Column reduced rational matrix functions with given null-pole data in the complex plane (Q1329972)

The problem of constructing a column reduced rational matrix function with prescribed null-pole data is discussed. This problem arises in the framework of parametrization of rational matrices meeting a number of prescribed bitangential interpolation conditions and it is related to Wiener-Hopf factorization at infinity. A regular \(m \times m\) rational matrix \(W(s)\) can be represented as \(W(s) = E(s) \cdot \text{diag} (s^{k_ 1}, \dots, s^{k_ m})\) where \(E(s) = E_ 0 + s^{-1} E_ 1 + s^{-2} E_ 2 + \cdots\); the integers \(k_ j\), \(j=1, \dots, m\) denote the highest power of \(s\) in the \(j\)th column of \(W(s)\). \(W(s)\) is said to be column reduced at infinity if \(E_ 0\) is an invertible matrix. A quintet \(\sigma = (C, A_ \pi; A_ \zeta, B; \Gamma)\) of matrices of appropriate dimensions is called an admissible data set if the pairs \((C,A_ \pi)\) and \((A_ \zeta, B)\) are respectively observable and controllable and \(\Gamma\) satisfies the Sylvester equation \(\Gamma A_ \pi - A_ \zeta \Gamma = BC\). The main problem is the following: Given an admissible Sylvester data set \(\sigma\) construct all regular rational matrix functions \(W\) such that \(W\) has \(\sigma\) as its \(\mathbb{C}\)-null-pole triple and it is column reduced at infinity. This problem is completely solved. Explicit formulas are given for a solution \(W\) and up to a certain constant invertible factor on the right one obtains a full parametrization of such solutions. The results are specified and developed in the last section for matrix polynomials.