Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

TokaMark: A Comprehensive Benchmark for MAST Tokamak Plasma Models

T0 review · 3 major / 3 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read TokaMark gives AI tokamak models a common benchmark on real data

desk verdict TokaMark is a genuinely useful benchmark artifact that deserves serious refereeing, but the evaluation protocol has fixable soft spots in missing-data handling and rare-event metrics. read the letter →

arxiv 2602.10132 v3 pith:VRCLCQ43 submitted 2026-02-05 physics.plasm-ph cs.AI

Cécile Rousseau , Samuel Jackson , Rodrigo H. Ordonez-Hurtado , Nicola C. Amorisco , Tobia Boschi , George K. Holt , Andrea Loreti , Eszter Székely
show 8 more authors
Alexander Whittle Adriano Agnello Stanislas Pamela Alessandra Pascale Robert Akers Juan Bernabe Moreno Sue Thorne Mykhaylo Zayats
This is my paper · ORCID
classification physics.plasm-phcs.AI
keywords tokamakplasmamodelingbenchmarkMASTdata-drivenfusionNRMSEmulti-modaldiagnosticsdisruptionprediction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces TokaMark, an open benchmark built from real MAST tokamak diagnostic data, with 14 standardized tasks across four difficulty groups, a hierarchical evaluation protocol, and a released multi-branch convolutional baseline. It aims to fix the fragmentation of fusion datasets by giving fusion and machine-learning researchers a shared measuring stick for comparing AI plasma models. If correct, it means models trained on different tokamaks or by different labs can be scored on the same tasks, and the baseline results show which plasma-prediction problems are currently tractable and which are not. The benchmark is open-sourced along with tooling and baseline training scripts.

What carries the argument

The benchmark itself is the central object: a signal taxonomy (time series, profiles, videos) with origins (diagnostics, actuators, derived), plus a windowing scheme that defines tasks as reconstruction, autoregressive forecasting, or reconstructive forecasting. The evaluation machinery is hierarchical NRMSE aggregation—samples to windows to signals to tasks to shots to groups—where each signal error is normalized by its empirical standard deviation. The baseline is a multi-branch convolutional encoder–decoder with a shared latent vector, using 1D, 2D, or 3D convolutions matched to input modality. This machinery converts the heterogeneity of multi-rate, incomplete tokamak data into comparabl

What would settle it

Run a trivial mean-predictor on the Group 4 targets: if it achieves NRMSE close to or below the reported baseline on any task, the headline numbers for that task do not demonstrate predictive skill. The benchmark could also be validated by recomputing scores with event-based metrics (detection rate vs. false alarms) on the same held-out windows; if rankings change materially, the NRMSE protocol is not capturing what Group 4 claims to measure.

Watch

Extended reading notes

Core claim

TokaMark is the first large, open benchmark for evaluating AI models on real tokamak diagnostic data. The authors curate 39 signals from 11,573 MAST discharges and define 14 tasks organized into four groups—equilibrium reconstruction, magnetics dynamics, profile evolution, and MHD/disruption forecasting—each with standardized input and output windows. Performance is summarized by a normalized root-mean-square error (NRMSE) computed per signal and aggregated hierarchically to tasks and groups. A generic multi-branch convolutional encoder–decoder baseline, trained separately per task, achieves group NRMSEs of 0.163, 0.126, 0.339, and 0.476, with the hardest task (Task 4-5, locked-mode precurso

Load-bearing premise

The load-bearing premise is that the normalized RMSE, computed after excluding test windows whose outputs are not fully available, faithfully captures predictive skill in the sparse, noisy regime; for rare-event tasks, most of a signal's variance comes from non-event periods, so a model that never predicts the event can still score near NRMSE=1.

