Computation of large-scale quadratic forms and transfer functions using the theory of moments, quadrature and Padé approximation
Large-scale problems in scientific and engineering computing often require solutions involving large-scale matrices. In this paper, we survey numerical techniques for solving a variety of large-scale matrix computation problems, such as computing the entries and trace of the inverse of a matrix, computing the determinant of a matrix, and computing the transfer function of a linear dynamical system.
Most of these matrix computation problems can be cast as problems of computing quadratic forms u T ƒ(A)u involving a matrix functional ƒ(A). It can then be transformed into a Riemann-Stieltjes integral, which brings the theory of moments, orthogonal polynomials and the Lanczos process into the picture. For computing the transfer function, we focus on numerical techniques based on Padé approximation via the Lanczos process and the moment-matching property. We will also discuss issues related to the development of efficient numerical algorithms, including using Monte Carlo simulation.
KeywordsKrylov Subspace Linear Dynamical System Tridiagonal Matrix Hankel Matrix Expansion Point
Unable to display preview. Download preview PDF.
- A. C. Antoulas and D. C. Sorensen, Approximation of large-scale dynamical systems: An overview, Technical report, Electrical and Computer Engineering, Rice University, Houston, TX, Feb. 2001.Google Scholar
- Z. Bai, Krylov subspace techniques for reduced-order modeling of large-scale dynamical systems, to appear in Appl. Numer. Math. (2002).Google Scholar
- Z. Bai, M. Fahey, G. Golub, E. Menon, and M. Richter, Computing partial eigenvalue sum in electronic structure calculations, Scientific Computing and Computational Mathematics Program SCCM-98-03, Stanford University, 1998.Google Scholar
- R. R. Craig, Jr., Structural Dynamics: An Introduction to Computer Methods, Wiley, New York, 1981.Google Scholar
- M. Fahey, Numerical computation of quadratic forms involving large scale matrix functions, Ph.D. thesis, University of Kentucky, 1998.Google Scholar
- W. Gautschi, A survey of Gauss-Christoffel quadrature formulae, in: E. B. Christoffel-the Influence of His Work on Mathematics and the Physical Sciences (P. L. Bultzer and F. Feher, eds.), Birkhäuser, Boston, 1981, 73–157.Google Scholar
- G. Golub and G. Meurant, Matrics, moments and quadrature, in: Proc. 15th Dundee Conference, June 1993 (D. F. Griffiths and G. A. Watson, eds.), Longman Sci. Tech., Harlow, 1994.Google Scholar
- E. Grimme, Krylov projection methods for model reduction, Ph.D. thesis, University of Illinois at Urbana-Champaign, 1997.Google Scholar
- L. Komzsik, MSC/NASTRAN, Numerical Methods User’s Guide, Version 70.5, MacNeal-Schwendler Corp., Los Angeles, 1998.Google Scholar
- H. Madrid, private communication, 1995.Google Scholar
- A. Odabasioglu, M. Celik, and L. T. Pileggi, Practical considerations for passive reduction of RLC circuits, in: Proc. Internat. Conf, on Computer-Aided Design, 1999, 214–219.Google Scholar
- J. C. Sexton and D. H. Weingarten, The numerical estimation of the error induced by the valence approximation, Nuclear Phys. B Proc. Suppl. (1994), xx.Google Scholar
- J. Vlach and K. Singhal, Computer Methods for Circuit Analysis and Design, Van Nostrand, New York, 1994.Google Scholar
- K. Willcox, J. Peraire, and J. White, An Arnoldi approach for generalization of reduced-order models for turbomachinery, FDRL TR-99-1, Fluid Dynamics Research Lab., Massachusetts Institute of Technology, 1999.Google Scholar
- M. N. Wong, Finding tr(A -1) for large A by low discrepancy sampling, Presentation at the Fourth International Conference on Monte Carlo and Quasi-Monte Carlo Methods in Scientific Computing, Hong Kong, China, 2000.Google Scholar