Extrapolating the Arnoldi Algorithm To Improve Eigenvector Convergence

We consider extrapolation of the Arnoldi algorithm to accelerate computation of the dominant eigenvalue/eigenvector pair. The basic algorithm uses sequences of Krylov vectors to form a small eigenproblem which is solved exactly. The two dominant eigenvectors output from consecutive Arnoldi steps are then recombined to form an extrapolated iterate, and this accelerated iterate is used to restart the next Arnoldi process. We present numerical results testing the algorithm on a variety of cases and find on most examples it substantially improves the performance of restarted Arnoldi. The extrapolation is a simple post-processing step which has minimal computational cost.

[1]  Sara N. Pollock.  Using small eigenproblems to accelerate power method iterations , 2021 .

[2]  Vicente Hernández,et al.  SLEPc: A scalable and flexible toolkit for the solution of eigenvalue problems , 2005, TOMS.

[3]  L. R. Scott.  Kinetic Energy Flow Instability With Application to Couette Flow , 2020 .

[4]  M. Rozložník.  Numerics of Gram-Schmidt orthogonalization , 2007 .

[5]  Ioannis Mitliagkas,et al.  Accelerated Stochastic Power Iteration , 2017, AISTATS.

[6]  Y. Saad.  On the Rates of Convergence of the Lanczos and the Block-Lanczos Methods , 1980 .

[7]  Peter Benner,et al.  Locally Optimal Block Preconditioned Conjugate Gradient Method for Hierarchical Matrices , 2011 .

[8]  Nilima Nigam,et al.  A simple extrapolation method for clustered eigenvalues , 2020, ArXiv.

[9]  Miroslav Rozlozník,et al.  Modified Gram-Schmidt (MGS), Least Squares, and Backward Stability of MGS-GMRES , 2006, SIAM J. Matrix Anal. Appl..

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

[11]  Danny C. Sorensen,et al.  Deflation Techniques for an Implicitly Restarted Arnoldi Iteration , 1996, SIAM J. Matrix Anal. Appl..

[12]  L. Ridgway Scott,et al.  Van der Waals Interactions Between Two Hydrogen Atoms: The Slater-Kirkwood Method Revisited , 2018, SIAM J. Math. Anal..

[13]  Andrew Knyazev.  Recent implementations, applications, and extensions of the Locally Optimal Block Preconditioned Conjugate Gradient method (LOBPCG) , 2017, ArXiv.

[14]  Y. Saad.  Variations on Arnoldi's method for computing eigenelements of large unsymmetric matrices , 1980 .

[15]  Andrew V. Knyazev,et al.  Toward the Optimal Preconditioned Eigensolver: Locally Optimal Block Preconditioned Conjugate Gradient Method , 2001, SIAM J. Sci. Comput..

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