Automatic Selection of Active Orbitals from Generalized Valence Bond Orbitals

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Materials Science

Background:

• Accurate multireference (MR) calculations for strongly correlated systems require precise selection of active orbitals.
• Current trial-and-error methods for active orbital selection are often tedious, unreliable, and a bottleneck for large systems.

Purpose of the Study:

• To develop an automated, reliable method for selecting active orbitals in MR calculations.
• To improve the efficiency and accuracy of constructing active spaces for large-scale electronic structure computations.

Main Methods:

• Proposed an automated active orbital selection based on natural orbital occupation numbers.
• Utilized black-box generalized valence bond (GVB) calculations for orbital selection.
• Tested the method on various systems including diatomic molecules, conjugated molecules, and transition-metal complexes.

Main Results:

• The automated method successfully identified high-quality initial active orbitals.
• Results were validated against complete active space self-consistent field (CASSCF) and density matrix renormalization group (DMRG)-CASSCF references.
• The proposed method demonstrated accuracy comparable to or exceeding other inexpensive strategies.

Conclusions:

• The automated active orbital selection method significantly benefits large active space ground-state MR calculations.
• This approach is expected to facilitate large-scale computations, such as those using DMRG.
• The method provides a robust and efficient alternative to manual active space construction.