- Physical Review Journals
All JournalsPhysics Magazine

Physical Review Research

Reuse & Permissions

It is not necessary to obtain permission to reuse this article or its components as it is available under the terms of the Creative Commons Attribution 4.0 International license. This license permits unrestricted use, distribution, and reproduction in any medium, provided attribution to the author(s) and the published article's title, journal citation, and DOI are maintained. Please note that some figures may have been included with permission from other third parties. It is your responsibility to obtain the proper permission from the rights holder directly for these figures.

Export citation

Export citation

Choose format for download:

Download Citation
  • Open Access

AI-Newton: A concept-driven physical law discovery system without prior physical knowledge

You-Le Fang*, Dong-Shan Jian†, and Xiang Li‡

Yan-Qing Ma§

  • *Contact author: eden@stu.pku.edu.cn
  • †Contact author: dsjian@stu.pku.edu.cn
  • ‡Contact author: lix-PHY@pku.edu.cn
  • §Contact author: yqma@pku.edu.cn
PDF

Phys. Rev. Research 8, 043129 – Published 4 November, 2026

DOI: https://doi.org/10.1103/dvvc-4rsr

Abstract

While current artificial intelligence (AI)-driven methods excel at deriving empirical models from individual experiments, a significant challenge remains in uncovering the common fundamental physics that underlie these models—a task at which human physicists are adept. To bridge this gap, we introduce AI-Newton, a framework for concept-driven scientific discovery. Our system autonomously derives general physical laws directly from raw, multiexperiment data, operating without supervision or prior physical knowledge. Its core innovations are twofold: (1) proposing interpretable physical concepts to construct laws and (2) progressively generalizing these laws to broader domains. Applied to a large, noisy dataset of mechanics experiments, AI-Newton successfully rediscovers foundational and universal laws, such as Newton's second law, the conservation of energy, and the universal gravitation. This work represents a proof-of-concept step toward autonomous, humanlike scientific discovery.

View figure in article

Physics Subject Headings (PhySH)

Article Text

Supplemental Material

References (55)

