Finite groups have more conjugacy classes (Q516938)

Finding lower bounds for the number \(k(G)\) of conjugacy classes of a finite group \(G\) in terms of \(|G|\) is a venerable problem in finite group theory. It all started with double logarithmic bounds in the 1960s and 70s, but today it is conjectured that the asymptotically best possible lower bound is logarithmic in \(|G|\). The last significant progress in this direction has been obtained by \textit{L. Pyber} [J. Lond. Math. Soc., II. Ser. 46, No. 2, 239--249 (1992; Zbl 0712.20016)], who was able to prove that \(k(G)\geq C\log |G|/(\log\log |G|)^8\) for some fixed \(C>0\). In lack of new ideas, since then people have tried to improve pieces of Pyber's proof to improve his bound a little bit. The reviewer improved the solvable case in Pyber's proof [Isr. J. Math. 181, 433--444 (2011; Zbl 1261.20030)], which led to a replacement of the exponent 8 in Pyber's result by 7. In the paper under review, the authors improve the part of Pyber's proof dealing with finite groups with a trivial solvable radical, and this ultimately allows them to show that for every \(\varepsilon >0\) there exists a \(\delta >0\) such that if \(|G|>3\), then \(k(G)\geq \delta\log |G|/(\log\log |G|)^{3+\varepsilon}\). They also prove the desired logartihmic bound for groups with trivial solvable radical by proving that \(k(G)>\log_3 |G|\) for such groups. The proofs in this paper are technical and depend on CFSG.