Pith. sign in

REVIEW 4 major objections 5 minor 64 references

Lund jet images from generative and cycle-consistent adversarial networks

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read This paper claims that a neural generative model trained on Lund jet images reproduces the two-dimensional radiation-pattern density within a few percent, and that a cycle-consistent network can map images between jet categories.

desk verdict A genuinely useful and honest paper: gLund and CycleJet bring GANs to the Lund plane with open code and data, and the few-percent claim holds for average densities, with the authors upfront that event-level correlations are a known limitation. read the letter →

arxiv 1909.01359 v2 pith:CZNFJD7K submitted 2019-09-03 hep-ph cs.LGeess.IVhep-exstat.ML

classification hep-phcs.LGeess.IVhep-exstat.ML PACS 12.38.-t07.05.Mh
keywords generativeadversarialnetworksLundjetplanesubstructurecycle-consistentfastsimulationdataaugmentationdeeplearninginparticlephysics
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 is trying to establish that neural generative models can serve as fast, accurate stand-ins for Monte Carlo simulation of jet substructure. The authors encode each jet's radiation pattern as a Lund jet plane image, train a Least-Squares GAN on probabilistically averaged images, and report that the generated samples reproduce the average reference density to within a few percent in the bulk of the image. They also train a cycle-consistent adversarial network to translate Lund images between jet categories, showing that parton-level images can be promoted to detector-level images and that QCD jets can be mapped to boosted W jets (and back). If these claims hold, event generation for jet-substructure studies could be replaced by near-instant neural inference, and existing simulated samples could be repurposed to new processes or simulation settings without regeneration.

What carries the argument

The central object is the primary Lund jet plane: reclustering a jet's constituents with the Cambridge/Aachen algorithm and stepping down the hardest branch assigns each emission a pixel at coordinates ($\ln(1/\Delta_{ab})$, $\ln k_t$), producing a 24 by 24 binary image of the jet's radiation pattern. The paper's training trick is to average $n_{\rm avg}$ such sparse images and treat each averaged pixel as an activation probability, which makes the discrete images learnable by a Least-Squares GAN, a generator/discriminator pair with a quadratic loss and a minibatch-discrimination layer. For the mapping half, a cycle-consistent adversarial network learns paired forward and inverse translations between two jet-image domains, enforcing that translating an image to the other domain and back recovers the original.

What would settle it

Compute the joint probability that two pixels separated widely in $\ln k_t$ or $\ln(1/\Delta_{ab})$ are simultaneously activated, for both generated and reference samples; if the single-pixel densities agree to a few percent but the joint probabilities disagree by more than the statistical uncertainty, the model's independent-pixel sampling is falsified as a description of individual jets.

Watch

Extended reading notes

Core claim

The central discovery claimed is twofold. First, with a stochastic probabilistic interpretation of Lund images, averaging $n_{\rm avg}$ input images and sampling each pixel as a probability, an LSGAN with a minibatch-discrimination layer and ZCA whitening reproduces the average Lund jet plane density of detector-level QCD jets to within 3 to 5 percent in the bulk region, matching a WGAN-GP and clearly outperforming a VAE whose accuracy saturates around 20 percent because of posterior collapse. Second, a CycleGAN-based model, CycleJet, learns unpaired translations between parton-level and detector-level Lund images and between QCD-jet and W-jet images, with the translated average images matching their target domains well enough to be used for retroactive changes of simulation settings or underlying process.

Load-bearing premise

The load-bearing premise is that averaging $n_{\rm avg}$ images and sampling each pixel independently as a probability gives a faithful model of individual jet radiation patterns, not just of the average density.

Editorial extensions

If this is right

  • Jet-substructure samples for the bulk of phase space can be produced by fast network inference instead of full Monte Carlo generation, reducing simulation time and storage.
  • Parton-level samples can be upgraded to detector-level Lund images with a learned mapping, avoiding a full detector-simulation pass.
  • An existing QCD dijet sample can be reinterpreted as a sample of boosted hadronically decaying W jets (and vice versa), enabling process remapping without new event generation.
  • Among the tested alternatives, the LSGAN is the practical choice: the VAE is limited to roughly 20 percent accuracy by posterior collapse, while the WGAN-GP matches distributions but produces less realistic individual images.
  • Groomed observables reconstructed from generated images, such as soft-drop multiplicity and jet mass, follow the reference distributions, supporting the use of these samples in physics analyses.

