Pith. sign in

REVIEW 4 major objections 3 minor 49 references

Neural Posterior Unfolding claims that a normalizing flow trained on simulated truth-versus-measured histogram pairs yields a full, calibrated posterior over unfolded cross sections, with accurate maximum-likelihood readout and correct unce

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-04 23:42 UTC pith:6S6YOHC4

load-bearing objection NPU is a credible proof-of-concept for flow-based binned unfolding, but the training prior is unspecified and the validation is mostly same-prior, so the UQ claims outrun the evidence. the 4 major comments →

arxiv 2509.06370 v1 pith:6S6YOHC4 submitted 2025-09-08 hep-ph hep-ex

Neural Posterior Unfolding

Fernando Torales Acosta , Jay Chan , Krish Desai , Vinicius Mikuni , Benjamin Nachman , Jingjing Pan , Francesco Rubbo This is my paper
classification hep-ph hep-ex
keywords unfoldingdeconvolutionnormalizing flowsBayesian inferencedifferential cross sectionsuncertainty quantificationjet substructureLHC
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Neural Posterior Unfolding (NPU) is proposed as a way to correct measured histograms for detector distortions while keeping the uncertainty faithful. The paper's central claim is that a normalizing flow trained on simulated pairs of particle-level and detector-level histograms can learn the conditional density of the truth given the measured histogram, and that reading out this learned density with maximum likelihood recovers the true unfolded cross section. This matters because standard iterative Bayesian unfolding returns only a point estimate, and in degenerate bins—where two different truth spectra would produce identical detector data—it can report zero uncertainty when the true ambiguity is maximal. The paper shows, on a two-bin toy, an eight-bin Gaussian problem, and LHC-like jet substructure simulations, that NPU returns accurate central values and a full posterior, and that amortized training makes later uses much faster than MCMC-based Fully Bayesian Unfolding.

Core claim

The paper's central claim is that binned unfolding can be cast as an amortized neural posterior estimation problem. A normalizing flow is trained on pairs of particle-level and detector-level histograms to approximate p(t|m), the density of the true histogram t given the measured histogram m; after training, unfolding a new measurement is a maximum-likelihood optimization over t under the learned density. The claim is that this gives a full, properly calibrated posterior with correlations, that it reproduces the true spectrum in Gaussian and LHC-like tests, and that it handles degenerate response regions correctly by returning a broad posterior where iterative Bayesian unfolding would report

What carries the argument

The central object is a conditional normalizing flow—an invertible neural network that maps a simple base distribution to a flexible probability density—trained with a negative log-likelihood loss on (truth, measured) histogram pairs to approximate p(t|m). The measured histogram serves as the conditional input, and the output is a density over truth histograms. After training, a maximum-likelihood estimation step optimizes the learned log-density with respect to the truth vector to extract the unfolded spectrum from new data. This flow does the work that MCMC does in Fully Bayesian Unfolding, but amortized: the expensive density learning happens once, and each new dataset only needs the fast

Load-bearing premise

The load-bearing premise is that the simulated (truth, detector-level) pairs used to train the flow describe the same relationship as the real data, and that the distribution of truth histograms in training is close enough to the real spectrum that the likelihood dominates; otherwise the learned p(t|m) can be wrong and the unfolded result silently biased.

