A note on majorization and range inclusion of adjointable operators on Hilbert \(C^\ast\)-modules (Q730616): Difference between revisions

\textit{X.-C. Fang} et al. [Linear Algebra Appl. 431, No. 11, 2142--2153 (2009; Zbl 1175.47014)] extended the celebrated majorization theorem of \textit{R. G. Douglas} [Proc. Am. Math. Soc. 17, 413--415 (1966; Zbl 0146.12503)] as follows: Let \(C\in \mathcal{L}(\mathscr{G},\mathscr{F})\), \(A\in \mathcal{L}(\mathscr{E},\mathscr{F})\) and \(\overline{\mathcal{R}(A^*)}\) be orthogonally complemented. Then the following statements are equivalent: {\parindent=0.6cm\begin{itemize}\item[(1)] \(CC^*\leq \lambda AA^*\) for some \(\lambda > 0.\) \item[(2)] There exists \(\mu > 0\) such that \(\| C^*z\|\leq \mu \| A^*z\|\) for any \(z\in \mathscr{F}\). \item[(3)] There exists \(D\in \mathcal{L}(\mathscr{G},\mathscr{E})\) such that \(AD=C\), i.e., \(AX=C\) has a solution. \item[(4)] \(\mathcal{R}(C)\subseteq \mathcal{R}(A)\). \end{itemize}} In this paper, the authors give a nice counterexample to show that the conditions (1) and (4) are not equivalent. For another counterexample showing that (2) and (4) are not equivalent and some related discussion, see also [\textit{Z. Mousavi} et al., Linear Algebra Appl. 517, 85--98 (2017; Zbl 1355.15013)].