Diagonalizing matrices over \(AW^\ast\)-algebras (Q1948123): Difference between revisions
From MaRDI portal
Set profile property. |
Set OpenAlex properties. |
||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W2034272279 / rank | |||
Normal rank |
Revision as of 19:25, 19 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Diagonalizing matrices over \(AW^\ast\)-algebras |
scientific article |
Statements
Diagonalizing matrices over \(AW^\ast\)-algebras (English)
0 references
30 April 2013
0 references
Let \(M_n({\mathcal A})\) be the matrix algebra of \(n\times n\) matrices with coefficients from a unital \(AW^\ast\) algebra \({\mathcal A}\). The main result shows that for each integer \(n\geq 2\) every commuting subset \(X\subseteq M_n({\mathcal A})\) of normal matrices is simultaneously diagonalizable in a sense that there exists a unitary \(u\in M_n({\mathcal A})\) such that \(uxu^\ast\) are diagonal matrices for all \(x\in X\). For a proof, the authors study the dimension function, \(d(e)\), of a properly infinite projection \(e\), which is defined to be the least cardinal strictly larger than the cardinal of any orthogonal family of projections equivalent to \(e\). It is shown, for example, that for two properly infinite projections \(e,f\) which have the same central cover, \(e\sim f'\leq f\) implies \(d(e)\leq d(f)\). This may no longer be true if their central covers, \(c(e)\) and \(c(f)\), differ. However, if \(e,f\) belong to a certain large class of properly infinite projections, then \(e\sim f'\leq f\) is equivalent to the fact that \(d(e)\leq d(f)\) and \(c(e)\leq c(f)\). This class will not be defined in the present review; let us just mention that every properly infinite projection in \({\mathcal A}\) is a sum of orthogonal projections from the class. As an application it is shown that if \(\phi:{\mathcal A}\to{\mathcal B}\) is a \(\ast\)-homomorphism between unital \(AW^\ast\)-algebras, which preserves the suprema of subsets of projections, then the same is true for a map \(M_n(\phi):M_n({\mathcal A})\to M_n({\mathcal B})\), defined by \((a_{ij} )_{ij}\mapsto (\phi(a_{ij}) )_{ij}\). It follows that the assignment \(\mathbb M_n:{\mathcal A}\mapsto M_n({\mathcal A})\), which transforms the morphism \({\mathcal A}\buildrel\phi\over\longrightarrow {\mathcal B}\) into \(M_n({\mathcal A})\buildrel M_n(\phi)\over\longrightarrow M_n({\mathcal B})\), is a functor in the category AWstar whose objects are unital \(AW^\ast\)-algebras and whose morphisms are \(\ast\)-homomorphisms that preserve the suprema of subsets of projections.\smallskip The converse implication, that is, if a unital \(C^\ast\)-algebra \({\mathcal C}\) must be an \(AW^\ast\)-algebra when every commuting subset \(X\subseteq M_n({\mathcal C})\) of normal matrices is, for each integer \(n\geq 2\), simultaneously diagonalizable, remains unknown at the time of writing the report. It is known, however, to be true for abelian \(C^\ast\)-algebras by \textit{K. Grove} and \textit{G. K. Pedersen} [J. Funct. Anal 59, 65--89 (1984; Zbl 0554.46026)].
0 references
diagonalization
0 references
\(AW^*\)-algebra
0 references
dimension
0 references
matrix algebra
0 references
normal matrix: properly infinite projection
0 references
\(C^*\)-algebra
0 references
linear preserver
0 references