adaptivebandit a multi armed bandit framework for adaptive sampling in molecular simulations

• DOI: 10.48550/arxiv.2002.12582
• Publication Date: 2020-06-15
• Authors: Pablo Herrera-Nieto; Adrià Pérez; Gianni De Fabritiis; Gianni De Fabritiis; Stefan Doerr
• Keywords: 0301 basic medicine; Protein Folding; 03 medical and health sciences; Microfilament Proteins; Proteins; FOS: Physical sciences; Molecular Dynamics Simulation; Computational Physics (physics.comp-ph); Physics - Computational Physics; Algorithms

Sampling from the equilibrium distribution has always been a major problem in molecular simulations due to the very high dimensionality of conformational space. Over several decades, many approaches have been used to overcome the problem. In particular, we focus on unbiased simulation methods such as parallel and adaptive sampling. Here, we recast adaptive sampling schemes on the basis of multi-armed bandits and develop a novel adaptive sampling algorithm under this framework, \UCB. We test it on multiple simplified potentials and in a protein folding scenario. We find that this framework performs similarly or better in every type of test potentials compared to previous methods. Furthermore, it provides a novel framework to develop new sampling algorithms with better asymptotic characteristics.