MACE-MatPES-r2SCAN-0
Foundation model trained on the MatPES database using the r2SCAN meta-GGA functional. Provides improved accuracy over GGA functionals for many materials properties including better treatment of strongly correlated systems. Covers 89 elements.
| Version | 1.0.0 |
|---|---|
| Creator | ACEsuit |
| Delivery | executable |
| Requirements | ram: 8GB; disk: 500MB |
Features
- Single-point energy calculations
- Force predictions
- Stress tensor calculations
- Geometry optimization
- Periodic boundary conditions
- r2SCAN meta-GGA accuracy
- Better for strongly correlated systems
Keywords
matpesr2SCANmeta-GGAmaterials-scienceVASPstrongly-correlated
Citation
Batatia, I., et al. (2024). MACE-MatPES: Machine Learning Potentials for Materials at the PES Level.