The linear pencil approach to rational interpolation (Q979046): Difference between revisions

The authors start from a general bounded \(MP\)-fraction [\textit{E. Hendriksen} and \textit{O. Njåstad}, J. Math. Anal. Appl. 142, No. 2, 508--520 (1989; Zbl 0721.30025)] and introduce the associated linear pencils together with the rational solutions of some underlying three-term recurrence relations. Generalising the work of \textit{A. I. Aptekarev, V. Kaliaguine} and \textit{W. van Assche} [Proc. Am. Math. Soc. 123, No. 8, 2423--2430 (1995; Zbl 0824.47017)], the authors show how the asymptotic behavior of these rational solutions allows them to decide whether the linear pencil is boundedly invertible. Certain results of \textit{B. Beckermann} [J. Approximation Theory 99, No. 2, 369--408 (1999; Zbl 0931.30002)] and \textit{B. Beckermann} and \textit{V. Kaliaguine} [Constructive Approximation 13, No. 4, 481--510 (1997; Zbl 0897.39003)] on the convergence of Padé approximants at infinity in terms of complex Jacobi matrices have been generalized to the case of multipoint Padé approximants in terms of linear pencils. Finally, for a Markov function and for pairwise conjugate interpolation points tending to infinity, the spectrum and the numerical range of the underlying linear pencil have been computed explicitly.