Automated Selection of Active Orbital Spaces

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Theoretical Chemistry

Background:

• Manual active orbital selection is a significant hurdle in quantum-chemical multi-configuration methods.
• Expertise and experience are required, limiting the application of ab initio methods.
• Incorrect orbital selection can lead to qualitatively inaccurate results, hindering systematic improvement.

Purpose of the Study:

• To develop an automated method for active orbital space selection.
• To overcome the limitations of manual selection in multi-configuration methods.
• To enhance the applicability and reliability of ab initio quantum chemistry.

Main Methods:

• Utilizing the iterative nature of the density matrix renormalization group (DMRG).
• Exploiting DMRG's capacity to include a large number of orbitals (up to ~100) in the active space.
• Implementing a systematic assessment and selection protocol for active orbitals.

Main Results:

• Demonstrated a systematic approach for active orbital selection.
• Showcased the potential of DMRG for automated orbital space definition.
• Enabled multi-configuration models to function as 'black box' approaches.

Conclusions:

• Automated active orbital selection significantly improves the usability of multi-configuration methods.
• The proposed method enhances the reliability and accuracy of ab initio calculations.
• This advancement makes advanced quantum chemistry more accessible to researchers.