Generalized autonomous optimization for quantum transmitters with deep reinforcement learning

Precise control of system parameters and extensive optimization play a crucial role in enabling quantum information technologies. As a further challenge, when targeting practical manufacturable systems, the presence of manufacturing variations in components necessitates individual optimization for each system. To address this challenge, we develop a generalisable optimisation framework based on deep reinforcement learning (RL). By applying our method to real-world quantum transmitters based on optical injection locking (OIL), we demonstrate that our RL agent can autonomously identify the optimal operating regions, and generalise its knowledge for new quantum transmitters of the same type. This work presents a new avenue for efficient optimisation of complex systems using modern RL algorithm.