Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

HIPED: Machine Learning Framework for Spherical Tokamak Pedestal Prediction and Optimization

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Random-forest models predict spherical-tokamak pedestal height at 76 percent variance explained.

desk verdict Useful MAST-U pedestal database study with a real scaling result, but the headline predictive accuracy is not yet established because the train/test split ignores discharge grouping. read the letter →

arxiv 2504.19861 v1 pith:FIRGSAKX submitted 2025-04-28 physics.plasm-ph

J.F. Parisi , J. G. Clark , J. W. Berkery , C. Bowman , C. J. Fitzpatrick , S. M. Kaye , M. Lampert This is my paper · ORCID
classification physics.plasm-ph
keywords pedestalpredictionrandomforestsphericaltokamakH-modeedge-localizedmodesParetooptimizationMAST-Ucontrolroomparameters
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

This paper introduces HIPED, a machine-learning framework that predicts and optimizes the edge pedestal in a spherical tokamak using thousands of H-mode time slices. Its central claim is that the standard power-law scaling used on conventional tokamaks—pedestal width proportional to the square root of normalized pedestal pressure—does not hold at low aspect ratio, where a random-forest model using normalized plasma pressure, elongation, and Greenwald fraction predicts pedestal height with $R^2\approx 0.71$–$0.76$. The paper also shows that a model trained on control-room parameters alone reaches $R^2\approx 0.674$, and that Pareto optimization over the fitted models identifies real discharges and the control settings that best trade competing objectives such as time to the next edge-localized mode and normalized plasma pressure. If the framework is sound, it gives experiments a practical way to estimate pedestal performance from knobs they can actually turn and to plan discharges against explicit priorities.

What carries the argument

The machinery is a Random Forest ensemble—hundreds of decision trees trained on bootstrapped data subsets, whose averaged output captures nonlinear interactions among many inputs—combined with SHAP feature attribution to say which drivers matter. The pedestal labels come from hyperbolic-tangent fits to density and temperature profiles within a Bayesian multi-diagnostic inference system. For optimization, the paper samples the input space synthetically, identifies Pareto-front points across four objectives, and scores real discharges with a weighted sum $S_j=\sum_i w_i \hat{s}_{ij}$, time-averaged over a 0.05 s window to favor sustained performance.

What would settle it

Retrain the same random-forest models with a grouped split that keeps every time slice of a discharge in either training or testing; if the held-out $R^2$ for $\beta_{\theta,\mathrm{ped}}$ falls toward the 0.62 level of the power-law regressions rather than the claimed 0.76, the accuracy claim is refuted. A second check is to test the trained models on a completely new experimental campaign without any retraining.

Watch

Extended reading notes

Core claim

Using data from the third MAST-U campaign, the paper fits electron density and temperature pedestals with hyperbolic-tangent profiles and finds that fitting pedestal width $\Delta_{\mathrm{ped}}$ to normalized pedestal pressure $\beta_{\theta,\mathrm{ped}}$ gives $R^2 < 0.13$, so the conventional $\Delta_{\mathrm{ped}}\propto\sqrt{\beta_{\theta,\mathrm{ped}}}$ scaling fails at low aspect ratio. A random-forest regressor predicts $\beta_{\theta,\mathrm{ped}}$ on a held-out 70 percent of the data with $R^2\approx 0.755$ (0.713 for pre-ELM slices, 0.753 for ELM-free slices), with $\beta_N$ the dominant feature; removing $\beta_N$ drops $R^2$ to 0.585. Separate forests predict pedestal density with $R^2=0.64$ and pedestal temperature with $R^2=0.73$. Restricting inputs to seven control-room parameters still gives $R^2\approx 0.674$. The same models are used to generate 50,000 synthetic points, build Pareto fronts over normalized pressure, Greenwald fraction, line-averaged density, and time-to-next-ELM, and rank actual discharges with a weighted Multi-Criteria Decision-Making score, yielding the control-room trajectories that approach each Pareto-optimal regime.

Load-bearing premise

The load-bearing assumption is that the 7,481 H-mode time slices are independent samples, because the 30/70 train/test split is random across time slices rather than grouped by discharge, and slices from the same shot therefore appear on both sides of the split and can inflate the reported $R^2$ values.

Editorial extensions

If this is right

  • At low aspect ratio, pedestal width cannot be treated as a single-variable function of pedestal pressure; models that include global pressure, elongation, and Greenwald fraction are needed.
  • Operators can estimate pedestal height to about 67 percent variance explained using only control-room quantities, so real-time or near-real-time guidance is plausible without full profile reconstruction.
  • Pareto-optimal discharges that maximize time to the next ELM systematically pay for it with lower normalized pressure and Greenwald fraction, while prioritizing pressure tends to require lower elongation and higher south-beam power.
  • The framework's ranking and local derivative of the MCDM score indicate which parameters to change next to a high-performing discharge, pointing to concrete experiment design choices.
  • Because the method is data-driven and device-agnostic, the same pipeline can be retrained on later campaigns or other tokamaks to produce analogous predictors and Pareto fronts.

Reading between the lines

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

  • The reported accuracy probably overstates true generalization because the random 30/70 split lets time slices from the same discharge appear in both training and testing; a shot-grouped split would give the honest number.
  • A testable next step would be to use the random-forest model as an emulator for symbolic regression, distilling the low-aspect-ratio width–height relation into a compact scaling law that the paper does not provide.
  • The Pareto-optimal control-room prescriptions could be turned into a prospective experimental test: set the identified current, shaping, and beam powers on a new shot and check whether the predicted performance and ELM timing are realized.
  • Because the paper notes the random-forest model struggles at very small time-to-ELM values, Pareto fronts that rely on short ELM intervals should be treated as the least trustworthy region of the optimization.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper introduces HIPED, a Random Forest (RF) machine-learning framework trained on MAST-U pedestal data from the third campaign. The models predict pedestal height measures (β_θ,ped, n_e,ped, T_e,ped) from engineering and physics parameters, report test-set R² around 0.71–0.76 for β_θ,ped (0.674 with 'control room' inputs only), and show that simple power-law scalings such as Δ_ped~√β_θ,ped do not reproduce the data. The paper also applies the RF surrogates to construct synthetic Pareto fronts for multi-objective optimization, ranks actual MAST-U discharges by their proximity to those fronts, and identifies control-room settings associated with the best discharges for four weightings. The central claims are accurate pedestal-height prediction, the value of additional shaping and pressure parameters, and a practical pathway for experiment planning.

