Systematic Study of Selected Diagonalization Methods for Configuration Interaction Matrices

Several modifications to the Davidson algorithm are systematically explored to establish their performance for an assortment of configuration interaction (CI) computations. The combination of a generalized Davidson method, a periodic two‐vector subspace collapse, and a blocked Davidson approach for multiple roots is determined to retain the convergence characteristics of the full subspace method. This approach permits the efficient computation of wave functions for large‐scale CI matrices by eliminating the need to ever store more than three expansion vectors (bi) and associated matrix‐vector products (σi), thereby dramatically reducing the I/O requirements relative to the full subspace scheme. The minimal‐storage, single‐vector method of Olsen is found to be a reasonable alternative for obtaining energies of well‐behaved systems to within μEh accuracy, although it typically requires around 50% more iterations and at times is too inefficient to yield high accuracy (ca. 10−10 Eh) for very large CI problems. Several approximations to the diagonal elements of the CI Hamiltonian matrix are found to allow simple on‐the‐fly computation of the preconditioning matrix, to maintain the spin symmetry of the determinant‐based wave function, and to preserve the convergence characteristics of the diagonalization procedure. © 2001 John Wiley & Sons, Inc. J Comput Chem 22: 1574–1589, 2001

[1]  Per E. M. Siegbahn,et al.  Generalizations of the direct CI method based on the graphical unitary group approach. II. Single and double replacements from any set of reference configurations , 1980 .

[2]  A. Stathopoulos,et al.  Solution of large eigenvalue problems in electronic structure calculations , 1996 .

[3]  Ernest R. Davidson,et al.  Matrix Eigenvector Methods , 1983 .

[4]  Stefano Evangelisti,et al.  Computation and analysis of the full configuration interaction wave function of some simple systems , 1993 .

[5]  T. H. Dunning.  Gaussian basis sets for use in correlated molecular calculations. I. The atoms boron through neon and hydrogen , 1989 .

[6]  H. Schaefer,et al.  Structures and vibrational frequencies in the full configuration interaction limit: Predictions for four electronic states of methylene using a triple-zeta plus double polarization (TZ2P) basis , 1998 .

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

[8]  William H. Press,et al.  Numerical recipes in C , 2002 .

[9]  Ernest R. Davidson,et al.  Super-matrix methods , 1989 .

[10]  Nicholas C. Handy,et al.  Multi-root configuration interaction calculations , 1980 .

[11]  E. Davidson,et al.  Improved Algorithms for the Lowest Few Eigenvalues and Associated Eigenvectors of Large Matrices , 1992 .

[12]  Roberto Ansaloni,et al.  A full CI algorithm on the CRAY T3D. Application to the NH3 molecule , 1995 .

[13]  J. H. van Lenthe,et al.  A space‐saving modification of Davidson's eigenvector algorithm , 1990 .

[14]  Stefano Evangelisti,et al.  Benchmark full-CI calculation on C2H2: comparison with (SC)2-CI and other truncated-CI approaches , 1998 .

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

[16]  Stefano Evangelisti,et al.  A vector and parallel full configuration interaction algorithm , 1993 .

[17]  Per E. M. Siegbahn,et al.  A new direct CI method for large CI expansions in a small orbital space , 1984 .

[18]  Stefano Evangelisti,et al.  Full configuration interaction algorithm on a massively parallel architecture: Direct‐list implementation , 1998 .

[19]  Isaiah Shavitt,et al.  The history and evolution of configuration interaction , 1998 .

[20]  Roberto Ansaloni,et al.  A parallel Full-CI algorithm ✩ , 2000 .

[21]  E. Davidson.  The iterative calculation of a few of the lowest eigenvalues and corresponding eigenvectors of large real-symmetric matrices , 1975 .

[22]  Matthew L. Leininger,et al.  Is Mo/ller–Plesset perturbation theory a convergent ab initio method? , 2000 .

[23]  A. Zunger,et al.  A new method for diagonalising large matrices , 1985 .

[24]  J. Olsen,et al.  Passing the one-billion limit in full configuration-interaction (FCI) calculations , 1990 .

[25]  Per-Åke Malmqvist.  Mathematical Tools in Quantum Chemistry , 1992 .

[26]  L. Gagliardi,et al.  Full configuration interaction calculations on Be2 , 1994 .

[27]  Jeppe Olsen,et al.  Determinant based configuration interaction algorithms for complete and restricted configuration interaction spaces , 1988 .

[28]  G. Diercksen,et al.  Methods in Computational Molecular Physics , 1983 .

[29]  J. M. Bofill,et al.  Some remarks on the use of the three-term recurrence method in the configuration interaction eigenvalue problem , 1994 .

[30]  H. A. V. D. Vorsty.  University Utrecht a Generalized Jacobi-davidson Iteration Method for Linear Eigenvalue Problems a Generalized Jacobi-davidson Iteration Method for Linear Eigenvalue Problems , 1994 .

[31]  H. Schaefer,et al.  Full Configuration Interaction Energies, Geometries, and Quartic Force Fields of the Nitrenium Ion , 1998 .

[32]  Holger Dachsel,et al.  An efficient data compression method for the Davidson subspace diagonalization scheme , 1995 .

[33]  Peter J. Knowles,et al.  A new determinant-based full configuration interaction method , 1984 .

[34]  Matthew L. Leininger,et al.  Benchmark configuration interaction spectroscopic constants for X 1Σg+ C2 and X 1Σ+ CN+ , 1998 .

[35]  T. H. Dunning.  Gaussian Basis Functions for Use in Molecular Calculations. III. Contraction of (10s6p) Atomic Basis Sets for the First‐Row Atoms , 1970 .

[36]  B. Roos,et al.  A new method for large-scale Cl calculations , 1972 .

[37]  H.J.J. Van Dam,et al.  An improvement of Davidson's iteration method: Applications to MRCI and MRCEPA calculations , 1996 .

[38]  A. Mitrushenkov.  Passing the several billions limit in FCI calculations on a mini-computer , 1994 .

[39]  Ron Shepard,et al.  A data compression method applicable to first‐order convergent iterative procedures , 1990 .

[40]  Stefano Evangelisti,et al.  A full-configuration benchmark for the N2 molecule , 1999 .

[41]  K. Hirao,et al.  A generalization of the Davidson's method to large nonsymmetric eigenvalue problems , 1982 .

[42]  R. Morgan,et al.  Generalizations of Davidson's method for computing eigenvalues of sparse symmetric matrices , 1986 .

[43]  E. Davidson,et al.  A modified Ho for Epstein-Nesbet Rayleigh-Schrōdinger pertubation theory , 1979 .

[44]  Roberto Ansaloni,et al.  A one billion determinant full CI benchmark on the Cray T3D , 1996 .

[45]  M. G. Feler.  Calculation of eigenvectors of large matrices , 1974 .

[46]  A. Császár,et al.  The barrier to linearity of hydrogen sulphide , 2000 .

[47]  B. Roos,et al.  Lecture notes in quantum chemistry , 1992 .

[48]  Ruben Pauncz,et al.  Spin Eigenfunctions: Construction and Use , 1979 .