Adjoining an order unit to a matrix ordered space (Q850554)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Adjoining an order unit to a matrix ordered space |
scientific article |
Statements
Adjoining an order unit to a matrix ordered space (English)
0 references
3 November 2006
0 references
A matrix ordered space \(V\) is a vector space equipped with a \(*\) and a sequence of cones \(M_n(V)^+\) in the self-adjoint part \(M_n(V)_{sa}\) of \(M_n(V)\) such that \(M_n(V)\) is an ordered vector space and the cones satisfy invariance with respect to \(v \mapsto \gamma^* v \gamma\) for rectangular scalar matrices \(\gamma\). If such a \(V\) has an abstract operator space structure (a sequence of norms on \(M_n(V)\) making \(V\) an \(L^\infty\)-matricially normed space), then the combination is called Riesz normed if for \(v \in M_n(V)\) we have \(\| v\| _n \leq 1 \iff \forall \varepsilon > 0, \exists u_1, u_2 \in M_n(V)\) with \(\max(\| u_1\| _n, \| u_2\| _n) < 1 + \varepsilon\) and \({\begin{pmatrix} u_1 & v\cr v^* & u_2 \end{pmatrix}} \in M_{2n}^+\). \(V\) is called a matrix order unit space if there is \(e \in V^+\) of norm 1, so that we may take \(u_1 = u_2 = (1 + \varepsilon)(e \oplus e \oplus \cdots \oplus e)\) above. The first result is that an \(L^\infty\)-matricially Riesz normed space \(V\) can be embedded as a subspace of codimension 1 in a matrix order unit space \(\widetilde{V}\). This seems to apply when \(V\) already has a matrix order unit. A number of characterisations of \(\widetilde{V}\) are given and some extension theorems for completely bounded maps. The author mentions another approach by \textit{W.\,Werner} [J.~Funct.\ Anal.\ 193, No.\,2, 207--223 (2002; Zbl 1020.46015)] where the order on a self-adjoint subspace \(V \subset {\mathcal B}(H)\) comes from the usual order on \({\mathcal B}(H)\). However, the connection between the two approaches seems not to be explained. Moreover, there are sufficiently many typographical omissions in the paper so that the intended hypotheses are sometimes not clear to this reviewer. Some of these issues are addressed in a corrigendum by the author [Positivity 11, No.\,2, 369--374 (2007; Zbl 1116.46047)].
0 references
adjoining an order unit
0 references
decomposition of completely bounded maps
0 references
extension of completely positive maps
0 references
matrix order unit spaces
0 references
order unit
0 references