Efficient Sampling of Equilibrium States using Boltzmann Generators
Image of the energy landscape corresponding to protein folding from Dill & MacCallum (2012) ‘The Protein-Folding Problem, 50 Years On’ Science 338 (6110).Discovering low energy states (e.g. the folded state of a protein) is a difficult task, owing to how rare these states are in all of configuration space. In this project, we used a neural network to convert a rugged energy landscapes into one that is easier to sample to enable exploration of these low energy states.
