DMFF: An Open-Source Automatic Differentiable Platform for Molecular Force Field Development and Molecular Dynamics Simulation

XY Wang and JC Li and L Yang and FY Chen and YZ Wang and JH Chang and JM Chen and W Feng and LF Zhang and K Yu, JOURNAL OF CHEMICAL THEORY AND COMPUTATION, 19, 5897-5909 (2023).

DOI: 10.1021/acs.jctc.2c01297

Inthe simulation of molecular systems, the underlying force field(FF) model plays an extremely important role in determining the reliabilityof the simulation. However, the quality of the state-of-the-art molecularforce fields is still unsatisfactory in many cases, and the FF parameterizationprocess largely relies on human experience, which is not scalable.To address this issue, we introduce DMFF, an open-source molecularFF development platform based on an automatic differentiation technique.DMFF serves as a powerful tool for both top-down and bottom-up FFdevelopment. Using DMFF, both energies/forces and thermodynamic quantitiessuch as ensemble averages and free energies can be evaluated in adifferentiable way, realizing an automatic, yet highly flexible FFoptimization workflow. DMFF also eases the evaluation of forces andvirial tensors for complicated advanced FFs, helping the fast validationof new models in molecular dynamics simulation. DMFF has been releasedas an open-source package under the LGPL-3.0 license and is availableat https://github.com/deepmodeling/DMFF.

Return to Publications page