AI2DFT — Auto-Differentiable Density Functional Theory
An auto-differentiable DFT code compatible with pseudo-atomic orbital basis and norm-conserving pseudopotentials.
Code Development
Modified SIESTA for DeepH-DFPT
In-house modifications of the SIESTA DFT code for deep-learning density functional perturbation theory (DeepH-DFPT).
DeepH-pack — Graph Neural Network for DFT Hamiltonians
Interface of DeepH-pack with multiple DFT codes including SIESTA, FHI-aims, ABACUS, pyscf, and GPAW. Also developed several post-processing tools for DeepH.