CoChat
Sign Up
Cited by 2,285Machine Learning in Materials ScienceAdvanced Chemical Physics StudiesModel Reduction and Neural Networks

The DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.

E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials

Open access
Batzner, Simon, Musaelian, Albert, Sun, Lixin, Geiger, Mario, Mailoa, Jonathan P., Kornbluth, Mordechai, Molinari, Nicola, Smidt, Tess E., Kozinsky, Boris·2022·Nature Communications
DOI 10.1038/s41467-022-29939-5PMC PMC9068614

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

This work presents Neural Equivariant Interatomic Potentials (NequIP), an E(3)-equivariant neural network approach for learning interatomic potentials from ab-initio calculations for molecular dynamics simulations. While most contemporary symmetry-aware models use invariant convolutions and only act on scalars, NequIP employs E(3)-equivariant convolutions for interactions of geometric tensors, resulting in a more information-rich and faithful representation of atomic environments. The method achieves state-of-the-art accuracy on a challenging and diverse set of molecules and materials while exhibiting remarkable data efficiency. NequIP outperforms existing models with up to three orders of magnitude fewer training data, challenging the widely held belief that deep neural networks require massive training sets. The high data efficiency of the method allows for the construction of accurate potentials using high-order quantum chemical level of theory as reference and enables high-fidelity molecular dynamics simulations over long time scales.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-022-29939-5

Cites

7 of 7 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Existing MLIP efforts typically train models per material, with data often sampled from ab initio molecular dynamics, which markedly limits their general applicability and adoption.

    “Existing efforts typically train models per material, with data often sampled from ab initio molecular dynamics (AIMD). This markedly limits their general applicability and adoption, requiring expensive data collection and training a new potential from scratch for each system.”

    From Scaling deep learning for materials discovery (2023) · cites this paper as [30]

    The Introduction says NN interatomic potentials typically require large sets of ab initio calculations and that collecting them has severely limited adoption. The Results and Methods describe several training datasets drawn from AIMD. Our copy does not establish that existing models are typically trained separately for each material, require retraining from scratch for every system, or are generally inapplicable to other systems. (LLM, conf=0.84)

    Experimental — accuracy not yet measured

  • The structures and relaxation trajectories present a large and diverse dataset to enable training of learned, equivariant interatomic potentials with unprecedented accuracy and zero-shot generalization.

    “The structures and relaxation trajectories present a large and diverse dataset to enable training of learned, equivariant interatomic potentials30,31 with unprecedented accuracy and zero-shot generalization.”

    From Scaling deep learning for materials discovery (2023) · cites this paper as [30]

  • The zero-shot GNoME potential, not fine-tuned on any data from this composition, outperforms a state-of-the-art NequIP model trained on hundreds of structures.

    “The zero-shot GNoME potential, not fine-tuned on any data from this composition, outperforms even a state-of-the-art NequIP model trained on hundreds of structures.”

    From Scaling deep learning for materials discovery (2023) · cites this paper as [30]

  • The pretrained GNoME potential shows remarkable accuracy when evaluated on downstream data sampled from the new distribution of AIMD in a zero-shot manner.

    “Despite only being trained on ionic relaxations and not on molecular-dynamics data, the pretrained GNoME potential shows remarkable accuracy when evaluated on downstream data sampled from the new distribution of AIMD in a zero-shot manner, that is, in which no training data originate from AIMD simulations (see Fig. 3).”