Editorial extensions

If this is right

  • If adopted, TokaMark gives fusion-AI papers a common yardstick: any model can be scored on the same 14 tasks and compared against published baseline numbers.
  • The baseline's strong Group 1 and Group 2 scores (NRMSE below 0.17) suggest that equilibrium reconstruction and short-horizon magnetics forecasting are practical targets for real-time neural surrogates.
  • The weak Group 3 and Group 4 scores, including Task 4-5 above 1.0, mark where generic feed-forward architectures currently fail and where physics-informed or temporally structured models would need to improve.
  • The hierarchical metric isolates which signals drive a task's difficulty, so researchers can tell whether a model struggles with fast magnetics, slow transport, or rare-event precursors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension the authors leave implicit is adding event-based metrics—detection rate and false-alarm rate—alongside NRMSE for Group 4, since a model that never predicts a rare event can still score near NRMSE=1.
  • The selective filtering of test windows with incomplete outputs softens the benchmark's missing-data robustness claim; a stress test that deliberately masks random inputs at evaluation time would directly probe that property.
  • The baseline's fixed 50 ms input truncation for non-Markovian tasks may understate what longer-context models could achieve, so varying context length is a cheap way to test the benchmark's sensitivity to temporal memory.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 3 minor

Summary. The paper introduces TokaMark, a benchmark suite for evaluating AI models on real MAST tokamak data. It defines 14 standardized tasks in 4 groups (equilibrium reconstruction, magnetics dynamics, profile dynamics, and long-horizon forecasting of MHD/rare events), packages 39 diagnostic/actuator signals from 11,573 FAIR-MAST shots, and provides a shot-level split, window-based data loading, a hierarchical evaluation protocol with NRMSE aggregation, and a released multi-branch convolutional encoder–decoder baseline. Baseline results show low NRMSE for Groups 1–2 (0.163 and 0.126) and higher for Groups 3–4 (0.339 and 0.476).

Significance. If the evaluation protocol is valid, TokaMark would be a valuable community resource: it is openly released, includes standardized task definitions, a reproducible baseline, and a clean shot-level split, which are exactly what the fusion-ML community lacks. The shipment of code, data loaders, and baseline model is a concrete strength. However, the benchmark's central claim—that its scores provide a reliable common measuring stick—depends on the evaluation protocol faithfully measuring the capabilities the paper emphasizes: robustness to missing data and rare-event forecasting. The protocol has load-bearing weaknesses in both areas, so the current results must be treated with caution until those are addressed.

major comments (3)
  1. [§3.2 (Data Preparation) and §2.3 (Data-driven Challenges)] The test-set filtering rule in §3.2 excludes windows with all-NaN inputs and requires all output signals to be fully available over the prediction horizon. This directly contradicts §2.3's statement that 'Naively discarding shots or windows with missing components wastes valuable examples and can introduce distributional bias.' The held-out evaluation never exercises the sparse/incomplete-data regime that the paper motivates as a core challenge ('robustness to incomplete state information' in §1.2, Group 3 in §3.1.3). Consequently, the reported NRMSE values describe performance on a cleaned distribution and cannot substantiate the benchmark's robustness objective. Please either include missing-data windows with appropriate masking or partial-output metrics, or explicitly re-scope the robustness claim and report the fraction of excluded windows.
  2. [§3.3, Eq. (2)] The metric NRMSE_k = RMSE_k / σ_k, with σ_k computed over evaluation shots, is not a valid measure of early-warning skill for Group 4 rare-event tasks. The paper states that 'NRMSE_k = 1 corresponds to a model no better than approximating the signal by its mean'—this is only true for a constant prediction. For signals whose variance is dominated by quiescent periods (soft X-ray, Mirnov, plasma current), a model emitting a slowly varying baseline can achieve NRMSE well below 1 without ever predicting the event. Thus Task 4-1's NRMSE ≈ 0.344 does not by itself demonstrate MHD/early-warning capability. Please add event-focused metrics (e.g., AUC, detection latency, false-alarm rate) for Group 4, and present NRMSE only as a secondary signal-fidelity measure.
  3. [§4.2 (Experimental Settings) vs Table 2] The baseline for non-Markovian tasks truncates input context to 50 ms ('inputs for non-Markovian tasks are truncated to a duration of 50ms'), while Table 2 specifies 100 ms input windows for all Group 4 tasks, and §3.1.5 argues that long context is essential for these tasks. The baseline therefore does not evaluate the defined tasks; its scores are for a truncated variant. This undermines the 'realistic lower bound' claim in §4.3 for Group 4 and prevents fair comparison with future models that use the full specified context. Please either extend the baseline to accept the full specified windows, or explicitly report the truncation as an architectural limitation and re-state the baseline results as corresponding to that truncated version.
minor comments (3)
  1. [Abstract] The arXiv abstract states that the dataset and tooling 'are open-sourced' with a GitHub link, while the body abstract says 'The benchmark, documentation, and tooling will be fully open sourced upon acceptance.' Please reconcile these statements and confirm the public availability status.
  2. [Table 2] The table is poorly formatted in the manuscript text: the row for Task 3-3 is garbled (output window is missing), and the column alignment for Group 3 is unclear. Please fix the table so each task has explicit input and output window values.
  3. [§3.3, Eq. (2)] Equation (2) defines both RMSE_k and NRMSE_k in a single line, which is confusing. Consider splitting into two equations or adding a sentence clarifying the relationship.

