Pith. sign in

REVIEW 3 major objections 6 minor 21 references

Deep Learning Models for ADITYA-U MHD Equilibrium

T0 review · 3 major / 6 minor · reviewed 2026-07-11 · grok-4.5

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

Pith's one-line read Deep-learning surrogates predict ADITYA-U MHD equilibrium parameters and profiles from a large synthetic free-boundary dataset within the machine’s circular-limiter flat-top domain.

desk verdict Solid first-for-ADITYA-U equilibrium surrogate suite: careful methods and honest scoping, limited mainly by synthetic-only evaluation. read the letter →

arxiv 2607.04865 v1 pith:6WIRAPOV submitted 2026-07-06 physics.plasm-ph

Udaya Maurya , Suman Aich , Indranil Bandyopadhyay , Daniel Raju This is my paper · ORCID
classification physics.plasm-ph
keywords MHDequilibriumDeeplearningReal-timecontrolADITYA-UTokamakGrad-ShafranovPhysics-informedneuralnetworksSafetyfactorprofilePoloidalflux
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

Tokamak control and analysis need the Grad-Shafranov equilibrium solution (plasma boundary, magnetic axis, safety factor, flux, inductance, and related scalars), but classical free-boundary solvers are too slow for real-time use. This paper builds a 100,760-case synthetic library of free-boundary equilibria for the ADITYA-U tokamak, generated from experimentally motivated inputs drawn from 766 discharges and filtered to circular limiter plasmas near flat-top. Dense networks, PCA reduced-order models, 1-D and 2-D CNNs, and physics-informed networks that penalize Grad-Shafranov residuals are trained to map readily available inputs (coil currents, Ip, Bφ, magnetic probes, loop voltage, and cascaded scalars) onto magnetic-axis location, βp, internal inductance, edge safety factor, full q(ρ), full ψ(R,Z), and selected poloidal-field coil currents. Within the domain spanned by the library the surrogates recover the target quantities accurately, with the largest model evaluating in roughly one millisecond. The practical claim is that these models can serve as fast substitutes for conventional equilibrium estimation in ADITYA-U real-time control, rapid discharge analysis, and experiment planning.

What carries the argument

A large, physics-filtered synthetic free-boundary equilibrium library (pyIPREQ solutions of the Grad-Shafranov equation with a three-parameter Jφ profile, Bayesian-optimized γ, and a probabilistic βp prior) that supplies training targets for dense, PCA, CNN and Grad-Shafranov-residual PINN models.

What would settle it

Apply the trained forward models to a set of real ADITYA-U flat-top discharges that have independent equilibrium reconstructions (or diamagnetic βp and magnetic-axis measurements) never used in the synthetic generation pipeline; systematic errors larger than the reported synthetic percentiles would falsify the claimed transferability.

Watch

Extended reading notes

Core claim

Key ADITYA-U MHD equilibrium parameters and profiles—Rax, Zax, βp, ℓi, q1, ψaxs/ψlim, the full q(ρ) profile, the full ψ(R,Z) map, and selected PF coil currents—can be accurately estimated by deep-learning surrogates trained on a 100,760-case pyIPREQ free-boundary synthetic dataset, with inference times of order 1 ms, inside the circular-limiter flat-top operational domain represented by that dataset.

Load-bearing premise

The synthetic library—built with a fixed parametric current-density form, a linear βp model fitted on only a few dozen measured discharges, and hard filters on axis shift, inductance and safety factor—must faithfully cover the equilibria that real ADITYA-U flat-top plasmas actually produce, so that synthetic test errors transfer to experiment.

Editorial extensions

If this is right

  • Magnetic-axis position, βp, ℓi and edge q can be obtained in ~1 ms from magnetic diagnostics and coil currents, enabling real-time equilibrium feedback on ADITYA-U.
  • Full q(ρ) and ψ(R,Z) maps become available for rapid post-shot analysis without repeated free-boundary solves.
  • The inverse coil-current model supplies candidate PF actuator settings for desired plasma parameters inside the trained domain, supporting experimental planning.
  • Physics-informed residual losses keep the predicted flux maps approximately consistent with the Grad-Shafranov equation, reducing the chance of non-physical reconstructions.
  • The same library-plus-surrogate pattern can be reused for other circular-limiter machines once an analogous synthetic database is generated.

