Groups in which the centralizer of any non-central element is maximal (Q2197977)

This paper is concerned with finite \(\mathscr{MC}\)-groups, which are defined as finite groups \(G\) such that the centralizer \(\mathrm{C}_G(g)\) of each noncentral element \(g\in G\) is a maximal subgroup of \(G\). The authors provide the following structural restrictions on the central quotient \(\overline{G}=G/Z(G)\) of a finite \(\mathscr{MC}\)-group \(G\): \begin{itemize} \item Theorem B: If \(\overline{G}\) is a \(p\)-group for some prime \(p\), then it is in fact an elementary abelian \(p\)-group. \item Theorem A: If \(\overline{G}\) is not of prime power order, then it is either abelian, or there are primes \(p\not=q\) such that \(\overline{G}\) is an inner abelian \(\{p,q\}\)-group of the form \(\overline{P}\rtimes\overline{Q}\) for some abelian \(p\)-group \(\overline{P}\) and a cyclic group \(\overline{Q}\) of prime order \(q\). \end{itemize} An ``inner abelian'' group is defined as a nonabelian group all of whose proper subgroups are abelian. Here are two technical comments on the paper which the reviewer thought to be noteworthy: \begin{itemize} \item In Lemma 3.2, \(x\) is intended to be a \(p\)-element. Moreover, this lemma includes the assumption from Theorem A that \(|\overline{G}|\) shall not be a prime power, although its proof does not make use of this assumption. \item The authors' proof of Theorem A also yields the following information in case \(\overline{G}=\overline{P}\rtimes\overline{Q}\) is nonabelian: \(\overline{P}\) is elementary abelian and \(\overline{Q}=\langle\overline{y}\rangle\) acts irreducibly on \(\overline{P}\). Indeed, at the end of the proof, apply its Step 1 to see that if \(y\in G\) has order a power of \(q\) and maps to \(\overline{y}\) in \(\overline{G}\), then \(\overline{\operatorname{C}_G(y)}\) is a Sylow \(q\)-subgroup of \(\overline{G}\). Hence \(\overline{\operatorname{C}_G(y)}=\overline{Q}\). But \(\operatorname{C}_G(y)\) is a maximal subgroup of \(G\) containing \(Z(G)\), so its image \(\overline{Q}\) must be a maximal subgroup of \(\overline{G}\). This is only possible if there are no proper nontrivial \(\overline{y}\)-invariant subgroups of \(\overline{P}\), which implies the above statements on \(\overline{P}\) and \(\overline{Q}\). \end{itemize}