High-level coupled-cluster energetics by merging moment expansions with selected configuration interaction

Area of Science:

• Quantum chemistry
• Computational physics
• Theoretical chemistry

Background:

• Coupled-cluster (CC) theory is essential for accurate electronic structure calculations.
• Previous work utilized semi-stochastic methods to improve CC energetics.
• Identifying dominant excited determinants is crucial for high-level CC calculations.

Purpose of the Study:

• To develop a novel, more efficient CC(P; Q) theory.
• To replace stochastic Quantum Monte Carlo with a selected configuration interaction approach.
• To accurately compute electronic energies for challenging molecular systems.

Main Methods:

• The proposed method integrates the selected configuration interaction approach using the perturbative selection made iteratively (CIPSI) algorithm into the CC(P; Q) framework.
• Stochastic Quantum Monte Carlo propagations are replaced by CIPSI for identifying important determinants.
• The methodology is tested on molecular examples like F2 dissociation and cyclobutadiene automerization.

Main Results:

• The CIPSI-driven CC(P; Q) methodology successfully recovers electronic energies comparable to full singles, doubles, and triples coupled-cluster calculations.
• Accurate energetics were obtained using compact configuration interaction wave functions derived from efficient Hamiltonian diagonalizations.
• The method demonstrates its efficacy on challenging chemical problems.

Conclusions:

• The new CIPSI-driven CC(P; Q) method provides a computationally efficient and accurate way to obtain high-level coupled-cluster energetics.
• This approach offers a viable alternative to stochastic methods for electronic structure calculations.
• The study highlights the power of combining selected configuration interaction with coupled-cluster theory for complex molecular systems.