Reading between the lines

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

  • Because the models rely on magnetic-probe and loop-voltage inputs, they cannot yet replace free-boundary solvers that operate from coil currents alone; a pure-actuator forward map would be a natural next library.
  • The low-dimensional PCA manifolds for both q and ψ suggest that simple parametric families already capture most ADITYA-U flat-top equilibria, so uncertainty-aware or active-learning extensions could focus data collection on the residual high-order modes.
  • Localized CNN distortions versus globally smooth PCA errors imply that hybrid PCA–CNN or ensemble predictors may be needed before the maps are trusted for stability calculations that depend on local shear or curvature.
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

3 major / 6 minor

Summary. The manuscript develops deep-learning surrogates for ADITYA-U free-boundary MHD equilibria. A 100,760-case synthetic library is generated with pyIPREQ from 766 experimental discharges (circular limiter, near flat-top, Ip>100 kA), using a fixed parametric Jφ form (Eq. 2), Bayesian-optimized γ, a linear probabilistic βp prior fitted on 39 diamagnetic shots plus noise, and physics-informed filters on ΔR, Zax, ℓi, q0, q1. Dense networks predict scalars (Rax, Zax, βp, ℓi, q1, ψaxs/ψlim) and an inverse map from desired equilibrium parameters to selected PF coil currents. PCA and 1d-CNN models reconstruct q(ρ) (the latter via softplus-constrained q′); PCA and 2d-CNN PINNs reconstruct ψ(R,Z) with Grad–Shafranov residual losses and simultaneous prediction of α,β,γ. Shot-wise 80/10/10 splits, Hyperband search, and 95th/99th-percentile absolute errors plus median/worst-case profile figures are reported. Inference of the largest model is ~1 ms on CPU. Claims are scoped to accuracy inside the synthetic operational domain.

Significance. If the reported synthetic accuracy transfers, the work supplies the first large-scale, multi-architecture ML equilibrium framework for ADITYA-U, with practical utility for real-time control, rapid discharge analysis, and actuator planning. Strengths include the experimentally motivated input ranges, shot-wise splits that avoid temporal leakage, systematic PCA-vs-CNN comparison, physics-informed constraints (monotonicity of q, progressive GS residual with uncertainty weighting), and explicit quantitative error distributions rather than only mean metrics. The inverse coil-current model and cascade-ready scalar predictors are useful engineering contributions within the stated domain. The principal limitation is that all validation remains synthetic; experimental transfer is left for future work and is already acknowledged.

major comments (3)
  1. The central claim is carefully scoped to the synthetic domain, yet the abstract, introduction and conclusion repeatedly advertise utility for real-time control and experimental planning. No experimental reconstruction or even a single pyIPREQ-vs-diagnostic comparison on held-out ADITYA-U shots is shown. A minimal experimental sanity check (or a clearly labeled “synthetic-only” caveat in the abstract) is needed so that the transfer premise of §2.1–2.2 does not over-extend the reported numbers.
  2. §2.2: the βp prior is a linear model trained on only 39 diamagnetic discharges; Fig. 3 shows clear bias at high/low βp and the residual noise N(0,0.04) is then injected into all 100k cases. Because βp directly sets the Jφ parameter β (Eq. 2) and therefore shapes ℓi and q, the sensitivity of the downstream scalar and profile errors to this prior should be quantified (e.g., by re-training with a wider or alternative βp distribution).
  3. §4.2 and cascade discussion: Rax/Zax (and later q1, ψaxs/ψlim) are used as inputs to subsequent models, yet training uses ground-truth values; cascading error is never measured. Because the intended real-time pipeline is cascaded, the reported test-set errors for βp/ℓi, q-profile and ψ-profile are optimistic. A short end-to-end cascade evaluation on the test shots is required.
minor comments (6)
  1. Table 1 lists coil geometry but omits the actual current ranges present in the dataset; a short summary row would help readers judge the inverse-model domain.
  2. Figs. 9–22 report absolute errors with medians and 95th percentiles in the captions; adding the same numbers to the main text or a summary table would improve readability.
  3. Eq. (3) and the progressive factor f are introduced without a short derivation or reference to the uncertainty-weighting paper beyond [21]; a one-sentence justification of the numerical constants (−50, 0.15) would help reproducibility.
  4. The 1d-CNN predicts 100 values of q′ and reconstructs q by integrating from the separately predicted q1; the accumulation of integration error should be stated explicitly when comparing core-region accuracy with the PCA model.
  5. Minor typographical issues: “two-dimensionalpoloidalfluxprofiles” (abstract), “physics-informedneuralnetworks” (abstract), and occasional missing spaces after commas in the introduction.
  6. Acknowledgment notes LLM rephrasing; a brief statement that scientific content and all numerical results were verified by the authors would be appropriate.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: models are intentional surrogates of pyIPREQ on a filtered synthetic library; accuracy claims are scoped to that domain and do not reduce by construction to inputs.

