|
Orbital-Free Density Functional Theory with Small Datasets and Deep Learning
K. Ryczko, S.J. Wetzel, R.G. Melko, I. Tamblyn Journal of Chemical Theory and Computation, 18, 2, 1122-1128 (2022) Orbital-free DFT is an electronic structure simulation method which describes electrons only based on their density. Unfortunately, parts of the theory are not yet fully developed (the exact functional is not known), meaning the accuracy of the method is not as high as one would hope. He were have used another, related form of electronic structure Kohn-Sham DFT to provide some training data to fill in the gap. We train a deep neural network on these data and show that this may be a viable way to improve Orbital-free DFT. |
