Commutativity of projectors (Q5957188)

Suppose that \(P_1\) and \(P_2\) are (orthogonal) projectors in an \(n\)-dimensional (unitary) complex vector space. There are many papers that give necessary and/or sufficient conditions for the sum \(P_1+P_2\), the difference \(P_1-P_2\) or the product \(P_1P_2\) to be a projector or an orthogonal projector; see, among others, \textit{J. Groß} and \textit{G. Trenkler} [Linear Multilinear Algebra 44, No. 3, 247-259 (1998; Zbl 0929.15016)] or \textit{Y. Takane} and \textit{H. Yanai} [Linear Algebra Appl. 289, 297-310 (1999; Zbl 0930.15003)]. In particular, the commutativity property \(P_1P_2=P_2P_1\) is a necessary and sufficient condition for the product \(P_1P_2\) of orthogonal projectors to be a projector, but for arbitrary projectors this condition is no longer necessary. The theme of the present paper to investigate similarities and dissimilarities of this type. The interesting results are too technical to be explained here in detail. Loosely speaking, each of the theorems involves a series of statements on projectors which are equivalent for orthogonal projectors whereas in the general case certain statements will no longer be equivalent. Some of these statements are in terms of generalized inverses, others involve subspaces like range and kernel.