A Riesz theory in von Neumann algebras (Q1175123)

Let \({\mathcal A}\) be a von Neumann algebra. An operator \(T\in{\mathcal A}\) is called a Riesz operator relative to \({\mathcal A}\) if \(\lambda\cdot1- T\in\Phi\) for every \(\lambda\neq 0\). By \(\Phi\) is meant the class of Fredholm operators in \({\mathcal A}\). This is equivalent to \(\sigma^ e(T)\) (the Wolf essential spectrum)\(=0\), i.e. \(\lim_{n\to\infty}(\inf_{k\in{\mathcal A}}\| T^ n- K\|)^{1/n}=0\). \(\Phi_ 0\) shall denote the class of Fredholm operators \(T\) with index 0, i.e. \(N_ T\sim N_{T^*}\). Let \({\mathcal R}\) be the set of Riesz operators. One has the generalized Fredholm alternative: Let \(T\in{\mathcal R}\). Then \((1-T)\in\Phi_ 0\). The ideal \({\mathcal K}\) of compact operators is the largest two-sided ideal consisting of Riesz operators only (since Riesz projections are finite).