full rationale

The paper's derivation chain is standard surrogate modeling: experimental coil/Ip/Bphi ranges (plus a linear beta_p prior fitted on 39 diamagnetic shots) seed pyIPREQ free-boundary solutions under a fixed J_phi ansatz (Eq. 2) and hard filters; dense/PCA/CNN/PINN networks then learn the resulting input-to-output map and are evaluated on shot-wise held-out synthetic cases. Reported errors (Tables 9/11/13/15, Figs. 9-42) therefore measure fidelity to the solver library, not an independent physical derivation. The beta_p linear model (Sec. 2.2) is an acknowledged stochastic prior used only for dataset generation, not a fitted parameter later re-labeled as a first-principles prediction. Self-citations to pyIPREQ [13,14] simply identify the data-generation tool; no uniqueness theorem or load-bearing external result is imported from the authors. Inverse coil-current mapping is explicitly noted as non-unique and statistical. PINN GS residuals act as regularization, not as a circular definition of psi. All central claims are carefully limited to 'within the operational domain represented by the dataset' (Abstract, Sec. 7). No step reduces a claimed prediction to its own inputs by construction.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard MHD equilibrium theory plus a large set of modeling and sampling choices that define the synthetic training domain. No new physical entities are postulated; free parameters are the usual ML hyperparameters plus the probabilistic βp prior and profile-sampling distributions that shape the library.

