Princeton Plasma Physics Laboratory is leading a new initiative with the goal of using AI technology to accelerate the development of fusion energy research through high-fidelity computer simulations. The project includes national laboratories, universities, technology companies, and other partners.

Simulation, Technology, and Experiment Leveraging Learning-Accelerated Research enabled by AI (STELLAR-AI) has been developed as part of the Department of Energy's Genesis Mission, which was established by presidential executive order last year to speed up the application of AI in scientific research.

Connection to NSTX-U: The platform is expected to substantially reduce the time required to run fusion-related simulations and develop AI programs "capable of designing an ideal fusion system," according to PPPL. It connects computing resources directly to experimental devices, such as PPPL's National Spherical Torus Experiment-Upgrade (NSTX-U), allowing researchers to analyze data as the experiments are being conducted.

PPPL deputy director for research Jonathan Menard said that fusion is "a complex system of systems. We need AI and high-performance computing to really optimize the design for economic construction and operation. We want to link simulation technology and experiments-in particular, NSTX-U-with AI and partnerships to get to accelerated fusion."

One of the many research projects for which STELLAR-AI is being used is the creation of a digital twin of NSTX-U. Another project, StellFoundry, is using AI for the design of stellarators-fusion devices that have certain advantages over tokamaks. These and other research undertakings with STELLAR-AI "span simulation, design and real-time experiment support," in the words of PPPL.

Computing power: STELLAR-AI's capabilities will rely on an integrated system of central processing units (CPUs), graphics processing units (GPUs), and quantum processing units (QPUs) in a configuration specifically designed for fusion research. CPUs are standard computer chips; GPUs are specialized chips for the parallel calculations needed in AI; and QPUs are units based on the principles of quantum physics to deal with extremely complex calculations.

Partners, a mission, and a road map: Among PPPL's partners in the STELLAR-AI project are other national labs, Princeton University, the Massachusetts Institute of Technology, the University of Wisconsin-Madison, Nvidia, Microsoft, Commonwealth Fusion Systems, General Atomics, Type One Energy, Realta Fusion, and the U.K. Atomic Energy Authority.

The expressed objective of the Genesis Mission is the development of an "integrated platform that connects the world's supercomputers, experimental facilities, AI systems, and unique datasets across every major scientific domain to double the productivity and impact of American research and innovation within a decade." STELLAR-AI will contribute to the Genesis Mission by adding fusion-specific computer codes, data, and scientific models into the national system.

Shantenu Jha, the head of PPPL's Computational Sciences Department, said, "The Genesis platform is an integrated, ambitious system that will bring together the various unique DOE assets: experimental and user facilities, the supercomputers, data archives and, importantly, the AI models."

STELLAR-AI is also part of the DOE's Fusion Science and Technology Roadmap, which sets objectives for building an AI-Fusion Digital Convergence platform to accelerate fusion power commercialization.