Significance. If the predictive accuracy claims survive proper evaluation, HIPED would be a useful, interpretable tool for MAST-U experimental planning and for comparing spherical-tokamak pedestal behavior with conventional-aspect-ratio scalings. The paper provides a clearly described database filtering procedure and a GitLab repository for the pedestal database tools, which will aid reproducibility. The feature-importance and SHAP analyses give concrete, physically interpretable hypotheses about which parameters control the MAST-U pedestal. However, the central quantitative claim—that the RF accurately predicts pedestal height on unseen data—is currently undermined by the data-splitting methodology, so the significance is conditional on re-analysis with a discharge-grouped split.

major comments (4)
  1. [Section V.A] The random train/test split is performed per time slice rather than per discharge. Section IV states that the 'H-mode all' category contains 7,481 time slices from 711 shots, so consecutive time slices from the same discharge almost certainly appear in both the 30% training set and the 70% test set. Time slices within a discharge are strongly autocorrelated in both the input features (P_beam, I_p, shaping, f_G) and the target β_θ,ped, so the test set contains near-duplicates of training points. A Random Forest can effectively interpolate along each shot's trajectory, inflating the test R² values reported in Figures 3, 5, 6, and Appendix D. The authors must re-run the analysis with a shot-grouped split (e.g., split by discharge ID, or leave-one-shot-out cross-validation) and report the resulting R² values before the predictive accuracy claim can be accepted.
  2. [Section IV and Figure 2] The comparison between the RF models and the linear/power-law regressions is asymmetric. The simple fits in Figure 2(a)–(f) are in-sample regressions on the full dataset, whereas the RF performance in Figures 3, 5, and 6 is evaluated on a held-out test set. This discrepancy by construction favors the RF (in-sample fits are not penalized for overfitting, while the RF is evaluated on unseen data). To support the statement that Random Forests are 'significantly better than linear approaches in Figure 2', the authors should evaluate the linear and power-law models on the same train/test splits (or with cross-validation) and report the out-of-sample R² for both families of models.
  3. [Abstract and Section I] The abstract and introduction claim that HIPED provides 'accurate estimates of pedestal height and width', and the introduction states that the framework 'predicts pedestal parameters such as ... width'. However, no model is trained to predict the pedestal width: Δ_ped, Δ_n_e,ped, and Δ_T_e,ped appear only as input features to the β_θ,ped models (Table III and Section V.B). The absence of width prediction models means the title's and abstract's 'width' claim is not supported by any result in the paper. The authors should either train and evaluate explicit width-prediction models or revise the claims to state that the framework predicts pedestal height using the width as an input.
  4. [Section VI and Appendix C] The Pareto fronts are constructed from RF surrogate predictions on 50,000 synthetic samples 'based on the underlying MAST-U database' (Section VI.A), and the RF models for the Pareto objectives (f_G, β_N, T_ELM, ⟨n_e⟩_L) are trained on the same database with the random time-slice split. The resulting front is thereby partly an artifact of the training-data distribution and the surrogate model's extrapolation errors. For the optimization results to be useful, the authors should validate the surrogate predictions on held-out shots, and ideally compare the predicted Pareto-optimal points with actual experimental outcomes (e.g., whether the identified discharges indeed have the claimed T_ELM, f_G, and β_N). The time-window average in Eq. (8) with T_window = 0.05 s is also a free parameter; its influence on the selected optimal discharges should be discussed or shown to be robust.
minor comments (6)
  1. [Section IV] The abstract and Section V.C refer to 'control room parameters' but f_G is not directly controllable as it depends on the achieved density; the text does note 'We keep the f_G parameter to approximate fueling control', but this approximation should be stated earlier and more prominently.
  2. [Figure 1 caption] The caption says 'Pbeam is the total neutral beam bower'; 'bower' should be 'power'.
  3. [Section V.A] The feature-selection step 'Features with Pearson correlation > 0.85 are removed' is not documented in detail: the list of removed features and the thresholds used are not provided, which affects reproducibility.
  4. [Section VI.B] The statement 'Solutions that ignore T_ELM will achieve higher β_N, f_G, and ⟨n_e⟩_L' is presented as a finding, but it is a direct consequence of removing an objective from the Pareto optimization; rephrasing as a definitional expectation would be clearer.
  5. [Appendix A] The hyperparameter description says 'maximum depth of 30 to mitigate overfitting', but a maximum depth of 30 is generally not an effective overfitting mitigation for random forests; the authors likely rely on bagging and feature subsampling. This sentence could be clarified.
  6. [Throughout] Reference [25] is cited as 'In Preparation'; while acceptable in a preprint, the authors should update it to a published or archived version when available.

Circularity Check

1 steps flagged · score 4.0 of 10

Reported test R² is inflated by a per-time-slice random split without shot grouping; the central prediction claim is partially forced by within-shot leakage, though not by definitional identity.

  1. fitted input called prediction [Section IV (MAST-U Pedestal Database) and Section V.A (Data Input and Preparation)]
    "Our data comes from the third MAST-U campaign, initially containing 711 shots and 65,393 equilibria ... H-mode all: Data from 25% to 99% of the ELM cycle (7481 points). ... The remaining data is randomly split: 30% for training (with cross-validation for hyperparameter selection) and 70% for testing."

