mlsanityData to 13 Oct 2026

Hypothesize, Evaluate, Refine: A Scientific Agent for PDE Discovery with Unknown Spatial Coefficient Fields

arXiv 2608.27475 ↗First posted 19 Aug 2026cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Discovering PDEs in heterogeneous media requires jointly identifying the governing operator and the unknown spatial fields that parameterize it. These tasks are coupled: changing field placement changes the differential law, while a sufficiently flexible field can conceal structural error on a single trajectory. We present Hypothesize, Evaluate, Refine for PDE Discovery (HER-PDE), a scientific-agent framework that discovers compositional PDE structure together with nonparametric, time-invariant coefficient fields. The Agent analyzes two noisy trajectories generated by different excitations, proposes complete expression-tree hypotheses, and combines creative structural exploration with local candidate refinement. Its Hypothesis Evaluation Interface (HEI) estimates only the fields explicitly declared in each hypothesis, never adds missing terms, and scores structures by bidirectional cross-excitation transfer. The selected law is subsequently audited on a sealed temporal interval. Across five controlled two-dimensional systems observed with 5 percent relative Gaussian state noise, the Agent recovers the generating operator in all five cases, including equivalent signed-field and product-rule parameterizations. Across nine unknown coefficient fields, the recovered fields attain a median Pearson correlation of approximately 0.85 and a median relative L2 error of approximately 0.28. These results show that agent-guided hypothesis refinement can recover heterogeneous governing laws without prescribing a parametric form for their spatial coefficients.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 49 of its 91 references that are in the map. First posted on arXiv 19 Aug 2026.

Builds on

49 of 91 references.

References · 49

  1. 2019AI Feynman: a Physics-Inspired Method for Symbolic Regressioncited by 540
  2. 2020Discovering Symbolic Models from Deep Learning with Inductive Biasescited by 339
  3. 2017PDE-Net: Learning PDEs from Datacited by 306
  4. 2018PDE-Net 2.0: Learning PDEs from Data with A Numeric-Symbolic Hybrid Deep Networkcited by 208
  5. 2020Physics-informed learning of governing equations from scarce datacited by 167
  6. 2020Weak SINDy For Partial Differential Equationscited by 154
  7. 2024LLM-SR: Scientific Equation Discovery via Programming with Large Language Modelscited by 123
  8. 2024LLM and Simulation as Bilevel Optimizers: A New Paradigm to Advance Physical Scientific Discoverycited by 84
  9. 2024Symbolic Regression with a Learned Concept Librarycited by 78
  10. 2019DeepMoD: Deep learning for Model Discovery in noisy datacited by 70
  11. 2022Symbolic Physics Learner: Discovering governing equations via Monte Carlo tree searchcited by 63
  12. 2023ODEFormer: Symbolic Regression of Dynamical Systems with Transformerscited by 60
37 more
  1. 2024KAN-ODEs: Kolmogorov-Arnold Network Ordinary Differential Equations for Learning Dynamical Systems and Hidden Physicscited by 58
  2. 2020DLGA-PDE: Discovery of PDEs with incomplete candidate library via combination of deep learning and genetic algorithmcited by 45
  3. 2021Any equation is a forest: Symbolic genetic algorithm for discovering open-form partial differential equations (SGA-PDE)cited by 38
  4. 2025SR-Scientist: Scientific Equation Discovery With Agentic AIcited by 30
  5. 2021Discovering Sparse Interpretable Dynamics from Partial Observationscited by 25
  6. 2021PDE-READ: Human-readable Partial Differential Equation Discovery using Deep Learningcited by 24
  7. 2022Discovering Nonlinear PDEs from Scarce Data with Physics-encoded Learningcited by 24
  8. 2022Deep Learning and Symbolic Regression for Discovering Parametric Equationscited by 22
  9. 2021Robust discovery of partial differential equations in complex situationscited by 21
  10. 2022PDE-LEARN: Using Deep Learning to Discover Partial Differential Equations from Noisy, Limited Datacited by 20
  11. 2020Deep-learning of Parametric Partial Differential Equations from Sparse and Noisy Datacited by 19
  12. 2022Discovery of partial differential equations from highly noisy and sparse data with physics-informed information criterioncited by 16
  13. 2023Weak-PDE-LEARN: A Weak Form Based Approach to Discovering PDEs From Noisy, Limited Datacited by 16
  14. 2020Data-Driven Discovery of Coarse-Grained Equationscited by 14
  15. 2020Variational system identification of the partial differential equations governing microstructure evolution in materials: Inference over sparse and spatially unrelated datacited by 13
  16. 2025Generative Discovery of Partial Differential Equations by Learning from Math Handbookscited by 12
  17. 2026Influence-Guided Symbolic Regression: Scientific Discovery via LLM-Driven Equation Search with Granular Feedbackcited by 11
  18. 2024Discovering physical laws with parallel symbolic enumerationcited by 9
  19. 2023Physics-constrained robust learning of open-form partial differential equations from limited and noisy datacited by 8
  20. 2021Physics-Guided Discovery of Highly Nonlinear Parametric Partial Differential Equationscited by 6
  21. 2023Discovering stochastic partial differential equations from limited data using variational Bayes inferencecited by 6
  22. 2026GENSR: Symbolic Regression Based in Equation Generative Spacecited by 6
  23. 2021Bayesian data-driven discovery of partial differential equations with variable coefficientscited by 4
  24. 2024OptPDE: Discovering Novel Integrable Systems via AI-Human Collaborationcited by 4
  25. 2025Mechanistic PDE Networks for Discovery of Governing Equationscited by 4
  26. 2026LLM-ODE: Data-driven Discovery of Dynamical Systems with Large Language Modelscited by 4
  27. 2023Machine-Guided Discovery of a Real-World Rogue Wave Modelcited by 3
  28. 2024A Data-Driven Framework for Discovering Fractional Differential Equations in Complex Systemscited by 3
  29. 2026Data-driven discovery of governing differential equations across physical systemscited by 3
  30. 2026Language models guide symbolic equation discovery by controlling searchcited by 3
  31. 2024Simplest Mechanism Builder Algorithm (SiMBA): An Automated Microkinetic Model Discovery Toolcited by 2
  32. 2025Neural Networks Enabled Discovery On the Higher-Order Nonlinear Partial Differential Equation of Traffic Dynamicscited by 2
  33. 2025Beyond empirical models: Discovering new constitutive laws in solids with graph-based equation discoverycited by 2
  34. 2026Data-driven sparse identification of governing PDEs via knockoff filters and multi-criteria trade-offscited by 2
  35. 2026Joint discovery of governing partial differential equations from multi-source datasets by competitive optimizationcited by 1
  36. 2026LLM-PDESR: Robust PDE Discovery via Subdomain Weighted Residuals and LLM-Guided Symbolic Hypothesis Generationcited by 1
  37. 2026Dynamics-aware identification of governing equations from sparse and noisy datacited by 1

Built on by

Not cited yet.