scientific article; zbMATH DE number 17335

From MaRDI portal
Publication:3974035





[For the entire collection see Zbl 0722.00021.] Let \(W\) be a rational matrix function (i.e., \(W\) is a matrix whose entries are quotients of scalar polynomials). The pole structure of \(W\) over a subset \(\sigma\) in the complex plane may be described in terms of a pair of matrices \((C_ \pi,A_ \pi)\), called a right pole pair for \(W\) over \(\sigma\), which encodes data about poles and partial pole multiplicities of \(W\) at points in \(\sigma\) and information concerning the principal parts of the Laurent expansions of \(W\) at these poles. Similarly, the null structure is encoded in a pair of matrices \((A_ \zeta,B_ \zeta)\), called a left null pair for \(W\) over \(\sigma\), which gives the zeros and partial multiplicities of zeros of \(W\) in \(\sigma\) together with directional information about these zeros. These notions, which have proved to be very useful in applications to factorization, interpolation and control theory, have been developed mainly for regular rational matrix functions, i.e., for the case when \(W\) is square and \(\text{det} W\) does not vanish identically [see, e.g., \textit{J. A. Ball}, \textit{I. Gohberg} and \textit{L. Rodman}, Interpolation of rational matrix functions (Birkhäuser 1990; Zbl 0708.15011)]. In this paper the authors develop systematically from first principles analogous notions of zero and pole structure for the nonregular case. The subset \(\sigma\) is allowed to contain the point infinity. In the nonregular case the zero-pole structure of \(W\) over \(\sigma\) involves the pole and zero pairs referred to above and an additional triple of matrices \((A_ \kappa,B_ \kappa,D_ \kappa)\), called a left kernel triple, which describes a matrix polynomial whose rows form a minimal polynomial basis for the set of all rational row vectors \(r\) such that \(rW=0\). The paper explains how to compute these objects from the Smith- McMillan factorization of \(W\). Also connections with realizations of \(W\) are discussed. In the analysis orthogonality with respect to a valuation on the field of scalar rational functions turns out to be a useful tool. The authors also solve the corresponding inverse problem. They present a necessary and sufficient condition for existence of a rational matrix function \(W\) with a given left kernel triple \((A_ \kappa,B_ \kappa,D_ \kappa)\), with a left null pair over a subset \(\sigma\) equal to \((A_ \zeta,B_ \zeta)\), and with a right pole pair over \(\sigma\) equal to \((C_ \pi,A_ \pi)\). Also an algorithm is given for constructing a rational matrix function \(W\) with these data from a regular rational matrix function which has \((A_ \zeta,B_ \zeta)\) and \((C_ \pi,A_ \pi)\) as left zero pair and right pole pair, respectively.




Cited in
(31)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3974035)