Reinforcement learning steers a tokamak away from tearing instabilities
A controller trained on past DIII-D shots forecast tearing-mode instabilities up to 300 ms ahead and adjusted the plasma in real time to avoid them, holding high-performance conditions that would otherwise have collapsed.
- Model
- Deep reinforcement learning controller over a learned plasma model
- Field
- Physics
- Date
- 2024-02-21
- Human collaborators
- Jaemin Seo, Egemen Kolemen
What was found
Tearing modes break up the magnetic surfaces confining a fusion plasma and can end a discharge. Existing approaches suppress them once they appear; this one predicts them. A network trained on past DIII-D discharges estimates the instability probability, and a reinforcement-learning policy trained against a learned plasma model adjusts beam power and plasma shape to keep that probability low while holding pressure high. It was tested live on DIII-D, not only in simulation. Published in Nature.
Novelty check
Distinct from the 2022 DeepMind–TCV work already in this registry, which learned magnetic shape control rather than instability avoidance. Tearing-mode prediction from machine learning had been published before; closing the loop so the controller acts on the prediction during a live shot is the new part.
Caveats and known objections
Demonstrated on one machine in specific scenarios. A data-driven controller trained on DIII-D discharges has no guarantee of transferring to another tokamak, still less to ITER-scale burning plasma where the training data does not exist. Tearing modes are one disruption pathway among several. The physics model, actuators and safety limits are human-designed; the policy operates inside them.
Nobody outside the lab has checked this yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Peer reviewed and Search scaffold.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Graded peer reviewed for verification and search scaffold for autonomy. What these mean.
Cite this entry
whataifound.org. (2024). Reinforcement learning steers a tokamak away from tearing instabilities. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2024-02-21-tearing-mode-avoidanceBibTeX
@misc{whataifound-princetonuniversity-2024-avoidance,
title = {Reinforcement learning steers a tokamak away from tearing instabilities},
author = {{whataifound.org}},
year = {2024},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by Princeton University / PPPL / DIII-D National Fusion Facility. Verification: Peer reviewed. Autonomy: Search scaffold.},
url = {https://whataifound.org/finding/2024-02-21-tearing-mode-avoidance}
}