Polynomial interpolation and Gaussian quadrature for matrix-valued functions (Q1334911): Difference between revisions

The authors develop an analogue of Lagrange interpolation for matrix polynomials. The interpolation conditions on the matrix polynomial \(P\) are taken in the tangential form \(u_ i P(z_ i)= v_ i\) where \(u_ i\) is a nonzero row vector, \(v_ i\) is a row vector and \(z_ i\) a complex number; more intricate versions involving derivatives of \(P\) are also considered. The number and form of interpolation conditions are arranged so as to guarantee a unique \(p\times p\) matrix polynomial solution of polynomial degree \(n-1\). A number of closed formulas for the unique solution are obtained, some connected with matrix orthogonal polynomials; these formulas differ somewhat from earlier results on polynomial interpolation by Fuhrmann, Gohberg-Kaashoek-Lerer-Rodman and Ball-Kang. The authors then apply these formulas to obtain Gaussian quadrature formulas of prescribed order for matrix-valued functions. The authors use extensively the approach to the results on matrix polynomials developed in the (1982; Zbl 0482.15001) book entitled `Matrix polynomials' by \textit{I. Gohberg}, \textit{P. Lancaster} and \textit{L. Rodman}.