Boltzmann Generators -- Sampling Equilibrium States of Many-Body Systems with Deep Learning
This provides a tool for researchers in condensed-matter physics and biochemistry to avoid rare events in sampling without prior knowledge, though it builds incrementally on existing deep learning and statistical mechanics methods.
The paper tackles the challenge of generating statistically independent equilibrium samples for many-body systems like proteins by introducing Boltzmann Generators, which use deep learning to produce unbiased one-shot samples, enabling accurate free energy computations and new configuration discovery.
Computing equilibrium states in condensed-matter many-body systems, such as solvated proteins, is a long-standing challenge. Lacking methods for generating statistically independent equilibrium samples in "one shot", vast computational effort is invested for simulating these system in small steps, e.g., using Molecular Dynamics. Combining deep learning and statistical mechanics, we here develop Boltzmann Generators, that are shown to generate unbiased one-shot equilibrium samples of representative condensed matter systems and proteins. Boltzmann Generators use neural networks to learn a coordinate transformation of the complex configurational equilibrium distribution to a distribution that can be easily sampled. Accurate computation of free energy differences and discovery of new configurations are demonstrated, providing a statistical mechanics tool that can avoid rare events during sampling without prior knowledge of reaction coordinates.