Pith. sign in

REVIEW 3 major objections 5 minor 86 references

Generation of cosmic ray trajectories by a Diffusion Model trained on test particles in 3D magnetohydrodynamic turbulence

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a diffusion model trained only on velocity trajectories from test particles in 3D magnetohydrodynamic turbulence can synthesize new cosmic-ray trajectories whose statistics match the baseline at fixed particle…

desk verdict A solid proof-of-concept for diffusion-based cosmic-ray trajectory generation, but the in-sample evaluation leaves the generalization claim unproven. read the letter →

arxiv 2412.12923 v2 pith:RL7JSP6D submitted 2024-12-17 physics.flu-dyn astro-ph.HEastro-ph.SRphysics.plasm-ph

Johannes Martin , Jeremiah Lübke , Tianyi Li , Michele Buzzicotti , Rainer Grauer , Luca Biferale This is my paper · ORCID
classification physics.flu-dynastro-ph.HEastro-ph.SRphysics.plasm-ph
keywords cosmicraysdiffusionmodelsmagnetohydrodynamicturbulencetestparticletrajectoriesLagrangianstatisticsgyro-centercurvatureanomalousscalinggenerativemachinelearning
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 tries to establish that a generative diffusion model, trained only on the velocity components of cosmic-ray test-particle trajectories extracted from a 3D magnetohydrodynamic turbulence simulation, can synthesize new trajectories whose statistics match the simulated baseline at fixed particle energy. The test covers three coupling strengths, $\alpha = 32, 128, 512$, corresponding to different particle energies, and checks short-time velocity increments, fourth-order flatness, long-time mean squared displacement, and the curvature of the gyro-center motion including magnetic mirror events. The authors compare against two synthetic turbulence generators and find that the diffusion model tracks the MHD baseline much more closely for transport and curvature statistics. If correct, this provides a fast black-box stochastic generator for charged-particle trajectories in a fixed turbulent magnetic environment, avoiding the expense of direct numerical simulation for the trained energies. The paper states clearly that the model does not yet generalize to unseen energies or background fields; each $\alpha$ requires its own trained network.

What carries the argument

The central object is a denoising diffusion probabilistic model applied to trajectory segments of the velocity vector $V(t)$. The forward process gradually adds Gaussian noise to a training trajectory, and a U-Net with residual and attention blocks learns to reverse this process, so that sampling from pure noise yields new trajectories from the learned distribution. To enforce energy conservation and reduce dimensionality, the velocity is converted to spherical coordinates $(\theta, \phi)$ with radius fixed to one, so the generated signals automatically satisfy $\|V\| = 1$. Spatial trajectories are recovered by integrating the velocity signal, and the gyro-center curvature is computed after low-pass filtering with a window of $1.5\,T_g$ to isolate motion on scales above one gyration. The load-bearing statistics are the Lagrangian structure functions, the fourth-order flatness, the mean squared displacement, and the curvature distribution of the gyro-center motion.

What would settle it

Train the same diffusion model on trajectories from a subset of the MHD snapshots and compare generated trajectories against the remaining held-out snapshots; if the velocity structure functions, mean squared displacement, or gyro-center curvature deviate from the held-out statistics by more than the in-training agreement, the claim of faithful synthesis fails. A second, already bounded check is to generate trajectories at an untrained energy such as $\alpha = 64$ and test whether the statistics interpolate between $\alpha = 32$ and $128$; the paper reports this currently fails, which delimits the claim to the trained energies.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a diffusion model trained exclusively on velocity signals from test particles in 3D MHD turbulence generates cosmic-ray trajectories whose physical statistics agree with the baseline for the trained particle energies. The model reproduces velocity statistics at intermediate and long times, the long-time transport behavior measured by mean squared displacement, and the geometry of the gyro-center motion, including the $\kappa^{-2.5}$ high-curvature tail associated with sharp turns and magnetic mirror events. At very short lags the generated signals lack the small-scale smoothness of the baseline and require a three-point low-pass filter. The paper also reports that the model cannot interpolate or extrapolate to particle energies not seen in training, so a separate network is needed for each $\alpha$. Compared with the continuous cascade and Lagrangian mapping synthetic turbulence fields, the diffusion model is the only generator that reproduces the MHD diffusion coefficients and gyro-center curvature.

Load-bearing premise

The evaluation treats the same 96,000 MHD test-particle trajectories that trained the diffusion model as the independent baseline truth, and no held-out trajectories are used, so the reported agreement could partly reflect the model reproducing statistics of its own training data.

Editorial extensions

If this is right

  • For the three trained values of $\alpha$, the diffusion model can act as a fast generator of cosmic-ray trajectories whose bulk transport matches direct MHD simulation, without re-running the turbulence solver.
  • Because it matches gyro-center curvature and mirror-event geometry, the model captures coherent-structure effects that spectrum-based synthetic fields (CC and LM) miss in the transport statistics.
  • Training on velocity alone is sufficient to encode the magnetic field's influence on particle motion at the trained energies, so no explicit field information is needed at generation time.
  • A separate network is required for each energy; making $\alpha$ a conditioning variable of a single model is the paper's stated next step.
  • The same approach could interpolate or impute missing segments of observed cosmic-ray time series once energies and background fields are made controllable.