    From Scaling deep learning for materials discovery (2023) · cites this paper as [30]

    The Results (MD-17 small molecule dynamics; page unavailable) evaluate NequIP using data generated by ab-initio molecular dynamics, not a pretrained GNoME potential. A search for GNoME, ionic-relaxation pretraining, and zero-shot AIMD evaluation found no passage addressing the attributed claim. (LLM, conf=0.76)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 54

  1. Richards, W. D. et al. Design and synthesis of the superionic conductor na 10 snp 2 s 12. Nat. Commun. 7, 1–8 (2016). · DOI
  2. Boero, M., Parrinello, M. & Terakura, K. First principles molecular dynamics study of ziegler- natta heterogeneous catalysis. J. Am. Chem. Soc. 120, 2746–2752 (1998). · DOI
  3. Lindorff-Larsen, K., Piana, S., Dror, R. O. & Shaw, D. E. How fast-folding proteins fold. Science 334, 517–520 (2011). · DOI · Paper page
  4. Behler, J. & Parrinello, M. Generalized neural-network representation of high-dimensional potential-energy surfaces. Phys. Rev. Lett. 98, 146401 (2007). · DOI · Paper page
  5. Bartók, A. P., Payne, M. C., Kondor, R. & Csányi, G. Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons. Phys. Rev. Lett. 104, 136403 (2010). · DOI · Paper page
  6. Shapeev, A. V. Moment tensor potentials: a class of systematically improvable interatomic potentials. Multiscale Model. Simul. 14, 1153–1173 (2016). · DOI
  7. Thompson, A. P., Swiler, L. P., Trott, C. R., Foiles, S. M. & Tucker, G. J. Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials. J. Comput. Phys. 285, 316–330 (2015). · DOI
  8. Vandermause, J. et al. On-the-fly active learning of interpretable bayesian force fields for atomistic rare events. Npj Comput. Mater. 6, 1–11 (2020). · DOI · Paper page
  9. Schütt, K. et al. Schnet: A continuous-filter convolutional neural network for modeling quantum interactions. Adv. Neural Inf. Process. Syst. 30, 991–1001 (2017).
  10. Unke, O. T. & Meuwly, M. Physnet: A neural network for predicting energies, forces, dipole moments, and partial charges. J. Chem. Theory Comput. 15, 3678–3693 (2019). · DOI
  11. Klicpera, J., Groß, J. & Günnemann, S. Directional message passing for molecular graphs. arXiv preprint arXiv:2003.03123 (2020). · arXiv
  12. Mailoa, J. P. et al. A fast neural network approach for direct covariant forces prediction in complex multi-element extended systems. Nat. Mach. Intell. 1, 471–479 (2019). · DOI
  13. Park, C. W. et al. Accurate and scalable multi-element graph neural network force field and molecular dynamics with direct force architecture. arXiv preprint arXiv:2007.14444 (2020). · DOI
  14. Artrith, N. & Kolpak, A. M. Understanding the composition and activity of electrocatalytic nanoalloys in aqueous solvents: A combination of dft and accurate neural network potentials. Nano Lett. 14, 2670–2676 (2014). · DOI
  15. Zhang, L., Han, J., Wang, H., Car, R. & Weinan, E. Deep potential molecular dynamics: a scalable model with the accuracy of quantum mechanics. Phys. Rev. Lett. 120, 143001 (2018). · DOI
  16. Gilmer, J., Schoenholz, S. S., Riley, P. F., Vinyals, O. & Dahl, G. E. Neural message passing for quantum chemistry. arXiv preprint arXiv:1704.01212 (2017). · arXiv
  17. Anderson, B., Hy, T. S. & Kondor, R. Cormorant: Covariant molecular neural networks. Adv. Neural Inf. Process. Syst. 14537–14546 (2019).
  18. Townshend, R. J., Townshend, B., Eismann, S. & Dror, R. O. Geometric prediction: Moving beyond scalars. arXiv preprint arXiv:2006.14163 (2020). · arXiv
  19. Thomas, N. et al. Tensor field networks: Rotation-and translation-equivariant neural networks for 3d point clouds. arXiv preprint arXiv:1802.08219 (2018). · arXiv
  20. Batzner, S. et al. Se(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. arXiv preprint arXiv:2101.03164v1 (2021). · DOI
  21. Schütt, K. T., Unke, O. T. & Gastegger, M. Equivariant message passing for the prediction of tensorial properties and molecular spectra. arXiv preprint arXiv:2102.03150 (2021). · arXiv
  22. Haghighatlari, M. et al. Newtonnet: A newtonian message passing network for deep learning of interatomic potentials and forces. arXiv preprint arXiv:2108.02913 (2021). · DOI
  23. Klicpera, J., Becker, F. & Günnemann, S. Gemnet: Universal directional graph neural networks for molecules. arXiv preprint arXiv:2106.08903 (2021). · arXiv
  24. Unke, O. T. et al. Spookynet: Learning force fields with electronic degrees of freedom and nonlocal effects. Nat. Commun. 12, 7273 (2021). · DOI
  25. Qiao, Z. et al. Unite: Unitary n-body tensor equivariant network with applications to quantum chemistry. arXiv preprint arXiv:2105.14655 (2021). · arXiv
