Experiences with a Lanczos Eigensolver in High-Precision Arithmetic

We investigate the behavior of the Lanczos process when it is used to find all the eigenvalues of large sparse symmetric matrices. We study the convergence of classical Lanczos (i.e., without re-orthogonalization) to the point where there is a cluster of Ritz values around each eigenvalue of the input matrix A. At that point, convergence to all the eigenvalues can be ascertained if A has no multiple eigenvalues. To eliminate multiple eigenvalues, we disperse them by adding to A a random matrix with a small norm; using high-precision arithmetic, we can perturb the eigenvalues and still produce accurate double-precision results. Our experiments indicate that the speed with which Ritz clusters form depends on the local density of eigenvalues and on the unit roundoff, which implies that we can accelerate convergence by using high-precision arithmetic in computations involving the Lanczos iterates.

[1]  Y. Saad.  Numerical Methods for Large Eigenvalue Problems , 2011 .

[2]  L. Knizhnerman.  The quality of approximations to an isolated eigenvalue and the distribution of “Ritz numbers” in the sample Lanczos procedure , 1995 .

[3]  Shang-Hua Teng,et al.  Smoothed analysis: an attempt to explain the behavior of algorithms in practice , 2009, CACM.

[4]  Kesheng Wu,et al.  Thick-Restart Lanczos Method for Large Symmetric Eigenvalue Problems , 2000, SIAM J. Matrix Anal. Appl..

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

[6]  Gene H. Golub,et al.  Matrix computations (3rd ed.) , 1996 .

[7]  J. Grcar.  Analyses of the lanczos algorithm and of the approximation problem in richardson's method , 1981 .

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

[9]  Jack Dongarra,et al.  LAPACK Users' Guide, 3rd ed. , 1999 .

[10]  G. Meurant.  The Lanczos and Conjugate Gradient Algorithms: From Theory to Finite Precision Computations , 2006 .

[11]  G. Stewart.  Matrix Algorithms, Volume II: Eigensystems , 2001 .

[12]  B. Parlett.  The Symmetric Eigenvalue Problem , 1981 .

[13]  Zdeněk Strakoš,et al.  Open questions in the convergence analysis of the lanczos process for the real symmetric eigenvalue problem , 1992 .

[14]  D. Calvetti,et al.  AN IMPLICITLY RESTARTED LANCZOS METHOD FOR LARGE SYMMETRIC EIGENVALUE PROBLEMS , 1994 .

[15]  J. Cullum,et al.  Lanczos Algorithms for Large Symmetric Eigenvalue Computations, Vol. 1 , 2002 .

[16]  H. Simon.  The Lanczos algorithm with partial reorthogonalization , 1984 .

[17]  B. Parlett,et al.  The Lanczos algorithm with selective orthogonalization , 1979 .

[18]  C. Lanczos.  An iteration method for the solution of the eigenvalue problem of linear differential and integral operators , 1950 .

[19]  D. Sorensen.  Numerical methods for large eigenvalue problems , 2002, Acta Numerica.

[20]  Study of Cullum's and Willoughby's Lanczos method for Wilson fermions , 1996 .

[21]  Wolfgang Wülling.  On Stabilization and Convergence of Clustered Ritz Values in the Lanczos Method , 2005, SIAM J. Matrix Anal. Appl..

[22]  H. Simon.  Analysis of the symmetric Lanczos algorithm with reorthogonalization methods , 1984 .

[23]  W. Wülling,et al.  The Stabilization of Weights in the Lanczos and Conjugate Gradient Method , 2005 .

[24]  Inderjit S. Dhillon,et al.  The design and implementation of the MRRR algorithm , 2006, TOMS.

[25]  J. H. Wilkinson,et al.  Reliable Numerical Computation. , 1992 .

[26]  E. B. Davies.  Approximate Diagonalization , 2007, SIAM J. Matrix Anal. Appl..

[27]  L. Knizhnerman,et al.  Error bounds in the simple Lanczos procedure for computing functions of symmetric matrices and eigenvalues , 1992 .

[28]  Inderjit S. Dhillon,et al.  Glued Matrices and the MRRR Algorithm , 2005, SIAM J. Sci. Comput..

[29]  B. Parlett,et al.  On estimating the largest eigenvalue with the Lanczos algorithm , 1982 .

[30]  H. V. D. Vorst,et al.  The convergence behavior of ritz values in the presence of close eigenvalues , 1987 .

[31]  J. Cullum,et al.  Lanczos algorithms for large symmetric eigenvalue computations , 1985 .

[32]  A. Greenbaum.  Behavior of slightly perturbed Lanczos and conjugate-gradient recurrences , 1989 .

[33]  J. Edwards,et al.  Use of the Lanczos Method for Finding Complete Sets of Eigenvalues of Large Sparse Symmetric Matrices , 1979 .

[34]  J. Reid,et al.  Tracking the Progress of the Lanczos Algorithm for Large Symmetric Eigenproblems , 1981 .