Reading between the lines

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

  • Beyond the paper: the agreement is evaluated against the same trajectories used for training, since no held-out test-particle set is described; a validation set from a fresh MHD snapshot would test whether the model learned transport physics rather than reproducing its training statistics.
  • Beyond the paper: because the model already reproduces statistics at three discrete $\alpha$ values, conditioning the U-Net on a scalar $\alpha$ embedding and training jointly is the most direct route to energy interpolation, and its success or failure would sharply bound the approach.
  • Beyond the paper: the gyro-center curvature distribution, with its robust $\kappa^{-2.5}$ tail, could serve as a sensitive diagnostic for any generative model of cosmic-ray transport, since it reflects mirror events and fieldline structure rather than trivial one-point velocity statistics.
  • Beyond the paper: if extended to variable background magnetic fields, the generator could produce cheap synthetic cosmic-ray trajectory datasets for training surrogate models of diffusion coefficients in regimes where direct simulation is too expensive.
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 / 5 minor

Summary. The paper presents a diffusion-model-based generative approach for charged-particle trajectories in 3D MHD turbulence. It trains separate denoising diffusion probabilistic models on velocity-only segments of test-particle trajectories for three values of alpha, using a spherical-coordinate encoding with unit speed. Generated trajectories are evaluated through second-order velocity structure functions, fourth-order flatness, mean squared displacement, and gyro-center curvature/torsion, and are compared with the MHD baseline and with trajectories from continuous-cascade and Lagrangian-mapping synthetic turbulence models. The authors report excellent agreement at intermediate and long times, in transport statistics, and in curvature statistics, while conceding small-scale mismatch and the need for a separate model per alpha.

Significance. If the agreement were established on a held-out test set, the method would be a useful fast black-box generator of test-particle trajectories in a fixed turbulent magnetic environment, avoiding repeated Lorentz-force integration in the trained regime. The paper is careful in documenting network architecture, hyperparameters, and the training/sampling algorithms, and the cosine-similarity analysis in Appendix B is a useful guard against memorization of individual samples. The comparison with the CC and LM synthetic models gives the work useful external context. However, the absence of a train/test split and of uncertainty quantification means that the advertised agreement is not yet independent evidence of a correct physical model.

major comments (3)
  1. [Section 2.2 and Section 3] The reference MHD statistics are computed from the same 96,000 trajectories per alpha that are used for training (Sections 2.1 and 2.2); no held-out test trajectories are described. Because the training objective in Eq. (9) minimizes the mismatch with the training distribution, agreement between generated samples and the training-set statistics is a consistency check rather than independent validation that the model synthesizes new physical trajectories. Please add a train/test split (e.g., train on one subset and evaluate all Section 3 statistics on a disjoint held-out subset) and repeat the same figures for the held-out baseline. The cosine-similarity check in Appendix B addresses memorization of individual samples, but not overfitting to the empirical distribution.
  2. [Section 3, first paragraph; Figures 4 and 5] The DM outputs are post-processed with a low-pass filter of 3 grid points (0.03 Tg) before any statistics are computed, and the text states this is done because the model struggles with small-scale smoothness. The paper does not show unfiltered DM results or a sensitivity study of the filter width. Since short-time velocity statistics are part of the central claim, the reported agreement in S^(2)_tau and F^(4)_tau may be partly an artifact of this post-processing step. Please quantify the effect of the filter and present the raw generated trajectories as well.
  3. [Figures 3-7 and Section 3] No error bars or confidence intervals are given for any of the statistical comparisons, although the statistics are estimated from finite samples of 10,000 signals per model (Section 3, first paragraph). In particular, the deviations of the DM from the MHD baseline at small tau and alpha=512 in Figure 4 cannot be judged without uncertainty estimates. Please add bootstrap confidence bands over trajectories (and, if feasible, over MHD snapshots) for each reported statistic.
minor comments (5)
  1. [Section 3.1] The text says 'the scale-dependent flatness is plotted in Figure 11,' but the flatness results appear in Figure 5; Figure 11 belongs to Appendix B. Please correct the cross-reference.
  2. [Equations (14)-(15) and Section 2.2] The symbol theta is used both for the spherical polar angle in the input representation and for torsion in the curvature formulas; please disambiguate the notation.
  3. [Section 3.1] The wording 'the CC and LM models perform rather bad' and 'exhibits ... very well' should be corrected to 'rather poorly' and 'agrees very well' (or similar).
  4. [Figure 3 caption] The caption states 'Power-law tails for small tau,' while the surrounding text and the plotted distributions indicate exponential tails; the caption and text should be aligned.
  5. [Section 2.2 and Section 3] Please clarify whether the 10,000 evaluation signals are individual generated segments or concatenations of segments, and how continuity across segment boundaries is handled during sampling and integration to X(t).

Circularity Check

1 steps flagged · score 6.0 of 10

The reported agreement is between generated samples and the same MHD trajectories used for training; without a held-out split, the central claim reduces to an in-sample consistency check.