  26. Grisafi, A., Wilkins, D. M., Willatt, M. J. & Ceriotti, M. Atomic-scale representation and statistical learning of tensorial properties. In Machine Learning in Chemistry: Data-Driven Algorithms, Learning Systems, and Predictions, 1–21 (ACS Publications, 2019). · DOI
  27. Weiler, M., Geiger, M., Welling, M., Boomsma, W. & Cohen, T. S. 3d steerable cnns: Learning rotationally equivariant features in volumetric data. Adv. Neural Inf. Process. Syst. 31, 10381–10392 (2018).
  28. Kondor, R. N-body networks: a covariant hierarchical neural network architecture for learning atomic potentials. arXiv preprint arXiv:1803.01588 (2018). · arXiv
  29. Kondor, R., Lin, Z. & Trivedi, S. Clebsch–gordan nets: a fully fourier space spherical convolutional neural network. Adv. Neural Inf. Process. Syst. 31, 10117–10126 (2018).
  30. Hendrycks, D. & Gimpel, K. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415 (2016). · arXiv
  31. Larsen, A. H. et al. The atomic simulation environment—a python library for working with atoms. J. Phys. Condens. Matter 29, 273002 (2017). · DOI
  32. Chmiela, S. et al. Machine learning of accurate energy-conserving molecular force fields. Sci. Adv. 3, e1603015 (2017). · DOI
  33. Chmiela, S., Sauceda, H. E., Müller, K.-R. & Tkatchenko, A. Towards exact molecular dynamics simulations with machine-learned force fields. Nat. Commun. 9, 3887 (2018). · DOI
  34. Ko, H.-Y. et al. Isotope effects in liquid water via deep potential molecular dynamics. Mol. Phys. 117, 3269–3281 (2019). · DOI
  35. Christensen, A. S. & von Lilienfeld, O. A. On the role of gradients for machine learning of molecular energies and forces. Mach. learn.: sci. technol. 1, 045018 (2020).
  36. Devereux, C. et al. Extending the applicability of the ani deep learning molecular potential to sulfur and halogens. J. Chem. Theory Comput. 16, 4192–4202 (2020). · DOI
  37. Christensen, A. S., Bratholm, L. A., Faber, F. A. & Anatole von Lilienfeld, O. Fchl revisited: Faster and more accurate quantum machine learning. J. Chem. Phys. 152, 044107 (2020). · DOI
  38. Drautz, R. Atomic cluster expansion for accurate and transferable interatomic potentials. Phys. Rev. B 99, 014104 (2019). · DOI
  39. Kovács, D. P. et al. Linear atomic cluster expansion force fields for organic molecules: beyond rmse. J. Chem. Theory Comput. 17, 7696–711 (2021). · DOI
  40. Zhang, L. et al. End-to-end symmetry preserving inter-atomic potential energy model for finite and extended systems. Adv. Neural Inf. Process. Syst. 31, (2018).
  41. Sim, W. S., Gardner, P. & King, D. A. Multiple bonding configurations of adsorbed formate on ag111. J. Phys. Chem. 100, 12509–12516 (1996). · DOI
  42. Wang, G., Morikawa, Y., Matsumoto, T. & Nakamura, J. Why is formate synthesis insensitive to copper surface structures? J. Phys. Chem. B 110, 9–11 (2006). · DOI
  43. Yu, X., Bates, J. B., Jellison, G. E. & Hart, F. X. A stable thin-film lithium electrolyte: lithium phosphorus oxynitride. J. Electrochem. Soc. 144, 524–532 (1997). · DOI
  44. Westover, A. S. et al. Plasma synthesis of spherical crystalline and amorphous electrolyte nanopowders for solid-state batteries. ACS Appl. Mater. Interfaces 12, 11570–11578 (2020). · DOI
  45. Li, W., Ando, Y., Minamitani, E. & Watanabe, S. Study of li atom diffusion in amorphous li3po4 with neural network potential. J. Chem. Phys. 147, 214106 (2017). · DOI
  46. Cheng, B., Engel, E. A., Behler, J., Dellago, C. & Ceriotti, M. Ab initio thermodynamics of liquid and solid water. Proceedings of the National Academy of Sciences 116, 1110–1115 (2019). · DOI
  47. Hestness, J. et al. Deep learning scaling is predictable, empirically. arXiv preprint arXiv:1712.00409 (2017). · arXiv
  48. Paszke, A. et al. Pytorch: An imperative style, high-performance deep learning library. Adv. Neural Inf. Process. Syst. 32, 8026–8037 (2019).
  49. Fey, M. & Lenssen, J. E. Fast graph representation learning with pytorch geometric. arXiv preprint arXiv:1903.02428 (2019). · arXiv
  50. Hutter, J., Iannuzzi, M., Schiffmann, F. & VandeVondele, J. cp2k: atomistic simulations of condensed matter systems. WIREs Computational Molecular Science 4, 15–25 (2014). · DOI
  51. Kresse, G. & Hafner, J. Ab initiomolecular dynamics for liquid metals. Phys. Rev. B 47, 558–561 (1993). · DOI
  52. Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 15–50 (1996). · DOI
  53. Kingma, D. P. & Ba, J. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014). · arXiv
  54. Loshchilov, I. & Hutter, F. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101 (2017). · arXiv

