Word maps, conjugacy classes, and a noncommutative Waring-type theorem. (Q972565)

Let \(w=w(x_1,\dots,x_d)\) be a non trivial group word. The author proves that if \(G\) is a sufficiently large finite simple group, then every element \(g\in G\) can be expressed as a product of three values of \(w\) in \(G\). This improves many known results for powers, commutators, as well as a theorem on general words obtained by \textit{M. W. Liebeck} and \textit{A. Shalev}, [Ann. Math. (2) 154, No. 2, 383-406 (2001; Zbl 1003.20014)]. In fact for finite simple groups \(G\) of Lie type, the author proves even more: if \(w_1,w_2\) and \(w_3\) are any three nontrivial group words, then \(w_1(G)w_2(G)w_3(G)=G\) provided \(|G|\) is large enough. The main tools involved in the proof are algebraic geometry, character theory and probabilistic ingredients. In the first stage the proof relies on a result of Borel that a word map is a dominant map at the level of simple algebraic groups, and consequences proved by Larsen showing that word maps have large image in finite simple groups. This allows the author to show that if \(w\) is a nontrivial word and \(G\) is a large finite simple group, then the set \(w(G)\) of values of \(w\) in \(G\) contains a ``large'' conjugacy class \(C_w\). In the next stage of the proof, the author uses a probabilistic approach, studying the random variable \(y=y_1y_2y_3\) where \(y_i \in C_w\) are randomly chosen. This is done with the help of character theory and in particular of the recent work on the ``zeta function'' \(\zeta_G(s)=\sum_{\chi\in\text{Irr\,}G}\chi(1)^{-s}\) encoding character degrees. In the course of the proof of the main theorem, the author establishes a variety of results of independent interest, related to short products of conjugacy classes in finite Chevalley groups and the probability distributions they induce. For example it is proved that if \(G\) is a finite simple group and \(x\in G\) is chosen at random, then the probability that \((x^G)^3=G\) tends to 1 as \(|G|\to\infty\). Moreover, for every \(\delta>0\), there is \(\varepsilon>0\) such that if \(G\) is a finite simple group, and \(C\) is a conjugacy class of size at most \(|G|^{1-\delta}\), then \(|C^3|\geq |C|^{1+\varepsilon}\).