  1. Y. Xu, X. Liu, X. Cao, C. Huang, E. Liu, S. Qian, X. Liu, Y. Wu, F. Dong, C.-W. Qiu et al., Artificial intelligence: A powerful paradigm for scientific research, Innovation 2, 100179 (2021).
  2. H. Wang, T. Fu, Y. Du, W. Gao, K. Huang, Z. Liu, P. Chandak, S. Liu, P. Van Katwyk, A. Deac, et al., Scientific discovery in the age of artificial intelligence, Nature (London) 620, 47 (2023).
  3. X. Zhang, L. Wang, J. Helwig, Y. Luo, C. Fu, Y. Xie, M. Liu, Y. Lin, Z. Xu, K. Yan, et al., Artificial intelligence for science in quantum, atomistic, and continuum systems, Found. Trends Mach. Learn. 18, 385 (2025).
  4. C. Lu, C. Lu, R. T. Lange, J. Foerster, J. Clune, and D. Ha, The AI Scientist: Towards fully automated open-ended scientific discovery, arXiv:2408.06292.
  5. C. K. Reddy and P. Shojaee, Towards scientific discovery with generative AI: Progress, opportunities, and challenges, in Proceedings of the AAAI Conference on Artificial Intelligence (AAAI Press, Washington, DC, 2025), Vol. 39, pp. 28601–28609.
  6. M. Schmidt and H. Lipson, Distilling free-form natural laws from experimental data, Science 324, 81 (2009).
  7. S. Brunton, J. Proctor, and J. Kutz, Discovering governing equations from data: Sparse identification of nonlinear dynamical systems, Proc. Natl. Acad. Sci. USA 113, 3932 (2015).
  8. K. Champion, B. Lusch, J. N. Kutz, and S. L. Brunton, Data-driven discovery of coordinates and governing equations, Proc. Natl. Acad. Sci. USA 116, 22445 (2019).
  9. T. Wu and M. Tegmark, Toward an artificial intelligence physicist for unsupervised learning, Phys. Rev. E 100, 033311 (2019).
  10. S. Greydanus, M. Dzamba, and J. Yosinski, Hamiltonian Neural Networks (Curran Associates Inc., Red Hook, NY, 2019).
  11. M. Cranmer, S. Greydanus, S. Hoyer, P. Battaglia, D. Spergel, and S. Ho, Lagrangian neural networks, in ICLR 2020 Deep Differential Equations Workshop, arXiv:2003.04630.
  12. B. M. De Silva, D. M. Higdon, S. L. Brunton, and J. N. Kutz, Discovery of physics from data: Universal laws and discrepancies, Front. Artif. Intell. 3, 25 (2020).
  13. Z. Liu and M. Tegmark, Machine learning conservation laws from trajectories, Phys. Rev. Lett. 126, 180604 (2021).
  14. G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang, Physics-informed machine learning, Nat. Rev. Phys. 3, 422 (2021).
  15. Z. Liu, V. Madhavan, and M. Tegmark, Machine learning conservation laws from differential equations, Phys. Rev. E 106, 045307 (2022).
  16. G. Camps-Valls, A. Gerhardus, U. Ninad, G. Varando, G. Martius, E. Balaguer-Ballester, R. Vinuesa, E. Diaz, L. Zanna, and J. Runge, Discovering causal relations and equations from data, Phys. Rep. 1044, 1 (2023).
  17. C. Cornelio, S. Dash, V. Austel, T. R. Josephson, J. Goncalves, K. L. Clarkson, N. Megiddo, B. El Khadir, and L. Horesh, Combining data and theory for derivable scientific discovery with AI-Descartes, Nat. Commun. 14, 1777 (2023).
  18. P. Lemos, N. Jeffrey, M. Cranmer, S. Ho, and P. Battaglia, Rediscovering orbital mechanics with machine learning, Mach. Learn.: Sci. Technol. 4, 045002 (2023).
  19. Z. Liu, P. O. Sturm, S. Bharadwaj, S. J. Silva, and M. Tegmark, Interpretable conservation laws as sparse invariants, Phys. Rev. E 109, L023301 (2024).
  20. R. Cory-Wright, C. Cornelio, S. Dash, B. E. Khadir, and L. Horesh, Evolving scientific discovery by unifying data and background knowledge with AI Hilbert, Nat. Commun. 15, 5922 (2024).
  21. D. Zheng, V. Luo, J. Wu, and J. B. Tenenbaum, Unsupervised learning of latent physical properties using perception-prediction networks, in Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, Proceedings of Machine Learning Research Vol. R16 (PMLR, 2018), pp. 496–506.
  22. M. Tegmark, Latent representations of dynamical systems: When two is better than one, arXiv:1902.03364.
  23. R. Iten, T. Metger, H. Wilming, L. del Rio, and R. Renner, Discovering physical concepts with neural networks, Phys. Rev. Lett. 124, 010508 (2020).
  24. B. Chen, K. Huang, S. Raghupathi, I. Chandratreya, Q. Du, and H. Lipson, Automated discovery of fundamental variables hidden in experimental data, Nat. Comput. Sci. 2, 433 (2022).
  25. Q. Li, T. Wang, V. Roychowdhury, and M. K. Jawed, Metalearning generalizable dynamics from trajectories, Phys. Rev. Lett. 131, 067301 (2023).
  26. S.-M. Udrescu and M. Tegmark, AI Feynman: A physics-inspired method for symbolic regression, Sci. Adv. 6, eaay2631 (2020).
  27. S.-M. Udrescu, A. Tan, J. Feng, O. Neto, T. Wu, and M. Tegmark, AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity, Adv. Neural Inf. Process. Syst. 33, 4860 (2020).
