Assisting sampling of physical systems with generative models

Deep generative models parametrize very flexible families of distributions able to fit complicated datasets of images or text. These models provide independent samples from complex high-distributions at negligible costs. On the other hand, sampling exactly a target distribution, such as the Boltzmann distribution of a physical system, is typically challenging: either because of dimensionality, multi-modality, ill-conditioning or a combination of the previous. In this talk, I will discuss a recent line of work using generative models to accelerate sampling. While the approach shows promises, it still struggles as the system size gets large. When a coarse-graining resolving the metastability is known, I will also discuss how enhanced sampling can be revisited with generative models, and can ease this curse of dimensionality.