How Accurate are Current Band Gap Prediction Methods?
The accuracy of band gap prediction methods varies:
- DFT: While widely used, DFT often underestimates the band gap. Hybrid functionals and GW approximations can provide more accurate results but at higher computational costs. - ML Models: These can be highly accurate if trained on extensive and high-quality datasets. However, their accuracy can diminish when extrapolating to entirely new types of materials. - Empirical Methods: These are generally accurate for materials similar to those in the training dataset but may fail for novel compounds.
