GradDFT is a JAX-based library enabling the differentiable design and experimentation of exchange-correlation functionals using machine learning techniques.
Without much change, Grad DFT can now handle calculations with solids provided we use the gamma point only to sample the first BZ. Full BZ sampling will come later.
Without much change, Grad DFT can now handle calculations with solids provided we use the gamma point only to sample the first BZ. Full BZ sampling will come later.