Fast Reduction of Generalized Companion Matrix Pairs for Barycentric Lagrange Interpolants

For a barycentric Lagrange interpolant $p(z)$, the roots of $p(z)$ are exactly the eigenvalues of a generalized companion matrix pair $(\mathbf{A},\mathbf{B})$. For real interpolation nodes, the matrix pair $(\mathbf{A},\mathbf{B})$ can be reduced to a pair $(\mathbf{H},\mathbf{B})$, where $\mathbf{H}$ has tridiagonal plus rank-one structure. In this paper we propose two fast algorithms for reducing the pair $(\mathbf{A},\mathbf{B})$ to Hessenberg-triangular form. The matrix pair $(\mathbf{A},\mathbf{B})$ has two spurious infinite eigenvalues, and if the leading coefficients of the interpolant are zero, there will also be other infinite eigenvalues. We propose tools for detecting when the leading coefficients of $p(z)$ are zero, and describe a procedure to deflate all of the infinite eigenvalues from the reduced matrix pair $(\mathbf{H},\mathbf{B})$, while still maintaining the tridiagonal plus rank-one structure of the resulting standard eigenvalue problem. Since fast $QR$ algorithms exist for such struc...

[1]  E. E. Osborne On Pre-Conditioning of Matrices , 1960, JACM.

[2]  Jean-Paul Berrut,et al.  Rational functions for guaranteed and experimentally well-conditioned global interpolation , 1988 .

[3]  L. Trefethen Spectral Methods in MATLAB , 2000 .

[4]  Louis A. Romero,et al.  Roots of Polynomials Expressed in Terms of Orthogonal Polynomials , 2005, SIAM J. Numer. Anal..

[5]  Lloyd N. Trefethen,et al.  Stability of Barycentric Interpolation Formulas for Extrapolation , 2011, SIAM J. Sci. Comput..

[6]  I. J. Good THE COLLEAGUE MATRIX, A CHEBYSHEV ANALOGUE OF THE COMPANION MATRIX , 1961 .

[7]  Robert C. Ward,et al.  Balancing the Generalized Eigenvalue Problem , 1981 .

[8]  Lloyd N. Trefethen,et al.  Barycentric Lagrange Interpolation , 2004, SIAM Rev..

[9]  Paul Van Dooren,et al.  Balancing Regular Matrix Pencils , 2006, SIAM J. Matrix Anal. Appl..

[10]  R. M. Corless,et al.  Generalized Companion Matrices in the Lagrange Bases , 2004 .

[11]  Kai Hormann,et al.  Barycentric rational interpolation with no poles and high rates of approximation , 2007, Numerische Mathematik.

[12]  Daniel B. Szyld,et al.  The matrix eigenvalue problem: GR and Krylov subspace methods , 2009, Math. Comput..

[13]  Lloyd N. Trefethen,et al.  An Extension of MATLAB to Continuous Functions and Operators , 2004, SIAM J. Sci. Comput..

[14]  Marcel Riesz,et al.  Eine trigonometrische Interpolationsformel und einige Ungleichungen für Polynome. , 1914 .

[15]  David S. Watkins,et al.  Fundamentals of matrix computations , 1991 .

[16]  F. Tisseur,et al.  STRUCTURED TOOLS FOR STRUCTURED MATRICES , 2003 .

[17]  Gene H. Golub,et al.  Matrix computations , 1983 .

[18]  Israel Gohberg,et al.  On the fast reduction of a quasiseparable matrix to Hessenberg and tridiagonal forms , 2007 .

[19]  W. Specht,et al.  Die Lage der Nullstellen eines Polynoms. III , 1957 .

[20]  Victor Y. Pan,et al.  Fast and stable QR eigenvalue algorithms for generalized companion matrices and secular equations , 2005, Numerische Mathematik.

[21]  N. Higham The numerical stability of barycentric Lagrange interpolation , 2004 .

[22]  C. Reinsch,et al.  Balancing a matrix for calculation of eigenvalues and eigenvectors , 1969 .

[23]  Raf Vandebril,et al.  A Multiple Shift QR-step for Structured Rank Matrices ? , 2007 .

[24]  Paul Van Dooren,et al.  Implicit double shift QR-algorithm for companion matrices , 2010, Numerische Mathematik.

[25]  Heinz Rutishauser,et al.  Vorlesungen über numerische Mathematik , 1976 .