Second-order perturbation corrections to singles and doubles coupled-cluster methods: General theory and application to the valence optimized doubles model
- Department of Chemistry, University of California, and Chemical Sciences Division, Lawrence Berkeley Laboratory, Berkeley, California 94720-1460 (United States)
- Department of Chemistry, University of Southern California, Los Angeles, California 90089 (United States)
We present a general perturbative method for correcting a singles and doubles coupled-cluster energy. The coupled-cluster wave function is used to define a similarity-transformed Hamiltonian, which is partitioned into a zeroth-order part that the reference problem solves exactly plus a first-order perturbation. Standard perturbation theory through second-order provides the leading correction. Applied to the valence optimized doubles (VOD) approximation to the full-valence complete active space self-consistent field method, the second-order correction, which we call (2), captures dynamical correlation effects through external single, double, and semi-internal triple and quadruple substitutions. A factorization approximation reduces the cost of the quadruple substitutions to only sixth order in the size of the molecule. A series of numerical tests are presented showing that VOD(2) is stable and well-behaved provided that the VOD reference is also stable. The second-order correction is also general to standard unwindowed coupled-cluster energies such as the coupled-cluster singles and doubles (CCSD) method itself, and the equations presented here fully define the corresponding CCSD(2) energy. (c) 2000 American Institute of Physics.
- OSTI ID:
- 20217443
- Journal Information:
- Journal of Chemical Physics, Vol. 113, Issue 9; Other Information: PBD: 1 Sep 2000; ISSN 0021-9606
- Country of Publication:
- United States
- Language:
- English
Similar Records
A coupled cluster study of the spectroscopic properties and electric dipole moment functions of nitrous sulfide
Can coupled cluster singles and doubles be approximated by a valence active space model?