Extremal vectors for a class of linear operators

From MaRDI portal





\textit{S. Ansari} and \textit{P. Enflo} [Trans. Am. Math. Soc. 350, No. 2, 539-558 (1998; Zbl 0888.47002)] have recently introduced the notion of backward minimal vectors. Let \(H\) be a separable real or complex Hilbert space. Let \(T: H \to H\) be a bounded operator with dense range. Let \(x_0 \in H\) and \(\varepsilon > 0\) with \(\varepsilon < \| x_0 \|\). There is a unique vector \(y^\varepsilon_{n, x_{0}}\) such that \(\| T^n y^\varepsilon_{n, x_{0}} - x_0 \| \leq \varepsilon\) and \(\| y^\varepsilon_{n, x_{0}}\| = \inf \{ \| y \|: \| T^n y - x_0 \| \leq \varepsilon \}\). \(y^\varepsilon_{n, x_{0}}\) are called backward minimal vectors. Ansari and Enflo have studied backward minimal vectors and other types of extremal vectors, and established their connection to invariant subspaces. In this paper, the author gives results on backward minimal vectors -- or rather \(T^n y^\varepsilon_{n, x_{0}}\) -- for a class of linear operators. He conjectures that this class includes multiplication by \(1 - z\) in the space \(H^2\). For the entire collection see [Zbl 0893.00033].











This page was built for publication: Extremal vectors for a class of linear operators

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4213337)