Laurent interpolation for rational matrix functions and a local factorization principle (Q1193176)

Let \(\Sigma\) be a finite subset of \(\mathbb{C}\cup\{\infty\}\) and suppose that for every \(\mu\in\Sigma\) integers \(k(\mu)\geq 0\) and \(\ell(\mu)\geq-k(\mu)\) and \(n\times n\)-matrices \(U_{j\mu}\) and \(V_{j\mu}\), \(-k(\mu)\leq j\leq\ell(\mu)\), are given. It is shown that there exists a regular rational matrix function \(W(\lambda)\) such that the Laurent series of \(W(\lambda)\) and \(W^{-1}(\lambda)\) around \(\lambda=\mu\) begin with \(\sum^{\ell(\mu)}_{j=-k(\mu)}V_{j\mu}(\lambda-\mu)^ j\) and \(\sum^{\ell(\mu)}_{j=-k(\mu)}U_{j\mu}(\lambda-\mu)^ j\), respectively, provided a certain (obviously necessary) condition is satisfied. Moreover, \(W(\lambda)\) can be chosen such that it has at most one pole outside \(\Sigma\), and this pole may be specified in advance. Some applications of the result are given.