Generating semisimple groups by tori (Q536166): Difference between revisions

After the work of Kuranishi and Auerbach we know that a connected simple Lie group \(G\) has a dense subgroup with two generators. Also, Kuranishi has shown that if these two elements are closed to the identity then they generate \(G\) if and only if their logarithms generate the Lie algebra of \(G. \) In this note the authors show: a) When \(G\) has finite center, given a noncentral element \(g\), there exists an elliptic element \(h\) (i.e., an element of a compact subgroup) such that \(h\) and \(ghg^{-1}\) generate \(G.\) b) Let \(g\) be a noncentral element of \(G.\) Then there is a torus \(T\) such that the semigroup spanned by \(T \cup gTg^{-1}\) is \(G\). For such a torus, for every element except for a finite set of torsion elements, the subsemigroup generated by \(\{t, gtg^{-1}\}\) is dense in \(G\). As a consequence they show: 1) The set of pairs \((x,y)\in G\times G\) so that \(x\) and \(y\) generate \(G\) is Zariski dense in \(G\times G\). 2) Given \(g \in G\), the set of elliptic elements \(h\) such that \(h, ghg^{-1}\) generate \(G\) is topologically dense in the set of elliptic elements. The main result is deduced from the following: Let \(G\) be a simply connected semisimple complex Lie group, and let \(\sigma\) be an involutory automorphism of \(G\). Let \(G_0\) be the subgroup of fixed points of \(\sigma\) and let \(T_0\) be a maximal torus of \(G_0.\) Let \(C \subset G\) be a conjugacy class which is not contained in a proper normal subgroup of \(G.\) Then there exists an element \(g \in C\) for which \(T_0 \cup gT_0g^{-1}\) generates \(G.\) The set of such \(g \in C\) is Zariski open in \(C\).