    The random split is performed per time slice, not per discharge, so the 70% test set and 30% training set both contain time slices from the same 711 shots. Within a discharge, consecutive time slices are strongly autocorrelated in both features (Ip, Pbeam, shaping, fG) and target (βθ,ped), so a Random Forest can interpolate along shot trajectories already seen in training. The reported test R² ≈ 0.71–0.76 (and R² = 0.674 for the control-room model) therefore partly measures within-shot memorization rather than shot-to-shot generalization. The central claim of accurate pedestal-height prediction is thus partially forced by the construction of the data split, although the model retains some independent predictive content.

full rationale

The core derivation chain is not definitionally circular: pedestal targets are defined from tanh fits (Eqs. 1–5), direct pedestal-height measurements are excluded from the feature set, and the control-room model uses a reduced feature set. The main circularity is in the evaluation construction: Section V.A randomly splits 7481 time slices from only 711 shots into 30% training and 70% testing without grouping by discharge, so the same shot appears on both sides; RF test predictions can interpolate along already-seen shot trajectories, inflating the reported R² values. This makes the headline accuracy claim partially forced by construction. The Pareto front is built from RF surrogate predictions on synthetic samples, but actual MAST-U discharges are ranked using measured MCDM scores, providing an independent check. The self-citation of Ref. [25] for the fitting routine supplies input data rather than a conclusion, so it is not separately load-bearing. The abstract's claim of width prediction is unsupported because no width model is trained, but that is a completeness gap, not a circularity.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central analysis relies on standard pedestal profile fitting, random forest regression, and a synthetic sampling procedure that is not described. No new physical entities are introduced. The main free choices are the RF hyperparameters and the time-averaging window.

free parameters (2)
  • Random Forest hyperparameters = 100-300 trees, max_depth 30
    Chosen by hand in Appendix A; sensitivity checks are claimed but not shown.
  • Time window Twindow = 0.05 s
    Used for time-averaging scores in Section VI.B; chosen without stated justification.
assumptions (3)
  • domain assumption Pedestal density and temperature profiles follow the modified hyperbolic tangent forms in Equations (1) and (2).
    Standard in pedestal analysis, but the paper does not validate the functional form for MAST-U.
  • domain assumption The Bayesian multi-diagnostic inference system of Refs [23-25] produces correct pedestal fits.
    The core pedestal parameter extraction is delegated to that system, with Ref [25] listed as In Preparation.
  • standard math Random forest regression with bagging gives unbiased predictions for this dataset.
    In Appendix A the authors rely on standard RF behavior without a formal justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HIPED: Machine Learning Framework for Spherical Tokamak Pedestal Prediction and Optimization." pith.science (2026). https://pith.science/paper/FIRGSAKX

@misc{pith2026250419861,
  author       = {Pith},
  title        = {Pith review of: HIPED: Machine Learning Framework for Spherical Tokamak Pedestal Prediction and Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FIRGSAKX}},
  note         = {Machine review of arXiv:2504.19861}
}
read the original abstract

We introduce a Machine Learning framework, HIPED (HeIght and width Predictor for Edge Dynamics), for predicting and optimizing pedestal and core performance in spherical tokamak plasmas. Trained on pedestal and core datasets from the third MAST-U campaign, HIPED provides accurate estimates of pedestal height and width. The results reveal notable differences compared with conventional aspect-ratio studies; for instance, a simple power-law relation between pedestal width and height has very low accuracy. Instead, additional parameters such as normalized plasma pressure, elongation, and Greenwald fraction significantly improve accuracy. HIPED can also be trained only on `control room parameters' to inform experimentalists of which controllable parameters to adjust for improving core-integrated performance. The framework further includes a multi-objective optimization scheme that helps guide experimental planning and optimization. We find Pareto-optimal discharges with respect to various features, including distance from edge-localized modes and normalized plasma pressure, track their parameter trajectories over time, and identify the control room parameters required for these Pareto-optimal discharges. This provides a framework for systematically optimizing core and edge performance according to different experimental priorities.

Figures

Figures reproduced from arXiv: 2504.19861 by the authors.

Figure 1
Figure 1. FIG. 1: Histograms of representative pedestal and global [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2: (a)-(c) Scatter plots of ∆ [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3: Random Forest predictions of [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (15 more)
Figure 4
Figure 4. Figure 4: FIG. 4: SHAP analysis of the H-mode all RF model for [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5: For control room parameters only: (a) Target distribution for [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6: Random Forest predictions for (left) pedestal density [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7: Pareto front in ( [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8: Example Pareto front in (a) ( [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: FIG. 9: Pareto-optimal MAST-U points for Cases A and C. The marker transparency is inversely proportional to the square [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: FIG. 10: Pareto-optimal MAST-U points for Cases A and C. The marker transparency is inversely proportional to the square [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: FIG. 11: Control-room parameters for Pareto-optimal time points in MAST-U, Cases A and C: ( [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: FIG. 12: Distance from MCDM [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: FIG. 13: Sensitivity of MCDM score to input parameters in the vicinity of the Pareto maximum for Cases A and C. See [PITH_FULL_IMAGE:figures/full_fig_p014_13.png]
Figure 14
Figure 14. Figure 14: FIG. 14: Physics parameters for Pareto-optimal time points in MAST-U, Cases A-D (rows 1 -4) and quantities ( [PITH_FULL_IMAGE:figures/full_fig_p017_14.png]
Figure 15
Figure 15. Figure 15: FIG. 15: Control-room parameters for Pareto-optimal time points in MAST-U, Cases A- D (rows 1 - 4) and quantities [PITH_FULL_IMAGE:figures/full_fig_p018_15.png]
Figure 16
Figure 16. Figure 16: FIG. 16: Random Forest model performance on [PITH_FULL_IMAGE:figures/full_fig_p019_16.png]
Figure 17
Figure 17. Figure 17: FIG. 17: Random Forest model for [PITH_FULL_IMAGE:figures/full_fig_p019_17.png]
Figure 18
Figure 18. Figure 18: FIG. 18: SHAP analyses illustrating feature importances for (a) [PITH_FULL_IMAGE:figures/full_fig_p020_18.png]

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. Prediction of ELM-free Operation in Spherical Tokamaks With High Plasma Squareness

    physics.plasm-ph 2025-05 conditional novelty 6.0 of 10

    Increasing plasma squareness in spherical tokamaks is predicted to degrade kinetic-ballooning stability while barely moving the peeling-ballooning boundary, which could allow ELM-free H-mode operation.

