Error propagation in the fully self-consistent stochastic second-order Green's function method

Blair Winograd,Emanuel Gull,Dominika Zgid

DOI: https://doi.org/10.48550/arXiv.2009.10048

2020-09-22

Abstract:We present an implementation of a fully self-consistent finite temperature second order Green's function perturbation theory (GF2) within the diagrammatic Monte Carlo framework. In contrast to the previous implementations of stochastic GF2 ({\it J. Chem. Phys.},{\bf 151}, 044144 (2019)), the current self-consistent stochastic GF2 does not introduce a systematic bias of the resulting electronic energies. Instead, the introduced implementation accounts for the stochastic errors appearing during the solution of the Dyson equation. We present an extensive discussion of the error handling necessary in a self-consistent procedure resulting in dressed Green's function lines. We test our method on a series of simple molecular examples.

Chemical Physics,Computational Physics