Understanding Complex Quantum Dynamics Using Machine Learning

In recent years, quantum experiments have become increasingly complicated, with modern experiments pushing the limits of even our best supercomputers to simulate.  This increased complexity has made quantum devices challenging to model, which in turn makes them both difficult to control in quantum technologies and also exceedingly difficult to understand.  In this talk, I will show how ideas from machine learning and statistical inference can be used to probe complex quantum systems.  In particular, we will show how these methods can allow us to identify common pathologies in devices, as well as infer Hamiltonian dynamics for quantum systems.  This latter ability has led to a world record for room-temperature magnetometry using Nitrogen vacancy centers, which was achieved not by engineering a new device but rather through the use of machine learning to extract a weak signal from noisy data.  These results illustrate the growing importance of machine learning techniques to quantum science and technology and suggest a new methodology for analyzing as well as controlling physical systems.