Irreducible operator semigroups such that \(AB\) and \(BA\) are proportional (Q1826839): Difference between revisions

Let \(S\) be an irreducible semigroup of compact linear operators on a complex Banach space of dimension \(n\geq 2\). Suppose that, for all \(A,B\in S\), \(BA = \lambda AB\), where \(\lambda= \lambda_{A,B}\in\mathbb{C}\). The authors' central result is that \(n\) is finite and the nonzero elements of \(S\) form a group-necessarily nilpotent of class 2. This is proved in Theorem 2.1 by passing to the closure of the set of all positive multiples of the elements of \(S\) and showing that its nonzero elements form a group. The authors also show (Theorem 2.4) that the same conclusion holds when the condition \(BA= \lambda AB\) is weakened to \((BA- \lambda AB)A^k= 0\), where \(\lambda= \lambda_{A,B}\in\mathbb{C}\) and \(k= k_{A,B}\in\mathbb{N}\), provided that the scalars \(\lambda\) are restricted to lie in a nonvoid compact subset \(\Gamma\) of \(\mathbb{C}\setminus\{0\}\). The simple example of the semigroup formed by the \(n^2\) matrix units \(E_{ij}\) together with \(0\), in which the identities \(ABA^2= 0\) and \(A^2(AB- BA)A^2= 0\) hold, shows that the condition \(\lambda\neq 0\) is essential in Theorem 2.4 and warns that further generalization to identities \(A^h(BA- \lambda AB)A^k= 0\) is not straightforward.