Convergence and machine learning predictions of Monkhorst-Pack k-points and plane-wave cut-off in high-throughput DFT...

Area of Science:

• Computational Materials Science
• Solid State Physics
• Quantum Chemistry

Background:

• Density Functional Theory (DFT) calculations require careful convergence of k-points and plane wave cut-off for accuracy.
• Manual convergence is time-consuming and can lead to inconsistencies across different materials and studies.

Purpose of the Study:

• To develop an automated procedure for determining optimal k-point and plane wave cut-off parameters in DFT.
• To establish relationships between convergence parameters and material properties.
• To create machine learning models for predicting these parameters for new materials.

Main Methods:

• Developed an automatic convergence framework for k-points and plane wave cut-off in DFT.
• Applied the framework to over 30,000 materials, using energy per cell (EPC) and energy per atom (EPA) convergence criteria.
• Analyzed correlations between convergence parameters and material characteristics (e.g., density, band structure, crystal system).
• Investigated the influence of exchange-correlation functionals on convergence parameters.
• Trained machine learning models to predict k-point density and plane wave cut-off.

Main Results:

• Established relationships between k-point density, plane wave cut-off, and various material properties.
• Identified specific material types that necessitate more rigorous convergence checks.
• Statistically analyzed the impact of exchange-correlation functionals on convergence parameters.
• Developed predictive machine learning models for generalized materials.

Conclusions:

• The automated convergence procedure and predictive models offer a significant advancement for efficient and reliable DFT calculations.
• The findings provide a valuable starting point for users seeking converged DFT results, reducing computational cost and time.
• The developed code and data are publicly available to facilitate broader adoption and further research.