Moore-Penrose inverses and commuting elements of \(C^*\)-algebras (Q932347)

Let \(\mathcal A\) be a \(C^*\)-algebra with unit 1. The set of invertible elements of \(\mathcal A\) will be denoted by \({\mathcal A}^{-1}\). An element \(p\in {\mathcal A}\) is a projection if it satisfies \(p=p^2=p^*\). An element \(a\in {\mathcal A}\) is said to have a Moore-Penrose inverse if there exists \(x\in {\mathcal A}\) such that \[ (ax)^*=ax, \quad (xa)^*=xa, \quad axa=a, \quad xax=x. \] The Moore-Penrose inverse of \(a\) is unique and is denoted by \(a^\dag\). The set of all elements in \(\mathcal A\) having a Moore-Penrose inverse will be denoted by \({\mathcal A}^\dag\). An element \(a\in{\mathcal A}^\dag\) is EP if it commutes with its Moore-Penrose inverse. The set of all such elements will be denoted by \({\mathcal A}^{\text{EP}}\). The author gives the following characterization of EP elements: Theorem 1. Let \(\mathcal A\) be a unital \(C^*\)-algebra and let \(a\in{\mathcal A}\). The following conditions are equivalent: {\parindent6mm \begin{itemize}\item[[1]] there exists a unique projection \(a^\pi\) such that \(a+a^\pi\in{\mathcal A}^{-1}\) and \(aa^\pi=a^\pi a=0\); \item[[2]] \(a\in {\mathcal A}^{\text{EP}}\). \end{itemize}} The paper is devoted to a study of commutativity of EP elements. The main results are as follows: Theorem 2. Let \(\mathcal A\) be a unital \(C^*\)-algebra, \(a\in {\mathcal A}^{\text{EP}}\), and let \(b\in{\mathcal A}\) be such that \(ab=ba=0\). Then: {\parindent6mm \begin{itemize}\item[[1]] \(a^\pi b=b=ba^\pi\); \item[[2]] \(a^\dag b=ba^\dag=0\); \item[[3]] \(b\in {\mathcal A}^\dag\) implies \(ab^\dag=b^\dag a=0\); \item[[4]] \(b\in {\mathcal A}^\dag\) implies \(a+b\in{\mathcal A}^\dag\) and \((a+b)^\dag=a^\dag+b^\dag\); \item[[5]] \(b\in {\mathcal A}^{\text{EP}}\) implies \(a+b\in{\mathcal A}^{\text{EP}}\) and \((a+b)^\pi=a^\pi+b^\pi-\text{ 1}\). \end{itemize}} Theorem 3. Let \(\mathcal A\) be a unital \(C^*\)-algebra and let \(a, b\in{\mathcal A}^{\text{EP}}\) be such that \(ab=ba\). Then: {\parindent6mm \begin{itemize}\item[[1]] \(a^\pi b^\pi=b^\pi a^\pi\); \item[[2]] \(a b^\dag=b^\dag a\) and \(b a^\dag =a^\dag b\); \item[[3]] \(a^\dag b^\dag=b^\dag a^\dag=(ab)^\dag\). \end{itemize}} Some results concerning EP matrices are deduced from the theorems .