What would settle it

Take a trained NPU model and unfold pseudo-experiments whose true spectrum is deliberately far from the training prior, e.g. a Gaussian with a different mean and width under the same detector response; if the maximum-likelihood readout is visibly pulled toward the training distribution, or the posterior's credible intervals do not contain the true spectrum at the claimed rate, the accuracy claim fails outside the training regime.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Every unfolded cross section comes with a full posterior over bins, including correlations, rather than a single point estimate, so downstream physics statements can propagate the complete statistical uncertainty.
  • In detector-degenerate regions, NPU reports wide credible intervals reflecting the true ambiguity, correcting the zero-uncertainty failure of iterative Bayesian unfolding.
  • The amortized training makes re-unfolding with bootstrapped or newly arriving data cheap after one training pass: the paper reports about five minutes of training versus roughly 67 minutes for 100 pseudo-experiments with MCMC-based unfolding.
  • NPU and FBU both pass pull-based calibration tests with means near zero and widths near one across tested smearing parameters, indicating the posterior is statistically well calibrated in the tested cases.
  • The method reproduces truth for LHC-like jet substructure observables in simulation, making it a plausible candidate for analysis-ready unfolding at the LHC and beyond.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Inference: The two-bin degenerate argument implies that any regularized matrix-inversion method whose regularization acts on the truth vector—not just IBU—will understate uncertainty where response rows nearly coincide; NPU-style posterior estimation is a generic fix for that failure mode.
  • Inference: Since the paper does not specify the training prior over truth histograms in detail, the method's reliability on data whose true spectrum is far outside that prior is untested; an obvious stress test is to train on one prior and unfold pseudo-data drawn from a different one.
  • Inference: The flow's conditional density could be extended to include detector-systematic parameters such as smear widths and efficiencies by conditioning on them, turning the method into a joint statistical-plus-systematic unfolding.
  • Inference: Because the flow provides a density over histograms, the same machinery is naturally suited to unbinned unfolding if the conditional density is defined over individual event-level observables instead of bin counts; the paper names this as a future direction.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 3 minor

Summary. The paper introduces Neural Posterior Unfolding (NPU), a normalizing-flow-based method that learns the posterior p(t|m) over binned particle-level histograms t conditional on measured detector-level histograms m, using simulated (truth, measured) pairs. The unfolded result is obtained by maximizing the learned posterior over t. The paper also provides a modern Python implementation of Fully Bayesian Unfolding (FBU). The methods are first tested on a 2-bin degenerate-response example, where NPU returns a broad posterior while IBU incorrectly reports zero uncertainty on the bin ratio; then on an 8-bin Gaussian example with pull-based calibration checks; and finally on LHC-like jet substructure observables, using Herwig as pseudo-data and Pythia to build the response. The central claims are that NPU provides accurate unfolded central values, full posterior uncertainties that remain well-calibrated in weakly constrained regions, and fast amortized inference.

Significance. If the central claims hold, NPU would be a useful contribution to the unfolding literature: it offers amortized inference, direct posterior access, and a principled treatment of degenerate response regions where classical methods such as IBU are known to underestimate uncertainty. The paper also provides a modern, maintained FBU implementation and releases code, which are practical assets for the community. The conceptual demonstration on the 2-bin degenerate example is clear and valuable. However, the paper's load-bearing claims about posterior fidelity and calibration are not yet fully supported: the training prior for NPU is never specified, the Gaussian validation is same-prior, the acknowledged undercoverage at low smearing is not reconciled with the calibration conclusion, and the LHC example lacks quantitative agreement tests. These gaps are fixable, but they are central to the advertised advantages of the method.