Reading between the lines

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

  • The paper's per-pixel independence assumption implies that event-level observables sensitive to correlations between widely separated emissions are not guaranteed by the few-percent density agreement; testing such observables would probe the boundary of the claim.
  • The same cycle-consistency recipe could be applied to other domain pairs, such as quark- versus gluon-initiated jets or different pileup conditions, provided the two domains share the Lund-plane support; the paper's two examples motivate but do not establish that generality.
  • Because CycleJet is trained on averages and then samples pixels, applying it to real data would require the preprocessing to be invertible or the mapping to be trained directly on individual events.
  • If the few-percent density accuracy persists at higher jet transverse momenta or in rarer kinematic corners, fast inference could be combined with traditional generators in a hybrid scheme, using the network for bulk phase space and the generator for tails.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces two neural-network-based tools for Lund jet images. The first, gLund, is an LSGAN trained on 500k Pythia+Delphes QCD jets. To handle sparsity, training images are produced by averaging navg=32 individual Lund images, and generated images are converted back to discrete images by independent per-pixel Bernoulli sampling. The paper reports agreement with the reference sample within a few percent in the bulk of the Lund plane, and compares gLund with a VAE and a WGAN-GP on density slices and derived observables. The second tool, CycleJet, is a CycleGAN trained to map between parton-level and detector-level images and between QCD and boosted-W jets, with validation based on comparisons of average images. Code and data are released openly.

Significance. If the reported accuracy extends to event-level jet substructure, the paper would provide a fast, storage-efficient alternative to full Monte Carlo simulation and a new way to reinterpret event samples. The strengths of the work are its open-source implementation, reproducible data release, systematic hyperparameter optimization, and a useful benchmark comparison among LSGAN, WGAN-GP, and VAE. However, as presented, the evidence supports reproduction of the one-point Lund-plane density, not the joint distribution of activated pixels that defines individual jet radiation patterns. The authors explicitly concede in Section 4 that their preprocessing loses wide-separation correlations and that the formal logarithmic accuracy of generated samples is hard to evaluate. The paper is therefore a useful contribution to fast approximate simulation of averaged jet densities, but its central event-level claims need either additional validation or a careful reframing.

major comments (4)
  1. [Section 2.3 and Section 4] The stochastic generation step treats each pixel as an independent Bernoulli random variable with probability equal to the averaged pixel value. This preserves per-pixel marginal rates but cannot reproduce correlations between emissions that are well separated in the Lund plane; real Lund images have structured activation patterns inherited from the clustering history. The few-percent agreement shown in Figures 7 through 10 therefore constrains only the average density, not the joint distribution over individual jet images. The paper's own Section 4 concession, that the preprocessing 'looses information on correlations between emissions at wide angular and transverse momentum separation' and that it is 'difficult to evaluate or improve the formal logarithmic accuracy of the generated samples,' confirms this limitation. The authors should either validate event-level observables sensitive to inter-pixel correlations or explicitly restrict the central claim to reproduction of the averaged Lund-plane density.
  2. [Sections 2.4 and 2.5] No train/test split is described anywhere in the experimental setup. Hyperparameter selection is performed using the loss Lh in Eq. (3), which compares generated images with the reference preprocessed images, and the final validation in Figures 7-10 again compares generated samples against the same reference sample used for training. Without an independent test set and without statistical uncertainties on the ratio plots, the reported 'few percent' accuracy may reflect memorization or overfitting rather than generalization. The authors should specify the split between training, validation, and test samples, and report quantitative errors with uncertainties for Figures 7-10.
  3. [Section 3.2, Eq. (11)] The CycleJet validation is based on the loss Lh = ||RA - P_{B->A}|| + ||RB - P_{A->B}||, which compares only the average reference images with the average transformed images. Consequently, the claim that one can 'retroactively change simulation settings or the underlying process on an existing sample' is demonstrated only at the level of mean densities. Individual event mappings may not preserve the correlations required for physics applications. The authors should add per-event validation metrics, such as distributions of pixel counts, two-point correlations, or downstream substructure observables evaluated on individual transformed events, or substantially weaken the corresponding claims in the abstract and Section 3.
  4. [Figure 7(c) and abstract] The abstract states that the model 'retrieves the underlying two-dimensional distribution to within a few percent,' but Figure 7(c) shows deviations well beyond a few percent near the boundaries of the Lund plane, and the 'bulk' region is never quantitatively defined. The authors should define the region over which the few-percent claim holds and report a concrete aggregate metric, e.g., a maximum or average relative deviation over that region, rather than relying on visual inspection of ratio plots.
minor comments (5)
  1. [Section 4] The sentence 'loosing information on correlations' should read 'losing information on correlations.'
  2. [Section 3.2] The caption of Figure 12 refers to 'delphes-level sample'; the capitalization should be 'Delphes-level' for consistency with the rest of the text.
  3. [Equation (6)] The quantities z(i) and Delta(i) in Eq. (6) are not defined in the text; please define them explicitly or refer to the original mMDT definition.
  4. [Appendix A] The hyperparameter tables would be more reproducible if they included the random seed and the number of evaluation samples used for each reported metric.
  5. [References] Reference [40] is incomplete: it lists only '(2019), 1909.04451' without authors or title. Please complete the bibliographic entry.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: the generative models are trained and evaluated on the same Pythia/Delphes reference, so the few-percent agreement is a consistency/fit-quality statement rather than a disguised prediction; the only self-citation is the externally published Lund-plane encoding.

