Fast and Accurate Equivariant Foundation Models for Atomistic Simulation
6 affiliations
- 1Yusuf Hamied Department of Chemistry , University of Cambridge , Cambridge , UK
- 2John A. Paulson School of Engineering and Applied Sciences , Harvard University , Cambridge , MA , USA
- 3Department of Molecular Science and Software Engineering , University of California , Berkeley , CA , USA
- 4Sandia National Laboratories , Albuquerque , NM , USA
- 5Mirian Technologies Inc. , Boston , MA , USA
- 6Robert Bosch LLC Research and Technology Center , Watertown , MA , USA
Abstract
Machine-learned interatomic potentials (MLIPs) have emerged as a transformative tool for computational materials science and chemistry, with universal potentials trained on large and diverse datasets now routinely deployed as 'foundation models' for downstream fine-tuning in targeted chemical spaces. Many scientific applications of the resulting models, such as molecular dynamics (MD), require high inference and training speeds as well as accuracy. In this work, we examine the limits of equivariant MLIPs, which directly encode physical symmetries in model architectures, to achieve these competing targets -- particularly in the regime of extremely large datasets where data efficiency is less critical. We show how this trade-off can be addressed, and present a family of foundation potentials in the NequIP and Allegro equivariant MLIP architectures which achieve leading inference speeds and strong scalability as well as excellent accuracies across a range of community benchmarks -- spanning materials discovery, thermal conductivity prediction, and near-equilibrium mechanical and thermodynamic properties. Accelerations implemented within the NequIP infrastructure now permit training of high-accuracy foundation potentials on ultra-large datasets with dramatically reduced computational cost. Alongside, we show that efforts to improve model accuracy for materials discovery should focus on dataset diversity and improved, consistent descriptions of transition metal compound energy surfaces.
Related
- Fast and accurate simulation of Raman spectra of gold-organic systems
- Fast and accurate simulation of Raman spectra of gold–organic systems
- Multiscale Neural Network Potential with Anisotropic Message Passing for the Fast and Accurate Simulation of Protein Dynamics and Enzymatic Reactions
- NEMORA: Neural Equivariant Multipole Operators for Long-Range Atomistic Learning
- Unconstrained learning of equivariant atomistic properties
BibTeX
@misc{kavanagh2026fast,
title = {{Fast and Accurate Equivariant Foundation Models for Atomistic Simulation}},
author = {Seán R. Kavanagh and Chuin Wei Tan and Menghang Wang and Marc Descoteaux and Gabriel de Miranda Nascimento and Ulrik Unneberg and Laura Zichi and Francesco Libbi and Norma Rivano and Austin Glover and Vivek Bharadwaj and Anders Johansson and William C. Witt and Albert Musaelian and Boris Kozinsky},
howpublished = {arXiv},
year = {2026},
month = jul,
doi = {10.48550/arxiv.2607.28461},
eprint = {2607.28461},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.28461},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Fast and Accurate Equivariant Foundation Models for Atomistic Simulation
AU - Seán R. Kavanagh
AU - Chuin Wei Tan
AU - Menghang Wang
AU - Marc Descoteaux
AU - Gabriel de Miranda Nascimento
AU - Ulrik Unneberg
AU - Laura Zichi
AU - Francesco Libbi
AU - Norma Rivano
AU - Austin Glover
AU - Vivek Bharadwaj
AU - Anders Johansson
AU - William C. Witt
AU - Albert Musaelian
AU - Boris Kozinsky
PY - 2026
DA - 2026/07/30
PB - arXiv
DO - 10.48550/arxiv.2607.28461
UR - https://arxiv.org/abs/2607.28461
AB - Machine-learned interatomic potentials (MLIPs) have emerged as a transformative tool for computational materials science and chemistry, with universal potentials trained on large and diverse datasets now routinely deployed as 'foundation models' for downstream fine-tuning in targeted chemical spaces. Many scientific applications of the resulting models, such as molecular dynamics (MD), require high inference and training speeds as well as accuracy. In this work, we examine the limits of equivariant MLIPs, which directly encode physical symmetries in model architectures, to achieve these competing targets -- particularly in the regime of extremely large datasets where data efficiency is less critical. We show how this trade-off can be addressed, and present a family of foundation potentials in the NequIP and Allegro equivariant MLIP architectures which achieve leading inference speeds and strong scalability as well as excellent accuracies across a range of community benchmarks -- spanning materials discovery, thermal conductivity prediction, and near-equilibrium mechanical and thermodynamic properties. Accelerations implemented within the NequIP infrastructure now permit training of high-accuracy foundation potentials on ultra-large datasets with dramatically reduced computational cost. Alongside, we show that efforts to improve model accuracy for materials discovery should focus on dataset diversity and improved, consistent descriptions of transition metal compound energy surfaces.
SN - 2331-8422
ER -