Inversion of a generalized block Loewner matrix, the minimal partial realization, and matrix rational interpolation problem (Q1301283)

Inversion theorems for generalized block Loewner matrices with non-square blocks are presented. Connections are made to the general matrix rational interpolation problem, the theory of partial realizations and its extensions, theoretical and computational aspects of many other special types of matrices such as block Hankel, block Toeplitz, Vandermonde, block Bézout matrices, etc. In Section 2 the authors present two inversion results for generalized block Loewner matrices. The aim of Section 3 is to specify two theorems of Section 2 to the block Hankel case. To a given general matrix rational interpolation problem (GMRIP) with multiple modes, there corresponds a class of generalized block Loewner matrices, denoted by \({\mathcal L}\). The question arises as to whether inversions for \(L\in{\mathcal L}\) and for block Hankel matrices can be treated in a unifying framework. In Section 4, the authors prove that this is indeed the case by establishing a very rigid relationship between these two inversion problems in terms of the block Hankel vector of a GMRIP. Finally, in Section 5, the authors consider the connection between the inversion problem for a matrix \(L\in{\mathcal L}\) and the solution of the minimal partial realization, and the solution of the GMRIP.