Circularity Check

0 steps flagged · score 1.0 of 10

No material circularity; EFIT-derived targets are transparent surrogates, not hidden predictions.

full rationale

TokaMark is an empirical benchmark rather than a derivation, so the circularity patterns involving fitted parameters or self-citation chains do not apply. The split is at shot level, the test set is held out, and the reported NRMSE values are measured on that test set. The EFIT-derived equilibrium targets are explicitly labeled as 'derived signals' in Table 1 and the tasks are framed as 'fast, numerics-free surrogates'; using the same magnetics that EFIT consumes is a disclosed benchmark design, not a reduction of the prediction to its input. The test-set filtering (Sec. 3.2) and sigma-normalized NRMSE (Sec. 3.3, Eq. 2) are evaluation-interpretation concerns for rare-event tasks, not circularity: the 'mean predictor gives NRMSE=1' statement is definitionally true, and filtering affects distributional representativeness, not logical independence. No load-bearing self-citation appears: FAIR-MAST [17,18] is an open, externally accessible dataset and the baseline architecture cites external prior work [31,32]. No circular step is therefore identified.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The ledger is dominated by design choices and availability assumptions rather than physical free parameters. The central claim rests on the representativeness of FAIR-MAST, the validity of EFIT-derived targets, and the assumption that the evaluation protocol's filtering and normalization do not distort the intended missing-data and rare-event regime.

free parameters (4)
  • Baseline architecture hyperparameters = D=16, N=3, K=3, stride=3, padding=1, output_padding=1
    Hand-selected in Section 4.2 and identical across tasks; they affect the reported NRMSE scores but not the benchmark task definitions.
  • Training downsampling stride = 0.005 s (Markovian), 0.025 s (non-Markovian)
    Chosen in Section 4.2 to reduce training data; affects baseline results and effective time resolution.
  • Non-Markovian input truncation length = 50 ms
    Fixed by the architecture in Section 4.2; truncates long-horizon context for Group 3-4 tasks despite Table 2's input-window notation.
  • Task input/output window lengths and sliding stride = 5-100 ms windows; 0.001 s stride
    Hand-chosen in Table 2 and Section 3.2 to define task difficulty; they are load-bearing for what the benchmark measures.
assumptions (6)
  • domain assumption FAIR-MAST is the only openly available dataset of real tokamak diagnostics
    Stated in Section 1.2 without a systematic survey; if false, the 'first' claim weakens.
  • domain assumption The selected 39 signals sufficiently characterize plasma state for all 14 tasks
    Underlies task definitions in Section 2.1/Table 1; unmeasured physics is acknowledged but assumed learnable from these signals.
  • domain assumption Shot-level random split prevents information leakage from overlapping windows
    Section 3.2; reasonable but unvalidated, and windows from the same shot are highly correlated.
  • ad hoc to paper Test-set exclusions of incomplete windows do not bias evaluation
    Section 3.2 'Handling of missing information' applies constraints only to the test set, changing the distribution relative to real operational data.
  • domain assumption EFIT-derived equilibrium outputs are valid ground truth for reconstruction tasks
    Section 2.1; outputs are produced by EFIT from magnetic diagnostics, so Group 1-2 tasks may reward imitating EFIT rather than learning physics.
  • ad hoc to paper Baseline model results represent a realistic lower bound
    Section 4.3; no comparison to prior single-task models or tuned variants is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TokaMark: A Comprehensive Benchmark for MAST Tokamak Plasma Models." pith.science (2026). https://pith.science/paper/VRCLCQ43

