Real-space Hubbard-corrected density functional theory

Area of Science:

• Computational materials science
• Quantum chemistry
• Solid-state physics

Background:

• Density functional theory (DFT) is a powerful quantum mechanical modeling method.
• Accurate DFT calculations often require significant computational resources, especially for strongly correlated materials.
• Real-space methods offer an alternative to traditional reciprocal-space (plane-wave) approaches.

Purpose of the Study:

• To develop an accurate and efficient real-space framework for Hubbard-corrected DFT.
• To implement and parallelize the framework for large-scale computations.
• To investigate the impact of exchange-correlation inconsistency and optimize Hubbard parameters.

Main Methods:

• Formulation of energy, atomic forces, and stress tensor for real-space finite-difference discretization.
• Development of a large-scale parallel implementation of the framework.
• Verification against established plane-wave methods and application to TiO2 polymorphs.

Main Results:

• The real-space Hubbard-corrected DFT framework provides accurate results comparable to plane-wave methods.
• The implementation demonstrates high efficiency and scalability, outperforming plane-wave codes by over an order of magnitude in time to solution.
• Performance advantages increase with system size and processor count.
• The framework was used to study exchange-correlation inconsistency and optimize Hubbard parameters for TiO2.

Conclusions:

• The developed real-space Hubbard-corrected DFT framework is accurate, efficient, and highly scalable.
• This computational tool significantly accelerates materials simulations, particularly for large and complex systems.
• The approach facilitates advanced investigations into electronic structure and materials properties, aiding in the discovery of new materials.