Applications of machine learning to quantum physics

ABSTRACT

Machine learning is attracting tremendous interests in physics in the last several years. In this talk, I will start with a brief overview about machine learning application to quantum physics. Then I will describe our recent works on using machine learning in the study of quantum many-body physics and quantum computing. For the former, I will present a rigorous correspondence between short-range coupled restricted Boltzman machine and area-law entangled quantum many-body states. Examples for this correspondence based on stabiliser codes will be discussed. For the latter, I will present our recent work proposing a reinforcement learning architecture for quantum algorithm design. Our approach is applicable to problems which are hard-to-solve but easy-to-verify. We show this approach provides improved quantum adiabatic algorithms when applied to Grover search and 3-SAT problems.

BIOGRAPHY

Dr Xiaopeng Li got his Ph.D. from University of Pittsburgh in 2013, where he worked on optical lattice quantum simulations under the support of US DARPA-OLE program. He was a postdoctoral researcher at Maryland University---College Park from 2013 to 2016 supported by JQI Theoretical Fellowship. He joined Fudan University as a faculty in late 2016, was awarded the 1000-Young-Talent grant in 2017, and is currently an associate professor. His research interests include machine learning applications to field theories and quantum information, and quantum simulations with atomic systems.