Learning Non-Markovian Noise Parameters Dynamically with Ensemble Optimal Control

ABSTRACT

We considered the estimation of non-Markovian noise spectrum parameters as a dynamical process. An operational challenge of this task is to determine an optimal time to make the measurement such that the system has evolved long enough to acquire sufficient information on the quantum bath, but not too long for the information to be lost due to dissipation, without knowing the parameters beforehand. Inspired by machine learning techniques, an optimized control scheme is designed to run over a representative ensemble and train a control field so that the optimal time for the measurement is at a prescribed runtime. This protocol demonstrates robustness to errors in the assumptions in the training process, while also enhancing measurement precision with non-Markovian memory effects, and the measurement uncertainty may approach the limits imposed by the Cramér–Rao bound.

BIOGRAPHY

Dr. Luo is a postdoc researcher in Prof. Ting Yu's group. He has carried out research work as a postdoc in China, Spain, and the US, and his main research topics lie in quantum information, quantum control, and quantum dynamics, especially non-Markovian open systems. He is now working on using novel quantum effects for real-world metrology tasks such as parameter estimation and weak signal sensing.