Variational Localized Active Space Self-Consistent Field Method.

Area of Science:

• Computational chemistry
• Quantum chemistry
• Theoretical chemistry

Background:

• Fragmentation methods for multireference wave functions are crucial for accurate ab initio calculations on large systems.
• Reproducibility and transferability in fragmentation schemes depend on minimal dependence on initial guesses and user-defined parameters.
• Ensuring energy obeys a variational principle enhances the robustness of fragmentation schemes.

Purpose of the Study:

• To extend the localized active space self-consistent field (LASSCF) method to be fully variational.
• To develop a new method, vLASSCF, that minimizes energy with respect to all orbital rotations for improved robustness and reproducibility.
• To analyze the computational cost scaling of vLASSCF compared to standard CASSCF.

Main Methods:

• Extended the theory of the localized active space self-consistent field (LASSCF) method.
• Developed the variational localized active space self-consistent field (vLASSCF) method, ensuring full minimization of energy with respect to orbital rotations.
• Analyzed storage and operation cost scaling and performed calculations on test systems.

Main Results:

• The vLASSCF method significantly improves the robustness and reproducibility of the localized active space wave function.
• vLASSCF is energetically equivalent to complete active space self-consistent field (CASSCF) in the limit of one active subspace.
• vLASSCF enhances the reliability of energy differences, enabling more meaningful analysis of potential energy curves for dissociating molecules.
• All forms of LASSCF exhibit lower operation cost scaling than the orbital optimization part of CASSCF.

Conclusions:

• The vLASSCF method offers a more robust and reproducible approach to ab initio wave function calculations using fragmentation.
• vLASSCF provides more reliable energy differences, facilitating detailed analysis of molecular behavior, especially during dissociation.
• The computational efficiency of LASSCF methods makes them attractive for large-scale quantum chemistry applications.