Markov's Theorem for Orthogonal Matrix Polynomials

From MaRDI portal





Markov's theorem for orthogonal polynomials on the real line is about the convergence of the rational function \(Q_n(x)/P_n(x)\), with \(P_n\) orthogonal polynomials with respect to a measure \(\mu\) and \(Q_n\) the associated polynomials, to the Stieltjes transform of the measure \(\mu\). This rational function is also the \(n\)th convergent of the corresponding continued fraction. In the present paper this theorem is extended to matrix polynomials on the real line. Let \(P_n\) be (left) orthogonal matrix polynomials satisfying \(\int P_n(x) dW(x) P_m^*(x) = \delta_{m,n}I\), where \(W\) is a \(N\times N\) positive definite matrix of measures, and let \(Q_n(x)\) be the associated polynomials, then the author proves that \(P_n^{-1}(z) Q_n(z)\) converges to \(\int dW(t)/(z-t)\) uniformly on compact subsets avoiding the accumulation points of the zeros of \(P_n\). The order in which \(P_n^{-1}\) and \(Q_n\) are multiplied is important. The proof is based on the Gauss quadrature formula for orthogonal matrix polynomials, which is described in section 3. The author also gives the asymptotic behavior of the ratio \(P_n^{-1}(z) P_{n-k}^{[k]}(z)\), where \(P_{n-k}^{[k]}\) are the \(k\)th associated polynomials. The reference [V] to the reviewer's paper \textit{``Orthogonal polynomials, associated polynomials and functions of the second kind} [J. Comput.\ Appl.\ Math. 37, No. 1-3, 237-249 (1991; Zbl 0744.42012)]\ is missing from the references.




Cited in
(64)








This page was built for publication: Markov's Theorem for Orthogonal Matrix Polynomials

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3128435)