New Offline Reinforcement Learning Benchmark Released for Nuclear Fusion Plasma Control
A team of researchers has introduced RL4F, a standardized offline reinforcement learning benchmark for plasma control in nuclear fusion tokamaks, built on historical data from the DIII-D reactor. The benchmark addresses a gap in the field where progress has been hard to measure due to the absence of a common evaluation framework for multi-actuator, long-horizon control problems. It provides the fusion and machine learning communities with shared tools to accelerate algorithm development without requiring costly or risky experiments on real devices.
Researchers have published RL4F, an open-source benchmark designed to evaluate offline reinforcement learning algorithms for controlling plasmas in nuclear fusion tokamaks. Because running trial-and-error experiments directly on real fusion devices is expensive and potentially damaging, offline RL — which learns from pre-collected historical data rather than live interaction — is an attractive alternative. The benchmark includes closed-loop evaluation environments and baseline comparisons across four plasma profile tracking tasks: rotation, density, temperature, and pressure. The underlying dynamics model is derived from historical discharge data from DIII-D, a real-world tokamak operated in the United States. Evaluations of a broad set of imitation learning and offline RL methods found that offline model-based RL approaches achieved the best average performance across most objectives, though no single algorithm dominated every task. The authors open-source the codebase, datasets, and evaluation framework, aiming to serve both the fusion energy community and the broader offline RL research community. The paper spans 23 pages and was submitted to arXiv in May 2026.
What's missing
The paper has not yet undergone formal peer review, as it is a preprint posted to arXiv. Key open questions include how well the DIII-D-derived dynamics model generalizes to other tokamak designs, whether the benchmark tasks capture the full complexity of real-time plasma control, and how performance on the benchmark translates to actual deployment on live fusion devices.
What different sources said
- arXiv cs.LGCenter
Offline Reinforcement Learning for Rotation Profile Control in Tokamaks
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.