On the commuting probability of \(p\)-elements in a finite group (Q6158355)

A fairly well-known result due to \textit{W. H. Gustafson} [Am. Math. Mon. 80, 1031--1034 (1973; Zbl 0276.60013)] says: The probability that two random elements in a finite group commute is greater than \(\frac{5}{8}\) this is equivalent to \(G\) being abelian, in which case the probability is 1. The paper under review deals with a local version of this result. Let \(p\) be a prime which divides \(|G|\). Then the probability \(\operatorname{Pr}_p(G)\) that two random \(p\)-elements commute is defined by \(\frac{|\{(x,y) \in G_p \times G_p: xy=yx\}|}{|G_p|^2}\), where \(G_p\) is the set of \(p\)-elements in \(G\). Set \(f(p) = \frac{p^2+p-1}{p^3}\). Recall that \(f(2) =\frac{5}{8}\), the Gustafson bound. The main result of the paper is that \(G\) has an abelian normal Sylow \(p\)-subgroup if and only if \(\operatorname{Pr}_p(G) > f(p)\). The authors also give a complete classification in the boundary case that \(\operatorname{Pr}_p(G) = f(p)\) and \(G = O^{p^\prime}(G)\). In fact, if \(p \geq 5\), \(\operatorname{Pr}_p(G) = f(p)\) and \(G\) is non abelian simple, then \(G \cong \mathrm{PSL}_2(p)\). There are no non solvable examples \(G = O^{p^\prime}(G)\) for \(p = 2\) or \(3\). There is a further result in the paper of independent interest. Let \(G\) be a finite group, \(p\) a prime and \(x \in G \setminus O_p(G)\) be a \(p\)-element. Then \(\frac{|C_G(x)_p|}{|G_p|} \leq \frac{1}{p}\). The proof of the main result depends on the classification of the finite simple groups. In fact, it uses a result on fixed point ratios due to \textit{T. C. Burness} and \textit{R. M. Guralnick} [Adv. Math. 411 A, Article ID 108778, 90 p. (2022; Zbl 1511.20010)] which depends on the classification of the finite simple groups. But in case of \(F^\ast(G)\) is a \(p^\prime\)-group, the classification is not needed. The authors also claim that for \(G\) simple, \(\operatorname{Pr}_p(G)\) goes to zero with \(|G|\) to infinity. They prove this for \(G\) an alternating group. They also give a sketch of a proof of this result for groups of Lie type. There are many further results in this worthwhile reading paper.