Table 8 Performance of regression machine learning models in JARVIS-ML with JARVIS-DFT data using OptB88vdW (OPT) and TBmBJ (MBJ) with mean absolute error (MAE).
Property | Training data | MAE | MAD |
|---|---|---|---|
Formation energy (eV/atom) | 24549 | 0.12 | 0.81 |
OPT bandgap (eV) | 22404 | 0.32 | 1.05 |
MBJ bandgap (eV) | 10499 | 0.44 | 1.60 |
Bulk mod., Kv (GPa) | 10954 | 10.5 | 49.95 |
Shear mod., Gv (GPa) | 10954 | 9.5 | 23.26 |
Refr. Index(x) (OPT) | 12299 | 0.54 | 1.15 |
Refr. Index(x) (MBJ) | 6628 | 0.45 | 1.03 |
IR mode (OPT) (cm−1) | 3411 | 77.84 | 316.7 |
Max. Born eff. charge (OPT) (e) | 3411 | 0.60 | 1.48 |
Plane-wave cutoff (OPT) (eV) | 24549 | 85.0 | 370.6 |
K-point length (OPT) (Å) | 24549 | 9.09 | 22.23 |
2D-Exfoliation energy(OPT) (eV/atom) | 616 | 37.3 | 46.09 |