@misc{pith2026260210132,
  author       = {Pith},
  title        = {Pith review of: TokaMark: A Comprehensive Benchmark for MAST Tokamak Plasma Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VRCLCQ43}},
  note         = {Machine review of arXiv:2602.10132}
}
read the original abstract

Development and operation of commercially viable fusion energy reactors such as tokamaks require accurate predictions of plasma dynamics from sparse, noisy, and incomplete sensors readings. The complexity of the underlying physics and the heterogeneity of experimental data pose formidable challenges for conventional numerical methods, and highlight the promise of modern data-native approaches. A major obstacle in realizing this potential is, however, the lack of curated, openly available datasets and standardized benchmarks. Existing fusion datasets are scarce, fragmented across institutions, facility-specific, and inconsistently annotated, which limits reproducibility and prevents a fair and scalable comparison of AI approaches. In this paper, we introduce TokaMark, a structured benchmark to evaluate AI models on real experimental data collected from the Mega Ampere Spherical Tokamak (MAST). TokaMark provides a comprehensive suite of tools designed to unify access to multi-modal fusion data and standardize evaluation protocols. The benchmark includes a curated list of 14 tasks spanning a range of physical mechanisms, exploiting a variety of diagnostics and covering multiple operational use cases. A baseline model is provided to facilitate transparent comparison and validation within a unified framework. By establishing a unified benchmark, TokaMark aims to accelerate progress in data-driven AI-based plasma modeling, contributing to the broader goal of achieving sustainable and stable fusion energy. The dataset, benchmark, documentation, and tooling are open-sourced under https://github.com/UKAEA-IBM-STFC-Fusion-FMs/tokamark_baseline.

Figures

Figures reproduced from arXiv: 2602.10132 by the authors.

Figure 1
Figure 1. Examples of multi-modal signals from FAIR-MAST [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Multi-branch convolutional encoder–decoder. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. TokaMind: A Multi-Modal Transformer Foundation Model for Tokamak Plasma Dynamics

    physics.plasm-ph 2026-02 conditional novelty 6.0 of 10

    A multi-modal transformer pretrained on MAST tokamak data beats the TokaMark CNN baseline on 13 of 14 tasks, and warm-start fine-tuning beats from-scratch training on the hardest forecasting and equilibrium tasks.

Reference graph

Works this paper leans on

54 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    Abbate, R

    J. Abbate, R. Conlin, and E. Kolemen. 2021. Data-driven profile prediction for DIII-D.Nuclear Fusion61, 4 (March 2021), 046027. doi:10.1088/1741-4326/abe08d Publisher: IOP Publishing

    work page doi:10.1088/1741-4326/abe08d 2021

  2. [2]

    Joseph Abbate, Rory Conlin, Ricardo Shousha, Keith Erickson, and Egemen Kolemen. 2023. A general infrastructure for data-driven control design and im- plementation in tokamaks.Journal of Plasma Physics89, 1 (Feb. 2023), 895890102. doi:10.1017/S0022377822001040

    work page doi:10.1017/s0022377822001040 2023

  3. [3]

    Abbate, E

    J. Abbate, E. Fable, G. Tardini, R. Fischer, E. Kolemen, and the ASDEX Upgrade Team. 2025. Combining physics-based and data-driven models for quantitatively accurate plasma profile prediction that extrapolates well; with application to DIII-D, AUG, and ITER tokamaks.Nuclear Fusion65, 5 (April 2025), 056014. doi:10.1088/1741-4326/adc283 Publisher: IOP Publishing

    work page doi:10.1088/1741-4326/adc283 2025

  4. [4]

    Appel and I

    L.C. Appel and I. Lupelli. 2018. Equilibrium reconstruction in an iron core tokamak using a deterministic magnetisation model.Computer Physics Communications 223 (2018), 1–17. doi:10.1016/j.cpc.2017.09.016

    work page doi:10.1016/j.cpc.2017.09.016 2018

  5. [5]

