A criterion for solvability of a finite group by the sum of element orders (Q1799013)

Denote by \(C_n\) the cyclic group of order \(n\). Moreover, for a finite group \(G\) and an element \(g\in G\), denote by \(o(g)\) the order of \(g\) in \(G\), and set \(\psi(G):=\sum_{g\in G}{o(g)}\), the sum of element orders in \(G\). The function \(\psi\) was introduced by \textit{H. Amiri} et al. [Commun. Algebra 37, No. 9, 2978--2980 (2009; Zbl 1183.20022)], where they proved the interesting result that if \(G\) is a finite group maximizing \(\psi(G)\) among all groups of order \(|G|\), then \(G\) is cyclic [loc. cit.]. Since then, many other interesting results involving \(\psi\) have been published, such as the following two results (both by Herzog, Longobardi and Maj): (1) For every non-cyclic finite group \(G\), one has \(\psi(G)\leq\frac{7}{11}\psi(C_{|G|})\), and the constant \(\frac{7}{11}\) is optimal, see [\textit{M. Herzog} et al., J. Pure Appl. Algebra 222, No. 7, 1628--1642 (2018; Zbl 1486.20030), Theorem 1 and Proposition 2]. (2) If \(G\) is a finite nonsolvable group, then \(\psi(G)<\frac{1}{6.68}\psi(C_{|G|})\), see [\textit{M. Herzog} et al., J. Algebra 511, 215--226 (2018; Zbl 1436.20034), Theorem 4]. Note that \(\frac{1}{6.68}\approx 0.14970\). Herzog, Longobardi and Maj also made the conjecture that the optimal value of the constant in their solvability criterion is \(\frac{211}{1617}\approx 0.13049\), obtained by comparing \(\psi(\operatorname{Alt}(5))\) and \(\psi(C_{60})\), see [Zbl 1436.20034, loc. cit., Conjecture 5]. In the paper at hand, the authors show that this conjecture is true, i.e., they show that a finite group \(G\) with \(\psi(G)>\frac{211}{1617}\psi(C_{|G|})\) is solvable. In what follows, we attempt to give a sketch of their proof, conveying the main ideas. Denote by \(\pi(G)\) the set of prime divisors \(|G|\). The proof is by induction on \(|\pi(G)|\). The induction step is easy by [Amiri et al., loc. cit., Corollary B] if \(G\) has a normal cyclic Sylow subgroup, so assume otherwise. Then by [Herzog et al., Zbl 1486.20030, loc. cit., Proposition 2.5], \(G\) is solvable if it has a cyclic subgroup of index less than \(2\cdot\max{\pi(G)}\), which can be shown to be true if \(\max{\pi(G)}\geq13\) using the assumption that \(\psi(G)>\frac{211}{1617}\psi(C_{|G|})\) and an analogue of [Herzog et al., Zbl 1436.20034, loc. cit., Lemma 2.4], see Lemma 2.4 in the paper at hand. Hence, one may assume that \(\pi(G)\subseteq\{2,3,5,7,11\}\), and, of course, also that \(|\pi(G)|\geq 3\). Since the order of a nonabelian finite simple group \(S\) is divisible by \(6\) if \(\max{\pi(S)}\leq 11\), one may also assume that \(\{2,3\}\subseteq\pi(G)\). The authors go through the (finitely many) possibilities for \(\pi(G)\) based on these restrictions. In each case, since the prime divisors of \(|G|\) are fixed, the assumption that \(\psi(G)>\frac{211}{1617}\psi(C_{|G|})\) implies that \(G\) has a cyclic subgroup of index bounded by some concrete constant, see Lemma 2.1 in the paper at hand. From there, using various structural considerations and known solvability criteria for finite groups (such as Herstein's classical result that the existence of an abelian maximal subgroup implies solvability [\textit{I. N. Herstein}, Proc. Am. Math. Soc. 9, 255--257 (1958; Zbl 0089.01601), Theorem], or the recent result [Herzog et al., Zbl 1436.20034, loc. cit., Theorem 1]), the authors succeed in showing that \(G\) is solvable in each case.