ES22-Shepherd,

2022 Workshop on Recent Developments in Electronic Structure (ES22) Poster Session

View Poster

Author: James J. Shepherd, University of Iowa

TItle: A piecewise propagator for stochastic simulations of electrons with finite temperature

Abstract: Density matrix quantum Monte Carlo is a method to sample the N-body interacting density matrix for finite-temperature electronic structure. Switching the propagator to the interaction picture has been found to allow for the more effective sampling of a specific target inverse temperature, similar to how starting from a mean-field wavefunction speeds up the convergence in ground-state quantum Monte Carlo calculations. However, this has the drawback that every calculation only targets one temperature. To overcome this, we developed a piecewise propagator that switches the equations of motion at a target temperature [1]. The first phase of the calculation is performed at the same temperature and allows the energy to vary between the mean-field energy and the fully-interacting energy. The second phase of the calculation then allows temperature to change with the fully-interacting energy being calculated at every temperature point. The result is a significant improvement in the speed of our calculations. Example applications are shown for a variety of molecular Hamiltonians. [1] Van Benschoten, W. Z., Shepherd, J. J. (2022). Piecewise interaction picture density matrix quantum Monte Carlo. The Journal of Chemical Physics, 156, 184107; https://doi.org/10.1063/5.0094290

Other authors: William Z. Van Benschoten, University of Iowa