Machine learning for simulations of gauge theories

As a statistical tool, external page lattice QCD has been successfully used to determine many parameters of the Standard Model, including quark masses and the strong interaction coupling constant, see external page [2], external page [3] and external page [4]. Despite the success of lattice QCD, limitations of the current statistical algorithms still exist, leading to problems such as critical slowing down of the simulations external page [5]. New approaches are required to circumvent these limitations. Machine learning algorithms provide a viable approach to address some of these difficulties. We explore deep generative models, such as normalizing flows external page [6] and external page [7], to develop alternatives to standard algorithms for generating lattice field configurations external page [8].