free parameters (7)
  • βp linear-model residual noise = N(0, 0.04), clip (0.05, 0.4)
    Noise N(0,0.04) (and clip to [0.05,0.4]) applied to the linear βp surrogate when generating the library; chosen from residual analysis on 39 discharges (§2.2).
  • α sampling distribution for Jφ = N(4.5, 0.5)
    α ~ N(4.5, 0.5) chosen empirically to produce relevant q0 and convergence (§2.1).
  • Rax target sampling distribution = N(0.7525, 0.0075) m
    Rax,target ~ N(0.7525, 0.0075) m matched to Sine-Cosine centroid data (§2.1, Fig. 1).
  • Equilibrium filtering thresholds = as listed in §2.1
    Hard cuts |ΔR|≤0.04 m, |Zax|≤0.02 m, ℓi∈(0.7,1.7), γ<8.9, q0∈(0.9,1.5), q1>2.0 define the retained domain (§2.1).
  • 2% uniform noise on coil currents, Ip, Bφ = 2% uniform
    Applied during generation to simulate measurement error and broaden sampling (§2.1).
  • PINN loss weights / uncertainty sigmas and progressive factor f = f=sigmoid(-50(MSEψ+MSEJ-0.15)); σ trainable
    MSE+2·δGS for PCA-ψ; uncertainty-weighted multi-task loss with shifted sigmoid f for CNN-ψ (§6.1–6.2); trainable σψ, σJ, σGS.
  • Network architectures and hyperparameters = Tables 3–14
    Layer widths, activations, dropout rates, PCA mode counts (4 for q; 5/8/13 for ψ) selected by keras-tuner Hyperband on validation loss.
assumptions (6)
  • domain assumption Axisymmetric ideal MHD equilibrium is described by the Grad–Shafranov equation (Eq. 1).
    Foundational PDE used both to generate the library and as residual constraint in PINN models (§1, §6).
  • domain assumption Toroidal current density follows the three-parameter form Jφ ∝ (β R/R0 + (1-β) R0/R) (1-(1-ψ̄)^α)^γ (Eq. 2).
    All synthetic equilibria are solutions under this adopted profile; shapes outside this family are excluded by construction (§2.1).
  • domain assumption Eddy currents can be neglected for the selected flat-top windows because the vessel is toroidally discontinuous and temporal variations are weak.
    Stated in §2; removes vessel-current degrees of freedom from the free-boundary problem.
  • ad hoc to paper Circular limiter plasmas near flat-top with Ip>100 kA for ≥80 ms adequately represent the operational domain of interest.
    Dataset construction filter (§2); excludes diverted, shaped, ramp-up/down and low-current phases.
  • standard math Shot-wise train/val/test splits prevent temporal leakage and give a realistic generalization estimate to unseen discharges.
    Standard ML practice for correlated time series; used throughout §3.
  • ad hoc to paper A linear regression on transformed Ip, coil currents, Vloop and timing features plus Gaussian noise is a sufficient stochastic prior for βp when diamagnetic data are sparse.
    §2.2; authors note bias at high/low βp and that high accuracy is not required for the intended use.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning Models for ADITYA-U MHD Equilibrium." pith.science (2026). https://pith.science/paper/6WIRAPOV

@misc{pith2026260704865,
  author       = {Pith},
  title        = {Pith review of: Deep Learning Models for ADITYA-U MHD Equilibrium},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6WIRAPOV}},
  note         = {Machine review of arXiv:2607.04865}
}
read the original abstract

This work presents deep learning models to predict magnetohydrodynamic equilibrium parameters and profiles for the ADITYA-U tokamak. A synthetic free-boundary equilibrium dataset consisting of 100,760 cases was generated using the pyIPREQ Grad-Shafranov solver, with inputs derived from 766 ADITYA-U plasma discharges and constrained to experimentally relevant circular limiter plasmas near the flat-top phase. Several deep learning approaches were investigated for predicting scalar equilibrium quantities, one-dimensional safety factor profiles and two-dimensional poloidal flux profiles. These approaches included Dense neural networks, principal component analysis based reduced-order models, one-dimensional and two-dimensional convolutional neural networks, and physics-informed neural networks incorporating Grad-Shafranov residual constraints. In addition, an inverse model was developed to estimate poloidal field coil currents from desired plasma equilibrium conditions. The results demonstrate that key equilibrium parameters and profiles can be accurately estimated within the operational domain represented by the dataset. The developed models provide a computationally efficient alternative to conventional equilibrium estimation and can be useful for real-time plasma control, rapid equilibrium analysis, and experimental planning in ADITYA-U operations.

Figures

Figures reproduced from arXiv: 2607.04865 by the authors.

