Abstract root subgroups and quadratic action. With an Appendix by A. E. Zalesskii (Q1283575)

Since several years the author investigates abstract groups by what he calls general root subgroups. One of the highlights in this work is certainly the paper under review. Unfortunately it would take to much place to describe the six theorems proved in the paper. The author himself uses 8 pages to state these and give the necessary definitions and notation. But I try to give a rough idea of what the paper is about. Let \(M\) be a finite dimensional \(k\)-vectorspace (with \(\text{char }k\neq 1\), \(k\neq GF(3)\)). An element \(\sigma\in\text{GL}(M)\) is said to act quadratically on \(M\) if \(\sigma\) has minimal polynomial \((x-1)^2\). A quadratic pair \((G,M)\) is a subgroup \(G\) of \(\text{GL}(M)\) generated by quadratic elements \(\sigma\) with \(k\circ\sigma\leq G\). The purpose is to determine quadratic pairs. The story goes back to J. Thompson who classified quadratic pairs for \(k=\text{GF}(p)\), \(p\leq 5\). Here \(G\) is a central product of finite Lie-type groups in characteristic \(p\). For \(K=\text{GF}(3)\) some work was done by \textit{C.-Y. Ho} [J. Algebra 43, 338-358 (1976; Zbl 0385.20006)]. Here also some sporadic groups show up, which shows that there is some principal difference between \(\text{GF}(3)\) and the other fields. For \(k=\text{GF}(2)\) the definition is meaningless as all involutions are quadratic. As in Thompson's paper the first reduction is to prove a ``Central Product-Theorem'' which basically tells us that we have a central product of quasisimple groups \(G_i\) and each \(G_i\) acts irreducible on some module \(N_i\) as a quadratic pair. So basically just irreducible actions have to be treated. \quad Let \(d\) be the minimal dimension of the commutator of a quadratic \(\sigma\) with \(M\). Now as in Thompson's paper for any such quadratic element \(\sigma\) we can introduce the set \(E_\sigma\) of all quadratic elements having the same centralizer and commutator on \(M\) as \(\sigma\). Then it turns out that \(E_\sigma\cup\{1\}=E(\sigma)\) is an Abelian group. Now the next theorem says that if \((G,M)\) is a quadratic pair with \(G\) quasisimple, then the \(E(\sigma)\) defined above form a class of abstract root subgroups of \(G\) satisfying the strong unipotent condition (unipotent sets of root groups generate nilpotent groups). Now the remainder of the paper is devoted to the classification of groups generated by abstract root subgroups in a fairly general setting. Here also there are reduction theorems to the quasisimple case. Under some additional assumptions (which are satisfied by the root subgroups which come from quadratic pairs) the author is able to determine the quasisimple groups generated by such abstract root subgroups. These roughly are classical groups defined over some division algebras. But also groups like \(F_4(k)\), \(E_6(k)\), \(E_7(k)\), \(E_8(k)\), \(^2E_6(k)\), \(E^9_{7,4}(k)\), \(E^{28}_{8,4}(k)\), \(F_4(k,K)\) show up. The paper contains much more information on quadratic pairs and groups generated by abstract root subgroups as can be said in this review.