  1. fitted input called prediction [Section 2.2, Eq. (9); Section 3 opening; Section 4 summary]
    "In total we simulate 96 000 particles per α ... and record their trajectories ... (Sec. 2.1); 'our training dataset consists of 96 000 trajectories, each sampled at 100 000 grid points' (Sec. 2.2). The training loss is Lsimple = E_{n,V0∼q(V0), ϵ∼N(0,Id)} [∥ϵ − ϵθ(Vn,n)∥²], 'where ..."

    Eq. (9) optimizes the model to denoise samples drawn from q(V0), which is the empirical distribution of the 96,000 baseline MHD trajectories. The reference statistics in Section 3 (structure functions, flatness, MSD, curvature) are computed from that same baseline. A successful diffusion model is therefore expected, by construction, to generate samples whose statistics are close to the training distribution; the reported 'excellent agreement' is a fit/consistency check, not an independent validation. No train/test split is described, and the Appendix B cosine-similarity test only excludes memorization of individual samples, not overfitting to the ensemble distribution. Some independent content remains because MSD and gyro-center curvature are not directly minimized by Eq.

full rationale

The paper contains no load-bearing self-citation chain: the diffusion-model architecture is attributed to Ho et al. and Dhariwal & Nichol, not to the authors' own prior work, and the cited Li et al. and Lübke et al. results are used as context or comparison, not as justification for the central claim. The curvature-scaling discussion is presented as an explanatory model, not as a derived prediction. The main circularity concern is the evaluation protocol: the 96,000 MHD trajectories are both the training data (Section 2.2) and the baseline against which the DM's generated trajectories are judged (Section 3). Since the diffusion objective is a denoising loss over exactly this training distribution, agreement of generated samples with summary statistics of the same distribution is a measure of how well the model fitted its training set. The additional low-pass filtering applied only to DM trajectories and the absence of any held-out MHD trajectories further weaken the external character of the validation. The cosine-similarity analysis in Appendix B is a useful memorization check but does not establish out-of-sample generalization. Thus the headline 'excellent agreement with the baseline trajectories' is partially circular: it demonstrates in-sample reconstruction rather than independent prediction of MHD transport statistics.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central empirical claim rests on a trained generative model, so the main free parameters are the post-processing filters and network hyperparameters chosen to make the statistics agree. The domain assumptions specify the physical environment (static, incompressible, isotropic MHD snapshots) and the choice of summary statistics as the definition of fidelity. No new physical entities are introduced.

free parameters (5)
  • DM output low-pass filter window = 3 grid points (0.03 Tg)
    Applied to generated DM trajectories in Section 3 to compensate for the model's small-scale smoothness deficit; directly affects the reported small-lag velocity statistics.
  • Gyro-center velocity filter window = 1.5 Tg (150 grid points)
    Chosen in Sections 2.3 and 3.3 as a proxy for the gyro-center motion; all curvature statistics depend on this choice.
  • Trajectory segment lengths = 8192 (alpha=512), 2048 (alpha=128), 1024 (alpha=32) grid points
    Section 2.2: chosen so that generalized flatness is constant at larger scales; determines the training distribution and available context.
  • Diffusion noise schedule endpoints = beta_1=1e-4 to beta_N=0.02, linear
    Table 3 fixed hyperparameters; part of the generative model training setup.
  • UNet depth and training hyperparameters = layers 6/10/9 for alpha=512/128/32; 1000 diffusion steps; 64 initial channels; learning rate 1e-4; batch size 64
    Tables 2 and 3; fine-tuned per particle energy, chosen by training experiments in Appendix B.
assumptions (6)
  • domain assumption Static snapshots of incompressible, Pm=1, 1024^3 MHD turbulence with periodic boundaries and random forcing adequately represent the magnetic-field environment for cosmic-ray transport.
    Section 2.1; the whole baseline dataset is built on these snapshots, and any transport conclusions inherit this representation.
  • domain assumption Test particles obey the Newton-Lorentz equation in a static magnetic field with no electric fields, no back-reaction, and normalized speed |V|=1, so particle energy is conserved.
    Equation (2) and surrounding text; this is the model of charged-particle motion used for both training and evaluation.
  • domain assumption The diffusion-model training objective L_simple (Eq. 9) yields a generator whose samples reproduce the true multi-time velocity distribution of the training data.
    Section 2.2; the paper relies on standard diffusion-model theory, but its convergence for this trajectory distribution is only checked through a few summary statistics.
  • domain assumption The statistical benchmarks considered (second and fourth order structure functions, MSD, gyro-center curvature) are sufficient to establish that the synthetic model is quantitatively correct for transport.
    Section 2.3; all conclusions are drawn from these summary statistics, not from full trajectory distribution comparisons.
  • standard math The asymptotic curvature scalings follow from assuming independent distributions for acceleration and velocity (chi-squared, chi-cubed, uniform), following Scagliarini (2011).
    Appendix C; the derivation is standard but the applicability to gyro-center motion is asserted from observations.
  • domain assumption Ten statistically independent MHD snapshots provide enough sampling for converged reference statistics.
    Section 2.1; no convergence analysis or error bars are given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generation of cosmic ray trajectories by a Diffusion Model trained on test particles in 3D magnetohydrodynamic turbulence." pith.science (2026). https://pith.science/paper/RL7JSP6D

