<br/>Stochastic density functional theory (sDFT) succeeds in achieving linear and even sub-linear scaling enabling the studies of very large systems. We introduce a new stochastic density functional scheme using high temperature embedding (ht-sDFT) that reduces the statistical deviation of observables in sDFT. Specifically, we embed an energy-filter function at a target temperature with a higher temperature filter. This leads to a rewriting of the electronic density as a large high-temperature density combined with smaller correction densities. The large term is faster to evaluate than the correction term(s) and is sampled with more orbitals than the correction. This scheme allows for more orbitals to be used at the same level of computational effort as sDFT reducing the deviation of observables. We demonstrate ht-sDFT on silicon supercells. We show that the scheme reduces the deviation in the total energy but not in the forces, in contrast to an energy-windows scheme which reduces the error in the forces but not in the energy.

High temperature embedding stochastic density functional theory

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.