Matrix-valued interpolation and hyperconvex sets (Q5945059)

Given a uniform algebra \(A\) on a compact Hausdorff space \(X\), the set \(M_{n}(A)\) of \(n \times n\) matrices with entries in \(A\) and distinct points \(x_{1},\dots , x_{k}\in X\), the \(n-\)th interpolation set is defined by \(\mathcal D^{n}(A;x_{1},\dots , x_{k})\)=\(\{(F(x_{1}),\dots ,F(x_{k})): F\in M_{n}(A), \|F\|\leq 1\}^{-}\). It is shown, that the scalar-valued interpolation set (\(n=1\)) does not uniquely determine the matrix-valued interpolant. The author have introduced the concept of matricially hyperconvex family (m.h.f.) \(\{ \mathcal D^{n}\}\) of closed sets with non-empty interiors, \(\mathcal D^{n}\subseteq M_{n}(\mathcal{C^{k}})\), provided that for every \(n,m\) and \(n \times n\) matrix \(P=(p_{ij})\) of polynomials in \(m^{2}\) variables which maps the ball of \(M_{m}\) into the ball of \(M_{n}\), we have \((P(W_{1}),\dots ,P(W_{k}))\in\mathcal D^{n}\) and \((V_{1}\bigoplus W_{1},\dots, V_{k}\bigoplus W_{k}) \in \mathcal D^{n+m}\), whenever \((W_{1},\dots, W_{k})\in \mathcal D^{m}\) and \((V_{1},\dots, V_{k})\in \mathcal D^{n}\). It is proved that \(\{ \mathcal D^{n}\}\) is m.h.f. iff \( \mathcal D^{n}\)=\(\mathcal D^{n}(A;x_{1},\dots , x_{k})\) for some \(A\) and \(x_{1},\dots , x_{k}\in X\). The duality theory for hyperconvex sets [see \textit{B. Cole}, \textit{K. Lewis}, and \textit{J. Wermer}, J. Funct. Anal. 107, No. 2, 235-254 (1992; Zbl 0762.46053)] is developed, for a tensor product of uniform algebras the Agler's interpolation formula [see \textit{J. Agler}, ``Some interpolation theorems of Nevanlinna-Pick type'', J. Oper. Theory (to appear)] is generalized.