IGNITE Tokamak World Model Architecture

2026-10-01

AI Summary

This study addresses the challenges of efficient simulation and experimental planning for tokamak plasmas by proposing a self-supervised generative world model. The approach introduces a novel multimodal spatiotemporal tokenizer that fuses heterogeneous data, including temporal sequences, images, and spectroscopic signals. Integrated with an autoregressive Transformer architecture, the model is pretrained in a self-supervised manner on a decade of unlabeled DIII-D operational data, enabling infinite-horizon predictions of plasma dynamics. Furthermore, it supports text- or goal-driven generation of actuator trajectories to facilitate full-discharge simulations. This work achieves high-fidelity modeling of plasma behavior, establishing a new paradigm for AI-driven nuclear fusion experimental planning.

Abstract

We introduce IGNITE, a generative world foundation model for fusion plasma behavior simulation trained in a self-supervised manner from over a decade of unlabeled experimental data at the DIII-D National Fusion Facility. The core of IGNITE is a dynamics model that can simulate DIII-D discharges from a given set of actuator trajectories. These trajectories can be supplied or generated on-the-fly from a textual prompt or from desired experimental outcomes. The model architecture consists of several spatio-temporal tokenizers that embed the different input modalities, including time-series like spatio-temporal measurement data, image sequences, and high-resolution spectrograms, each of which collected at vastly different time scales. The backbone is composed of an auto-regressive dynamics model that has the capacity to predict entire DIII-D discharges given initial latent plasma states and actuator trajectories over a theoretical infinite horizon. IGNITE paves the way towards efficient AI-driven experimental planning and world modeling for nuclear fusion.

Problem

• nuclear fusion
• plasma simulation
• world model
• tokamak
• multi-modal data

Innovation

• World Model
• Self-supervised Learning
• Spatio-temporal Tokenizers
• Autoregressive Dynamics Model
• Fusion Plasma Simulation