    Enrico Aymerich, Giuliana Sias, Fabio Pisano, Barbara Cannas, Sara Carcangiu, Carlo Sozzi, Chris Stuart, PJ Carvalho, Alessandra Fanni, and JET Contributors

  6. [6]

    Ian Char, Youngseog Chung, Joseph Abbate, Egemen Kolemen, and Jeff Schneider

  7. [7]

    R Michael Churchill. 2025. AI foundation models for experimental fusion tasks. Frontiers in Physics12 (2025), 1531334

    2025

  8. [8]

    R. M. Churchill, B. Tobias, Y. Zhu, and DIII-D team. 2020. Deep convolutional neural networks for multi-scale time-series classification and application to tokamak disruption prediction using raw, high temporal resolution diagnostic data.Physics of Plasmas27, 6 (June 2020), 062510. doi:10.1063/1.5144458

    work page doi:10.1063/1.5144458 2020

Show all 54 references
  1. [9]

    GF Counsell, RJ Akers, Lynton C Appel, D Applegate, KB Axon, Y Baranov, C Brickley, C Bunting, RJ Buttery, PG Carolan, et al . 2005. Overview of MAST results.Nuclear fusion45, 10 (2005), S157

    2005

  2. [10]

    PC De Vries, MF Johnson, B Alper, P Buratti, TC Hender, HR Koslowski, V Riccardo, Jet-Efda Contributors, et al. 2011. Survey of disruption causes at JET. Nuclear fusion51, 5 (2011), 053018

    2011

  3. [11]

    Jonas Degrave, Federico Felici, Jonas Buchli, Michael Neunert, Brendan Tracey, Francesco Carpanese, Timo Ewalds, Roland Hafner, Abbas Abdolmaleki, Diego de las Casas, Craig Donner, Leslie Fritz, Cristian Galperti, Andrea Huber, James Keeling, Maria Tsimpoukelli, Jackie Kay, An...

    2022

  4. [12]

    Qianyun Dong, Zhengwei Chen, Rongpeng Li, Zongyu Yang, Feng Gao, Yi- hang Chen, Fan Xia, Wulyu Zhong, and Zhifeng Zhao. 2025. Adapted Swin Transformer-based real-time plasma shape detection and control in HL-3.Nu- clear Fusion65, 2 (jan 2025), 026031. doi:10.1088/1741-4326/ada2fe

    2025 doi

  5. [13]

    AJH Donné, M Cox, N Sauthoff, and K Schoenberg. 2025. Beyond power gain: Toward a comprehensive milestone framework for all fusion energy concepts. Physics of Plasmas32, 9 (2025)

    2025

  6. [14]

    Ferreira, Pedro J

    Diogo R. Ferreira, Pedro J. Carvalho, and Horácio Fernandes. 2020. Deep Learning for Plasma Tomography and Disruption Prediction from Bolometer Data.IEEE Transactions on Plasma Science48, 1 (Jan. 2020), 36–45. doi:10.1109/TPS.2019. 2947304 arXiv:1910.13257 [physics]

    2020 arXiv

  7. [15]

    Freidberg

    Jeffrey P. Freidberg. 2007.Plasma Physics and Fusion Energy. Cambridge Univer- sity Press

    2007

  8. [16]

    B H Guo, D L Chen, B Shen, C Rea, R S Granetz, L Zeng, W H Hu, J P Qian, Y W Sun, and B J Xiao. 2021. Disruption prediction on EAST tokamak using a deep learning algorithm.Plasma Physics and Controlled Fusion63, 11 (Sept. 2021), 115007. doi:10.1088/1361-6587/ac228b Publisher: ...

    2021 doi

  9. [17]

    Samuel Jackson, Saiful Khan, Nathan Cummings, James Hodson, Shaun de Witt, Stanislas Pamela, Rob Akers, and Jeyan Thiyagalingam. 2025. An Open Data Service for Supporting Research in Machine Learning on Tokamak Data.IEEE Transactions on Plasma Science(2025). doi:10.1109/TPS.20...

