Functional decompositions on vector-valued function spaces via operators (Q764958): Difference between revisions

A classical theorem of \textit{J.\ Dixmier} [Ann.\ Math.\ 389, 1173--1190 (1950; Zbl 0036.35801)] states that each bounded linear functional \(f\) on the algebra \(\mathcal{B}(\ell_2)\) of bounded linear operators on the Hilbert sequence space \(\ell_2\) admits a unique decomposition \(f= g+h\), where \(g\) is a trace functional given by a trace class operator, and \(h\) is a singular functional, vanishing on the ideal \(\mathcal{K}(\ell_2)\) of compact operators; moreover, it holds that \(\| f \| = \| g \| + \| h \|\). Generalizations to a much wider class of Banach spaces have led to the vast literature on \(M\)-ideals introduced by \textit{E. M.\ Alfsen} and \textit{E. G.\ Effros} [Ann.\ Math.\ 96, 98--173 (1972; Zbl 0248.46019)]. The authors of the paper under review consider in the same spirit functional decompositions on vector-valued function spaces. More precisely, let \(X\) be a Banach space with a generalized basis. The Banach algebra \(\mathcal B(X)\) of bounded linear operators on \(X\) is used to construct Banach spaces, \(\mathcal M\) and \(\mathcal K\), of weak\(^{\star}\) continuous functions from the state space of a \(C^{\star}\)-algebra to \(\mathcal B(X)\). If the basis satisfies certain properties, the authors prove that the dual space of \(\mathcal M\) has a decomposition analogous to that of the dual space of \(\mathcal B(X)\). In terms of the notion of \(M\)-ideals, the subspace \(\mathcal K\) is an \(M\)-ideal in the Banach space \(\mathcal M\). For the cases of \(c_{0}\) and \(\ell ^p\), \(1<p<\infty \), they also prove an analogue of the result that trace\((AB)\)=trace\((BA)\) for a trace class operator \(A\) and a bounded operator \(B\) on a Hilbert space.