Kavli Affiliate: David T. Limmer
| First 5 Authors: Jorge L. Rosa-Raíces, Jorge L. Rosa-Raíces, , ,
| Summary:
We propose a method to compute free energies from nonadiabatic alchemical
transformations using diffusion-denoising generative models. The method,
nonadiabatic force matching, hinges on estimating — rather than minimizing —
the dissipation along an alchemical transition as the functional of a
nonadiabatic potential, which plays the role of a diffusion-denoising protocol.
Applying the algorithm to compute alchemical free energies of simple atomistic
models shows it can significantly cut the simulation cost of a free-energy
estimate at no loss of accuracy compared with thermodynamic integration.
| Search Query: ArXiv Query: search_query=au:”David T. Limmer”&id_list=&start=0&max_results=3