Key points are not available for this paper at this time.
Flow-based machine learning techniques have demonstrated effectiveness in tackling significant computational obstacles, including critical slowing-down and topological freezing, encountered in the sampling of gauge field configurations within lattice field theories. We investigate the viability of this approach for simulations of gauge theories at finite temperature. Several tests are performed on two dimensional U(1) gauge theory at different temporal extents.
Kirwan et al. (Fri,) studied this question.