Paper Title:
FeNNol: an Efficient and Flexible Library for Building Force-field-enhanced Neural Network Potentials
Published on:
2 May 2024
Primary Category:
Chemical Physics
Paper Authors:
Thomas Plé,
Olivier Adjoua,
Louis Lagardère,
Jean-Philip Piquemal
FeNNol provides flexible building blocks for constructing hybrid ML potentials
It combines neural networks with force field terms like electrostatics and dispersion
The library uses Jax for fast evaluation and differentiation
Popular ANI-2x model reaches similar speeds as AMOEBA force field
Library for building force-field-enhanced neural network potentials
FeNNol is a new Python library for easily constructing and training hybrid machine learning potentials that combine neural networks with traditional force field terms. It leverages Jax for fast evaluation and differentiation. The paper shows simulation speeds close to standard force fields.
Fast physics-informed neural networks in TensorFlow
Accelerating neural network training by pre-teaching boundary conditions
Learning interatomic potentials with graph networks
pfl-research: Fast simulator for federated learning research
Machine learning improves molecular force fields
Future interaction feedback for motion forecasting
No comments yet, be the first to start the conversation...
Sign up to comment on this paper