Comparison of Methods for Active Orbital Selection in Multiconfigurational Calculations

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Molecular Modeling

Background:

• Constructing accurate active orbital spaces is crucial for multiconfigurational wave function calculations.
• Strong electron correlation is prevalent in various molecular systems, including conjugated systems, transition states, and transition metal complexes.
• Existing methods for active space selection have limitations in accuracy, computational cost, or applicability.

Purpose of the Study:

• To compare different methods for constructing active orbital spaces in multiconfigurational calculations.
• To evaluate the performance of the unrestricted Hartree-Fock natural orbital (UNO) criterion for selecting active spaces.
• To address challenges associated with the UNO criterion, particularly in obtaining broken spin symmetry solutions.

Main Methods:

• Comparison of active space selection methods on moderately to strongly correlated ground-state molecules.
• Application of the unrestricted Hartree-Fock natural orbital (UNO) criterion.
• Development and application of an analytical method to overcome difficulties in finding broken spin symmetry unrestricted Hartree-Fock solutions.

Main Results:

• The UNO criterion successfully identifies active spaces comparable to more computationally expensive approximate full configuration interaction (CI) methods for the studied molecules.
• A novel analytical method accurately determines broken spin symmetry unrestricted Hartree-Fock solutions, resolving a key disadvantage of the UNO criterion.
• The UNO criterion considers both energetic proximity to the Fermi level and exchange interactions, providing a measure of correlation strength for orbital selection.

Conclusions:

• The UNO criterion is a simple, effective, and computationally advantageous method for selecting active spaces in multiconfigurational calculations.
• The developed analytical method enhances the reliability and applicability of the UNO criterion.
• UNO criterion offers a robust approach for estimating correlation strength, valuable for Restricted Active Space methods.