full rationale

The paper makes no claim to derive a physical constant or first-principles result. The gLund and CycleJet networks are trained on Lund images from Pythia v8.223 + Delphes and then compared with averages from the same simulation; agreement to a few percent is therefore a measure of how well the trained generative model represents its training distribution, not an external prediction. This is a consistency check, and the paper does not relabel fitted hyperparameters or the training objective as an independent result: the hyperopt procedure in Sec. 2.4 explicitly minimizes Lh = I + 5S, an average-image discrepancy, and the CycleJet scan in Sec. 3.2 minimizes ||RA - PB->A|| + ||RB - PA->B||, so the subsequent agreement figures are evaluations on the same metrics used for model selection. That is model-selection and validation practice rather than circular derivation. The one self-citation, the Lund-plane representation [22], is an externally published coordinate construction used as input encoding; it is not invoked as a uniqueness theorem or as evidence for the generative claims. Section 4's explicit concession that the averaging preprocessing loses correlations between emissions at wide angular and transverse-momentum separation is a real limitation on event-level fidelity, but it is a scope limitation, not circular reasoning. No load-bearing circular step is present; the modest score reflects only the presence of a minor, non-load-bearing self-citation to the Lund-plane encoding.

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

The central experiments rest on the Lund-plane encoding, the navg-averaging and pixel-wise resampling, the use of Pythia 8 plus Delphes as the reference distribution, and the CycleGAN cycle-consistency assumption. None of these are proven by the paper, and the first two are explicitly acknowledged as losing correlation information. No new physical entities are introduced; gLund and CycleJet are trained neural networks.

free parameters (6)
  • navg (gLund averaging window) = 32
    Number of input images averaged to form each training image; a model choice with a trade-off: too low gives sparse inputs, too high increases variance in generated samples (Section 2.3).
  • navg (CycleJet averaging window) = 20
    Averaging parameter for the CycleJet domain mapping, listed in Table 2.
  • Latent dimension (gLund) = 500
    Dimension of the generator input noise, chosen by hyperparameter optimization (Table 1).
  • ZCA whitening = True
    Preprocessing transform applied before training, selected via hyperopt (Table 1, Section 2.3).
  • Cycle consistency weight lambda = 10
    Weight of the cycle consistency loss in the CycleJet objective, Eq. (10) and Table 2.
  • Lh loss weights = I + 5*S
    Hyperparameter optimization objective combines an average-image norm and a structural similarity difference; the factor 5 is chosen by hand in Section 2.4.
assumptions (4)
  • domain assumption The primary Lund plane, encoded as a 24x24 binary image, is an adequate representation of jet radiation patterns for generative modeling.
    Used throughout Section 2; the authors footnote that binary on/off pixels are equivalent to hit-counting only at large image resolution, and correlations may be lost.
  • ad hoc to paper Averaging navg images and treating pixel values as independent activation probabilities produces samples that are physically faithful.
    Core preprocessing in Section 2.3; Section 4 concedes this loses information on correlations between emissions at wide angular and transverse momentum separation.
  • domain assumption Pythia 8 v8.223 plus Delphes 3 v3.4.1 CMS particle-flow is a valid reference for detector-level jet substructure.
    All training and validation data are simulated with this setup (Section 2.2), so the underlying distribution claimed is the simulation's distribution, not an independent truth.
  • domain assumption Cycle-consistency loss guarantees the learned domain mapping is physically meaningful rather than a pixel-space artifact.
    CycleJet relies on Eq. (7) and Eq. (10); the paper validates by comparing average images, but there is no independent check that the mapped image corresponds to the target process event-by-event.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Lund jet images from generative and cycle-consistent adversarial networks." pith.science (2026). https://pith.science/paper/CZNFJD7K