    2025

  10. [18]

    Samuel Jackson, Saiful Khan, Nathan Cummings, James Hodson, Shaun de Witt, Stanislas Pamela, Rob Akers, Jeyan Thiyagalingam, and The MAST Team. 2024. FAIR-MAST: A fusion device data management system.SoftwareX27 (2024), 101869

    2024

  11. [19]

    Julian Kates-Harbeck, Alexey Svyatkovskiy, and William Tang. 2019. Predicting disruptive instabilities in controlled fusion plasmas through deep learning.Nature 568, 7753 (April 2019), 526–531. doi:10.1038/s41586-019-1116-4 Number: 7753 Publisher: Nature Publishing Group

    2019 doi

  12. [20]

    A. Kit, A. E. Järvinen, Y. R. J. Poels, S. Wiesen, V. Menkovski, R. Fischer, M. Dunne, and ASDEX-Upgrade Team. 2024. On learning latent dynamics of the AUG plasma state.Physics of Plasmas31, 3 (March 2024), 032504. doi:10.1063/5.0174128 arXiv:2308.14556 [physics]

    2024 arXiv

  13. [21]

    Spangher Lucas, Matteo Bonotto, William Arnold, Dhruva Chayapathy, Tommaso Gallingani, Alexander Spangher, Francesco Cannarile, Daniele Bigoni, Eliana De Marchi, and Cristina Rea. 2024. DisruptionBench: A robust benchmarking framework for machine learning-driven disruption pre...

    2024 doi

  14. [22]

    Hendrik Meyer, RJ Akers, F Alladio, Lynton C Appel, KB Axon, N Ben Ayed, P Boerner, RJ Buttery, PG Carolan, D Ciric, et al. 2009. Overview of physics results from MAST.Nuclear fusion49, 10 (2009), 104017

    2009

  15. [23]

    Kevin J. Montes. 2021.Interpretable Machine Learning for Prediction and A voidance of Disruptions in Tokamak Plasmas. Thesis. Massachusetts Institute of Technology. https://dspace.mit.edu/handle/1721.1/142684 Accepted: 2022-05-24T19:18:44Z

    2021

  16. [24]

    K. J. Montes, C. Rea, R. A. Tinguely, R. Sweeney, J. Zhu, and R. S. Granetz. 2021. A semi-supervised machine learning detector for physics events in tokamak discharges.Nuclear Fusion61, 2 (Jan. 2021), 026022. doi:10.1088/1741-4326/abcdb9 Publisher: IOP Publishing

    2021 doi

  17. [25]

    Morris and the MAST team

    A.W. Morris and the MAST team. 2002. Diagnostic Developments for the Mast Spherical Tokamak.Advanced Diagnostics for Magnetic and Inertial Fusion. Springer, Boston, MA.(2002). doi:10.1007/978-1-4419-8696-2_68

    2002 doi

  18. [26]

    David Orozco, Brian Sammuli, Jayson Barr, William Wehner, and David Humphreys. 2022. Neural Network-Based Confinement Mode Prediction for Real-Time Disruption Avoidance.IEEE Transactions on Plasma Science50, 11 (Nov. 2022), 4157–4164. doi:10.1109/TPS.2022.3198596 Conference Na...

    2022

  19. [27]

    Priyanka, J

    M. Priyanka, J. Sangeetha, and C. Jayakumar. 2024. A Review of Traditional and Data-Driven Approaches for Disruption Prediction in Different Tokamaks.E3S Web of Conferences477 (2024), 00039. doi:10.1051/e3sconf/202447700039

    2024

  20. [28]

    C. Rea, R. S. Granetz, K. Montes, R. A. Tinguely, N. Eidietis, J. M. Hanson, and B. Sammuli. 2018. Disruption prediction investigations using Machine Learning tools on DIII-D and Alcator C-Mod.Plasma Physics and Controlled Fusion60, 8 (June 2018), 084004. doi:10.1088/1361-6587...

