Iterative Configuration Interaction with Selection.

Area of Science:

• Quantum Chemistry
• Computational Many-Body Methods
• Electronic Structure Theory

Background:

• The iterative configuration interaction (iCI) method offers rapid convergence to full configuration interaction (FCI) for strongly correlated systems.
• Direct FCI solvers like iCI are computationally intensive due to the large number of electronic configurations.
• Identifying and managing unimportant configurations is key to mitigating computational cost.

Purpose of the Study:

• To develop a computationally efficient selected configuration interaction (sCI) approach based on iCI.
• To accurately describe static correlation using a compact variational space and residual dynamic correlation.
• To maintain spin symmetry and facilitate parallelization in the selected CI calculation.

Main Methods:

• Introduced an 'iCI with selection' (iCIPT2) scheme that iteratively selects important orbital configurations (oCFGs) and configuration state functions (CSFs).
• Maintained full spin symmetry using CSFs as the many-electron basis.
• Employed Epstein-Nesbet second-order perturbation theory (PT2) to estimate dynamic correlation after static correlation convergence.

Main Results:

• The iCIPT2 scheme effectively reduces the computational cost of FCI calculations by focusing on important configurations.
• The method demonstrated accurate treatment of both static and dynamic correlation.
• Numerical benchmarks on C2, O2, Cr2, and C6H6 validated the efficacy of the iCIPT2 approach.

Conclusions:

• The iCIPT2 method provides a computationally feasible and accurate alternative for studying strongly correlated electronic systems.
• This approach successfully balances the need for a compact variational space for static correlation with efficient treatment of dynamic correlation.
• The developed scheme offers significant advantages in terms of computational efficiency and parallelization potential.