An operator Arzelà--Ascoli theorem (Q943553): Difference between revisions

\textit{C.\,Antonescu} and \textit{E.\,Christensen} [J.~Funct.\ Anal.\ 214, No.\,2, 247--259 (2004; Zbl 1063.46059)] developed the following non-commutative version of the Arzelà-Ascoli theorem. Given a metric set \(K\) (i.e., a norm compact, balanced, convex set separating the states) in a unital C*-algebra \(A\), a subset \(H\) of \(A\) is relatively norm compact if and only if the set of affine functions \(\{\hat{h}\in\mathbb{A}(S):h\in H\}\) is bounded and w*-equicontinuous on the state space \(S\) of \(A\), or equivalently, the set \(H\) is bounded and for any \(\varepsilon>0\), there exists \(\lambda>0\) such that \(H\subseteq A_{\varepsilon }+\lambda K+\mathbb{C}1\), where \(A_{\varepsilon}:=\{a\in A:\|a\|\leq\varepsilon\}\). In this paper, the author extends this theorem to the more general context of a matrix metric set \(K\) in a matrix order unit space \((A,1)\), which includes the case of a metric set \(K\) in an operator system \(A\).