Reference graph

Works this paper leans on

103 extracted references · 55 canonical work pages · cited by 1 Pith paper

  1. [1]

    positive pedestal width)

    The fit must converge and produce physically valid parameters (e.g. positive pedestal width)

  2. [2]

    We discard noise-induced outliers based on residual thresholds

  3. [3]

    This yields a set of valid density and temperature pedestal parameters

    We only include fits taken in a suitable fraction of the ELM cycle so that each point genuinely repre- sents an H-mode interval. This yields a set of valid density and temperature pedestal parameters. See Ref. [25] for more details on the pedestal-fitting routines and [23, 24] for more details on the Bayesian multi-diagnostic inference system. IV. MAST-U ...

  4. [4]

    Shimada, D

    M. Shimada, D. J. Campbell, V. Mukhovatov, M. Fuji- wara, N. Kirneva, K. Lackner, M. Nagami, V. D. Pus- tovitov, N. Uckan, J. Wesley, N. Asakura, A. E. Cost- ley, A. J. Donn´ e, E. J. Doyle, A. Fasoli, C. Gormezano, Y. Gribov, O. Gruber, T. C. Hender, W. Houl- berg, S. Ide, Y. Kamada, A. Leonard, B. Lipschultz, A. Loarte, K. Miyamoto, V. Mukhovatov, T. H....

    2007

  5. [5]

    H-mode all : Data from 25% to 99% of the ELM cycle (7481 points)

  6. [6]

    H-mode pre-ELM : Data from 75% to 99% of the ELM cycle (2582 points)

  7. [8]

    A. J. Creely, M. J. Greenwald, S. B. Ballinger, D. Brun- ner, J. Canik, J. Doody, T. F¨ ul¨ op, D. T. Garnier, R. Granetz, T. K. Gray, C. Holland, N. T. Howard, J. W. Hughes, J. H. Irby, V. A. Izzo, G. J. Kramer, A. Q. Kuang, B. LaBombard, Y. Lin, B. Lipschultz, N. C. Logan, J. D. Lore, E. S. Marmar, K. Montes, R. T. Mumgaard, C. Paz-Soldan, C. Rea, M. L....

    2020

  8. [9]

    Simple power-law relation- ships, such as ∆ ped ∼ p βθ,ped, are not suitable in our dataset

    Core-edge coupling. Simple power-law relation- ships, such as ∆ ped ∼ p βθ,ped, are not suitable in our dataset. Additional parameters like βN,fG, κ, and Pbeam significantly improve accuracy

Show all 103 references
  1. [10]

    The RF models achieve robust accuracy (R2≈ 0.7–0.8) and outper- form simpler regressions

    Random Forest performance. The RF models achieve robust accuracy (R2≈ 0.7–0.8) and outper- form simpler regressions. Feature-importance [e.g. Figure 3] and SHAP analyses [e.g. Figure 4] clarify which parameters dominate

  2. [11]

    We use Pareto meth- ods to identify high-value discharges and detail the control room parameters required to access them

    Multi-objective optimization. We use Pareto meth- ods to identify high-value discharges and detail the control room parameters required to access them. This multi-objective approach helps design experiments with complex performance targets. As demonstrated in this paper, this ...

  3. [12]

    Wagner, G

    F. Wagner, G. Becker, K. Behringer, D. Campbell, A. Eberhagen, W. Engelhardt, G. Fussmann, O. Gehre, J. Gernhardt, G. v. Gierke, G. Haas, M. Huang, F. Karger, M. Keilhacker, O. Kl¨ uber, M. Kornherr, K. Lackner, G. Lisitano, G. G. Lister, H. M. Mayer, D. Meisel, E. R. M¨ uller...

    1982

  4. [13]

    S. M. Kaye, M. G. Bell, K. Bol, D. Boyd, K. Brau, D. Buchenauer, R. Budny, A. Cavallo, P. Couture, T. Crowley, D. S. Darrow, H. Eubank, R. J. Fonck, R. Goldston, B. Grek, K. P. Jaehnig, D. Johnson, R. Kaita, H. Kugel, B. Leblanc, J. Manickam, D. Manos, D. Mansfield, E. Mazzuca...

    1984

  5. [14]

    Ryter, Nuclear Fusion 36, 1217 (1996)

    F. Ryter, Nuclear Fusion 36, 1217 (1996)

    1996

  6. [15]

    R. J. Groebner, M. A. Mahdavi, A. W. Leonard, T. H. Osborne, G. D. Porter, R. J. Colchin, and L. W. Owen, Physics of Plasmas 9 (2002), 10.1063/1.1462032

    2002 doi

  7. [16]

    for parameter definitions, which motivates studying pedestal width ∆ ped in relation to normalized pedestal pressure βθ,ped. The edge electron density and temper- ature profiles, ne(ψN) and Te(ψN), are often fitted to hyperbolic tangent (tanh) forms for the electron density ne...

  8. [17]

    Wenninger, F

    R. Wenninger, F. Arbeiter, J. Aubert, L. Aho-Mantila, R. Albanese, R. Ambrosino, C. Angioni, J. F. Artaud, M. Bernert, E. Fable, A. Fasoli, G. Federici, J. Garcia, G. Giruzzi, F. Jenko, P. Maget, M. Mattei, F. Maviglia, E. Poli, G. Ramogida, C. Reux, M. Schneider, B. Sieglin, ...

    2015

  9. [18]

    Sorbom, J

    B. Sorbom, J. Ball, T. Palmer, F. Mangiarotti, J. Sier- chio, P. Bonoli, C. Kasten, D. Sutherland, H. Barnard, C. Haakonsen, J. Goh, C. Sung, and D. Whyte, Fusion Engineering and Design 100, 378 (2015)

    2015

  10. [19]

    P. B. Snyder, J. W. Hughes, T. H. Osborne, C. Paz- Soldan, W. M. Solomon, M. Knolker, D. Eldon, T. Evans, T. Golfinopoulos, B. A. Grierson, R. J. Groebner, A. E. Hubbard, E. Kolemen, B. Labombard, F. M. Laggner, O. Meneghini, S. Mordijck, T. Petrie, S. Scott, H. Q. Wang, H. R....

