Princeton Plasma Physics Laboratory (PPPL) Colloquium

Advancements in AI are driving transformative progress in fusion energy. I will provide an overview and highlight the achievements of our group in this area: 1) Robust plasma state prediction even when there is diagnostic failure 2) Finding the minimal set of diagnostics needed to operate a reactor 3) Fusing data from multiple diagnostics (new Multimodal Super-Resolution technique) to discover hidden physics insights 4) Prediction of plasma evolution in future fusion reactors such as ITER by combining experimental data and simulations from multiple machines (DIII-D, AUG; ASTRA, TRANSP, TGLF,...) 5) Reinforcement learning control that achieves high performance fusion reactor operation without instabilities (tearing modes and ELMs) both at KSTAR and DIII-D.

I will then talk about our recent work to push AI for fusion forward with foundation models, AI-boosted diagnostics, automated plasma analysis.