Density-functional theory vs density-functional fits: The best of both

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Materials Science

Background:

• Kohn-Sham density functionals are crucial for electronic structure calculations.
• Previous work compared theory-based functionals with power-series fits, showing competitive performance.
• The potential for further improvement by integrating power-series terms was unexplored.

Purpose of the Study:

• To investigate the performance enhancement of theory-based density functionals by adding power-series terms.
• To introduce a novel series expansion variable capturing more local physics.
• To develop an improved hybrid Kohn-Sham density functional.

Main Methods:

• Developed a new series expansion variable with enhanced local physics.
• Performed one-dimensional fits to the GMTKN55 database using a theory-based functional augmented with power-series exchange and correlation terms.
• Optimized the number of power-series terms (12) and parent terms (6).

Main Results:

• Achieved the lowest WTMAD2 error on the GMTKN55 database reported to date for a hybrid Kohn-Sham functional.
• The new functional, B22plus, demonstrated superior performance compared to previous models.
• The novel series expansion variable proved effective in improving functional accuracy.

Conclusions:

• The integration of power-series terms significantly enhances the performance of physically motivated density functionals.
• B22plus represents a substantial advancement in the accuracy of hybrid Kohn-Sham density functionals.
• The developed series expansion variable offers a promising direction for future functional development.