Intriguing Accuracies of the Exponential Wave Function Expansions Exploiting Finite Two-Body Correlation Operators in Calculations for Many-Electron Systems
|Intriguing Accuracies of the Exponential Wave Function Expansions Exploiting Finite Two-Body Correlation Operators in Calculations for Many-Electron Systems
|Year of Publication
|Fan, P-D, Piecuch, P
|Journal of Molecular Structure: THEOCHEM
|xact many-electron wave functions; Generalized coupled-cluster methods; Two-body correlation operators; Nooijen's conjecture; Variational calculations; Multi-determinantal reference states; Excited states
Following the ideas laid down by Nooijen and Nakatsuji, several authors have considered an intriguing possibility of representing the exact many-electron wave functions by the exponential cluster expansions involving two-body correlation operators. In particular, inspired by the symmetric form of the Horn–Weinstein exact energy formula, and exploiting the variational principle and numerical analysis, we have demonstrated that one can obtain nearly exact ground-state wave functions for a few many-electron systems using the exponential cluster expansion involving a finite two-body operator acting on the Hartree–Fock determinant [P. Piecuch et al., Phys. Rev. Lett. 90 (2003) 113001]. After summarizing these earlier findings and making some additional comments on the nature of the exponential cluster expansions involving two-body correlation operators, we examine the following issues: (i) the improvements in the accuracy and convergence toward the full configuration interaction (CI) limit offered by cluster operators containing two-body as well as one-body components, (ii) the improvements in the accuracy resulting from the use of multi-determinantal reference states, and (iii) the potential accuracy of the exponential wave function expansions involving finite one- and two-body cluster operators in excited-state calculations. All calculations are performed for an eight electron model system, which is simple enough to allow for the exact, full CI, and other electronic structure calculations, which has fewer independent parameters in the Hamiltonian than the dimension of the corresponding full CI problem, and which enables one to examine ground and excited states with a varying degree of configurational quasi-degeneracy by simple changes in the corresponding nuclear geometry.