Selected configuration interaction dressed by perturbation

Area of Science:

• Computational chemistry
• Quantum chemistry
• Electronic structure theory

Background:

• Selected configuration interaction (sCI) methods approximate full configuration interaction (FCI) energies.
• These methods use a small fraction of the FCI determinant space for efficiency.
• Perturbative corrections enhance the accuracy of sCI energy calculations.

Purpose of the Study:

• Introduce novel state-specific and multi-state sCI methods.
• Develop methods based on the iterative perturbative selection (CIPSI) algorithm.
• Address computational bottlenecks in large-scale electronic structure problems.

Main Methods:

• Utilize the configuration interaction using a perturbative selection made iteratively (CIPSI) algorithm.
• Construct an effective Hamiltonian by revising the reference space.
• Employ the shifted-Bk philosophy for Hamiltonian construction.
• Implement a low-rank factorization of the dressing matrix for the multi-state approach.

Main Results:

• Demonstrate near FCI quality energies with sCI methods.
• Successfully developed and applied state-specific and multi-state sCI algorithms.
• The multi-state algorithm overcomes storage limitations of the effective Hamiltonian.
• Illustrative examples validate the performance of the new methods.

Conclusions:

• The new sCI methods provide accurate energies with reduced computational cost.
• The multi-state approach effectively handles storage challenges in large CI spaces.
• These advancements offer efficient solutions for complex quantum chemistry problems.