(536d) Recent Advances in Maximum Entropy Methods

White, A., University of Rochester
Making experimental data a first-class input like temperature or pressure is the goal of maximum entropy methods. This is important where force fields are approximate, such as when employing coarse-grained models or exploring new chemistry. In this talk I will review our recent advances in maximum entropy methods including treating uncertainty, large amounts of experimental data, and distributions. Maximum entropy has been used to simulate highly accurate ab initio molecular dynamics of water, incorporate environmental fluctuations inside of a macromolecular protein complex, improve RNA force fields, and to combine enhanced sampling with minimal biasing to model peptides. Maximum entropy is seeing renewed interest beyond molecular simulation. We have shown that maximum entropy can be used to connect physics-based models like molecular simulations with AI black-box predictions. Maximum entropy is even being applied to epidemiology, where complex disease models need to match current case statistics. Maximum entropy methods have revolutionized complex models by making it possible to rapidly match data without parameter optimization.