(551a) Systematic Multiscale Models and Physics Using the Relative Entropy

Many molecular process involve ranges of length and time scales that cannot be tackled with atomic-scale simulations. Instead, coarse-grained molecular models are necessary to interrogate such systems from a theoretical point of view. We discuss a fundamental approach to the identification of such models and more generally of emergent physical behavior from many-body systems. We proposed that a quantity called the relative entropy measures the information lost upon coarse graining and is the natural thermodynamic metric for such tasks: its minimization provides a universal variational principle for coarse-graining. This broad statistical-mechanical framework can improve and even detect both analytical and simulation models of complex systems. We discuss conceptual and numerical aspects of this approach and illustrate its use in several case studies, including our recent efforts that create greatly improved coarse models with unusual interaction potentials.