Process tensor simulations of non-Markovian open quantum systems with large uncoupled subspaces

Abstract

Process tensors in matrix product operator representation (PT-MPOs) are a powerful tool to calculate the temporal evolution and multi-time correlations of open quantum systems. While much progress has been made in making the calculation of process tensors for larger system dimensions possible, the numerical cost of the subsequent propagation of the system typically scales as $\mathcal{O}(d^6)$, which renders PT-MPO methods prohibitively expensive for system dimensions above several dozens. Here, we show how PT-MPO methods can be used efficiently when only some of the system states couple to the environment, and how combination with a Taylor expansion method allows for a reduction of the numerical complexity down to $\mathcal{O}(d^3)$. We show that these methods enable efficient simulations of systems with several hundreds of states. This is particularly relevant for solid-state cavity-QED systems with multiple modes, or when calculating frequency-filtered correlation functions by coupling sensor systems to the system of interest.