Entanglement distillation for continuous variables in a thermal environment: Effectiveness of a non-Gaussian operation

Abstract

We study the task of distilling entanglement by a coherent superposition operation $t\stackrel{\ifmmode \hat{}\else \^{}\fi{}}{a}+r{\stackrel{\ifmmode \hat{}\else \^{}\fi{}}{a}}^{\ifmmode\dagger\else\textdagger\fi{}}$ applied to a continuous-variable state under a thermal noise. In particular, we compare the performances of two different strategies; i.e., the non-Gaussian operation $t\stackrel{\ifmmode \hat{}\else \^{}\fi{}}{a}+r{\stackrel{\ifmmode \hat{}\else \^{}\fi{}}{a}}^{\ifmmode\dagger\else\textdagger\fi{}}$ is applied before or after the noisy Gaussian channel. This is closely related to a fundamental problem of whether Gaussian or non-Gaussian entanglement can be more robust under a noisy channel and also provides a useful insight into the practical implementation of entanglement distribution for a long-distance quantum communication. We specifically look into two entanglement characteristics, the logarithmic negativity as a measure of entanglement and the teleportation fidelity as a usefulness of entanglement, for each distilled state. We find that the non-Gaussian operation after (before) the thermal noise becomes more effective in the low- (high-) temperature regime.