Library

AI Poincaré: Machine Learning Conservation Laws from Trajectories

We present AI Poincaré, a machine learning algorithm for autodiscovering conserved quantities using trajectory data from unknown dynamical systems. We test it on five Hamiltonian systems, including the gravitational three-body problem, and find that it discovers not only all exactly conserved quantities, but also periodic orbits, phase transitions, and breakdown timescales for approximate conservation laws.

cs.LGastro-ph.EPnlin.SIphysics.class-phPhysicsComputer ScienceMedicine
Source

Paper

References (37)

01Discovering Symbolic Models from Deep Learning with Inductive BiasesMiles Cranmer, Alvaro Sanchez-Gonzalez, P. Battaglia et al.2020 · Neural Information Processing Systems · 687 citations In Library
02AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularityS. Udrescu, A. Tan, Jiahai Feng et al.2020 · Neural Information Processing Systems · 257 citations In Library
03Lagrangian Neural NetworksMiles Cranmer, S. Greydanus, Stephan Hoyer et al.2020 · International Conference on Learning Representations · 617 citations In Library
04Discovering Symmetry Invariants and Conserved Quantities by Interpreting Siamese Neural NetworksS. Wetzel, R. Melko, Joseph Scott et al.2020 · Physical Review Research · 77 citations In Library
05Interpretable Conservation Law Estimation by Deriving the Symmetries of Dynamics from Trained Deep Neural NetworksYoh-ichi Mototake2019 · Physical Review E · 19 citations In Library
06Integration of Neural Network-Based Symbolic Regression in Deep Learning for Scientific DiscoverySamuel Kim, Peter Y. Lu, Srijon Mukherjee et al.2019 · IEEE Transactions on Neural Networks and Learning Systems · 241 citations In Library
07Extracting Interpretable Physical Parameters from Spatiotemporal Systems using Unsupervised LearningPeter Y. Lu, Samuel Kim, M. Soljačić2019 · Physical Review X · 70 citations In Library
08Deep Lagrangian Networks: Using Physics as Model Prior for Deep LearningMichael Lutter, Christian Ritter, Jan Peters2019 · International Conference on Learning Representations · 495 citations In Library
09Learning Symmetries of Classical Integrable SystemsRoberto Bondesan, A. Lamacraft2019 · arXiv.org · 44 citations In Library
10Hamiltonian Neural NetworksS. Greydanus, Misko Dzamba, J. Yosinski2019 · Neural Information Processing Systems · 1.3k citations In Library
11AI Feynman: A physics-inspired method for symbolic regressionS. Udrescu, Max Tegmark2019 · Science Advances · 1.3k citations In Library
12Flow-based generative models for Markov chain Monte Carlo in lattice field theoryM. Albergo, G. Kanwar, P. Shanahan2019 · Physical Review D · 282 citations In Library

Scroll for more · 25 remaining

Similar papers

AI Poincaré 2.0: Machine Learning Conservation Laws from Differential EquationsZiming Liu, Varun Madhavan, M. Tegmark2022 · Physical Review E · 39 citations In Library
Discovering conservation laws from trajectories via machine learningSeungwoong Ha, Hawoong Jeong2021 · arXiv.org · 12 citations In Library
Machine Learning Conservation Laws of Dynamical systemsMebratie, Meskerem Abebaw, Rüdiger Nather, Guido Falk von Rudorff et al.2024 · arXiv (Cornell University) · 4 citations In Library
Model-agnostic machine learning of conservation laws from dataShivam Arora, Alexander Bihlo, Rudiger Brecht et al.2023 · arXiv.org · 3 citations In Library
From Data to Laws: Neural Discovery of Conservation Laws Without False PositivesRahul Ray2026 · 0 citations In Library
Machine learning independent conservation laws through neural deflationWei Zhu, Hong-Kun Zhang, P. Kevrekidis2023 · 5 citations In Library
Learning Poisson systems and trajectories of autonomous systems via Poisson neural networksPengzhan Jin, Zhen Zhang, I. Kevrekidis et al.2020 · IEEE Transactions on Neural Networks and Learning Systems · 65 citations In Library
Machine learning of independent conservation laws through neural deflation.Wei Zhu, Hong-Kun Zhang, P. Kevrekidis2023 · Physical Review E · 6 citations In Library
Reconstruction of observed mechanical motions with Artificial Intelligence toolsA. Jakovác, M. T. Kurbucz, P. Pósfay2022 · New Journal of Physics · 10 citations In Library
AI may help predict previously unseen states in dynamical systemsAlane Lim2021 · Scilight In Library
© 2026 NYSGPT2525 LLC