  28. T. Bendinelli, L. Biggio, and P.-A. Kamienny, Controllable neural symbolic regression, in International Conference on Machine Learning (PMLR, 2023), pp. 2063–2077.
  29. W. Tenachi, R. Ibata, and F. I. Diakogiannis, Deep symbolic regression for physics guided by units constraints: Toward the automated discovery of physical laws, Astrophys. J. 959, 99 (2023).
  30. Y. Tian, W. Zhou, M. Viscione, H. Dong, D. Kammer, and O. Fink, Interactive symbolic regression with co-design mechanism through offline reinforcement learning, Nat. Commun. 16, 3930 (2025).
  31. M. Cranmer, Interpretable machine learning for science with PySR and SymbolicRegression.jl, arXiv:2305.01582.
  32. M. Du, Y. Chen, Z. Wang, L. Nie, and D. Zhang, Large language models for automatic equation discovery of nonlinear dynamics, Phys. Fluids 36, 097121 (2024).
  33. B. Romera-Paredes, M. Barekatain, A. Novikov, M. Balog, M. P. Kumar, E. Dupont, F. J. R. Ruiz, J. S. Ellenberg, P. Wang, O. Fawzi, P. Kohli, and A. Fawzi, Mathematical discoveries from program search with large language models, Nature (London) 625, 468 (2024).
  34. M. Valipour, B. You, M. Panju, and A. Ghodsi, SymbolicGPT: A generative transformer model for symbolic regression, in Efficient Natural Language and Speech Processing (NeurIPS 2022 Workshop) (NeurIPS, 2022).
  35. X. Chu, H. Zhao, E. Xu, H. Qi, M. Chen, and H. Shao, Scalable neural symbolic regression using control variables, arXiv:2306.04718.
  36. S. Mežnar, S. Džeroski, and L. Todorovski, Efficient generator of mathematical expressions for symbolic regression, Mach. Learn. 112, 4563 (2023).
  37. Y.-L. Fang, D.-S. Jian, X. Li, and Y.-Q. Ma, AI-Newton [software and data], GitHub, 2025, https://github.com/Science-Discovery/AI-Newton.
  38. C. Wang, H. Zhai, and Y.-Z. You, Emergent Schrödinger equation in an introspective machine learning architecture, Sci. Bull. 64, 1228 (2019).
  39. B.-B. Li, Y. Gu, and S.-F. Wu, Discover physical concepts and equations with machine learning, Phys. Scr. 100, 086011 (2025).
  40. See Supplemental Material at http://link.aps.org/supplemental/10.1103/dvvc-4rsr for additional details on physical domain-specific language, recommendation engine, and computational engine.
  41. G. Pólya, Mathematics and Plausible Reasoning: Induction and Analogy in Mathematics (Princeton University Press, Princeton, NJ, 1990), Vol. 1.
  42. G. Pólya, Mathematics and Plausible Reasoning: Patterns of Plausible Inference (Princeton University Press, Princeton, NJ, 1990), Vol. 2.
  43. T. Lai and H. Robbins, Asymptotically efficient adaptive allocation rules, Adv. Appl. Math. 6, 4 (1985).
  44. T. L. Lai, Adaptive treatment allocation and the multi-armed bandit problem, Ann. Stat. 15, 1091 (1987).
  45. R. Agrawal, Sample mean based index policies by O(log n) regret for the multi-armed bandit problem, Adv. Appl. Probab. 27, 1054 (1995).
  46. M. N. Katehakis and H. Robbins, Sequential choice from several populations, Proc. Natl. Acad. Sci. USA 92, 8584 (1995).
  47. P. Auer, Using confidence bounds for exploitation-exploration trade-offs, J. Mach. Learn. Res. 3, 397 (2002).
  48. K. Pearson, LIII. On lines and planes of closest fit to systems of points in space, London, Edinburgh Dublin Philos. Mag. J. Sci. 2, 559 (1901).
  49. L. Wang, Discovering phase transitions with unsupervised learning, Phys. Rev. B 94, 195105 (2016).
  50. H. Kiwata, Deriving the order parameters of a spin-glass model using principal component analysis, Phys. Rev. E 99, 063304 (2019).
  51. D. Yevick, Conservation laws and spin system modeling through principal component analysis, Comput. Phys. Commun. 262, 107832 (2021).
  52. F. Boulier, D. Lazard, F. Ollivier, and M. Petitot, Representation for the radical of a finitely generated differential ideal, in Proceedings of the 1995 International Symposium on Symbolic and Algebraic Computation, ISSAC ′95 (Association for Computing Machinery, New York, 1995), pp. 158–166.
  53. F. Boulier, D. Lazard, F. Ollivier, and M. Petitot, Computing representations for radicals of finitely generated differential ideals, applicable algebra in engineering, Commun. Comput. 20, 73 (2009).
  54. Maplesoft, Different algebra in Maple, Maplesoft Help Center, 2024, https://cn.maplesoft.com/support/help/Maple/view.aspx?path=DifferentialAlgebra.
  55. Maplesoft, Maple 2024, https://www.maplesoft.com/products/maple/.

Outline

Information

Phys. Rev. Research 8, 043129– Published 4 November, 2026
  • Received 8 December 2025
  • Revised 15 June 2026
  • Accepted 9 October 2026
Crossmark - Check for updates button
Creative Commons Logo

Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.

Published by the American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation