2-Norm Error Bounds and Estimates for Lanczos Approximations to Linear Systems and Rational Matrix Functions

The Lanczos process constructs a sequence of orthonormal vectors $v_m$ spanning a nested sequence of Krylov subspaces generated by a hermitian matrix $A$ and some starting vector $b$. In this paper we show how to cheaply recover a secondary Lanczos process starting at an arbitrary Lanczos vector $v_m$. This secondary process is then used to efficiently obtain computable error estimates and error bounds for the Lanczos approximations to the action of a rational matrix function on a vector. This includes, as a special case, the Lanczos approximation to the solution of a linear system $Ax = b$. Our approach uses the relation between the Lanczos process and quadrature as developed by Golub and Meurant. It is different from methods known so far because of its use of the secondary Lanczos process. With our approach, it is now possible in particular to efficiently obtain upper bounds for the error in the 2-norm, provided a lower bound on the smallest eigenvalue of $A$ is known. This holds in particular for a lar...

[1]  I. Moret,et al.  RD-Rational Approximations of the Matrix Exponential , 2004 .

[2]  E. Hairer,et al.  Geometric Numerical Integration: Structure Preserving Algorithms for Ordinary Differential Equations , 2004 .

[3]  Andreas Frommer,et al.  BiCGStab(ℓ) for Families of Shifted Linear Systems , 2003, Computing.

[4]  Z. Strakos,et al.  On error estimation in the conjugate gradient method and why it works in finite precision computations. , 2002 .

[5]  VALERIA SIMONCINI,et al.  MATRIX FUNCTIONS , 2006 .

[6]  Gérard Meurant,et al.  On computing quadrature-based bounds for the A-norm of the error in conjugate gradients , 2012, Numerical Algorithms.

[7]  Gérard Meurant.  Estimates of the l2 norm of the error in the conjugate gradient algorithm , 2005, Numerical Algorithms.

[8]  G. Golub,et al.  Bounds for the error of linear systems of equations using the theory of moments , 1972 .

[9]  M. Hestenes,et al.  Methods of conjugate gradients for solving linear systems , 1952 .

[10]  Andreas Frommer,et al.  Fast CG-Based Methods for Tikhonov-Phillips Regularization , 1999, SIAM J. Sci. Comput..

[11]  Andreas Frommer,et al.  MONOTONE CONVERGENCE OF THE LANCZOS APPROXIMATIONS TO MATRIX FUNCTIONS OF HERMITIAN MATRICES , 2009 .

[12]  E. Hairer,et al.  Geometric Numerical Integration , 2022, Oberwolfach Reports.

[13]  Error Estimates in Linear Systems , 1999 .

[14]  Ernesto Estrada,et al.  Communicability in complex networks. , 2007, Physical review. E, Statistical, nonlinear, and soft matter physics.

[15]  Marlis Hochbruck,et al.  Preconditioning Lanczos Approximations to the Matrix Exponential , 2005, SIAM J. Sci. Comput..

[16]  P. Petrushev,et al.  Rational Approximation of Real Functions , 1988 .

[17]  G. Golub,et al.  Matrices, Moments and Quadrature with Applications , 2009 .

[18]  Yousef Saad,et al.  Iterative methods for sparse linear systems , 2003 .

[19]  Y. Saad.  Analysis of some Krylov subspace approximations to the matrix exponential operator , 1992 .

[20]  Valeria Simoncini,et al.  Acceleration Techniques for Approximating the Matrix Exponential Operator , 2008, SIAM J. Matrix Anal. Appl..

[21]  H. V. D. Vorst,et al.  Numerical methods for the QCDd overlap operator. I. Sign-function and error bounds , 2002, hep-lat/0202025.

[22]  Andreas Frommer,et al.  QCD and numerical analysis III : proceedings of the Third International Workshop on Numerical Analysis and Lattice QCD, Edinburgh, June-July 2003 , 2005 .

[23]  H. V. D. Vorst,et al.  An iterative solution method for solving f ( A ) x = b , using Krylov subspace information obtained for the symmetric positive definite matrix A , 1987 .

[24]  G. Golub,et al.  Matrices, moments and quadrature II; How to compute the norm of the error in iterative methods , 1997 .

[25]  Gene H. Golub,et al.  Estimates in quadratic formulas , 1994, Numerical Algorithms.

[26]  Gérard Meurant,et al.  Numerical experiments in computing bounds for the norm of the error in the preconditioned conjugate gradient algorithm , 1999, Numerical Algorithms.

[27]  G. Golub,et al.  Bounds for the error in linear systems , 1979 .

[28]  Gerard L. G. Sleijpen,et al.  Accurate conjugate gradient methods for families of shifted systems , 2004 .

[29]  G. A. Baker,et al.  Padé Approximants Second Edition: Extensions of Padé approximants , 1996 .

[30]  B. Parlett.  A new look at the Lanczos algorithm for solving symmetric systems of linear equations , 1980 .

[31]  Timothy A. Davis,et al.  The university of Florida sparse matrix collection , 2011, TOMS.

[32]  R. Freund.  On conjugate gradient type methods and polynomial preconditioners for a class of complex non-hermitian matrices , 1990 .

[33]  Henk A. van der Vorst,et al.  Approximate solutions and eigenvalue bounds from Krylov subspaces , 1995, Numer. Linear Algebra Appl..

[34]  Valeria Simoncini,et al.  Analysis of Projection Methods for Rational Function Approximation to the Matrix Exponential , 2006, SIAM J. Numer. Anal..

[35]  ANDREAS FROMMER,et al.  Stopping Criteria for Rational Matrix Functions of Hermitian and Symmetric Matrices , 2008, SIAM J. Sci. Comput..

[36]  Valeria Simoncini,et al.  Error Bounds for Lanczos Approximations of Rational Functions of Matrices , 2009, Numerical Validation in Current Hardware Architectures.

[37]  Vladimir Druskin,et al.  On monotonicity of the Lanczos approximation to the matrix exponential , 2008 .

[38]  Z. Strakos,et al.  Error Estimation in Preconditioned Conjugate Gradients , 2005 .

[39]  Gérard Meurant.  The computation of bounds for the norm of the error in the conjugate gradient algorithm , 2004, Numerical Algorithms.