A generalization of the cellular indecomposable property via fiber dimension (Q664720)

The authors examine the cellular indecomposable property (CIP), introduced by \textit{R. F. Olin} and \textit{J. E. Thomson} [Integral Equations Oper. Theory 7, 392--430 (1984; Zbl 0539.47016)]. We say that a bounded linear operator \(T\) on a Hilbert space \(H\) is cellular indecomposable if any two nontrivial \(T\)-invariant subspaces \(\mathcal M_1,\mathcal M_2\subset H\) have a nontrivial intersection \(\mathcal M_1\cap\mathcal M_2\neq\{0\}\). The authors extend CIP to the vector-valued case as follows: Let \(\mathcal H(k)\) be a Hilbert space of analytic functions over a domain \(\Omega\subset\mathbb C\) containing the origin, determined by a reproducing kernel \(k\) with the complete NP kernel property. For any multiplier invariant subspace \(\mathcal M \subset\mathcal H(k)\otimes\mathbb C^N\) and two multiplier invariant subspaces \(\mathcal M_1,\mathcal M_2\subset\mathcal M\), \[ fd(\mathcal M_1\cap\mathcal M_2)\geq fd(\mathcal M_1)+ fd(\mathcal M_2)- fd(\mathcal M), \] where \(fd(\mathcal M):= \sup\{\dim\mathcal M(\lambda): \lambda\in\mathbb D\}\) denotes the fiber dimension of \(\mathcal M\) and \(\mathcal M(\lambda):= \{f(\lambda): f\in\mathcal M\}\subset\mathbb C^N\). In particular, if \(fd(\mathcal M_1)+ fd(\mathcal M_2)> fd(\mathcal M)\), then \(\mathcal M_1\cap\mathcal M_2\neq\{0\}\). Motivated by the theorem above, the authors introduce a stronger version of CIP. A Hilbert space \(H\) of analytic functions over a domain \(\Omega\subset\mathbb C\) is said to have the complete cellular indecomposable property (CCIP) if for any invariant subspace \(\mathcal M\subset H\otimes\mathbb C^N\), two invariant subspaces \(\mathcal M_1,\mathcal M_2\subset\mathcal M\) such that \(fd(\mathcal M_1)+ fd(\mathcal M_2)> fd(\mathcal M)\) have a nontrivial intersection \(\mathcal M_1\cap\mathcal M_2\neq\{0\}\). They show the following result: If \(H\) has CLIP, then any invariant subspace \(\mathcal M\subset H\otimes\mathbb C^N\) satisfies \[ \dim(\mathcal M\ominus z \mathcal M)= fd(\mathcal M), \] where \(z\) denotes the coordinate function.