@misc{pith2026190901359,
  author       = {Pith},
  title        = {Pith review of: Lund jet images from generative and cycle-consistent adversarial networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CZNFJD7K}},
  note         = {Machine review of arXiv:1909.01359}
}
read the original abstract

We introduce a generative model to simulate radiation patterns within a jet using the Lund jet plane. We show that using an appropriate neural network architecture with a stochastic generation of images, it is possible to construct a generative model which retrieves the underlying two-dimensional distribution to within a few percent. We compare our model with several alternative state-of-the-art generative techniques. Finally, we show how a mapping can be created between different categories of jets, and use this method to retroactively change simulation settings or the underlying process on an existing sample. These results provide a framework for significantly reducing simulation times through fast inference of the neural network as well as for data augmentation of physical measurements.

Figures

Figures reproduced from arXiv: 1909.01359 by the authors.

Figure 1
Figure 1. Average Lund jet plane density for QCD jets simulated with Pythia v8.223 and Delphes v3.4.1. to significantly reduce the computing resources required to simulate realistic LHC data [31–40]. In this paper, we in￾troduce a generative model to create new samples of the substructure of a jet from existing data. We use the Lund jet plane [22], shown in figure 1, as a visual representa￾arXiv:1909.01359v2 [hep-ph] 29 Nov 2… view at source ↗
Figure 2
Figure 2. Sample input images after averaging with navg =1, 5, 10 and 20. where kt = pt,b∆ab is the transverse momentum of particle b relative to a. 2 2.2 Input data The data sample used in this article consists of 500k jets, generated using the dijet process in Pythia v8.223. Jets are clustered using the anti-kt algorithm [5, 52] with radius R = 1.0, and are required to pass a selection cut, with transverse momentum pt > 500… view at source ↗
Figure 3
Figure 3. Hyperparameter scan results obtained with the hyperopt library. The first row shows the scan over image and optimiser related parameters while the second row plots correspond to the final architecture scan. Parameters Value Architecture LSGAN D units 384 G units 512 αD 0.129 αG 0.477 Aux ratio 12 Kernel dimension 1 Number of kernels 2 Minibatch discriminator Yes Epochs 5000 Batch size 32 Latent dimension 500 ZCA Yes… view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Loss of the LSGAN discriminator and generator throughout the training stage. During the training of the VAE, we use KL cost an￾nealing [62] to avoid a collapse of the VAE output to the prior distribution. This is a problem caused by the large value of the KL divergence…
Figure 5
Figure 5. Figure 5: Left two figures: Sample input images before and after preprocessing. Right two: sample generated by the LSGAN and the corresponding Lund image. Input samples Generated samples [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: A random selection of preprocessed input images (left), and of images generated with the LSGAN model (right). Axes and colour schemes are identical to figure 5. We determine the best hyperparameters for both of these models through a hyperopt parameter sweep, which is …
Figure 7
Figure 7. Figure 7: Average Lund jet plane density for (a) the reference sample and (b) a data set generated by the gLund model. (c) shows the ratio between these two densities. 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 ( ab, fixed kt) 7.4 < kt [GeV] < 25.8 LSGAN WGAN-GP VAE Pythia 8 1 0.5 …
Figure 8
Figure 8. Figure 8: Slice of the Lund plane along ∆ab with 7.4 GeV < kt < 25.8 GeV. 3 Reinterpreting events using domain mappings In this section, we will introduce a novel application of domain mappings to reinterpret existing event samples. To this end, we implement a cycle-consistent a…
Figure 10
Figure 10. Figure 10: Distribution of (a) the number of activated pixels per image, (b) the reconstructed soft-drop multiplicity for zcut = 0.007, β = −1 and θcut = 0, and (c) the jet mass after applying the modified Mass Drop Tagger with zcut = 0.1. parton level detector level parton leve…
Figure 11
Figure 11. Figure 11: Top: transition from parton-level to delphes-level and back using CycleJet. Bottom: corresponding sampled event. 3.1 CycleGANs and domain mappings A CycleGAN learns mapping functions between two do￾mains X and Y , using as input training samples from both domains. It …
Figure 12
Figure 12. Figure 12: Top: average of the parton-level sample before (left) and after (right) applying the parton-to-detector mapping. Bottom: average of the delphes-level sample before (left) and after (right) applying the detector-to-parton mapping. For both examples we observe a good le…
Figure 13
Figure 13. Figure 13: Top: average of the QCD sample before (left) and af￾ter (right) applying the QCD-to-W mapping. Bottom: average of the W sample before (left) and after (right) applying the W-to-QCD mapping. Secondly, a CycleGAN model was constructed to map different jet configurations…
Figure 15
Figure 15. Figure 15: A random selection of preprocessed input images (left), and of images generated with the WGAN-GP model (right). Axis and colour schemes are the same of figure 5. The VAE encoder consists of a dense layer with 384 units with ReLU activation function connected to a la￾t…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 16 canonical work pages

