, No. 4, 1733--1751 (2008; Zbl 1147.20004)]. Their motivation came from universal algebra. The present work provides a theory of semiprimitive groups. At the heart of the investigations are certain normal subgroups of the (transitive) permutation group called plinths. Plinths are introduced in this paper and are defined to be minimally transitive normal subgroups. The authors describe the structure of plinths as much as possible. Surprisingly, much of the theory works for infinite permutation groups. In Theorem 5.1 an Aschbacher-O'Nan-Scott-type structure theorem is given for finite semiprimitive groups. Semiprimitive groups can be described by a glueing process using innately transitive groups. Innately transitive permutation groups were introduced by \textit{J. Bamberg} and \textit{C. E. Praeger} [Proc. Lond. Math. Soc. (3) 89, No. 1, 71--103 (2004; Zbl 1069.20003)] and these are the finite permutation groups with a transitive minimal normal subgroup. The present paper contains a wealth of interesting examples for semiprimitive groups which shows that in some sense the theory developed is best possible. An application is given in Theorem 1.4. Let $\Gamma$ be a locally finite graph and let $G \leq \mathrm{Aut}(\Gamma)$ be vertex-transitive. The pair $(\Gamma,G)$ is called locally $L$ for a permutation group $L$ if for each vertex $x \in \Gamma$ we have that $G_{x}^{\Gamma(x)}$ is permutation isomorphic to $L$. (Here $G_{x}^{\Gamma(x)}$ is the permutation group induced by the stabilizer $G_x$ of $x$ on the neighbourhood of $x$ in $\Gamma$.) A finite permutation group $L$ is said to be graph-restrictive if there exists a constant $c = c(L)$ such that for every locally $L$ pair $(\Gamma,G)$ we have $|G_{x}| \leq c$. A famous conjecture of \textit{R. Weiss} [in:Algebraic methods in graph theory. Vol. I, II. Conference held in Szeged, August 24--31, 1978. Amsterdam-Oxford-New York: North-Holland Publishing Company. 827--847 (1981; Zbl 0475.05040)] from 1981 states that every finite primitive permutation group is graph-restrictive. This conjecture was generalized by Praeger by replacing primitive with quasiprimitive. Work of \textit{P. Potočnik} et al. [J. Comb. Theory, Ser. B 102, No. 3, 820--831 (2012; Zbl 1241.05044)] showed that it may be true that a finite permutation group is graph restrictive if and only if it is semiprimitive. They showed one direction of this statement. The so-called PSV conjecture is that every semiprimitive permutation group is graph-restrictive. Theorem 1.4 of the present paper proves this conjecture in case the finite semiprimitive group has at least two plinths.