On the splitting fields of generic elements in Zariski dense subgroups (Q279726)

Earlier, the authors along with C. Bhagwat considered the notion of representation equivalent lattices. They used it to deduce some of Prasad-Rapinchuk's results on commensurability unconditionally. Using Selberg's trace formula, it follows that representation equivalent uniform lattices are characteristic (polynomial) equivalent. The latter property for two Zariski dense subgroups \(\Gamma_i\) of \(G_i(K)\) (\(i=1,2\)) for absolutely simple \(G_i\) means that collections of characteristic polynomials of elements of \(\Gamma_i\) for the adjoint representations are the same. In the paper under review, the authors look at regular semisimple elements \(g\) of \(G(K)\) for semisimple \(G\) over infinite \(K\) such that the centralizer torus \(Z(g)^0\) has a splitting field \(K_g\) whose Galois group contains the Weyl group of \(G\). Such elements are called ``generic''. For a Zariski-dense \(\Gamma\) of \(G(K)\), the authors consider the compositum of all the splitting fields of characteristic polynomials of generic elements of \(G(K)\) which are contained in \(\Gamma\). They prove that this field determines \(\Gamma\) up to commensurability. One of the key results of Prasad-Rapinchuk the authors employ is the existence of generic elements of infinite order in Zariski dense subgroups. The authors define a notion of Weyl iso-trivial pairs \((G_1,G_2)\) of absolutely simple \(K\)-groups. These are the following pairs: {\parindent=0.6cm\begin{itemize}\item[{\(\bullet\)}] both the groups are of type \(B_n\) or \(C_n\) (\(n \geq 2\)); \item[{\(\bullet\)}] \(n \geq 5\), \(n\) is odd, one of the \(G_i\)'s is of type \(D_n\) and the other is of type \(B_n\) or \(C_n\);\item[{\(\bullet\)}] one of the \(G_i\)'s is of type \(A_2\) and the other is of type \(G_2\). \end{itemize}} The main result of the paper asserts: \noindent Let \(G_1,G_2\) be connected, absolutely almost simple groups over a finitely generated infinite field \(K\) and suppose that the pair \((G_1,G_2)\) is not a Weyl iso-trivial pair. For \(i=1,2\), let \(\Gamma_i \leq G_i(K)\) be Zariski-dense subgroups and \(F_i\), the compositum of the splitting fields of the centralizer tori of generic elements of \(G_i(K)\) contained in \(\Gamma_i\). If \(F_1\) and \(F_2\) are finite extensions of \(F_1 \cap F_2\), then \(G_1\) and \(G_2\) are of the same type over the algebraic closure of \(K\).