@misc{pith2026241212923,
  author       = {Pith},
  title        = {Pith review of: Generation of cosmic ray trajectories by a Diffusion Model trained on test particles in 3D magnetohydrodynamic turbulence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RL7JSP6D}},
  note         = {Machine review of arXiv:2412.12923}
}
read the original abstract

Models for the transport of high energy charged particles through strong magnetic turbulence play a key role in space and astrophysical studies, such as describing the propagation of solar energetic particles and high energy cosmic rays. Inspired by the recent advances in high-performance machine learning techniques, we investigate the application of generative diffusion models to synthesizing test particle trajectories obtained from a turbulent magnetohydrodynamics simulation. We consider velocity increment, spatial transport and curvature statistics, and find excellent agreement with the baseline trajectories for fixed particle energies. Additionally, we consider two synthetic turbulence models for comparison. Finally, challenges towards an application-ready transport model based on our approach are discussed.

Figures

Figures reproduced from arXiv: 2412.12923 by the authors.

Figure 1
Figure 1. Schematic illustration of the Diffusion Model (DM) process. Algorithm 1 Training 1: Randomly select a trajectory V0 from the training dataset distribution q(V0). 2: Randomly select an intermediate step n of the Markov chain from a uniform distribution {1, ..., N}. 3: Create noise according to ϵ ∼ N (0,Id). 4: Compute the noisy trajectory in the n-th step of the Markov chain as Vn = √ α¯nV0 + √ 1 − α¯nϵ. 5: Make a gr… view at source ↗
Figure 2
Figure 2. Example trajectories for different models of charged particles in magnetic turbulence, colored by the curvature of the gyro-center motion. The cases (a), (c) and (d) are obtained by solving the Newton-Lorentz equations in a turbulent magnetohydrodynamics (MHD) snapshot, a multifractal continuous cascade (CC) field and a structured Lagrangian mapping (LM) field. The case (b) is generated by our generative diffusion m… view at source ↗
Figure 3
Figure 3. Distributions of normalized increments δτV /σδτ V for different time lags τ . Shown are the MHD and DM cases with α = 512. Power-law tails for small τ with strong non-Gaussianity, and confined sub-Gaussian behavior for large τ due to Vi ∈ [−1, 1]. 10−2 10−1 100 101 τ /Tg 0.8 1.0 1.2 S(2) τ /S(2) τ,MHD α = 512 MHD DM CC LM 10−2 10−1 100 101 τ /Tg α = 128 MHD DM CC LM 10−2 10−1 100 101 τ /Tg α = 32 MHD DM CC LM [PITH… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Second-order velocity structure functions S (2) τ of particle trajectories, compensated by the MHD case to highlight differences. 10−2 10−1 100 101 τ /Tg 101 3 × 100 4 × 100 6 × 100 F(4) τ α = 512 MHD DM CC LM 10−2 10−1 100 101 τ /Tg α = 128 MHD DM CC LM 10−2 10−1 100 …
Figure 5
Figure 5. Figure 5: Fourth-order velocity flatness F (4) τ of particle trajectories, indicating the deviation from the Gaussian case F (4) τ = 3 [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Mean squared displacement as an indicator for the transport properties of charged particles. Compensated by the time τ to distinguish between ballistic and diffusive behavior. and averaging according to Equation (12), the effects of the intermittent short-time velocity…
Figure 7
Figure 7. Figure 7: Distributions of curvature of gyro-center trajectory, by averaging the velocity signals V(t) with a window of 1.5 Tg. The gyro-center motion exhibits an asymptotic tail with slope −2.5, and, for large α, an intermediate regime with slope −1.5. The low-curvature tails s…
Figure 8
Figure 8. Figure 8: Example of a mirror event in a DM-generated trajectory. The event is characterized by high curvature and a change of the handedness of the helix, as indicated by the sign of the torsion. The curvature and torsion are computed for the gyro-center motion. 10−1 100 101 10…
Figure 9
Figure 9. Figure 9: Distribution of fieldline curvature in turbulent vector fields. The −2.5 asymptotic scaling is only present in the MHD case, while the CC case is characterized by a −4 slope. All low-curvature tails scale also with slope 1. The distributions are again compensated with …
Figure 10
Figure 10. Figure 10: Example diagram of the employed UNet architecture with 5 layers. The upper row corresponds to the encoder, the lower row to the decoder, and the vertical part to the bottleneck. The dashed lines represent the connecting paths where the data is concatenated. of cosmic …
Figure 11
Figure 11. Figure 11: a) Training loss function against epochs. b) Fourth-order flatness F (4) τ of the particle trajectories generated by the diffusion model at 5 different training times (A-E, highlighted in the training loss function), compensated by the MHD case to highlight difference…
Figure 12
Figure 12. Figure 12: Distribution of the cosine similarity of the closest training trajectory for a generated sample (red) and training trajectory (black) for α = 128. (D-E). Additionally, it can be seen that the model prioritizes the accuracy of the intermediate and long scales, where a …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

86 extracted references · 39 canonical work pages

  1. [1]

    2025, A&A, 693, A15, 10.1051/0004-6361/202451765

    Aerdker, Sophie , Merten, Lukas , Effenberger, Frederic , Fichtner, Horst , & Becker Tjus, Julia . 2025, A&A, 693, A15, 10.1051/0004-6361/202451765

    doi:10.1051/0004-6361/202451765 2025

  2. [2]

