Physics-Informed Neural ODEs for IBR Dynamics
Neural ODEs and Learned Vector Fields
C3-03 · Module outline
Learn a derivative law, then integrate it to predict trajectories.
Learning objectives
- Distinguish local derivative fit from rollout accuracy.
- Separate integration error from model error.
Concept outline
- One-dimensional learned dynamics
- Two-dimensional GFM vector field
- Euler, RK4, and extrapolation
Experiment direction · Planned
Simulation and code experiment
Compare learned and reference phase portraits with held-out initial conditions.
Model explanations, parameter settings, runnable code, and result interpretation will be developed here.
Practice and discussion
Explain how a small derivative error can accumulate in a rollout.
Worked solutions and feedback will accompany the full lesson.
Material preparation
Related notes, models, or notebooks are available and need adaptation into a web lesson and experiment.
Adapt the existing material into a lesson, a reproducible experiment, and worked practice.