Minimal generation of transitive permutation groups (Q1643178): Difference between revisions

Let the finite group \(G\) be a transitive permutation group of degree \(n\). Denote by \(d(G)\) the minimal number of elements required to generate \(G\). Results of \textit{L.G. Kovács} and \textit{M. Newman} [Q. J. Math., Oxf. II. Ser. 39, No. 155, 361--372 (1988; Zbl.0668.20004)], \textit{R. M. Bryant} et al., [Q. J. Math., Oxf. II. Ser. 46, No. 184, 385--407 (1995; Zbl.0854.20001)], and \textit{A. Lucchini} et al. [Bull. Lond. Math. Soc. 32, No. 2, 191-195 (2000; Zbl. 1021.20002)] show that \(d(G) = O( \frac{n}{ \sqrt{\log_2 n}} )\); an example constructed by Kovács and Newman [loc. cit.], the construction of which is included in the paper (Example 6.1), shows that the bound is ``asymptotically sharp''. The paper is dedicated to calculate the constants involved in these bounds. The results are displayed in Theorem \(1.1\) or, in a more detailed version, Theorem \(5.3\). Except in some exceptional cases, described in detail in Theorem \(5.3\), when \(d((G) \leq \lfloor \frac{cn}{\sqrt{\log_2 n}} \rfloor\) with \(c = 1512660\frac{\sqrt{\log_2(2^{19}\cdot 15}}{(2^{19}\cdot 15}\) holds, \(d(G) \leq \lfloor \frac{c_1n}{\sqrt{\log_2 n}} \rfloor\) with \(c_1 = \frac{\sqrt{3}}{2}\). If \(G\) has a solvable transitive subgroup, the latter bound always applies (Corollary \(1.2\)). Theorem \(1.7\) is motivated by a conjecture of \textit{L.Pyber} [Ann. Math. (2) 137, No. 1, 203--220 (1993; Zbl. 0778.20012)] stating the number of subgroups of the symmetric group \(S_n\) is precisely \( 2^{\frac{1}{16}+o(1))n^2}\). In a private correspondence with the author, J. C. Schlage-Puchta has proven that if the quantity \( \max\{\frac{d(G) \log_2 |G|}{n^2} \, \vert \, G \leq S_n\) transitive \(\}\) approaches \(0\) as \(n\) tends to \(\infty\), then there exists an absolute constant \( c\) such that the number of subgroups of \(S_n\) is at most \(2^{o(n^2)}\) multiplied by the number of subgroups of \(S_n\) whose orbits lengths are at most \(c\). Theorem \(1.7\) provides an absolute constant \(c\) such that \(d(G) \leq \lfloor \frac{C n^2}{\log_2 \vert G \vert \sqrt{\log_2 n}} \rfloor\), from which the reduction of Pyber's conjecture follows. The displayed bound is asymptotically best possible, as Kovács and Newman's [loc. cit.] example again demonstrates. The final main result of the paper, whose proofs reveal many interesting facts, too numerous to mention, concerns minimally transitive permutation groups. Theorem \(1.8\): A nonsolvable minimally transitive permutation group has a unique nonabelian chief factor which is a direct product of copies of \(L_2(p)\), where \(p\) is a Mersenne prime. A word on the strategy of the proof: Theorem \(1.1\) for primitive groups follows from a result of \textit{D. Holt} and \textit{C. M. Rooney-Dougal} [J. Algebra 387, 195--214 (2013; Zbl.1293.20067)] stated in the paper as Theorem \(2.11\). The bulk of the work thus lies in deriving the results in the case of an imprimitive group \(G\). This amounts to finding upper bounds on \(d(G)\) for subgroups G of wreath products \( R \wr S\); the paper's principal strategy for doing this is to reduce modulo the base group \( B\) of \(R \wr S\) and use induction to bound \(d(G/G \cap B).\) According to Lemma \(5.8\), \(G \cap B\) is built, as a normal subgroup of \(G\), from submodules of induced modules for \(G\) and nonabelian chief factors of \( G\). Upper bounds for generator numbers in submodules of induced modules are derived in Theorem \(4.13\) and Theorem \(4.24\).