Linear operators on \(S\)-graded vector spaces (Q1874289)

Given a pair \(S=(A,*)\) where \(A\) is a finite poset and \(*\) an involution on \(A\), an \(S\)-graded vector space \(U\) over a field \(k\) is an outer direct sum of vector spaces \(U_a\) (over \(k\)), where \(a\in A\), such that \(U_a=U_{a^*}\). A \(k\)-linear operator \(\varphi\) on \(U\) is said to be an \(S\)-map if \(\varphi_{aa}=\varphi_{a^*a^*}\) for each \(a\in A\) and \(\varphi_{bc}=0\) for all \(b,c\in A\) subject to \(b>c\); here \(\varphi_{xy}\) denotes the mapping \(U_x\to U_y\) given by the product of the natural embedding of \(U_x\to U\), the mapping \(\varphi\) and the natural projection \(U\to U_y\). The main topic of the paper is, for a fixed polynomial \(f(t)\), the classification within \(S\)-similarity of linear operators \(\varphi\) acting on \(U\) such that \(f(\varphi)=0\). There are tame and wild cases, and for both a complete description is given. Furthermore, the authors discuss the ``most'' general case, where the objects of a Krull-Schmidt subcategory of \(\text{mod }k\) are considered instead of graded vector spaces. It is shown that there are no new tame cases in this situation. A major tool in many considerations are bundles of semichains. A short final section is devoted to ``dispersing representations of quivers'' a concept which extends the usual representations of quivers. In terms of dispersing representations of quivers one can formulate all the classification problems encountered in this paper.