arXiv subjects

Nathaniel Chen

Publications and source records attributed to Nathaniel Chen.

8 recordsLinked to original sources

arXiv2026

Is Your AI Fast Enough to Run a Fusion Reactor?

In magnetic confinement fusion, compact machine learning algorithms are often integrated into the control loop on CPU-based custom hardware stacks. Deterministic implementations of these algorithms target instabilities which can evolve over microseconds to milliseconds. Staying within this framework may therefore limit the pace of control-system development to the inference time available. With the expansion of hardware acceleration and large models, we aim to bring ML models for fusion control to modern inference engines and compare, based on the physics reactor, the suitability of each engine per model. This study implements a consistent real-time machine learning algorithm for GPUs and optimized CPU packages, including OpenVINO, XLA, and TVM, at DIII-D National Fusion Facility. For the smallest model, one CPU core responds within 3.6 $μ$s in 99% of control cycles and an NVIDIA A100 within 28 $μ$s. We also evaluate models running simultaneously, as they would run in a typical experiment, and find best use cases per framework.

physics.plasm-ph↗

arXiv2026

A general extractor for coherent patterns across time-frequency measurements

Scientific discovery increasingly depends on finding meaningful patterns within volumes of sensor data too large for manual inspection. Existing time-frequency event extractors are often engineered or trained on manually labeled examples for each instrument, which can be laborious and difficult to adapt to new environments. To address this, we introduce TokEye, a reference-free method that learns a general criterion for coherent structure from heterogeneous fusion measurements. A model developed under this method, without retraining, can extract signals from various other fields including marine bioacoustics and wideband radar. It recovers 82.5\% of expert-annotated stellarator signals, detects 78\% of dolphin whistles, and identifies radar events with 81.5\% average precision across six benchmarks, while processing full experimental spectrograms in under one second on a GPU. This transfer across unrelated sensors and physical systems reveals common exploitable structure in time-frequency measurements and establishes a scalable foundation for automated scientific signal discovery.

eess.SP↗

arXiv2026

Tokamak-SI: Interpreting Tokamak Discharges with Labeled Catalogs

Finding which phenomena occur in a tokamak discharge, and when, is often a manual task. Previous efforts at labeling events have been task-specific, each with its own corpus of data. We introduce Tokamak-SI, which combines the labels and labeling methods of many earlier studies into a single scalable, time-aligned, and categorical dataset covering observations from confinement, detachment, alfven eigenmodes (AE), edge localized modes (ELM), sawtooth, tearing modes (TM), and resistive wall modes in DIII-D tokamak. Models based on this, which can extend the labels to new shots, match or beat earlier detectors on its labels, including for AE, ELM, and TM detection. Tokamak-SI is intended to enable task-specific benchmarking, physics-searchable archives, and direct identification of key events during an experiment.

physics.plasm-ph↗

arXiv2026

IGNITE Tokamak World Model Architecture

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.

physics.plasm-ph↗

arXiv2026

Towards Large-Scale Heterogeneous Data Organization for Scientific Foundation Models: A Nuclear Fusion Case Study

Training effective foundation models requires massive and organized datasets, yet scientific domains such as nuclear fusion present unique challenges due to largely heterogeneous and sparse data. Here we characterize the data used in developing such a model: with over 20 sensor types spanning 5 orders of magnitude in sampling rate, mixed tensor structures (point measurements, spectrograms, images), and nonstationary physics. We analyze our input complexity and discuss trade-offs between temporal context and frequency resolution. Our analysis provides a template for representing multi-modal fluctuation data at scale, with implications for both multi-modal control systems and nuclear fusion.

physics.plasm-ph↗

arXiv2026

Compact Experimental Negative TriAngUlarity Reactor (CENTAUR): A design study for a compact, affordable breakeven tokamak

This work presents the compact experimental negative triangularity reactor (CENTAUR), a low overnight cost, high-field tokamak, breakeven reactor design, achieving a predicted total fusion power of 40MW and scientific energy gain of 1.3. Ballooning stability calculations confirm that the device's pedestal is within the first stability regime, which is consistent with the expected ELM-free operation associated with negative triangularity (NT) plasmas. The geometry of the NT divertor allows for high fraction of radiated power (13.5$\%$) between the separatrix and plasma facing components. Heat transport modeling based on simulations of the edge region show heat loads into plasma facing components well below material limits. The magnet system employs rare-earth barium copper oxide (REBCO) high-temperature superconductors in 18 toroidal field coils, an hourglass-shaped central solenoid, and six poloidal field coils to support high-field ($B_0=10.9$ T) plasma confinement, shaping, and current drive. Neutronics analysis shows that a 12 cm $B_4C$ shield keeps superconducting magnet heating below the 33~K quench limit during 10 s, 40 MW DT pulses. With this shielding, the modeled fluence indicates HTS components can survive more than ten times the 3000-pulse design lifetime. Iteration of economic analysis in tandem with the technical design process allows CENTAUR to achieve its overnight cost goal of $\$$2B determined using a custom costing model that predicts a total overnight cost of $1.6$B$\pm0.2$B.

physics.plasm-ph↗

arXiv2026

Beyond the Loss Curve: Scaling Laws, Active Learning, and the Limits of Learning from Exact Posteriors

How close are neural networks to the best they could possibly do? Standard benchmarks cannot answer this because they lack access to the true posterior p(y|x). We use class-conditional normalizing flows as oracles that make exact posteriors tractable on realistic images (AFHQ, ImageNet). This enables five lines of investigation. Scaling laws: Prediction error decomposes into irreducible aleatoric uncertainty and reducible epistemic error; the epistemic component follows a power law in dataset size, continuing to shrink even when total loss plateaus. Limits of learning: The aleatoric floor is exactly measurable, and architectures differ markedly in how they approach it: ResNets exhibit clean power-law scaling while Vision Transformers stall in low-data regimes. Soft labels: Oracle posteriors contain learnable structure beyond class labels: training with exact posteriors outperforms hard labels and yields near-perfect calibration. Distribution shift: The oracle computes exact KL divergence of controlled perturbations, revealing that shift type matters more than shift magnitude: class imbalance barely affects accuracy at divergence values where input noise causes catastrophic degradation. Active learning: Exact epistemic uncertainty distinguishes genuinely informative samples from inherently ambiguous ones, improving sample efficiency. Our framework reveals that standard metrics hide ongoing learning, mask architectural differences, and cannot diagnose the nature of distribution shift.

cs.LG↗

arXiv2025

Regulation Compliant AI for Fusion: Real-Time Image Analysis-Based Control of Divertor Detachment in Tokamaks

While artificial intelligence (AI) has been promising for fusion control, its inherent black-box nature will make compliant implementation in regulatory environments a challenge. This study implements and validates a real-time AI enabled linear and interpretable control system for successful divertor detachment control with the DIII-D lower divertor camera. Using D2 gas, we demonstrate feedback divertor detachment control with a mean absolute difference of 2% from the target for both detachment and reattachment. This automatic training and linear processing framework can be extended to any image based diagnostic for regulatory compliant controller necessary for future fusion reactors.

cs.LG↗