ABSTRACT

Brownian and Langevin dynamics have been the mainstay mathematical models in the natural sciences for well over a century. Thermostat methods like Nosé dynamics originated in the 1980s as simplified models to mimic properties of thermal baths in molecular simulation. In this talk I will address the unification of these different approaches, their discretization, and their properties as general purpose tools for large scale computation in the physical sciences, machine learning and statistics.