    2019

  11. [20]

    Rodriguez-Fernandez, A

    P. Rodriguez-Fernandez, A. J. Creely, M. J. Greenwald, D. Brunner, S. B. Ballinger, C. P. Chrobak, D. T. Gar- nier, R. Granetz, Z. S. Hartwig, N. T. Howard, J. W. Hughes, J. H. Irby, V. A. Izzo, A. Q. Kuang, Y. Lin, E. S. Marmar, R. T. Mumgaard, C. Rea, M. L. Reinke, V. Riccar...

    2022

  12. [21]

    Osborne and S

    T. Osborne and S. Saarelma, Plasma Physics and Con- trolled Fusion 65 (2023)

    2023

  13. [22]

    J. W. Hughes, N. T. Howard, P. Rodriguez-Fernandez, A. J. Creely, A. Q. Kuang, P. B. Snyder, T. M. Wilks, 17 0.0 0.2 0.4 0.6 0.8 1.0 fG TW 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0N TW 49387 (#1) 49415 (#2) 49425 (#3) 49386 (#4) 49256 (#5) 100 101 102 # of points (a) Case A 0.0000.0...

    2020

  14. [23]

    Kotschenreuther, X

    M. Kotschenreuther, X. Liu, D. R. Hatch, S. Mahajan, L. Zheng, A. Diallo, R. Groebner, J. C. Hillesheim, C. F. Maggi, C. Giroud, F. Koechl, V. Parail, S. Saarelma, E. Solano, A. Chankin, A. Chankin, and J. Contributors, Nuclear Fusion 59, 096001 (2019)

    2019

  15. [24]

    Mordijck, Nuclear Fusion 60 (2020), 10.1088/1741- 4326/ab8d04

    S. Mordijck, Nuclear Fusion 60 (2020), 10.1088/1741- 4326/ab8d04

    2020 doi

  16. [25]

    R. J. Groebner and S. Saarelma, Plasma Physics and Controlled Fusion 65, 073001 (2023)

    2023

  17. [26]

    A. Kirk, H. R. Wilson, G. F. Counsell, R. Akers, E. Arends, S. C. Cowley, J. Dowling, B. Lloyd, M. Price, M. Walsh, and M. Team, Physical Review Letters 92 19 0.2 0.4 0.6 0.8 fG 0 2 4 1 2 3 4 N 0.0 0.5 1.0 0.00 0.05 0.10 0.15 TELM/s 0 20 40 60 MAST-U Database Pareto-Generated ...

    2004

  18. [27]

    P. B. Snyder, N. Aiba, M. Beurskens, R. J. Groebner, L. D. Horton, A. E. Hubbard, J. W. Hughes, G. T. Huys- mans, Y. Kamada, A. Kirk, C. Konz, A. W. Leonard, J. L¨ onnroth, C. F. Maggi, R. Maingi, T. H. Osborne, N. Oyama, A. Pankin, S. Saarelma, G. Saibene, J. L. Terry, H. Ura...

    2009

  19. [28]

    P. B. Snyder, R. J. Groebner, J. W. Hughes, T. H. Os- borne, M. Beurskens, A. W. Leonard, H. R. Wilson, and X. Q. Xu, Nuclear Fusion 51, 103016 (2011)

    2011

  20. [29]

    Merle, O

    A. Merle, O. Sauter, and S. Y. Medvedev, Plasma Physics and Controlled Fusion 59, 104001 (2017)

    2017

  21. [30]

    Saarelma, L

    S. Saarelma, L. Frassinetti, P. Bilkova, C. D. Challis, A. Chankin, R. Fridstr¨ om, L. Garzotti, L. Horvath, C. F. 18 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 fG TW 0.0 0.2 0.4 0.6 0.8 1.0 Ip TW /MA 49387 (#1) 49415 (#2) 49425 (#3) 49386 (#4) 49256 (#5) 100 101 102 # of points (a) Case...

    2019

  22. [31]

    Saarelma, J

    S. Saarelma, J. Connor, P. B´ ılkov´ a, P. Bohm, C. Bow- man, A. Field, L. Frassinetti, R. Friedstr¨ om, S. Hender- son, K. Imada, A. Kirk, O. Kwon, T. Luda, R. Sarwar, R. Scannell, S. Smith, the ASDEX Upgrade Team, M.- U. team, S. team, J. Contributors, and the Eurofusion Tok...

    2024

  23. [32]

    Parisi, W

    J. Parisi, W. Guttenfelder, A. Nelson, R. Gaur, A. Kleiner, M. Lampert, G. Avdeeva, J. Berkery, C. Clauser, M. Curie, A. Diallo, W. Dorland, S. Kaye, J. McClenaghan, and F. Parra, Nuclear Fusion 64, 054002 (2024)

    2024

  24. [33]

    Harrison, A

    J. Harrison, A. Aboutaleb, S. Ahmed, M. Aljunid, S. Al- lan, H. Anand, Y. Andrew, L. Appel, A. Ash, J. Ashton, et al. , Nuclear Fusion 64, 112017 (2024)

    2024

  25. [34]

    Bowman, J

    C. Bowman, J. R. Harrison, B. Lipschultz, S. Orchard, K. Gibson, M. Carr, K. Verhaegh, and O. Myatra, Plasma Physics and Controlled Fusion62, 045014 (2020)

    2020

  26. [35]

    Greenhouse, C

    D. Greenhouse, C. Bowman, B. Lipschultz, K. Verhaegh, A. Fil, and J. Harrison, Plasma Physics and Controlled Fusion 67, 035006 (2025)

    2025

  27. [36]

    J. G. Clark, C. J. Fitzpatrick, J. W. Berkery, J. F. Parisi, C. Bowman, and R. Scannell, In Preparation (2025)

    2025

  28. [37]

    Cˆ andido and R

    J. Cˆ andido and R. Jorge, Design of quasisymmetric fu- sion devices using novel machine learning methods , Mas- ter’s thesis, Instituto Superior T´ ecnico, Universidade De Lisboa (2023)

    2023

  29. [38]

