Cosets, genericity, and the Weyl group. (Q731906)

The author uses ideas of \textit{G. Cherlin} and \textit{E. Jaligot} [J. Algebra 276, No. 1, 13-79 (2004; Zbl 1056.20020)] and himself [J. Symb. Log. 71, No. 2, 599-610 (2006; Zbl 1103.03035)] to prove a general result about non-generosity of proper cosets. Recall that a subset \(X\) of a group \(G\) is called `generic' if finitely many translates cover the group; it is `generous' if the union of its \(G\)-conjugates is generic. A typical example is a maximal torus in a reductive algebraic group; in the more general context of groups of finite Morley rank, one has `decent tori' on the one hand (definable divisible Abelian subgroups which are the definable envelope of their torsion points), and `Carter subgroups' on the other (definable connected nilpotent subgroups of finite index in their normalizer). While Carter subgroups always exist [\textit{O. Frécon} and \textit{E. Jaligot}, in Model theory with applications to algebra and analysis. Vol. 2, Lond. Math. Soc. Lect. Note Ser. 350, 1-58 (2008; Zbl 1173.03033)], they need not always be generous; however, generous Carter subgroups, as well as maximal decent tori, are always conjugate. One can thus define a generalized Weyl group, either as \(N_G(T)/C_G(T)^0\) for some maximal decent torus, or as \(N_G(C)/C\) for some Carter subgroup \(C\) (the first being a quotient of the latter). The author proves that in a connected group \(G\) of finite Morley rank such that generic elements lie in a connected nilpotent subgroup, a coset \(wH\) is never generous for any definable subgroup \(H\) and any \(w\in N_G(H)\setminus H\). He deduces that in such a group, if \(H\) is a definable connected generous subgroup and \(w\in N_G(H)\setminus H\) of finite order \(n\) modulo \(H\) with \(\{h^n:h\in H\}\) generic in \(H\), then \(C_H(w)<H\). The paper contains a detailed discussion of generosity, and the implications of the result for the Weyl group.