A note on norm estimates of the numerical radius (Q2482942)

As claimed by the author, most of the results in this note are known. His purpose here is to give alternative and simpler proofs for them. For a bounded linear operator \(A\) on a Hilbert space \(H\), its numerical radius \(w(A)\) is by definition the quantity \(\sup\{|\langle Ax, x\rangle|: x\in H\), \(\| x\|= 1\}\). The author first gives a proof of \(\| A\|= 2w(A)\) for an operator \(A\) satisfying \(A^2= 0\). In fact, in the literature more precise information is known on the numerical range \(W(A)\) \((\equiv\{\langle Ax, x\rangle: x\in H\), \(\| x\|= 1\})\) for any quadratic operator \(A\) (\(A\) satisfying \(A^2+ aA+ bI= 0\) for some scalars \(a\) and \(b\)). This is in \textit{S.-H. Tso} and \textit{P. Y. Wu's} matricial ranges of quadratic operators [Rocky Mt. J. Math. 29, No. 3, 1139--1152 (1999; Zbl 0957.47005)] and is proved along a similar line as here by using a block matrix representation of \(A\). In the second part of the note, the author considers normaloid operators \(A\) (\(A\) satisfying \(\| A\|= r(A)\equiv\sup\{|\lambda|: \lambda\in\Sigma(A)\})\). In particular, he shows that if \(A\) acts on a two-dimensional space, then \(\| A^2\|=\| A\|^2\) if and only if \(A\) is normal.