Machine Learning for Sign Problems

Yukari Yamauchi, Ph.D.; University of Washington

Sign problems in lattice QCD prevent us from non-perturbatively calculating many important properties of dense nuclear matter both in and out of equilibrium. In this talk, I will discuss numerical methods to alleviate these sign problems in lattice field theories: complex normalizing flows and subtractions. Both of the methods are the cousins of the so-called manifold deformation method, in which one deforms the manifold of integration in the path integral to the complex plane, aiming for a milder sign problem. I will demonstrate the method of complex normalizing flows with the Φ4 scalar field theory at complex coupling. The subtraction method will be demonstrated with the Thirring model in 1+1-dimensions at finite density, which possesses a fermion sign problem.

For those interested in attending via Zoom, please use the meeting ID 973 3626 5389.