Table 4 NequIP E/F MAE/RMSE for LiPS and Li4P2O7 for different data set sizes in units of [meV/Å] and [meV/atom].
From: E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials
System | Data set size | MAE | RMSE | |
|---|---|---|---|---|
LiPS | 10 | Energy | 2.03 | 2.54 |
Forces | 97.8 | 132.4 | ||
LiPS | 100 | Energy | 0.44 | 0.56 |
Forces | 25.8 | 35.0 | ||
LiPS | 1000 | Energy | 0.12 | 0.15 |
Forces | 7.7 | 10.8 | ||
LiPS | 2500 | Energy | 0.08 | 0.10 |
Forces | 4.7 | 6.5 | ||
Li4P2O7, melt | 1000 | Energy | 0.4 | 0.8 |
Forces | 34.0 | 59.5 | ||
Li4P2O7, quench | 1000 | Energy | 0.5 | 0.5 |
Forces | 21.3 | 34.9 |
