Adaptive Projection Subspace Dimension for the Thick-Restart Lanczos Method

The Thick-Restart Lanczos (TRLan) method is an effective method for solving large-scale Hermitian eigenvalue problems. The performance of the method strongly depends on the dimension of the projection subspace used at each restart. In this article, we propose an objective function to quantify the effectiveness of the selection of subspace dimension, and then introduce an adaptive scheme to dynamically select the dimension to optimize the performance. We have developed an open-source software package a--TRLan to include this adaptive scheme in the TRLan method. When applied to calculate the electronic structure of quantum dots, a--TRLan runs up to 2.3x faster than a state-of-the-art preconditioned conjugate gradient eigensolver.

[1]  Jack J. Dongarra,et al.  State-of-the-art eigensolvers for electronic structure calculations of large scale nano-systems , 2008, J. Comput. Phys..

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

[3]  Andreas Stathopoulos,et al.  Nearly Optimal Preconditioned Methods for Hermitian Eigenproblems under Limited Memory. Part I: Seeking One Eigenvalue , 2007, SIAM J. Sci. Comput..

[4]  Lin-wang Wang,et al.  Pseudopotential theory of nanometer silicon quantum dots , 1997 .

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

[6]  Andreas Stathopoulos,et al.  Nearly Optimal Preconditioned Methods for Hermitian Eigenproblems Under Limited Memory. Part II: Seeking Many Eigenvalues , 2007, SIAM J. Sci. Comput..

[7]  Robert H. Halstead,et al.  Matrix Computations , 2011, Encyclopedia of Parallel Computing.

[8]  Danny C. Sorensen,et al.  Implicit Application of Polynomial Filters in a k-Step Arnoldi Method , 1992, SIAM J. Matrix Anal. Appl..

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

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

[11]  G. W. Stewart.  Adjusting the Rayleigh Quotient in Semiorthogonal Lanczos Methods , 2002, SIAM J. Sci. Comput..

[12]  Lin-wang Wang,et al.  Parallel Empirical Pseudopotential Electronic Structure Calculations for Million Atom Systems , 2000 .

[13]  Lin-Wang Wang,et al.  First principle study of core/shell structure quantum dots , 2004 .

[14]  T. Arias,et al.  Iterative minimization techniques for ab initio total energy calculations: molecular dynamics and co , 1992 .

[15]  H. Simon,et al.  TRLAN User Guide , 1999 .

[16]  Ronald B. Morgan,et al.  On restarting the Arnoldi method for large nonsymmetric eigenvalue problems , 1996, Math. Comput..

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

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

[19]  Joshua Schrier,et al.  A systematic first principles study of nanocrystal quantum-dot quantum wells , 2006 .

[20]  Lin-Wang Wang,et al.  Electronic structure of nanocrystal quantum-dot quantum wells , 2006 .

[21]  Kesheng Wu,et al.  Dynamic Thick Restarting of the Davidson, and the Implicitly Restarted Arnoldi Methods , 1998, SIAM J. Sci. Comput..

[22]  Beresford N. Parlett,et al.  The Symmetric Eigenvalue Problem (Classics in Applied Mathematics, Number 20) , 1999 .

[23]  Chao Yang,et al.  ARPACK users' guide - solution of large-scale eigenvalue problems with implicitly restarted Arnoldi methods , 1998, Software, environments, tools.

[24]  James Demmel,et al.  Applied Numerical Linear Algebra , 1997 .

[25]  Lin-wang Wang,et al.  Solving Schrödinger’s equation around a desired energy: Application to silicon quantum dots , 1994 .

[26]  Bernard Philippe,et al.  The Davidson Method , 1994, SIAM J. Sci. Comput..

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

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