Figure 1
Figure 1. Distribution of plasma current centroid position [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Comparison between measured and predicted [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 2
Figure 2. for ADITYA-U shot 35356. The motivation for using these time-based parameters was the absence of temperature and density measurements, which directly influence βp. 0.00 0.05 0.10 0.15 0.20 0.25 time (s) 0.0e+00 5.0e+04 1.0e+05 Ip (A) Operational Duration in Dataset tbeg tend ADITYA-U Shot 35356 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figures from the paper (31 more)
Figure 4
Figure 4. Figure 4: Probability density distributions of experimentally [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: shows the distribution of equilibria in the (βp, ℓi) parameter space. This figure and [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Ensemble of q-profiles plotted with ρ = 1 − ψ¯, for randomly selected 2000 cases from the dataset. The markers at ρ = 0 and ρ = 1 indicate ranges of q0 and q1 in the dataset. 3 Deep Learning Methodology This section describes the common methodology adopted for training…
Figure 7
Figure 7. Figure 7: Ensemble of LCFS and magnetic axis position for [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 10
Figure 10. Figure 10: Zax (in meters) predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 9
Figure 9. Figure 9: Rax (in meters) predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 12
Figure 12. Figure 12: βp predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p008_12.png]
Figure 14
Figure 14. Figure 14: MSE Loss for Training and Validation Datasets [PITH_FULL_IMAGE:figures/full_fig_p008_14.png]
Figure 13
Figure 13. Figure 13: ℓi predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p008_13.png]
Figure 18
Figure 18. Figure 18: ψlim predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p009_18.png]
Figure 16
Figure 16. Figure 16: MSE Loss for Training and Validation Datasets [PITH_FULL_IMAGE:figures/full_fig_p009_16.png]
Figure 17
Figure 17. Figure 17: ψaxs predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p009_17.png]
Figure 19
Figure 19. Figure 19: MSE Loss for Training and Validation Datasets [PITH_FULL_IMAGE:figures/full_fig_p010_19.png]
Figure 6
Figure 6. Figure 6: This behavior suggests that the profiles may [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 21
Figure 21. Figure 21: IV F (in Ampere/turn) predicted versus actual values for Test dataset (left axis) and absolute difference between them (right axis), for model shown in [PITH_FULL_IMAGE:figures/full_fig_p010_21.png]
Figure 25
Figure 25. Figure 25: MSE Loss for Training and Validation Datasets [PITH_FULL_IMAGE:figures/full_fig_p011_25.png]
Figure 24
Figure 24. Figure 24: Mean absolute difference between actual and re [PITH_FULL_IMAGE:figures/full_fig_p011_24.png]
Figure 26
Figure 26. Figure 26: Predicted q-profile with median MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p012_26.png]
Figure 27
Figure 27. Figure 27: Predicted q-profile with 99th percentile in MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p012_27.png]
Figure 29
Figure 29. Figure 29: MSE Loss for Training and Validation Datasets [PITH_FULL_IMAGE:figures/full_fig_p012_29.png]
Figure 32
Figure 32. Figure 32: Predicted q-profile with highest MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p013_32.png]
Figure 30
Figure 30. Figure 30: Predicted q-profile with median MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p013_30.png]
Figure 31
Figure 31. Figure 31: Predicted q-profile with 99th percentile in MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p013_31.png]
Figure 33
Figure 33. Figure 33: 1−CEV with respect to number of retained PCA modes for the ψ-profiles. 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 R(m) 10 5 10 4 | a ctu al p c a| 4 5 7 8 12 13 [PITH_FULL_IMAGE:figures/full_fig_p014_33.png]
Figure 34
Figure 34. Figure 34: Mean absolute difference between actual and re [PITH_FULL_IMAGE:figures/full_fig_p014_34.png]
Figure 37
Figure 37. Figure 37: Predicted ψ-profile with 99th percentile in MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p015_37.png]
Figure 38
Figure 38. Figure 38: Predicted ψ-profile with highest MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p015_38.png]
Figure 36
Figure 36. Figure 36: Predicted ψ-profile with median MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p015_36.png]
Figure 40
Figure 40. Figure 40: Predicted ψ-profile with median MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p016_40.png]
Figure 41
Figure 41. Figure 41: Predicted ψ-profile with 99th percentile in MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p017_41.png]
Figure 42
Figure 42. Figure 42: Predicted ψ-profile with highest MSE, for model shown in [PITH_FULL_IMAGE:figures/full_fig_p017_42.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references

  1. [1]

    L.L. Lao, H. St. John, R.D. Stambaugh, A.G. Kellman, andW.Pfeiffer. Reconstructionofcurrentprofileparam- eters and plasma shapes in tokamaks.Nuclear Fusion, 25(11):1611, Nov 1985

