Improving the precision of forces in real-space pseudopotential density functional theory

Area of Science:

• Computational Chemistry
• Materials Science
• Condensed Matter Physics

Background:

• Density functional theory (DFT) is crucial for large-scale electronic structure calculations.
• The finite difference real-space pseudopotential approach faces challenges with "egg-box" errors due to coarse grid spacing.
• Minimizing computational cost while maintaining accuracy is a persistent challenge in DFT.

Purpose of the Study:

• To implement and evaluate a finite difference interpolation (FDI) scheme for reducing "egg-box" errors in real-space DFT.
• To assess the combined utility of FDI with other computational techniques for enhancing precision and efficiency.
• To demonstrate the practical application of these methods in accurate molecular simulations and dynamics.

Main Methods:

• Implementation of the FDI scheme within the PARSEC code.
• Combination of FDI with orbital-based forces, compensating charges, and modified spatial domains.
• Testing on small molecules and metallic Lithium to evaluate performance and accuracy.

Main Results:

• The FDI scheme effectively reduces "egg-box" errors by exploiting pseudopotential resolution.
• Combining FDI with other methods significantly reduces computational cost.
• High precision is maintained, supporting accurate structural, vibrational, and molecular dynamics calculations.

Conclusions:

• The developed FDI scheme offers a systematic way to mitigate "egg-box" artifacts in real-space DFT.
• The integration of FDI with complementary techniques provides a computationally efficient and accurate approach for large-scale DFT.
• This combined strategy enables reliable simulations of molecular structures, spectra, and dynamics.