Technology
Princeton researchers deploy AI system to manage fusion plasma
ScienceDaily reports that a newly developed artificial intelligence framework can control fusion plasma within milliseconds.
The short version
- Princeton researchers developed PACMAN, an artificial intelligence framework designed to monitor and manage fusion plasma in milliseconds.[ScienceDaily]
- The software routinely executes in roughly 20 milliseconds to oversee plasma and anticipate instabilities.[ScienceDaily]
- During tests across five physical experiments, the system forecast an impending tearing mode roughly 200 milliseconds before it occurred.[ScienceDaily]
Key facts
- Researchers at Princeton developed a system called PACMAN, which stands for Prediction And Control using MAchiNe learning, to manage fusion plasma.[ScienceDaily]
- The entire PACMAN framework generally completes its execution cycle in approximately 20 milliseconds.[ScienceDaily]
- The framework was tested on an active fusion apparatus across five distinct experiments.[ScienceDaily]
- In a trial run, a machine learning model forecast a tearing mode approximately 200 milliseconds ahead of time.[ScienceDaily]
What remains uncertain
- The reported findings and performance metrics reflect single-outlet coverage based on five initial experimental runs.[ScienceDaily]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.
- AI can now control fusion plasma faster than humans can reactScienceDaily - Science