Machine learning meets Monte Carlo sampling

Monte Carlo sampling is a ubiquitous method for numerically estimating high-dimensional integrals. Unfortunately, standard methods for Monte Carlo sampling sometimes become highly inefficient for certain problems, for example in important limits of lattice QCD simulations that are used to study the Standard Model. I will discuss my ongoing work to apply machine learning methods to this problem, which has the potential to accelerate numerical calculations in many domains without compromising exactness.