  1. [1]

    Sterman, S

    G.F. Sterman, S. Weinberg, Phys. Rev. Lett. 39, 1436 (1977)

  2. [2]

    Salam, Eur

    G.P. Salam, Eur. Phys. J. C67, 637 (2010), 0906.1833

  3. [3]

    Ellis, D.E

    S.D. Ellis, D.E. Soper, Phys. Rev. D48, 3160 (1993), hep-ph/9305266

  4. [4]

    Dokshitzer, G.D

    Y.L. Dokshitzer, G.D. Leder, S. Moretti, B.R. Webber, JHEP 08, 001 (1997), hep-ph/9707323

  5. [5]

    Cacciari, G.P

    M. Cacciari, G.P. Salam, G. Soyez, JHEP 04, 063 (2008), 0802.1189

  6. [6]

    Thaler, L.T

    J. Thaler, L.T. Wang, JHEP 07, 092 (2008), 0806.0023

  7. [7]

    Kaplan, K

    D.E. Kaplan, K. Rehermann, M.D. Schwartz, B. Tweedie, Phys. Rev. Lett. 101, 142001 (2008), 0806.0848

  8. [8]

    Ellis, C.K

    S.D. Ellis, C.K. Vermilion, J.R. Walsh, Phys. Rev. D80, 051501 (2009), 0903.5081

Show all 64 references
  1. [9]

    Ellis, C.K

    S.D. Ellis, C.K. Vermilion, J.R. Walsh, Phys. Rev. D81, 094023 (2010), 0912.0033

  2. [10]

    Plehn, G.P

    T. Plehn, G.P. Salam, M. Spannowsky, Phys. Rev. Lett. 104, 111801 (2010), 0910.5472

  3. [11]

    Thaler, K

    J. Thaler, K. Van Tilburg, JHEP 03, 015 (2011), 1011.2268

  4. [12]

    Larkoski, G.P

    A.J. Larkoski, G.P. Salam, J. Thaler, JHEP 06, 108 (2013), 1305.0007

  5. [13]

    Chien, Phys

    Y.T. Chien, Phys. Rev. D90, 054008 (2014), 1304.5240

  6. [14]

    Cacciari, G.P

    M. Cacciari, G.P. Salam, G. Soyez, Eur. Phys. J. C75, 59 (2015), 1407.0408

  7. [15]

    Larkoski, I

    A.J. Larkoski, I. Moult, D. Neill, JHEP 12, 009 (2014), 1409.6298

  8. [16]

    Moult, L

    I. Moult, L. Necib, J. Thaler, JHEP 12, 153 (2016), 1609.07483

  9. [17]

    Dasgupta, A

    M. Dasgupta, A. Fregoso, S. Marzani, G.P. Salam, JHEP 09, 029 (2013), 1307.0007

  10. [18]

    Larkoski, S

    A.J. Larkoski, S. Marzani, G. Soyez, J. Thaler, JHEP 05, 146 (2014), 1402.2657

  11. [19]

    Komiske, E.M

    P.T. Komiske, E.M. Metodiev, B. Nachman, M.D. Schwartz, JHEP 12, 051 (2017), 1707.08600

  12. [20]

    Komiske, E.M

    P.T. Komiske, E.M. Metodiev, J. Thaler, JHEP 04, 013 (2018), 1712.07124 Stefano Carrazza, Fr´ ed´ eric A. Dreyer: Lund jet images with GANs 11