    Alouani-Bibi, F., & le Roux, J. A. 2014, , 781, 93

    2014

  3. [3]

    1999, Physics of Fluids, 11, 1880, 10.1063/1.870050

    Alvelius, K. 1999, Physics of Fluids, 11, 1880, 10.1063/1.870050

    work page doi:10.1063/1.870050 1999

  4. [4]

    2008, Phys

    Arn\`eodo, A., Benzi, R., Berg, J., et al. 2008, Phys. Rev. Lett., 100, 254504, 10.1103/PhysRevLett.100.254504

    work page doi:10.1103/physrevlett.100.254504 2008

  5. [5]

    H., et al

    Bandyopadhyay, R., Yang, Y., Matthaeus, W. H., et al. 2020, ApJL, 893, L25, 10.3847/2041-8213/ab846e

    doi:10.3847/2041-8213/ab846e 2020

  6. [6]

    M., Xu, S., & Lazarian, A

    Barreto-Mota, L., de Gouveia Dal Pino, E. M., Xu, S., & Lazarian, A. 2024, Cosmic Ray Diffusion in the Turbulent Interstellar Medium: Effects of Mirror Diffusion and Pitch Angle Scattering. 2405.12146

    arXiv 2024

  7. [7]

    D., Lalescu, C

    Bentkamp, L., Drivas, T. D., Lalescu, C. C., & Wilczek, M. 2022, Nat. Commun., 13, 2088, 10.1038/s41467-022-29422-1

    work page doi:10.1038/s41467-022-29422-1 2022

  8. [8]

    2008, Physics of Fluids, 20, 065103, 10.1063/1.2930672

    Biferale, L., Bodenschatz, E., Cencini, M., et al. 2008, Physics of Fluids, 20, 065103, 10.1063/1.2930672

    doi:10.1063/1.2930672 2008

Show all 86 references
  1. [9]

    1998, Phys

    Biferale, L., Boffetta, G., Celani, A., Crisanti, A., & Vulpiani, A. 1998, Phys. Rev. E, 57, R6261, 10.1103/PhysRevE.57.R6261

    1998 doi

  2. [10]

    2004, Phys

    Biferale, L., Boffetta, G., Celani, A., et al. 2004, Phys. Rev. Lett., 93, 064502, 10.1103/PhysRevLett.93.064502

    2004 doi

  3. [11]

    Bilodeau, C., Jin, W., Jaakkola, T., Barzilay, R., & Jensen, K. F. 2022, WIREs Computational Molecular Science, 12, e1608, https://doi.org/10.1002/wcms.1608

    2022 doi

  4. [12]

    2003, Magnetohydrodynamic Turbulence (Cambridge University Press)

    Biskamp , D. 2003, Magnetohydrodynamic Turbulence (Cambridge University Press)

    2003

  5. [13]

    Boris, J. P. 1970, Proceeding of Fourth Conference on Numerical Simulations of Plasmas

    1970

  6. [14]

    S., Hopkins, P

    Butsky, I. S., Hopkins, P. F., Kempski, P., et al. 2024, Monthly Notices of the Royal Astronomical Society, 528, 4245, 10.1093/mnras/stae276

    2024 doi

  7. [15]

    2001, Phys

    Casse, F., Lemoine, M., & Pelletier, G. 2001, Phys. Rev. D, 65, 023002, 10.1103/PhysRevD.65.023002

    2001 doi

  8. [16]

    E., Bott, A

    Chen, L. E., Bott, A. F. A., Tzeferacos, P., et al. 2020, The Astrophysical Journal, 892, 114, 10.3847/1538-4357/ab7a19

    2020 doi

  9. [17]

    G., Lev\^eque, E., et al

    Chevillard, L., Roux, S. G., Lev\^eque, E., et al. 2003, Phys. Rev. Lett., 91, 214502, 10.1103/PhysRevLett.91.214502

    2003 doi

  10. [18]

    , & Marcowith, A

    Cohet, R. , & Marcowith, A. 2016, , 588, A73, 10.1051/0004-6361/201527376

    2016 doi

  11. [19]

    1990, Mathematical geology, 22, 239

    Cressie, N. 1990, Mathematical geology, 22, 239

    1990

  12. [20]

    2021 a , in Advances in Neural Information Processing Systems, ed

    Dhariwal, P., & Nichol, A. 2021 a , in Advances in Neural Information Processing Systems, ed. M. Ranzato, A. Beygelzimer, Y. Dauphin, P. Liang, & J. W. Vaughan, Vol. 34 (Curran Associates, Inc.), 8780--8794. https://proceedings.neurips.cc/paper_files/paper/2021/file/49ad23d1ec...

