Variational Active Space Selection with Multiconfiguration Pair-Density Functional Theory.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Electronic Structure Theory

Background:

• Selecting active orbitals for strongly correlated electronic states is challenging and molecule-dependent.
• Existing methods lack universal applicability for automated active space selection.
• Accurate modeling of electronic states is crucial for understanding molecular properties.

Purpose of the Study:

• To introduce and validate a novel discrete variational selection (DVS) approach for automated active space selection.
• To assess the effectiveness of DVS using different energy functionals for selecting trial wave functions.
• To apply the optimized DVS approach to model vertical excitations in organic and inorganic molecules.

Main Methods:

• Generated multiple trial wave functions from systematically constructed active spaces.
• Employed variational selection between wave functions using energies from translated PBE (tPBE) functional within multiconfiguration pair-density functional theory (MC-PDFT).
• Applied the DVS-tPBE approach to 207 vertical excitations in small-to-medium-sized molecules from the QUESTDB database.

Main Results:

• DVS proved ineffective with density matrix renormalization group (DMRG) or complete active space self-consistent field (CASSCF) energies.
• DVS-tPBE demonstrated good performance, yielding a mean unsigned error of 0.17 eV for vertical excitations using hybrid MC-PDFT.
• The DVS-tPBE method achieved benchmark accuracy without filtering poor active spaces or requiring further orbital optimization.

Conclusions:

• The discrete variational selection (DVS) approach, particularly DVS-tPBE, offers a robust and effective method for active space selection.
• This method accurately models state-averaged DMRG wave functions and compares favorably with previous SA-CASSCF results.
• DVS-tPBE advances automated active space selection, enhancing the modeling of strongly correlated electronic states in computational chemistry.