    T. H. Osborne, G. L. Jackson, Z. Yan, R. Maingi, D. K. Mansfield, B. A. Grierson, C. P. Chrobak, A. G. McLean, S. L. Allen, D. J. Battaglia, A. R. Briesemeister, M. E. Fenstermacher, G. R. McKee, and P. B. Snyder, Nuclear Fusion 55 (2015)

    2015

  30. [39]

    Rea et al

    C. Rea et al. , Nature Communications 10, 1 (2019)

    2019

  31. [40]

    J. Vega, A. Murari, S. Dormido-Canto, G. A. Ratt´ a, and M. Gelfusa, Nature Physics 18, 741 (2022)

    2022

  32. [41]

    Sabbagh, J

    S. Sabbagh, J. Berkery, Y. Park, J. Butt, J. Riquezes, J. Bak, R. Bell, L. Delgado-Aparicio, S. Gerhardt, C. Ham, et al. , Physics of Plasmas 30 (2023)

    2023

  33. [42]

    Gambrioli, L

    M. Gambrioli, L. Piron, A. Pau, G. Cunningham, C. Piron, D. Ryan, P. Martin, team the MAST-U, and T. E. T. the EUROfusion, Plasma Physics and Controlled Fusion 67, 045007 (2025)

    2025

  34. [43]

    Ma et al

    Y. Ma et al. , Physics of Plasmas 27, 042305 (2020)

    2020

  35. [44]

    K. L. van de Plassche, J. Citrin, C. Bourdelle, Y. Came- nen, F. J. Casson, V. I. Dagnelie, F. Felici, A. Ho, S. Van Mulders, and J. Contributors, Physics of Plasmas 27 (2020)

    2020

  36. [45]

    Degrave, F

    J. Degrave, F. Felici, J. Buchli, M. Neunert, B. Tracey, F. Carpanese, T. Ewalds, R. Hafner, A. Abdolmaleki, D. de las Casas, C. Donner, L. Fritz, C. Galperti, A. Hu- ber, J. Keeling, M. Tsimpoukelli, J. Kay, A. Merle, J. M. Moret, S. Noury, F. Pesamosca, D. Pfau, O. Sauter, C...

    2022

  37. [46]

    Felici et al

    F. Felici et al. , Nuclear Fusion 51, 083052 (2011)

    2011

  38. [47]

    L. L. Lao, S. Kruger, C. Akcay, P. Balaprakash, T. A. Bechtel, E. Howell, J. Koo, J. Leddy, M. Leinhauser, Y. Q. Liu, S. Madireddy, J. McClenaghan, D. Orozco, A. Pankin, D. Schissel, S. Smith, X. Sun, and S. Williams, Plasma Physics and Controlled Fusion 64, 074001 (2022)

    2022

  39. [48]

    van Leeuwen, M

    L. van Leeuwen, M. Schoukens, J. Citrin, M. van Berkel, B. Duval, A. Perek, and the TCV Team, Plasma Physics and Controlled Fusion 67, 025024 (2025). 21

    2025

  40. [49]

    Wallace, Z

    G. Wallace, Z. Bai, R. Sadre, T. Perciano, N. Bertelli, S. Shiraiwa, E. Bethel, and J. Wright, Journal of Plasma Physics 88, 895880401 (2022). 0.0 0.1 0.2 0.3 0.4 , ped 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Predicted , ped R2 = 0.585 100 101 2 × 100 3 × 100 4 × 100 6 × ...

    2022

  41. [50]

    S´ anchez-Villar, Z

    ´A. S´ anchez-Villar, Z. Bai, N. Bertelli, E. Bethel, J. Hillairet, T. Perciano, S. Shiraiwa, G. Wallace, and J. Wright, Nuclear Fusion 64, 096039 (2024)

    2024

  42. [51]

    Piccione, J

    A. Piccione, J. Berkery, S. Sabbagh, and Y. Andreopou- los, Nuclear Fusion 60, 046033 (2020)

    2020

  43. [52]

    Piccione, J

    A. Piccione, J. Berkery, S. Sabbagh, and Y. Andreopou- los, Nuclear Fusion 62, 036002 (2022)

    2022

  44. [53]

    M. D. Boyer and J. Chadwick, Nuclear Fusion61, 046024 (2021)

    2021

  45. [54]

    Abbate, R

    J. Abbate, R. Conlin, and E. Kolemen, Nuclear Fusion 61, 046027 (2021)

    2021

  46. [55]

    Dubbioso, G

    S. Dubbioso, G. De Tommasi, A. Mele, G. Tartaglione, M. Ariola, and A. Pironti, Fusion Engineering and De- sign 194, 113725 (2023)

    2023

  47. [56]

    D. R. Smith, R. J. Fonck, G. R. McKee, D. S. Thompson, R. E. Bell, A. Diallo, W. Guttenfelder, S. M. Kaye, B. P. Leblanc, and M. Podesta, Physics of Plasmas 20, 055903 20 (a) ne,ped (b) Te,ped FIG. 18: SHAP analyses illustrating feature importances for (a) ne,ped and (b) Te,pe...

    2013

  48. [57]

    Meneghini, G

    O. Meneghini, G. Snoep, B. C. Lyons, J. McClenaghan, C. S. Imai, B. Grierson, S. P. Smith, G. M. Staebler, P. B. Snyder, J. Candy, E. Belli, L. Lao, J. M. Park, J. Citrin, T. L. Cordemiglia, A. Tema, and S. Mordijck, Nuclear Fusion 61 (2021)

    2021

  49. [58]

    A. Kit, A. E. J¨ arvinen, L. Frassinetti, and S. Wiesen, Plasma Physics and Controlled Fusion 65 (2023), 10.1088/1361-6587/acb3f7

    2023 doi

  50. [59]

    Frassinetti, S

    L. Frassinetti, S. Saarelma, G. Verdoolaege, M. Groth, J. Hillesheim, P. Bilkova, P. Bohm, M. Dunne, R. Frid- str¨ om, E. Giovannozzi,et al. , Nuclear Fusion 61, 016001 (2020)

    2020

  51. [60]