    2021

  13. [21]

    2021 b , Advances in Neural Information Processing Systems, 34, 8780

    Dhariwal, P., & Nichol, A. 2021 b , Advances in Neural Information Processing Systems, 34, 8780. https://proceedings.neurips.cc/paper_files/paper/2021/file/49ad23d1ec9fa4bd8d77d02681df5cfa-Paper.pdf

    2021

  14. [22]

    Dundovic, A., Pezzi, O., Blasi, P., Evoli, C., & Matthaeus, W. H. 2020, Phys. Rev. D, 102, 103016, 10.1103/PhysRevD.102.103016

    2020 doi

  15. [23]

    2022, Phys

    Durrive, J.-B., Changmai, M., Keppens, R., et al. 2022, Phys. Rev. E, 106, 025307, 10.1103/PhysRevE.106.025307

    2022 doi

  16. [24]

    2020, Monthly Notices of the Royal Astronomical Society, 496, 3015, 10.1093/mnras/staa1514

    Durrive, J.-B., Lesaffre, P., & Ferrière, K. 2020, Monthly Notices of the Royal Astronomical Society, 496, 3015, 10.1093/mnras/staa1514

    2020 doi

  17. [25]

    J., Reichherzer, P., Bott, A

    Ewart, R. J., Reichherzer, P., Bott, A. F. A., Kunz, M. W., & Schekochihin, A. A. 2024, Monthly Notices of the Royal Astronomical Society, 532, 2098, 10.1093/mnras/stae1578

    2024 doi

  18. [26]

    1995, Turbulence: the legacy of AN Kolmogorov (Cambridge University Press)

    Frisch, U. 1995, Turbulence: the legacy of AN Kolmogorov (Cambridge University Press)

    1995

  19. [27]

    Giacalone, J., & Jokipii, J. R. 1999, , 520, 204, 10.1086/307452

    1999 doi

  20. [28]

    2019, Magnetohydrodynamics of Laboratory and Astrophysical Plasmas (Cambridge University Press)

    Goedbloed, H., Keppens, R., & Poedts, S. 2019, Magnetohydrodynamics of Laboratory and Astrophysical Plasmas (Cambridge University Press)

    2019

  21. [29]

    2024, in ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 13136--13140, 10.1109/ICASSP48485.2024.10447612

    Grassucci, E., Marinoni, C., Rodriguez, A., & Comminiello, D. 2024, in ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 13136--13140, 10.1109/ICASSP48485.2024.10447612

    2024

  22. [30]

    1994, , 195, 335, https://doi.org/10.1016/0375-9601(94)90038-8

    Grauer, R., Krug, J., & Marliani, C. 1994, , 195, 335, https://doi.org/10.1016/0375-9601(94)90038-8

    1994 doi

  23. [31]

    2022, Advances in neural information processing systems, 35, 478

    Guth, F., Coste, S., De Bortoli, V., & Mallat, S. 2022, Advances in neural information processing systems, 35, 478

    2022

  24. [32]

    2020, ArXiv, abs/2006.11239

    Ho, J., Jain, A., & Abbeel, P. 2020, ArXiv, abs/2006.11239. https://api.semanticscholar.org/CorpusID:219955663

    2020 arXiv

  25. [33]

    2022, MNRAS, 512, 2111, 10.1093/mnras/stac319

    Hu, Y., Lazarian, A., & Xu, S. 2022, MNRAS, 512, 2111, 10.1093/mnras/stac319

    2022 doi

  26. [34]

    P., Sreenivasan, K

    Juneja, A., Lathrop, D. P., Sreenivasan, K. R., & Stolovitzky, G. 1994, Phys. Rev. E, 49, 5179, 10.1103/PhysRevE.49.5179

    1994 doi

  27. [35]

    Kamal Youssef, F. R. , & Grenier, I. A. 2024, A&A, 685, A102, 10.1051/0004-6361/202348299

    2024 doi

  28. [36]

    B., Quataert , E., et al

    Kempski , P., Fielding , D. B., Quataert , E., et al. 2023, arXiv e-prints, arXiv:2304.12335, 10.48550/arXiv.2304.12335

    2023 doi arXiv

  29. [37]

    Kolmogorov, A. N. 1941, in Dokl. Akad. Nauk SSSR, Vol. 30, 301--305

    1941

  30. [38]

    2012, , 749, 103, 10.1088/0004-637X/749/2/103

    Laitinen , T., Dalla , S., & Kelly , J. 2012, , 749, 103, 10.1088/0004-637X/749/2/103

    2012 doi

  31. [39]

    2021, ApJ, 923, 53, 10.3847/1538-4357/ac2de9

    Lazarian, A., & Xu, S. 2021, ApJ, 923, 53, 10.3847/1538-4357/ac2de9

    2021 doi

  32. [40]

    2023, arXiv e-prints, arXiv:2304.03023, 10.48550/arXiv.2304.03023

    Lemoine , M. 2023, arXiv e-prints, arXiv:2304.03023, 10.48550/arXiv.2304.03023

    2023 doi arXiv

  33. [41]

    2024, arXiv preprint arXiv:2405.03468

    Lempereur, E., & Mallat, S. 2024, arXiv preprint arXiv:2405.03468

    2024

  34. [42]

    2024 a , arXiv preprint arXiv:2410.23971

    Li, T., Biferale, L., Bonaccorso, F., Buzzicotti, M., & Centurioni, L. 2024 a , arXiv preprint arXiv:2410.23971

    2024 arXiv

  35. [43]

    A., & Buzzicotti, M

    Li, T., Biferale, L., Bonaccorso, F., Scarpolini, M. A., & Buzzicotti, M. 2024 b , Nature Machine Intelligence, 1

