The transvection free groups with polynomial rings of invariants (Q2627977)

Let \(k\) be a field and \(V=k^n\) a finite dimensional vector space over \(k\). A linear transformation \(s\in\text{GL}(V)\) is called a pseudoreflection if it has finite order, i.e. \(s^n=1\) for some integer \(n\), and if its fixed point set \(H_s\subset V\) is a codimension one subspace. A pseudoreflection \(s\) is diagonalizable if and only if the order of \(s\) is relatively prime to the characteristic of the field \(k\). In positive characteristic, a pseudoreflection which is not diagonalizable is called a transvection. For any finite group \(G\subset\text{GL}(V)\) let \(S(V^*)\) be the ring of polynomial functions on \(V\), equipped with the obvious \(G\) action, and let \(S(V^*)^G\subset S(V^*)\) be the subring of polynomials invariant under that action. The invariant subring \(S(V^*)^G\) is called polynomial if it is generated as an algebra by algebraically independent polynomials. The following is the main result of the paper under reveiw: Theorem 1: Suppose that \(G\) contains no transvections. Then \(S(V^*)^G\) is polynomial if and only if for each subspace \(U\subseteq V\), its pointwise stablizer subgroup \(\text{Fix}(U)\subseteq G\) is generated by pseudoreflections. Theorem 1 was formulated and proved for irreducible groups in an earlier paper of \textit{G. Kemper} and \textit{G. Malle}, Transform. Groups 2, No. 1, 57--89 (1997; Zbl 0899.13004)] (in characteristic zero, or more generally, for non-modular groups, Theorem 1 follows from classical results of Shephard-Todd, Chevalley, Steinberg, and Serre). The contribution of the paper under review is the extension of Theorem 1 to reducible groups.