More on \(\mathfrak R\) (Q1203775)

The author uses the most sophisticated tools of the theory of models to extend properties that were first observed for groups of finite Morley rank to a much more general stable context. The first attempt of this nature was made by Berline and Lascar in a superstable context: everything works when we consider groups of monomial \(U\)-rank; and if \(\text{RU}(G)= n\omega^ \alpha+\cdots\) we obtain a ``monomial component'' for \(G\) by considering the intersection of all definable subgroups \(H\) of \(G\) such that \(\text{RU}(G/H)< \omega^ \alpha\). Ideas of the same kind were later exploited by Hrushovki, in the presence of a regular type \(p\), when considering the \(p\)-component of \(G\), which is the intersection of the \(H\)'s such that \(G/H\) is foreign to \(p\). The author defines the substantial kernel of a stable group in an intrinsic way: he calls a formula ``Frattini'' if it can be discarded from an analysis of \(G\); \(G^ \Phi\), the ``Frattini free component'' of \(G\), is the intersection of all \(H\) such that \(G/H\) is Frattini. To establish properties of this \(G^ \Phi\), the \(R\) hypothesis (anything which algebraizes a generic is itself generic) must be invoked; also, one may think that the results on the structure of \(G^ \Phi\) are not so strong. But in the opinion of the reviewer these weaknesses are not the point: the value of the paper lies in the fact that previous results are extended in an adequate context, and are likely to receive here their ultimate setting. First result: A direct-product decomposition of \(G^ \Phi\) when \(G\) is semisimple. Second result: If \(G\) is a solvable \(R\)- group, then \(G^ \Phi\) is nilpotent by Abelian. Third result: In a stable group \(G\), the \(n\)-solvable normal subgroups generate a solvable group \(R_ n(G)\) (Lemma 41); therefore one asks the question whether this ascending chain of subgroups terminates at a maximal element \(R(G)\), which would be then the maximal solvable normal subgroup of \(G\) (the radical of \(G\)). It is easy to show, by induction on the rank, that \(R(G)\) exists when \(G\) is superstable. The author proves that if \(G\) is an \(R\)-group, then \(R(G^ \Phi)\) exists and \(G^ \Phi/R(G^ \Phi)\) is semisimple.