    2024

  36. [44]

    S., Buzzicotti, M., Bonaccorso, F., & Biferale, L

    Li, T., Lanotte, A. S., Buzzicotti, M., Bonaccorso, F., & Biferale, L. 2024 c , Atmosphere, 15, 10.3390/atmos15010060

    2024 doi

  37. [45]

    2024 d , International Journal of Multiphase Flow, 181, 104980, https://doi.org/10.1016/j.ijmultiphaseflow.2024.104980

    Li, T., Tommasi, S., Buzzicotti, M., Bonaccorso, F., & Biferale, L. 2024 d , International Journal of Multiphase Flow, 181, 104980, https://doi.org/10.1016/j.ijmultiphaseflow.2024.104980

    2024

  38. [46]

    2024, Europhysics Letters, 146, 43001, 10.1209/0295-5075/ad438f

    Lübke, J., Effenberger, F., Wilbert, M., Fichtner, H., & Grauer, R. 2024, Europhysics Letters, 146, 43001, 10.1209/0295-5075/ad438f

    2024 doi

  39. [47]

    2023, Journal of Physics: Complexity, 4, 015005, 10.1088/2632-072X/acb128

    Lübke, J., Friedrich, J., & Grauer, R. 2023, Journal of Physics: Complexity, 4, 015005, 10.1088/2632-072X/acb128

    2023 doi

  40. [48]

    2024, BxC Toolkit: Generating Tailored Turbulent 3D Magnetic Fields

    Maci, D., Keppens, R., & Bacchini, F. 2024, BxC Toolkit: Generating Tailored Turbulent 3D Magnetic Fields. 2405.09587

    2024 arXiv

  41. [49]

    Meneveau, C., & Sreenivasan, K. R. 1987, Phys. Rev. Lett., 59, 1424, 10.1103/PhysRevLett.59.1424

    1987 doi

  42. [50]

    2000, Physics Reports, 339, 1, https://doi.org/10.1016/S0370-1573(00)00070-3

    Metzler, R., & Klafter, J. 2000, Physics Reports, 339, 1, https://doi.org/10.1016/S0370-1573(00)00070-3

    2000 doi

  43. [51]

    2024, Applied and Computational Harmonic Analysis, 101724

    Morel, R., Rochette, G., Leonarduzzi, R., Bouchaud, J.-P., & Mallat, S. 2024, Applied and Computational Harmonic Analysis, 101724

    2024

  44. [52]

    2019, Phys

    Muzy, J.-F. 2019, Phys. Rev. E, 99, 042113, 10.1103/PhysRevE.99.042113

    2019 doi

  45. [53]

    Q., & Dhariwal, P

    Nichol, A. Q., & Dhariwal, P. 2021, in International Conference on Machine Learning, PMLR, 8162--8171

    2021

  46. [54]

    2024, Strong turbulence and magnetic coherent structures in the interstellar medium

    Ntormousi, E., Vlahos, L., Konstantinou, A., & Isliker, H. 2024, Strong turbulence and magnetic coherent structures in the interstellar medium. 2409.16699

    2024 arXiv

  47. [55]

    2024, Journal of Cosmology and Astroparticle Physics, 2024, 002, 10.1088/1475-7516/2024/01/002

    Palade, D. 2024, Journal of Cosmology and Astroparticle Physics, 2024, 002, 10.1088/1475-7516/2024/01/002

    2024 doi

  48. [56]

    M., Garban, C., & Chevillard, L

    Pereira, R. M., Garban, C., & Chevillard, L. 2016, , 794, 369–408, 10.1017/jfm.2016.166

    2016 doi

  49. [57]

    2007, , 14, 012311, 10.1063/1.2434795

    Pommois, P., Zimbardo, G., & Veltri, P. 2007, , 14, 012311, 10.1063/1.2434795

    2007 doi

  50. [58]

    2016, , 459, 3395, 10.1093/mnras/stw877

    Pucci, F., Malara, F., Perri, S., et al. 2016, , 459, 3395, 10.1093/mnras/stw877

    2016 doi

  51. [59]

    A., & Ni, R

    Qi, Y., Meneveau, C., Voth, G. A., & Ni, R. 2023, Phys. Rev. Lett., 130, 154001, 10.1103/PhysRevLett.130.154001

    2023 doi

  52. [60]

    H., & Bieber , J

    Qin , G., Matthaeus , W. H., & Bieber , J. W. 2002, , 578, L117, 10.1086/344687

    2002 doi

  53. [61]

    G., Merten , L., & Pueschel , M

    Reichherzer , P., Becker Tjus , J., Zweibel , E. G., Merten , L., & Pueschel , M. J. 2020, , 498, 5051, 10.1093/mnras/staa2533

    2020 doi

  54. [62]

    Reichherzer, P., Bott, A. F. A., Ewart, R. J., et al. 2023, Efficient micromirror confinement of sub-TeV cosmic rays in galaxy clusters. 2311.01497

    2023 arXiv

  55. [63]

    2020, A&A, 641, A138, 10.1051/0004-6361/201937085

    Robitaille, J.-F., Abdeldayem, A., Joncour, I., et al. 2020, A&A, 641, A138, 10.1051/0004-6361/201937085

