A note on \(k\)-generalized projections (Q861022)

Following \textit{H.--K.\ Du} and \textit{Y.~Li} [Linear Algebra Appl.\ 400, 313--318 (2005; Zbl 1067.47001)], a bounded linear operator \(A\) on a Hilbert space is called \(k\)-generalized projection if \(A^k=A^*\), where \(A^*\) denotes the adjoint of \(A\). A characterization of \(k\)-generalized \(n \times n\)-complex matrices was given by \textit{J.~Benítez} and \textit{N.~Thome} [ibid.\ 410, 150--159 (2005; Zbl 1086.15502)]. Given a bounded linear operator \(A\), the authors of the present paper prove that the following three statements are equivalent: (i) \(A\) is a \(k\)-generalized projection; (ii) \(A\) is normal and \(A^{k+2}=A\); (iii) \(A\) is normal and the spectrum of \(A\) is contained in \(\{0\} \cup \root{k+1} \of{1}\), where \(\root{k+1} \of {1}\) denotes the unit roots of order \(k+1\). The results generalize those of the above cited papers.