Login / Signup

Variational reduced density matrix method in the doubly occupied configuration interaction space using three-particle N-representability conditions.

Diego R AlcobaPablo CapuzziAlvaro Rubio-GarcíaJorge DukelskyGustavo E MassaccesiOfelia B OñaAlicia TorreLuis Lain
Published in: The Journal of chemical physics (2018)
Ground-state energies and two-particle reduced density matrices (2-RDMs) corresponding to N-particle systems are computed variationally within the doubly occupied configuration interaction (DOCI) space by constraining the 2-RDM to satisfy a complete set of three-particle N-representability conditions known as three-positivity conditions. These conditions are derived and implemented in the variational calculation of the 2-RDM with standard semidefinite programming algorithms. Ground state energies and 2-RDMs are computed for N2, CO, CN-, and NO+ molecules at both equilibrium and nonequilibrium geometries as well as for pairing models at different repulsive interaction strengths. The results from the full three-positivity conditions are compared with those from the exact DOCI method and with approximated 2-RDM variational ones obtained within two-positivity and two-positivity plus a subset of three-positivity conditions, as recently reported [D. R. Alcoba et al., J. Chem. Phys. 148, 024105 (2018) and A. Rubio-García et al., J. Chem. Theory Comput. 14, 4183 (2018)]. The accuracy of these numerical determinations and their low computational cost demonstrate the usefulness of the three-particle variational constraints within the DOCI framework.
Keyphrases
  • machine learning
  • density functional theory
  • magnetic resonance imaging
  • magnetic resonance
  • deep learning
  • diffusion weighted imaging