    1985

  2. [2]

    Bak, S.G

    Semin Joung, Jaewook Kim, Sehyun Kwak, J.G. Bak, S.G. Lee, H.S. Han, H.S. Kim, Geunho Lee, Daeho Kwon, and Y.-C. Ghim. Deep neural network Grad–Shafranov solver constrained with measured mag- netic signals.Nuclear Fusion, 60(1):016034, Dec 2019

    2019

  3. [3]

    Wai, M.D

    J.T. Wai, M.D. Boyer, and E. Kolemen. Neural net modeling of equilibria in NSTX-U.Nuclear Fusion, 62(8):086042, Jul 2022

    2022

  4. [4]

    EAST discharge prediction with- out integrating simulation results.Nuclear Fusion, 62(12):126060, Nov 2022

    Chenguang Wan, Zhi Yu, Alessandro Pau, Xiaojuan Liu, and Jiangang Li. EAST discharge prediction with- out integrating simulation results.Nuclear Fusion, 62(12):126060, Nov 2022

    2022

  5. [5]

    A machine-learning-based tool for last closed-flux surface reconstruction on tokamaks.Nuclear Fusion, 63(5):056019, Apr 2023

    Chenguang Wan, Zhi Yu, Alessandro Pau, Olivier Sauter, Xiaojuan Liu, Qiping Yuan, and Jiangang Li. A machine-learning-based tool for last closed-flux surface reconstruction on tokamaks.Nuclear Fusion, 63(5):056019, Apr 2023

    2023

  6. [6]

    Fast equilibrium reconstruction by deep learning on EAST tokamak.AIP Advances, 13(7):075007, Jul 2023

    Jingjing Lu, Youjun Hu, Nong Xiang, and Youwen Sun. Fast equilibrium reconstruction by deep learning on EAST tokamak.AIP Advances, 13(7):075007, Jul 2023

    2023

  7. [7]

    Reconstruction of tokamak plasma safety factor profile using deep learning.Nuclear Fusion, 63(8):086020, Jun 2023

    Xishuo Wei, Shuying Sun, William Tang, Zhihong Lin, Hongfei Du, and Ge Dong. Reconstruction of tokamak plasma safety factor profile using deep learning.Nuclear Fusion, 63(8):086020, Jun 2023

    2023

  8. [8]

    Madireddy, C

    S. Madireddy, C. Akçay, S. E. Kruger, T. Bechtel Amara, X. Sun, J. McClenaghan, J. Koo, A. Samad- dar, Y. Liu, P. Balaprakash, and L. L. Lao. EFIT- Prime: Probabilistic and physics-constrained reduced- order neural network model for equilibrium reconstruc- tion in DIII-D.Physics of Plasmas, 31(9):092505, Sep 2024

    2024

Show all 21 references
  1. [9]

    Novella Rutigliano, Andrea Murari, Pasquale Gaudio, Michela Gelfusa, Riccardo Rossi, on behalf of JET Con- tributors, and the EUROfusion Tokamak Exploita- tion Team. Multi-diagnostics reconstruction of magnetic equilibrium and kinetic profiles using physics-informed neural net...

    2026

  2. [10]

    Machine learning prediction of plasma behav- ior from discharge configurations on WEST.Nuclear Fusion, 66(4):044001, Mar 2026

    Chenguang Wan, Feda Almuhisen, Philippe Moreau, Rémy Nouailletas, Zhisong Qu, Youngwoo Cho, Robin Varennes, Kyungtak Lim, Kunpeng Li, Jia Huang, Wei- dong Chen, Jiangang Li, Xavier Garbet, and WEST Team. Machine learning prediction of plasma behav- ior from discharge configura...