major comments (4)
  1. [Sec. III and Sec. IV] The training prior for NPU is never specified. The method learns p(t|m) from (truth, measured) pairs, but the text never states how the truth histograms t are sampled in training. The only statement, in Sec. IV, that 'A uniform prior is also used for NPU (chosen based on the distribution of parameter points in the training sample)' is ambiguous and does not define a sampling distribution or range. Since the posterior in weakly constrained regions is dominated by the prior, an unspecified prior makes the uncertainty quantification claims unreproducible. Please specify the training pair generation procedure, including the prior over t, explicitly.
  2. [Sec. V.B, Table II] The Gaussian validation tests NPU only under the training prior. The simulation and natural datasets both have mu=0, sigma=1, so the test truth is drawn from the same distribution as the training samples. The pull tests in Fig. 4 therefore establish calibration under the training prior, not the ability to unfold a genuinely different true distribution. Since real data are not drawn from the simulation prior, this is a load-bearing gap. Please add a validation where the test truth is drawn from a different distribution (e.g., shifted mu or sigma) and report pulls/coverage, or otherwise demonstrate that the likelihood dominates the prior.
  3. [Sec. V.B, Fig. 4b and Sec. VI] The paper acknowledges that 'NPU appears to underestimate the variance at low smearing' (Fig. 4b), yet the conclusions state that both methods 'consistently produced pulls with mu=0 and sigma=1 across the tested smearing parameters.' These statements are contradictory. Undercoverage is a calibration failure that directly affects the central claim of reliable uncertainty quantification. Please quantify the pull standard deviations for each smearing value, report their statistical uncertainties, and either correct the conclusions or explain why this behavior is acceptable.
  4. [Sec. V.C, Fig. 5] The LHC example provides only a qualitative comparison between the unfolded result and the truth. The ratio-to-truth panels show agreement within bands, but no pulls, chi-square statistics, or coverage tests are reported for this mismatched-prior scenario. Given that this is the only external grounding of the method on a data distribution different from the training prior, a quantitative agreement test is necessary to support the claim that NPU 'accurately recovered their central values' and to validate the reported posterior widths.
minor comments (3)
  1. [Sec. II, Eq. (1)] The IBU update is written as t_j^{(n)} = sum_i Pr^{n-1}(truth_j|measure_i) Pr(measure_i). As written, this appears to be missing the normalization by total counts; the subsequent Eq. (2) is clearer. Please correct or clarify the notation.
  2. [Fig. 2(b) caption] The response matrix R is defined as R_ij = Pr(measure_i | truth_j), but the displayed matrix is labeled Pr(Truth | Reco). Please clarify whether the plotted matrix is the transpose of R or an element-wise conditional probability, and ensure the axes and colorbar labels are consistent.
  3. [Sec. VI] Typo: 'including including jet width' appears in the conclusions. Also, the references to 'T able' and 'Fig. 4' formatting should be checked for journal style.

Circularity Check

0 steps flagged

No significant circularity: NPU's learned posterior is trained on simulation pairs and applied to held-out/out-of-prior data; the main caveats are prior specification and validation scope, not circular reasoning.

full rationale

NPU trains a conditional normalizing flow on (truth, measured) simulation pairs to learn p(t|m), then obtains point estimates by maximizing this learned posterior over t (Algorithm 1). This is amortized neural posterior estimation; the training prior is an input, not a hidden re-use of the target. No equation in the paper equates a 'prediction' to a fitted parameter by construction. The Gaussian pull tests (Sec. V.B) use pseudo-experiments whose truth is drawn from the same Gaussian as the training sample; this checks calibration of the flow under the training prior, which is a legitimate self-consistency test, not a forced result—the flow could still be miscalibrated and the pulls could deviate from N(0,1). The physics example (Sec. V.C) provides an out-of-prior check by unfolding Herwig 'data' with a Pythia-based response, with agreement shown in Fig. 5. The paper's self-citations ([36,37] for datasets; [50] for an overview) are not load-bearing arguments. Sec. VI explicitly defers 'sensitivity to prior choices' to future work; that is a limitation of scope, but it does not make the derivation circular. Therefore no circular step can be exhibited with the specificity required.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The method's posterior is defined by the training distribution; the paper does not specify this distribution, making it the central unaccounted input. The other listed assumptions are standard for unfolding.

free parameters (3)
  • NPU training prior over truth histograms = not specified in paper
    The posterior learned by the flow is p(t|m) ∝ p_train(t) R(m|t); the training distribution p_train(t) is never defined in Sec III, yet it shapes the unfolded posterior. The claim of a 'uniform prior' in Sec IV is not connected to the training procedure.
  • Flow network hyperparameters = 3 layers + 1 conditional, 50-100 nodes, Swish, epochs 1000-1500, batch size 10^4, lr 10^-4
    Chosen by hand (Table I); authors state results are largely insensitive, but no ablation is shown. These affect the approximation quality and implicit regularization.
  • FBU uniform prior width = ±5√t around t
    The classical FBU prior range [t-5√t, t+5√t] is a subjective choice; it is not derived.
axioms (4)
  • domain assumption The response matrix R accurately models the detector response.
    Stated explicitly in Sec II: 'When the response matrix is correctly modeled (a necessary assumption)'.
  • standard math The observed counts follow a Poisson distribution given the true spectrum.
    Used in FBU likelihood in Sec IV, Eq. 6. NPU training also relies on simulated pairs which imply the same statistical model.
  • domain assumption The normalizing flow converges to a faithful approximation of the posterior p(t|m) on the training distribution.
    The method assumes the flow has enough capacity and is trained well; no convergence theorem is given (Sec III). Calibration is tested only empirically.
  • domain assumption The training pairs (truth, measured histograms) are sampled from the same prior and response model that govern the data being unfolded.
    NPU learns p(t|m) from simulation pairs; if the simulation truth prior differs from the data, the unfolded posterior is biased. The Gaussian validation in Sec V.B trains and tests on the same distribution.

