Research map / Bandits and online learning / Online convex optimization and games
Online Imitation Learning for Stabilizing Vlasov--Poisson Plasmas
Xiaofan Xia, Qin Li, Wenlong Mou
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Studies online imitation learning to stabilize Vlasov-Poisson plasmas under partial (macroscopic) observations, learning from a fully observable expert controller.
Problem
Controlling instability in Vlasov-Poisson plasma systems for nuclear fusion when only macroscopic measurements of the plasma state are available rather than the full state.
Approach
Imitation learning from a fully observable expert controller, run in an online fashion under macroscopic measurements; the paper theoretically separates online from offline imitation learning in this partially observed setting.
Key results
- Theoretically establishes a separation between online and offline imitation learning under partial observability (per abstract).
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase offline BC dataset from 400 to 1200 trajectories (ablation) | behavior cloning · 1D Vlasov-Poisson two-stream (electric energy) | long-horizon stability: 400 -> 1200 trajectories | No clear effect |
| add one DAgger iteration with 800 aggregated trajectories to BC-400 | BC-400 · 1D Vlasov-Poisson two-stream (electric energy) | electric energy over long horizon | Improved |
| online DAgger-style training instead of offline behavior cloning | history-based plasma controller · 1D Vlasov-Poisson phase-space evolution | phase-space distortion | Improved |
Closest papers by meaning
- Multi-Timescale Dynamics Model Bayesian Optimization for Plasma Stabilization in Tokamaks (ICML 2025 · similarity 0.85)
- A Memory Efficient Unified Algorithm for Online Learning of Linear Dynamical Systems (NeurIPS 2026 · similarity 0.83)
- Parametric model reduction of mean-field and stochastic systems via higher-order action matching (NeurIPS 2024 · similarity 0.82)
- GyroSwin: 5D Surrogates for Gyrokinetic Plasma Turbulence Simulations (NeurIPS 2025 · similarity 0.81)
- Stabilizing Backpropagation Through Time to Learn Complex Physics (ICLR 2024 · similarity 0.81)
- Stable Port-Hamiltonian Neural Networks (NeurIPS 2025 · similarity 0.80)
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.
