A Merger of the Spin-Flip ORMAS Approach and the MC-PDFT Method

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Accurately describing both static and dynamic electron correlation is crucial for multiconfigurational systems in quantum chemistry.
• Existing methods often struggle to efficiently handle both types of correlation simultaneously.
• The spin-flip (SF) approach and pair density functional theory (PDFT) are powerful tools for specific correlation types.

Purpose of the Study:

• To develop and validate a novel computational method, SF-ORMAS-PDFT, that combines the strengths of SF-CI and MC-PDFT.
• To investigate the efficiency and accuracy of SF-ORMAS-PDFT, particularly when augmented with virtual valence orbitals (VVOs).
• To assess the method's performance in calculating vertical excitation energies and rotational barriers.

Main Methods:

• The spin-flip occupation restricted multiple active space-CI (SF-ORMAS-CI) method was integrated with on-top pair density functional theory (MC-PDFT).
• The combined SF-ORMAS-PDFT approach utilizes a spin-flip strategy for static correlation and a translation scheme for GGA density functionals to capture dynamic correlation.
• Calculations were performed using both the full virtual space and a reduced subspace of virtual valence orbitals (VVOs) for comparison.

Main Results:

• The SF-ORMAS-PDFT method demonstrated comparable accuracy for vertical excitation energies of organic molecules whether using VVOs or the full virtual space.
• Calculations of ethylene's rotational barrier yielded results (65.5-65.9 kcal/mol) that closely match the experimental value (65 kcal/mol).
• The use of VVOs in SF-ORMAS-PDFT led to significant computational speedups compared to using the complete virtual orbital space.

Conclusions:

• SF-ORMAS-PDFT is a robust computational method capable of accurately treating both static and dynamic electron correlation in complex molecular systems.
• The integration of VVOs provides a computationally efficient alternative to using the full virtual space without compromising predictive accuracy.
• This method shows great promise for accurate and efficient calculations of electronic properties, including excitation energies and reaction barriers.