  13. [21]

    Dreyer, L

    F.A. Dreyer, L. Necib, G. Soyez, J. Thaler, JHEP 06, 093 (2018), 1804.03657

  14. [22]

    Dreyer, G.P

    F.A. Dreyer, G.P. Salam, G. Soyez, JHEP 12, 064 (2018), 1807.04758

  15. [23]

    Butter et al., SciPost Phys

    A. Butter et al., SciPost Phys. 7, 014 (2019), 1902.09914

  16. [24]

    Carrazza, F.A

    S. Carrazza, F.A. Dreyer (2019), 1903.09644

  17. [25]

    Berta, L

    P. Berta, L. Masetti, D.W. Miller, M. Spousta (2019), 1905.03470

  18. [26]

    Moreno, O

    E.A. Moreno, O. Cerri, J.M. Duarte, H.B. Newman, T.Q. Nguyen, A. Periwal, M. Pierini, A. Serikova, M. Spiropulu, J.R. Vlimant (2019), 1908.05318

  19. [27]

    Fischer, S

    N. Fischer, S. Gieseke, S. Pltzer, P. Skands, Eur. Phys. J. C74, 2831 (2014), 1402.3186

  20. [28]

    Les Houches 2017: Physics at TeV Colliders Standard Model Working Group Report (2018), 1803.07977

  21. [29]

    Goodfellow, J

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, Gen- erative adversarial nets, in Advances in neural information processing systems (2014), pp. 2672–2680

  22. [30]

    Kingma, M

    D.P. Kingma, M. Welling, Auto-Encoding Variational Bayes, in 2nd International Conference on Learn- ing Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings (2014), http://arxiv.org/abs/1312.6114

  23. [31]

    de Oliveira, M

    L. de Oliveira, M. Paganini, B. Nachman, Comput. Softw. Big Sci. 1, 4 (2017), 1701.05927

  24. [32]

    Paganini, L

    M. Paganini, L. de Oliveira, B. Nachman, Phys. Rev. Lett. 120, 042003 (2018), 1705.02355

  25. [33]

    Paganini, L

    M. Paganini, L. de Oliveira, B. Nachman, Phys. Rev. D97, 014021 (2018), 1712.10321

  26. [34]

    Otten, S

    S. Otten, S. Caron, W. de Swart, M. van Beekveld, L. Hen- driks, C. van Leeuwen, D. Podareanu, R. Ruiz de Austri, R. Verheyen (2019), 1901.00875

  27. [35]

    Musella, F

    P. Musella, F. Pandolfi, Comput. Softw. Big Sci. 2, 8 (2018), 1805.00850

  28. [36]

    Datta, D

    K. Datta, D. Kar, D. Roy (2018), 1806.00433

  29. [37]

    Cerri, T.Q

    O. Cerri, T.Q. Nguyen, M. Pierini, M. Spiropulu, J.R. Vli- mant, JHEP 05, 036 (2019), 1811.10276

  30. [38]

    Di Sipio, M

    R. Di Sipio, M. Faucci Giannelli, S. Ketabchi Haghighat, S. Palazzo (2019), 1903.02433

  31. [39]

    Butter, T

    A. Butter, T. Plehn, R. Winterhalder (2019), 1907.03764

  32. [41]

    collaboration (ATLAS) (2019)

    T.A. collaboration (ATLAS) (2019)

  33. [42]

    X. Mao, Q. Li, H. Xie, R.Y.K. Lau, Z. Wang, CoRR abs/1611.04076 (2016), 1611.04076

  34. [43]

    Sjstrand, S

    T. Sjstrand, S. Ask, J.R. Christiansen, R. Corke, N. De- sai, P. Ilten, S. Mrenna, S. Prestel, C.O. Rasmussen, P.Z. Skands, Comput. Phys. Commun. 191, 159 (2015), 1410.3012

  35. [44]

    de Favereau, C

    J. de Favereau, C. Delaere, P. Demin, A. Giammanco, V. Lematre, A. Mertens, M. Selvaggi (DELPHES 3), JHEP 02, 057 (2014), 1307.6346

  36. [45]

    Arjovsky, S

    M. Arjovsky, S. Chintala, L. Bottou, Wasser- stein Generative Adversarial Networks , in Pro- ceedings of the 34th International Conference on Machine Learning, ICML 2017, Sydney, NSW, Australia, 6-11 August 2017 (2017), pp. 214–223, http://proceedings.mlr.press/v70/arjovsky17a.html