    2026

  3. [11]

    Zheng, S.F

    G.H. Zheng, S.F. Liu, H.S. Xie, X. Gu, Z.Y. Chen, X.C. Lun, Y. Liu, J. Li, D. Guo, R.Y. Tao, et al. EFIT-mini: an embedded, multi-task neural network- driven equilibrium inversion algorithm.Nuclear Fusion, 65(10):106008, Sep 2025

    2025

  4. [12]

    Tanna, J

    R.L. Tanna, J. Ghosh, K.A. Jadeja, Rohit Kumar, Suman Aich, K.M. Patel, Harshita Raj, Kaushlender Singh, Suman Dolui, Kajal Shah, et al. Overview of physics results from the ADITYA-U tokamak and fu- ture experiments.Nuclear Fusion, 64(11):112011, Aug 2024

    2024

  5. [13]

    Singh, Suman Aich, Jaga- bandhu Kumar, Rohit Kumar, and Daniel Raju

    Udaya Maurya, Amit K. Singh, Suman Aich, Jaga- bandhu Kumar, Rohit Kumar, and Daniel Raju. Plasma position constrained free-boundary MHD equilibrium in Tokamaks using pyIPREQ.Physics of Plasmas, 32(10):102501, Oct 2025

    2025

  6. [14]

    pyIPREQ: Free Boundary MHD Equiliibrium Code.https://github.com/udy11/ pyIPREQ,https://gitlab.com/udy11/pyIPREQ, 2025

    Udaya Maurya. pyIPREQ: Free Boundary MHD Equiliibrium Code.https://github.com/udy11/ pyIPREQ,https://gitlab.com/udy11/pyIPREQ, 2025

    2025

  7. [15]

    Plasma column position measurements using magnetic diagnostics in ADITYA-U tokamak.Plasma Research Express, 3(3):035005, Sep 2021

    S Aich, R Kumar, T M Macwan, D Kumavat, S Jha, R L Tanna, K Sathyanarayana, J Ghosh, K A Jadeja, K Pa- tel, et al. Plasma column position measurements using magnetic diagnostics in ADITYA-U tokamak.Plasma Research Express, 3(3):035005, Sep 2021

    2021

  8. [16]

    S. Aich, T. M. Macwan, K. Galodiya, K. Singh, S. Dolui, J. Ghosh, R. L. Tanna, Abhijeet Kumar, H. Mandliya, E. V. Praveenlal, et al. Design and measurements of the diamagnetic loop in Aditya-U tokamak.Radiation Effects and Defects in Solids, 180(3-4):435–446, 2024

    2024

  9. [17]

    Nu- merical determination of axisymmetric toroidal magne- tohydrodynamic equilibria.Journal of Computational Physics, 32(2):212–234, 1979

    J.LJohnson, H.EDalhed, J.MGreene, R.CGrimm, Y.Y Hsieh, S.C Jardin, J Manickam, M Okabayashi, R.G Storer, A.M.M Todd, D.E Voss, and K.E Weimer. Nu- merical determination of axisymmetric toroidal magne- tohydrodynamic equilibria.Journal of Computational Physics, 32(2):212–234, 1979

    1979

  10. [18]

    Realtimeverticalpo- sition estimation of plasma column using fast imaging in Aditya-U tokamak.Nuclear Fusion, Jul 2025

    Suman Aich, Sharvil Patel, Laxmikanta Pradhan, Ashok Kumar Kumawat, Bharat Hegde, Kalpesh Ga- lodiya, RLTanna, KumarpalsinhAJadeja, MalayBikas Chowdhuri, NandiniYadava, etal. Realtimeverticalpo- sition estimation of plasma column using fast imaging in Aditya-U tokamak.Nuclear ...

    2025

  11. [19]

    Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, et al

    Martín Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S. Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, et al. Ten- sorFlow: Large-Scale Machine Learning on Heteroge- neous Systems, 2015. Software available from tensor- flow.org

    2015

  12. [20]

    KerasTuner

    Tom O’Malley, Elie Bursztein, James Long, François Chollet, Haifeng Jin, Luca Invernizzi, et al. KerasTuner. https://github.com/keras-team/keras-tuner, 2019

    2019

  13. [21]

    Multi- Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics

    Alex Kendall, Yarin Gal, and Roberto Cipolla. Multi- Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2018. 18

    2018

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.