Topological generation of simple algebraic groups (Q6963504)

The authors address a fundamental problem in the theory of algebraic groups: the topological generation of a simple algebraic group \(G\) over an algebraically closed field \(k\) by elements lying in specified conjugacy classes. More precisely, given an irreducible subvariety \(X \subseteq G^r\), they study the existence of tuples \((x_1, \dots, x_r) \in X\) such that the Zariski-closure of the subgroup \(\langle x_1, \dots, x_r \rangle\) equals \(G\). The focus is on the case where \(X = C_1 \times \dots \times C_r\), with \(C_i\) conjugacy classes of elements of prime order modulo \(Z(G)\) and \(r \ge 2\).\N\NBuilding on earlier work of the second author [J. Algebra 569, 511--543 (2021; Zbl 1458.20043)] concerning the groups \(\mathrm{SL}_n(k)\) and of all three authors [Adv. Math. 369, Article ID 107177, 49 p. (2020; Zbl 1505.20041)] for the exceptional types, the article provides a complete classification for the classical groups \(\mathrm{Sp}_n(k)\) and \(\mathrm{SO}_n(k)\) in positive characteristic \(p > 0\). The main result (Theorem~7) identifies the precise obstructions to topological generation in this setting, showing that, except for a few explicitly listed exceptional configurations with \(r \le 4\), the only obstructions arise from linear-algebraic constraints on the eigenvalues or Jordan forms of the \(x_i\). These include bounds on the dimensions of their largest eigenspaces and the case where all \(x_i\) are quadratic elements acting on the natural module.\N\NIn addition to the classification, the paper establishes several structural results about the locus\N\[\N\Sigma = \{ x \in X : \langle x_1, \dots, x_r \rangle^{\mathrm{Zar}} = G \},\N\]\Nproving that \(\Sigma\) is either empty or generic in \(X\) and describes the behavior of generic stabilizers in linear representations (Theorem~13). Furthermore, the authors apply their results to random generation of finite simple groups of Lie type, completing a line of research initiated by \textit{M. W. Liebeck} and \textit{A. Shalev} [Ann. Math. (2) 144, No. 1, 77--125 (1996; Zbl 0865.20020); Bull. Lond. Math. Soc. 34, No. 2, 185--188 (2002; Zbl 1046.20046)]. In particular, they prove that for fixed primes \(r, s\) with \(s > 2\), the probability that a random pair \((x, y)\) of respective orders \(r\) and \(s\) topologically generates a finite simple group \(G\) tends to 1 as \(|G| \to \infty\), with the sole exception of families \(G = \mathrm{PSp}_4(q)\) and \((r, s) \in \{(2, 3), (3, 3)\}\).\N\NThe authors also provide a general result (Theorem~2) on topological generation in varieties, extending their previous work [loc. cit.] and connecting it to questions about generic stabilizers.