Dual operator algebras and a hereditary property of minimal isometric dilations (Q1199270)

A dual algebra \(A\) is a subalgebra of \({\mathcal L}(H)\) which contains the identity and is closed for the ultraweak operator topology. If \(C_ 1\) is the trace class, define \(Q_ A=C_ 1/^ \perp A\), where \(^ \perp A\) is the preannihilator of \(A\) in \(C_ 1\) under the pairing \(\langle T,[L]\rangle=\text{tr}(TL)\), \(T\in A\), \([L]\in Q_ A\). Given cardinals \(m\), \(n\) such that \(1\leq m,n\leq\aleph_ 0\), \(A\) is said to have the property \((\mathbf{A}_{m,n})\) if for every \(m\times n\) matrix \(L\) with entries from \(Q_ A\), there exist \(x_ i\), \(y_ j\) in \(H\) (\(0\leq i<m\), \(0\leq j<n)\) such that \(x_ i\otimes y_ j=L_{ij}\). (Recall that, for \(x\), \(y\) in \(H\) \(x\otimes y\) is the rank 1 operator defined by \((x\otimes y)u= \langle u,y\rangle x)\). For \(T\) in \({\mathcal L}(H)\), \(\text{Lat}(T)\) denotes the lattice of invariant subspaces of \(T\). If \(T\) is a contraction and \(B_ T\in {\mathcal L}(K)\) is a minimal isometric dilation of \(T\), then \(B_ T\) can be decomposed as \(B_ T=U_ T\oplus R_ T\), where \(U_ T\in {\mathcal L}({\mathcal U}_ T)\) is a unilateral shift operator and \(R_ T\in {\mathcal L}(R_ T)\) is unitary. If \({\mathcal M}\in\text{Lat}(T)\), \({\mathcal M}\neq 0\) then \(B_ T\) is an isometric dilation of \(T\mid{\mathcal M}\). Hence \(T\mid{\mathcal M}\) has a minimal isometric dilation \(B_{T\mid{\mathcal M}}\in {\mathcal L}(\widetilde K)\) such that \({\mathcal M}\subset \widetilde K\subset K\), \(\widetilde K\in\text{Lat}(B_ T)\) and \(B_{T\mid{\mathcal M}}=B_ T\mid\widetilde K\). A contraction \(T\in{\mathcal L}(H)\) is said to have property (\textbf{H}) if for every \({\mathcal M}\in\text{Lat}(T)\), \({\mathcal M}\neq 0\), \({\mathcal U}_{T\mid{\mathcal M}}\subset {\mathcal U}_ T\). Denote by \(\mathbf{A}\) the class of absolutely continuous contractions [for details on this class see the book ``Dual algebras with applications to invariant subspaces and dilation theory'' by \textit{H. Bercovici}, \textit{C. Foiaş} and \textit{C. Pearcy}, CBMS Reg. Conf. Ser. Math. 56 (1985; Zbl 0569.47007)]. The main result of the paper under review is the following theorem: For \(T\in\mathbf{A}\) there exists \({\mathcal M}\in \text{Lat}(T)\) such that \(T\mid{\mathcal M}\in\mathbf{A}\) and \(T\mid{\mathcal M}\) has property (\textbf{H}) if and only if \(T\) has property \(\mathbf{A}_{1,\aleph_ 0}\) if and only if there exists \({\mathcal M}\in\text{Lat}(T)\) such that \(T\mid{\mathcal M}\in\mathbf{A}\cap C_{.0}\) (here, \(S\in C_{.0}(H)\) means that \(\| S^{*n}x\|\to 0\) for all \(x\in H\)). The paper is a sequel of Apostol, Bercovici, S. Brown, Chevreau, Exner, Foias and Pearcy's contributions to the study of the structure of contractions.