    Pavone, A

    A. Pavone, A. Merlo, S. Kwak, and J. Svensson, Plasma Physics and Controlled Fusion 65, 053001 (2023)

    2023

  52. [61]

    Landreman, J

    M. Landreman, J. Y. Choi, C. Alves, P. Balaprakash, R. M. Churchill, R. Conlin, and G. Roberg-Clark, arXiv preprint arXiv:2502.11657 (2025)

    2025 arXiv

  53. [62]

    Kates-Harbeck, A

    M. Kates-Harbeck, A. Svyatkovskiy, and W. Tang, Na- ture 568, 526 (2019)

    2019

  54. [63]

    Gillgren, E

    A. Gillgren, E. Fransson, D. Yadykin, L. Frassinetti, P. Strand, and J. Contributors, Nuclear Fusion 62, 096006 (2022)

    2022

  55. [64]

    S. F. Smith, A. Kirk, B. Chapman-Oplopoiou, J. G. Clark, C. J. Ham, L. Horvath, C. F. Maggi, R. Scan- nell, and S. Saarelma, Plasma Physics and Controlled Fusion 64 (2022)

    2022

  56. [65]

    D. R. Smith, R. J. Fonck, G. R. McKee, A. Diallo, S. M. Kaye, B. P. LeBlanc, and S. A. Sabbagh, Plasma Physics and Controlled Fusion 58, 045003 (2016)

    2016

  57. [66]

    Berkery, P

    J. Berkery, P. Adebayo-Ige, H. A. Khawaldeh, G. Avdeeva, S.-G. Baek, S. Banerjee, K. Barada, D. Battaglia, R. Bell, E. Belli, E. Belova, N. Bertelli, N. Bisai, P. Bonoli, M. Boyer, J. Butt, J. Candy, C. Chang, C. Clauser, L. C. Rivera, M. Curie, P. de Vries, R. Diab, A. Diallo...

    2024

  58. [67]

    Viezzer, M

    E. Viezzer, M. Austin, M. Bernert, K. Burrell, P. Cano- Megias, X. Chen, D. J. Cruz-Zabala, S. Coda, M. Faitsch, O. F´ evrier,et al. , Nuclear Materials and Energy 34, 101308 (2023)

    2023

  59. [68]

    R. J. Groebner and T. H. Osborne, Physics of Plasmas 5, 1800 (1998)

    1998

  60. [69]

    P. B. Snyder, R. J. Groebner, A. W. Leonard, T. H. Os- borne, and H. R. Wilson, Physics of Plasmas 16, 056118 (2009)

    2009

  61. [70]

    Imada, T

    K. Imada, T. H. Osborne, S. Saarelma, A. Kirk, S. Black- more, M. Kn¨ olker, R. Scannell, P. B. Snyder, C. Vincent, and H. R. Wilson, Plasma Physics and Controlled Fusion (2024)

    2024

  62. [71]

    Cordey, for the ITPA H-Mode Database Work- ing Group, and the ITPA Pedestal Database Work- ing Group, Nuclear Fusion 43, 670 (2003)

    J. Cordey, for the ITPA H-Mode Database Work- ing Group, and the ITPA Pedestal Database Work- ing Group, Nuclear Fusion 43, 670 (2003)

    2003

  63. [72]

    Greenwald, J

    M. Greenwald, J. L. Terry, S. M. Wolfe, S. Ejima, M. G. Bell, S. M. Kaye, and G. H. Neilson, Nuclear Fusion 28, 2199 (1988)

    1988

  64. [73]

    could also be included. VII. DISCUSSION We have introduced HIPED, a new framework for pre- dicting and optimizing core and pedestal performance in spherical tokamaks using a large MAST-U database. Our main findings include:

  65. [74]

    J. W. Berkery, S. A. Sabbagh, L. Kogan, S. Gibson, D. Ryan, V. Zamkovska, J. Butt, J. Harrison, S. Hender- son, M.-U. team, et al. , Plasma Physics and Controlled Fusion 65, 045001 (2023)

    2023

  66. [75]

    Breiman, Machine Learning 45, 5 (2001)

    L. Breiman, Machine Learning 45, 5 (2001)

    2001

  67. [76]

    Liaw and M

    A. Liaw and M. Wiener, R News 2, 18 (2002)

    2002

  68. [77]

    J. F. Rivero-Rodr´ ıguez, K. McClements, M. Fitzger- ald, S. Sharapov, M. Cecconello, N. Crocker, I. Dolby, M. Dreval, N. Fil, J. Gald´ on-Quiroga,et al., Nuclear Fu- sion 64, 086025 (2024)

    2024

  69. [78]

    Taherdoost and M

    H. Taherdoost and M. Madanchian, Encyclopedia 3, 77 (2023)

    2023

  70. [79]

    J. F. Parisi, A. O. Nelson, R. Gaur, S. M. Kaye, F. I. Parra, J. W. Berkery, K. Barada, C. Clauser, A. J. Creely, A. Diallo, W. Guttenfelder, J. W. Hughes, L. A. Ko- gan, A. Kleiner, A. Q. Kuang, M. Lampert, T. Macwan, J. E. Menard, and M. A. Miller, Physics of Plasmas 31, 030...

    2024

  71. [80]

    Parisi, A

    J. Parisi, A. Nelson, W. Guttenfelder, R. Gaur, J. Berk- ery, S. Kaye, K. Barada, C. Clauser, A. Diallo, D. Hatch, A. Kleiner, M. Lampert, T. Macwan, and J. Menard, Nuclear Fusion 64, 086034 (2024)

    2024

  72. [81]

    Imada, T

    K. Imada, T. H. Osborne, S. Saarelma, J. G. Clark, A. Kirk, M. Knolker, R. Scannell, P. B. Snyder, V. C., and H. R. Wilson, Nuclear Fusion (2024)

    2024

  73. [82]

    Nelson, C

    A. Nelson, C. Vincent, H. Anand, J. Lovell, J. Parisi, H. Wilson, K. Imada, W. Wehner, M. Kochan, S. Black- more, G. McArdle, S. Guizzo, L. Rondini, S. Freiberger, C. Paz-Soldan, and the MAST-U Team, Nuclear Fusion 64, 124004 (2024)