    2020 doi

  56. [64]

    Ronneberger, O., Fischer, P., & Brox, T. 2015, in Medical Image Computing and Computer-Assisted Intervention--MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18, Springer, 234--241

    2015

  57. [65]

    2008, Phys

    Rosales, C., & Meneveau, C. 2008, Phys. Rev. E, 78, 016313, 10.1103/PhysRevE.78.016313

    2008 doi

  58. [66]

    L., Beattie, J

    Sampson, M. L., Beattie, J. R., Krumholz, M. R., et al. 2022, Monthly Notices of the Royal Astronomical Society, 519, 1503, 10.1093/mnras/stac3207

    2022 doi

  59. [67]

    2011, Journal of Turbulence, 12, N25, 10.1080/14685248.2011.571261

    Scagliarini, A. 2011, Journal of Turbulence, 12, N25, 10.1080/14685248.2011.571261

    2011

  60. [68]

    Schekochihin, A. A. 2022, Journal of Plasma Physics, 88, 155880501, 10.1017/S0022377822000721

    2022 doi

  61. [69]

    1994, Phys

    She, Z.-S., & Leveque, E. 1994, Phys. Rev. Lett., 72, 336, 10.1103/PhysRevLett.72.336

    1994 doi

  62. [70]

    P., Seta, A., Bushby, P

    Shukurov, A., Snodin, A. P., Seta, A., Bushby, P. J., & Wood, T. S. 2017, , 839, L16, 10.3847/2041-8213/aa6aa6

    2017 doi

  63. [71]

    A., Wan, M., & Matthaeus, W

    Subedi, P., Chhiber, R., Tessein, J. A., Wan, M., & Matthaeus, W. H. 2014, ApJ, 796, 97, 10.1088/0004-637X/796/2/97

    2014 doi

  64. [72]

    Tautz , R. C. 2010, , 181, 71, 10.1016/j.cpc.2009.09.002

    2010 doi

  65. [73]

    C., & Dosch, A

    Tautz, R. C., & Dosch, A. 2013, Physics of Plasmas, 20, 022302, 10.1063/1.4789861

    2013 doi

  66. [74]

    P., Sarson, G

    Tharakkal, D., Snodin, A. P., Sarson, G. R., & Shukurov, A. 2023, Phys. Rev. E, 107, 065206, 10.1103/PhysRevE.107.065206

    2023 doi

  67. [75]

    2009, Annual Review of Fluid Mechanics, 41, 375, https://doi.org/10.1146/annurev.fluid.010908.165210

    Toschi, F., & Bodenschatz, E. 2009, Annual Review of Fluid Mechanics, 41, 375, https://doi.org/10.1146/annurev.fluid.010908.165210

    2009

  68. [76]

    Braun, F

    W. Braun, F. D. L., & Eckhardt, B. 2006, Journal of Turbulence, 7, N62, 10.1080/14685240600860923

    2006 doi

  69. [77]

    2023, PhD thesis, Ruhr-Universit \"a t Bochum

    Wilbert, M. 2023, PhD thesis, Ruhr-Universit \"a t Bochum

    2023

  70. [78]

    2022, Physics of Fluids, 34, 096607, 10.1063/5.0110153

    Wilbert, M., Giesecke, A., & Grauer, R. 2022, Physics of Fluids, 34, 096607, 10.1063/5.0110153

    2022 doi

  71. [79]

    Williams, C. K. I., & Rasmussen, C. E. 2005, Gaussian processes for machine learning (The MIT Press)

    2005

  72. [80]

    2012, , 750, 150, 10.1088/0004-637x/750/2/150

    Wisniewski, M., Spanier, F., & Kissmann, R. 2012, , 750, 150, 10.1088/0004-637x/750/2/150

    2012 doi

  73. [81]

    T., & Bodenschatz, E

    Xu, H., Ouellette, N. T., & Bodenschatz, E. 2007, Phys. Rev. Lett., 98, 050201, 10.1103/PhysRevLett.98.050201

    2007 doi

  74. [82]

    2006, Physica D, 215, 166 , http://dx.doi.org/10.1016/j.physd.2006.01.012

    Yakhot, V. 2006, Physica D, 215, 166 , http://dx.doi.org/10.1016/j.physd.2006.01.012

    2006 doi

  75. [83]

    H., et al

    Yang, Y., Wan, M., Matthaeus, W. H., et al. 2019, Physics of Plasmas, 26, 072306, 10.1063/1.5099360

    2019 doi

  76. [84]

    H., & Lazarian, A

    Yuen, K. H., & Lazarian, A. 2020, ApJ, 898, 66, 10.3847/1538-4357/ab9360

    2020 doi

  77. [85]

    2023, The Astrophysical Journal Letters, 959, L8, 10.3847/2041-8213/ad0fe5

    Zhang, C., & Xu, S. 2023, The Astrophysical Journal Letters, 959, L8, 10.3847/2041-8213/ad0fe5

    2023 doi

  78. [86]

    Özbey, M., Dalmaz, O., Dar, S. U. H., et al. 2023, IEEE Transactions on Medical Imaging, 42, 3524, 10.1109/TMI.2023.3290149

    2023

Pith tools

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