Topological generation of exceptional algebraic groups (Q2182270): Difference between revisions

Let \(G\) be a simple algebraic group over an algebraically closed field \(k\), and let \(C_1, \ldots, C_t\) be a collection of non-central conjugacy classes from \(G\). It is of great interest in many settings (e.g., generic freeness of representations, topological generation of simple groups, random generation of finite simple groups) to know that there exists a tuple \((g_1,\ldots,g_t)\) with each \(g_i\in C_i\) such that the subgroup generated by the \(g_i\) is dense in \(G\). This paper gives many results along these lines; the following are representative. (Theorem 1 of the paper) Suppose \(G\) is simply connected and simple and \(k\) is not algebraic over a finite field. Suppose \(\Omega\) is an irreducible subvariety of the \(t\)-fold Cartesian product \(G^t\), and for \(x\in \Omega\) let \(G(x)\) denote the closure of the subgroup of \(G\) generated by the elements of \(x\). If we set \[ \Delta^+ = \{x\in \Omega\mid G(x) \text{ has positive dimension}\}, \] then either \(\Delta^+\) is dense in \(G\), or it contains no \(k'\)-points for any extensions \(k'/k\). (Theorem 7 of the paper) With notation as above, suppose also that \(G\) has exceptional type, and \(\Omega\) is a product of non-central conjugacy classes in \(G\). If \(t\geq 5\), then the set of \(x\in \Omega\) such that \(G(x) = G\) is dense. It is also shown that \(t\geq 4\) will do for \(G = G_2\), and that these bounds are best possible. The methods in the paper are a mixture of uniform proofs and case-by-case analysis for different Dynkin types, building on much previous work in the literature.