A bitangential interpolation problem on the closed unit ball for multipliers of the Arveson space (Q1402339): Difference between revisions

Let \({\mathcal H}(k_d)\) be the reproducing kernel Hilbert space associated with the kernel \(k_d(z,w)=\frac{1}{1-\langle z,w\rangle }\) \((z,w\in {\mathbb C}^d)\). Let \({\mathcal E}\), \({\mathcal E}_*\) be auxiliary Hilbert spaces and let \({\mathcal H}(k_d)\otimes {\mathcal E}\) be a tensor product Hilbert space. The Schur class \({\mathcal S}_d({\mathcal E},{\mathcal E}_*)\) is defined as the class of all \({\mathcal L}({\mathcal E},{\mathcal E}_*)\)-functions such that the multiplication operator \(S:f(\cdot)\mapsto S(\cdot)f(\cdot)\) maps contractively \({\mathcal H}(k_d)\otimes {\mathcal E}\) into \({\mathcal H}(k_d)\otimes {\mathcal E}_*\). The authors consider the bitangential interpolation problem in the class \({\mathcal S}_d({\mathcal E},{\mathcal E}_*)\), which includes both the interior and the boundary interpolation conditions. Its left-sided tangential version is formulated as follows. Let \(\mathbb{B}\) denote the unit ball in \(\mathbb{C}\), let \(\Omega_b\subset\Omega\) be two sets and let \(\sigma:\Omega\to\overline{\mathbb B^d}\) be an injective function which maps \(\Omega\setminus\Omega_b\) and \(\Omega_b\) into \(\mathbb B\) and \(\partial\mathbb B\), respectively. Let \({\mathcal E}_L\) be an auxiliary Hilbert space, let \(a:\Omega\to{\mathcal L}({\mathcal E}_L,{\mathcal E}_*)\), \(c: \Omega\to {\mathcal L}({\mathcal E}_L,{\mathcal E})\) be arbitrary functions and let \(\Psi:\Omega_b\to{\mathcal L}({\mathcal E}_L)\), \(\Psi(\xi)\geq 0\) for \(\xi\in\Omega_b\). Find all functions \(S\in{\mathcal S}_d({\mathcal E},{\mathcal E}_*)\) such that \(\lim_{r\to 1}S(r\sigma(\xi))^*a(\xi)=c(\xi)\) (\(\xi\in\Omega\)) and \(\lim_{r\to 1}a(\xi)^* \frac{I_{{\mathcal E}_*}-S(r\sigma(\xi)) S(r\sigma(\xi))^*}{1-r^2} a(\xi)\leq\Psi(\xi)\) \((\xi\in\Omega_b)\). The problem is shown to be solvable if the following kernel \[ \Lambda(\xi,\mu)=\begin{cases} \Psi(\xi), & \xi=\mu\in\Omega_b;\\ \frac{a(\xi)^*a(\mu)-c(\xi)^*c(\mu)}{1-\langle \sigma(\xi),\sigma(\mu)\rangle }, & \text{otherwise}, \end{cases} \] is positive on \(\Omega\). A full description of all the solutions of the problem is given as a linear-fractional transform of a free parameter from a Schur class \({\mathcal S}_d(\widetilde{\mathcal E},\widetilde{\mathcal E}_*)\) with \(\widetilde{\mathcal E}\supset{\mathcal E}\), \(\widetilde{\mathcal E}_*\supset{\mathcal E}_*\). In the case when \({\mathcal E}\), \({\mathcal E}_*\) are finite-dimensional spaces, \(\Omega\) is a finite set and \(\Omega_b=\emptyset\) this problem is reduced to the bitangential interpolation problem in the class \({\mathcal S}_d({\mathbb C}^p,{\mathbb C}^q)\) considered by \textit{J.A. Ball} and \textit{V. Bolotnikov} [Linear Algebra Appl. 353, 107--147 (2002; Zbl 1037.47011)]. In the case \(\Omega={\mathbb B}^d\) it contains the multi-dimensional version of Leech's theorem [see \textit{M. Rosenblum} and \textit{J. Rovnyak}, ``Hardy classes and operator theory'' (Oxford University Press) (1985; Zbl 0918.47001)].