    2018 doi

  21. [29]

    Riccardo Rossi, Michela Gelfusa, Andrea Murari, and on behalf of JET contributors

  22. [30]

    Raphael Schramm, Alexander Bock, Emiliano Fable, Joerg Stober, Marc Maraschek, Maximilian Reisner, Rainer Fischer, and Hartmut Zohm. 2024. Devel- opment and application of a predictive model for advanced tokamak scenario design.Nuclear Fusion(Jan. 2024). doi:10.1088/1741-4326/ad2062

    2024 doi

  23. [31]

    Jaemin Seo, Rory Conlin, Andrew Rothstein, SangKyeun Kim, Joseph Abbate, Azarakhsh Jalalvand, and Egemen Kolemen. 2023. Multimodal prediction of tearing instabilities in a tokamak. In2023 International Joint Conference on Neural Networks (IJCNN). IEEE, 1–8

    2023

  24. [32]

    Jaemin Seo, SangKyeun Kim, Azarakhsh Jalalvand, Rory Conlin, Andrew Roth- stein, Joseph Abbate, Keith Erickson, Josiah Wai, Ricardo Shousha, and Egemen Kolemen. 2024. Avoiding fusion plasma tearing instability with deep reinforce- ment learning.Nature626, 8000 (2024), 746–751

    2024

  25. [33]

    Jaemin Seo, Y.-S. Na, B. Kim, C. Y. Lee, M. S. Park, S. J. Park, and Y. H. Lee

  26. [34]

    Seo, Y.-S

    J. Seo, Y.-S. Na, B. Kim, C. Y. Lee, M. S. Park, S. J. Park, and Y. H. Lee. 2022. Development of an operation trajectory design algorithm for control of multiple 0D parameters using deep reinforcement learning in KSTAR.Nuclear Fusion62, 8 (July 2022), 086049. doi:10.1088/1741-...

    2022 doi

  27. [35]

    P Strand, DP Coster, M Plociennik, S de Witt, IA Klampanos, J Decker, F Imbeaux, JF Artaud, B Bosak, N Cummings, et al . 2022. A FAIR based approach to data sharing in Europe.Plasma Physics and Controlled Fusion64, 10 (2022), 104001

    2022

  28. [36]

    A Sykes, RJ Akers, LC Appel, ER Arends, PG Carolan, NJ Conway, GF Counsell, G Cunningham, A Dnestrovskij, Yu N Dnestrovskij, et al. 2001. First results from MAST.Nuclear Fusion41, 10 (2001), 1423

    2001

  29. [37]

    J. Vega, A. Murari, S. Dormido-Canto, G. A. Rattá, and M. Gelfusa. 2022. Disruption prediction with artificial intelligence techniques in tokamak plasmas.Nature Physics18, 7 (July 2022), 741–750. doi:10.1038/s41567-022-01602-2 Number: 7 Publisher: Nature Publishing Group

    2022 doi

  30. [38]

    J. T. Wai, M. D. Boyer, and E. Kolemen. 2022. Neural net modeling of equilibria in NSTX-U.Nuclear Fusion62, 8 (July 2022), 086042. doi:10.1088/1741-4326/ac77e6 Publisher: IOP Publishing

    2022 doi

  31. [39]

    Wakatsuki, M

    T. Wakatsuki, M. Yoshida, E. Narita, T. Suzuki, and N. Hayashi. 2023. Simulta- neous control of safety factor profile and normalized beta for JT-60SA using reinforcement learning.Nuclear Fusion63, 7 (May 2023), 076017. doi:10.1088/1741- 4326/acd393 Publisher: IOP Publishing

    2023 doi

  32. [40]

    Chenguang Wan, Zhi Yu, Alessandro Pau, Olivier Sauter, Xiaojuan Liu, Qiping Yuan, and Jiangang Li. 2023. A machine-learning-based tool for last closed-flux surface reconstruction on tokamaks.Nuclear Fusion63, 5 (April 2023), 056019. doi:10.1088/1741-4326/acbfcc Publisher: IOP ...

    2023 doi

  33. [41]

    Chenguang Wan, Zhi Yu, Feng Wang, Xiaojuan Liu, and Jiangang Li. 2021. Exper- iment data-driven modeling of tokamak discharge in EAST.Nuclear Fusion61, 6 (April 2021), 066015. doi:10.1088/1741-4326/abf419 Publisher: IOP Publishing

    2021 doi

  34. [42]

    Wang, Alessandro Pau, Cristina Rea, Oswin So, Charles Dawson, Olivier Sauter, Mark D

    Allen M. Wang, Alessandro Pau, Cristina Rea, Oswin So, Charles Dawson, Olivier Sauter, Mark D. Boyer, Anna Vu, Cristian Galperti, Chuchu Fan, Antoine Merle, Yoeri Poels, Cristina Venturini, Stefano Marchioni, and the TCV Team. 2025. Learning Plasma Dynamics and Robust Rampdown...