  37. [46]

    J.Y. Zhu, T. Park, P. Isola, A.A. Efros, Unpaired Image- to-Image Translation using Cycle-Consistent Adversarial Networkss, in Computer Vision (ICCV), 2017 IEEE In- ternational Conference on (2017)

  38. [47]

    Duarte et al., JINST 13, P07027 (2018), 1804.06913

    J. Duarte et al., JINST 13, P07027 (2018), 1804.06913

  39. [48]

    Dreyer, S

    F. Dreyer, S. Carrazza, Jetsgame/glund v1.0.0 (2019), https://doi.org/10.5281/zenodo.3384920

  40. [49]

    Dreyer, S

    F. Dreyer, S. Carrazza, Jetsgame/cyclejet v1.0.0 (2019), https://doi.org/10.5281/zenodo.3384918

  41. [50]

    Carrazza, F.A

    S. Carrazza, F.A. Dreyer, JetsGame/data v1.0.0 (2019), this repository is git-lfs., https://doi.org/10.5281/zenodo.2602514

  42. [51]

    Wobisch, T

    M. Wobisch, T. Wengler, Hadronization corrections to jet cross-sections in deep inelastic scattering , in Monte Carlo generators for HERA physics. Proceedings, Work- shop, Hamburg, Germany, 1998-1999 (1998), pp. 270–279, hep-ph/9907280

  43. [52]

    Cacciari, G.P

    M. Cacciari, G.P. Salam, G. Soyez, Eur. Phys. J. C72, 1896 (2012), 1111.6097

  44. [53]

    Goodfellow, J

    I.J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, inPro- ceedings of the 27th International Conference on Neural Information Processing Systems - Volume 2 (MIT Press, Cambridge, MA, USA, 2014), NIPS’14, pp. 2672–2680, http:...

  45. [54]

    Salimans, I.J

    T. Salimans, I.J. Goodfellow, W. Zaremba, V. Cheung, A. Radford, X. Chen, CoRR abs/1606.03498 (2016), 1606.03498

  46. [55]

    Bell, T.J

    A.J. Bell, T.J. Sejnowski, Vision Research 37, 3327 (1997)

  47. [56]

    Bergstra, D

    J. Bergstra, D. Yamins, D.D. Cox, Making a Science of Model Search: Hyperparameter Optimization in Hundreds of Dimensions for Vision Architectures , in Proceedings of the 30th International Conference on International Con- ference on Machine Learning - Volume 28 (JMLR.org, 201...

  48. [57]

    Bovik, H.R

    Zhou Wang, A.C. Bovik, H.R. Sheikh, E.P. Simoncelli, IEEE Transactions on Image Processing 13, 600 (2004)

  49. [58]

    Maas, A.Y

    A.L. Maas, A.Y. Hannun, A.Y. Ng, Rectifier nonlineari- ties improve neural network acoustic models , in in ICML Workshop on Deep Learning for Audio, Speech and Lan- guage Processing (2013)

  50. [59]

    Higgins, L

    I. Higgins, L. Matthey, X. Glorot, A. Pal, B. Uria, C. Blun- dell, S. Mohamed, A. Lerchner, CoRR abs/1606.05579 (2016), 1606.05579

  51. [60]

    Burgess, I

    C.P. Burgess, I. Higgins, A. Pal, L. Matthey, N. Wat- ters, G. Desjardins, A. Lerchner, CoRR abs/1804.03599 (2018), 1804.03599

  52. [61]

    Gulrajani, F

    I. Gulrajani, F. Ahmed, M. Arjovsky, V. Dumoulin, A.C. Courville, CoRR abs/1704.00028 (2017), 1704.00028

  53. [62]

    Bowman, L

    S.R. Bowman, L. Vilnis, O. Vinyals, A.M. Dai, R. J´ ozefowicz, S. Bengio, CoRRabs/1511.06349 (2015), 1511.06349

  54. [63]

    Frye, A.J

    C. Frye, A.J. Larkoski, J. Thaler, K. Zhou, JHEP 09, 083 (2017), 1704.06266

  55. [64]

    Andreassen, B

    A. Andreassen, B. Nachman (2019), 1907.08209

  56. [65]

    Ulyanov, A

    D. Ulyanov, A. Vedaldi, V.S. Lempitsky, CoRR abs/1607.08022 (2016), 1607.08022

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.