Fast evaluations for real-time Feynman diagrammatics in Anderson impurity models

We introduce a deterministic algorithm for evaluating Feynman diagrams arising from the strong-coupling hybridization expansion of real time steady-state Anderson impurity models. Our method is able to reach diagrammatic approximations beyond those in common use, such as the non-crossing approximation. The algorithm is numerically stable, parallelizable, offers controllable accuracy, and scales favorably towards low temperatures. We propose this method as a reliable and efficient impurity solver for real-time dynamical mean-field theory calculations.