    2025 doi arXiv

  35. [43]

    Wang, Oswin So, Charles Dawson, Darren T

    Allen M. Wang, Oswin So, Charles Dawson, Darren T. Garnier, Cristina Rea, and Chuchu Fan. 2024. Active Disruption Avoidance and Trajectory Design for Tokamak Ramp-downs with Neural Differential Equations and Reinforcement Learning. (2024). doi:10.48550/ARXIV.2402.09387 Publish...

    2024 doi arXiv

  36. [44]

    1987.Tokamaks

    J Wesson. 1987.Tokamaks. Oxford University Press,New York, NY. https: //www.osti.gov/biblio/5589784

    1987

  37. [45]

    Bin Yang, Zhenxing Liu, Xianmin Song, Xiangwen Li, and Yan Li. 2020. Modeling of the HL-2A plasma vertical displacement control system based on deep learning and its controller design.Plasma Physics and Controlled Fusion62, 7 (May 2020), 075004. doi:10.1088/1361-6587/ab8a64 Pu...

    2020 doi

  38. [46]

    Zongyu Yang, Zhenghao Yang, Wenjing Tian, Jiyuan Li, Xiang Sun, Guohui Zheng, Songfen Liu, Niannian Wu, Rongpeng Li, Zhaohe Xu, Bo Li, Zhongbing Shi, Zhe Gao, Wei Chen, Xiaoquan Ji, Min Xu, and Wulyu Zhong. 2025. FusionMAE: large- scale pretrained model to optimize and simplif...

    2025

  39. [47]

    Ming Zhang, Qiqi Wu, Wei Zheng, Yangxing Shang, and Yuxing Wang. 2020. A database for developing machine learning based disruption predictors.Fusion Engineering and Design160 (Nov. 2020), 111981. doi:10.1016/j.fusengdes.2020. 111981

    2020 doi

  40. [48]

    Jinxiang Zhu, Cristina Rea, Robert S Granetz, Earl S Marmar, Kevin J Montes, Ryan Sweeney, Roy A Tinguely, DL Chen, Biao Shen, BJ Xiao, et al. 2021. Sce- nario adaptive disruption prediction study for next generation burning-plasma tokamaks.Nuclear Fusion61, 11 (2021), 114005

    2021

  41. [49]

    JX Zhu, Cristina Rea, Kevin Montes, RS Granetz, Ryan Sweeney, and Roy Alexan- der Tinguely. 2021. Hybrid deep-learning architecture for general disruption prediction across multiple tokamaks.Nuclear Fusion61, 2 (2021), 026007

    2021

  42. [50]

    group name

    J. X. Zhu, C. Rea, R. S. Granetz, E. S. Marmar, R. Sweeney, K. Montes, and R. A. Tinguely. 2023. Integrated deep learning framework for unstable event identifica- tion and disruption prediction of tokamak plasmas.Nuclear Fusion63, 4 (March 2023), 046009. doi:10.1088/1741-4326/...

    2023 doi

  43. [419]

    doi:10.1038/s41586-021-04301-9 Number: 7897 Publisher: Nature Publishing Group

    doi

  44. [2021]

    2021), 106010

    Feedforward beta control in the KSTAR tokamak by deep reinforcement learning.Nuclear Fusion61, 10 (Sept. 2021), 106010. doi:10.1088/1741-4326/ac121b Publisher: IOP Publishing

    2021 doi

  45. [2022]

    Disruption prediction at JET through deep convolutional neural networks using spatiotemporal information from plasma profiles.Nuclear Fusion62, 6 (2022), 066005

    2022

  46. [2024]

    http://arxiv.org/abs/2404.12416 arXiv:2404.12416 [physics]

    Full Shot Predictions for the DIII-D Tokamak via Deep Recurrent Networks. http://arxiv.org/abs/2404.12416 arXiv:2404.12416 [physics]

    arXiv

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.