    2024

  74. [83]

    Zamkovska, S

    V. Zamkovska, S. Sabbagh, M. Tobin, J. Berkery, J. Riquezes, Y. Park, K. Erickson, J. Butt, J. Bak, J. Kim, K. Lee, J. Ko, S. Yoon, C. Ham, L. Kogan, and the MAST Upgrade Team, Nuclear Fusion 64, 066030 (2024)

    2024

  75. [84]

    P¨ utterich, R

    T. P¨ utterich, R. Dux, M. Janzer, R. McDermott, A. U. Team, et al. , Journal of Nuclear Materials 415, S334 (2011)

    2011

  76. [85]

    A. E. Hubbard, Plasma Physics and Controlled Fusion 42, A15 (2000)

    2000

  77. [86]

    Chang, S

    C.-S. Chang, S. Ku, and H. Weitzner, Physics of Plasmas 11, 2649 (2004)

    2004

  78. [87]

    Angioni, E

    C. Angioni, E. Fable, M. Greenwald, M. Maslov, A. G. Peeters, H. Takenaga, and H. Weisen, Plasma Physics and Controlled Fusion 51, 124017 (2009)

    2009

  79. [88]

    Callen, R

    J. Callen, R. Groebner, T. Osborne, J. Canik, L. W. Owen, A. Pankin, T. Rafiq, T. Rognlien, and W. Stacey, Nuclear fusion 50, 064004 (2010)

    2010

  80. [89]

    D. R. Hatch, M. Kotschenreuther, S. Mahajan, P. Valanju, and X. Liu, Nuclear Fusion 57, 036020 (2017)

    2017

  81. [90]

    Guttenfelder, R

    W. Guttenfelder, R. J. Groebner, J. M. Canik, B. A. Grierson, E. A. Belli, and J. Candy, Nuclear Fusion 61, 056005 (2021)

    2021

  82. [91]

    Chapman-Oplopoiou, D

    B. Chapman-Oplopoiou, D. Hatch, A. Field, L. Frassinetti, J. Hillesheim, L. Horvath, C. Maggi, J. Parisi, C. Roach, S. Saarelma, J. Walker, and J. Contributors, Nuclear Fusion 62, 086028 (2022)

    2022

  83. [92]

    E. A. Belli, J. Candy, and I. Sfiligoi, Plasma Physics and Controlled Fusion 65, 024001 (2022)

    2022

  84. [93]

    Hatch, M

    D. Hatch, M. Kotschenreuther, P.-Y. Li, B. Chapman- Oplopoiou, J. Parisi, S. Mahajan, and R. Groebner, Nu- clear Fusion 64, 066007 (2024)

    2024

  85. [94]

    Neuhauser, D

    J. Neuhauser, D. Coster, H. U. Fahrbach, J. C. Fuchs, G. Haas, A. Herrmann, L. Horton, M. Jakobi, A. Kallen- 22 bach, M. Laux, J. W. Kim, B. Kurzan, H. W. M¨ uller, H. Murmann, R. Neu, V. Rohde, W. Sandmann, W. Sut- trop, and E. Wolfrum, Plasma Physics and Controlled Fusion 44...

    2002

  86. [95]

    Lipschultz, X

    B. Lipschultz, X. Bonnin, G. Counsell, A. Kallenbach, A. Kukushkin, K. Krieger, A. Leonard, A. Loarte, R. Neu, R. Pitts, et al. , Nuclear Fusion 47, 1189 (2007)

    2007

  87. [96]

    T. Luda, C. Angioni, M. G. Dunne, E. Fable, A. Kallen- bach, N. Bonanomi, P. A. Schneider, M. Siccinio, and G. Tardini, Nuclear Fusion 60 (2020)

    2020

  88. [97]

    T. M. Wilks, M. Knolker, P. B. Snyder, D. Eldon, F. Scotti, C. Chrystal, F. M. Laggner, C. Lasnier, A. McLean, T. Osborne, C. Paz-Soldan, H. Wang, J. Watkins, L. Casali, B. Grierson, and J. W. Hughes, Nuclear Fusion 61 (2021)

    2021

  89. [98]

    Stagni, N

    A. Stagni, N. Vianello, C. Tsui, C. Colandrea, S. Gorno, M. Bernert, J. Boedo, D. Brida, G. Falchetto, A. Hakola, et al. , Nuclear Fusion 62, 096031 (2022)

    2022

  90. [99]

    Snyder, J

    P. Snyder, J. M. Park, H. Wilson, C. Collins, E. Hassan, J. Hughes, M. Knolker, T. Osborne, J. Parisi, M. Shafer, et al., Bulletin of the American Physical Society (2024)

    2024

  91. [100]

    Zhang, C

    X. Zhang, C. Marsden, M. Moscheni, E. Maartensson, A. Rengle, M. Robinson, T. O’Gorman, H. Lowe, E. Vek- shina, S. Janhunen, et al., Nuclear Materials and Energy 41, 101772 (2024)

    2024

  92. [101]

    Rodriguez-Fernandez, N

    P. Rodriguez-Fernandez, N. Howard, A. Saltzman, S. Kantamneni, J. Candy, C. Holland, M. Balandat, S. Ament, and A. White, Nuclear Fusion 64, 076034 (2024)

    2024

  93. [102]

    T. Eich, T. Body, M. Faitsch, O. Grover, M. Miller, P. Manz, T. Looby, A. Kuang, A. Redl, M. Reinke, A. Creely, D. Battaglia, J. Hillesheim, M. Wigram, and J. Hughes, Nuclear Materials and Energy 42, 101896 (2025)

    2025

  94. [103]

    Welsh, L

    A. Welsh, L. Casali, E. Fable, J. Smiskey, R. Mattes, G. Tardini, and B. T. Taczak, Nuclear Fusion65, 044002 (2025)

    2025

  95. [104]

    J. G. Cordey, J. C. Deboo, and O. Kardaun, in Plasma physics and controlled nuclear fusion research 1990. V. 3 (1991)

    1991

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.