Machine Learning Optimal Control Pulses in an Optical Quantum Memory

The use of machine learning (ML) methods is an attractive approach to abstract the control of experimental variables away from researchers, to an optimizable, automated process. ML has been used to optimize experimental processes, ranging from laser cooling for BEC creation [1], to super-resolution optical microscopy [2] and optical memory optimization [3]. Optical quantum memories have been identified as an essential component for the realisation of a quantum internet, when used for quantum repeater nodes. A key factor is the efficiency of the memory in the node, indeed, it has been shown with an efficiency of 0.8 one can achieve kHz key rates [4]. Current experimental realisations have an efficiency of 0.33 [5]; thus improvement of the internal memory efficiency remains an important research goal.