Application of machine learning to experimental design in quantum mechanics

Abstract: The recent advances in machine learning hold great promise for the fields of quantum sensing and metrology. With the help of reinforcement learning, we can tame the complexity of quantum systems and solve the problem of optimal experimental design. Reinforcement learning is a powerful model-free technique that allows an agent, typically a neural network, to learn the best strategy to reach a certain goal in a completely a priori unknown environment. However, in general, we know something about the quantum system with which the agent is interacting, at least that it follows the rules of quantum mechanics. In quantum metrology, we typically have a model for the system, and only some parameters of the evolution or the initial state are unknown. We present here a general machine learning technique that can optimize the precision of quantum sensors, exploiting the knowledge we have on the system through model-aware reinforcement learning. This framework has been implemented in the Python package qsensoropt, which is able to optimize a broad class of problems found in quantum metrology and quantum parameter estimation. The agent learns an optimal adaptive strategy that, based on previous outcomes, decides the next measurements to perform. We have explored some applications of this technique to NV centers and photonic circuits. So far, we have been able to certify better results than the current state-of-the-art controls for many cases. The machine learning technique developed here can be applied in all scenarios where the quantum system is well-characterized and relatively simple and small. In these cases, we can extract every last bit of information from a quantum sensor by appropriately controlling it with a trained neural network. The qsensoropt software is available on PyPI and can be installed with pip.