Research map / Time series / Causal and generative time series
Weak Form Generalized Hamiltonian Learning
Kevin Course, Trefor Evans, Prasanth Nair
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace state regression (adjoint ODE solve) with weak form regression | FCNN pendulum model · Nonlinear pendulum | training time: 25min 46s -> 34s | Improved |
| replace state regression with weak form regression | FCNN pendulum model · Nonlinear pendulum | state error: 0.48 -> 0.17 | Improved |
| replace derivative regression with weak form regression | FCNN pendulum model · Nonlinear pendulum, noise 0.1 | state error: Diverged -> 0.17 | Improved |
| replace derivative regression with weak form regression | FCNN pendulum model · Nonlinear pendulum | derivative error: 1.35 -> 0.15 | Improved |
| increase measurement frequency from 10Hz to 100Hz | state regression · Nonlinear pendulum | training time: 4:23:59 -> 1:45:22 | Improved |
| increase measurement frequency from 10Hz to 100Hz | weak form regression · Nonlinear pendulum | training time: 0:00:32 -> 0:00:32 | No clear effect |
| increase measurement frequency | weak form regression · Nonlinear pendulum | state error | Improved |
| increase measurement frequency | state regression · Nonlinear pendulum | state error | Hurt |
| use generalized Hamiltonian structure (GHNN with prior) instead of unstructured FCNN | GHNN · Nonlinear pendulum | derivative error: 0.43 -> 0.07 | Improved |
| use generalized Hamiltonian structure (GHNN with prior) instead of unstructured FCNN | GHNN · Nonlinear pendulum | state error: 0.04 -> 0.08 | Hurt |
| use GHNN instead of HNN, which assumes energy conservation | GHNN · Nonlinear pendulum | state error: 3.20 -> 0.08 | Improved |
| use GHNN instead of HNN | GHNN · Duffing | state error: 1.61 -> 0.31 | Improved |
Closest papers by meaning
- Physics-Informed Regularization for Domain-Agnostic Dynamical System Modeling (NeurIPS 2024 · similarity 0.88)
- Data-driven Prediction of General Hamiltonian Dynamics via Learning Exactly-Symplectic Maps (ICML 2021 · similarity 0.87)
- Learning long range dependencies through time reversal symmetry breaking (NeurIPS 2025 · similarity 0.86)
- HHD-GP: Incorporating Helmholtz-Hodge Decomposition into Gaussian Processes for Learning Dynamical Systems (NeurIPS 2024 · similarity 0.86)
- Towards Cross Domain Generalization of Hamiltonian Representation via Meta Learning (ICLR 2024 · similarity 0.85)
- A Riemannian Framework for Learning Reduced-order Lagrangian Dynamics (ICLR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