Cited by

Sorted by citations

Metadata

Authors
Batzner, S., Musaelian, A., Sun, L., Geiger, M., Mailoa, J. P., Kornbluth, M., Molinari, N., Smidt, T. E., & Kozinsky, B.
Affiliations
  • Harvard University (US)
  • École Polytechnique Fédérale de Lausanne (CH)
  • Robert Bosch (United States) (US)
  • Lawrence Berkeley National Laboratory (US)
Corresponding author
  • Batzner, Simon
  • Kozinsky, Boris
ORCID
Venue
Nature Communications
Publisher
Springer Science and Business Media LLC
Published
2022-05-04
Volume
13
Issue
1
Pages
2453-2453
ISSN
2041-1723 (ISSN-L)
DOI
10.1038/s41467-022-29939-5
PMC
PMC9068614
OpenAlex
W4225405705
Cited by
2,285
References
7
Citations by year
2026: 789 · 2025: 655 · 2024: 437 · 2023: 283 · 2022: 93
FWCI
97.08
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • eScholarship (California Digital Library) (submittedVersion)
  • Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion, cc-by-nc-nd)
  • OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion)
  • arXiv (Cornell University)
Funders
  • Robert Bosch (n/a)
  • U.S. Department of Energy (DE-SC0021110, DE-SC001257, DE-AR0000775, DE-AC02-05CH11231, DE-SC0022199, DE-SC0012573)
  • United States Department of Defense | United States Navy | Office of Naval Research (N00014-20-1-2418)
  • National Science Foundation (DMR-2011754)
Topics
  • Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Advanced Chemical Physics Studies
  • Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Model Reduction and Neural Networks
Keywords
interatomic potentials, E(3) equivariance, graph neural networks, NequIP, first-principles calculations, data efficiency, molecular dynamics
MeSH terms
Neural Networks, Computer, Molecular Dynamics Simulation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review