pith-pipeline@v1.3.0-alltime-deepseek · 11357 in / 14364 out tokens · 135083 ms · 2026-08-04T23:42:27.275623+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Neural Posterior Unfolding." pith.science (2026). https://pith.science/paper/6S6YOHC4

@misc{pith2026250906370,
  author       = {Pith},
  title        = {Pith review of: Neural Posterior Unfolding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6S6YOHC4}},
  note         = {Machine review of arXiv:2509.06370}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Differential cross section measurements are the currency of scientific exchange in particle and nuclear physics. A key challenge for these analyses is the correction for detector distortions, known as deconvolution or unfolding. Binned unfolding of cross section measurements traditionally rely on the regularized inversion of the response matrix that represents the detector response, mapping pre-detector (`particle level') observables to post-detector (`detector level') observables. In this paper we introduce Neural Posterior Unfolding, a modern, Bayesian approach that leverages normalizing flows for unfolding. By using normalizing flows for neural posterior estimation, NPU offers several key advantages including implicit regularization through the neural network architecture, fast amortized inference that eliminates the need for repeated retraining, and direct access to the full uncertainty in the unfolded result. In addition to introducing NPU, we implement a classical Bayesian unfolding method called Fully Bayesian Unfolding (FBU) in modern Python so it can also be studied. These tools are validated on simple Gaussian examples and then tested on simulated jet substructure examples from the Large Hadron Collider (LHC). We find that the Bayesian methods are effective and worth additional development to be analysis ready for cross section measurements at the LHC and beyond.

Figures

Figures reproduced from arXiv: 2509.06370 by Benjamin Nachman, Fernando Torales Acosta, Francesco Rubbo, Jay Chan, Jingjing Pan, Krish Desai, Vinicius Mikuni.

Figure 1
Figure 1. Figure 1: FIG. 1. Comparison between NPU and IBU using a two–bin example. The MLE estimate of NPU is represented with an orange [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Gaussian example. (a) the initial setup. The goal of any unfolding method is recover the truth distribution, or the [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. (a) The unfolded Gaussian distribution obtained via NPU (MLE as orange down triangles) compared with the truth [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Pull distributions from 100 pseudo–experiments for NPU (orange) and FBU (green) in the Gaussian example with [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. The unfolding result for (a) jet width [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references · 21 canonical work pages · 9 internal anchors

  1. [1]

    Adye, inPHYSTAT 2011(CERN, Geneva, 2011) pp

    T. Adye, inPHYSTAT 2011(CERN, Geneva, 2011) pp. 313–318, arXiv:1105.1160 [physics.data-an]

    Pith/arXiv arXiv 2011

  2. [2]

    Arratiaet al., (2021), arXiv:2109.13243 [hep-ph]

    M. Arratiaet al., (2021), arXiv:2109.13243 [hep-ph]

    Pith/arXiv arXiv 2021

  3. [3]

    Huetschet al., (2024), arXiv:2404.18807 [hep-ph]

    N. Huetschet al., (2024), arXiv:2404.18807 [hep-ph]

    Pith/arXiv arXiv 2024

  4. [4]

    Cowan, Conf

    G. Cowan, Conf. Proc.C0203181, 248 (2002)

    work page 2002

  5. [5]

    Blobel, PHYSTAT2011 Proceedings , 240 (2011)

    V. Blobel, PHYSTAT2011 Proceedings , 240 (2011)

    work page 2011

  6. [6]

    Blobel, Data Analysis in High Energy Physics , 187 (2013)

    V. Blobel, Data Analysis in High Energy Physics , 187 (2013). 3 Unlike cases of computationally expensive forward models, the generative model here is a simple product of Poisson distributions. However, the computational differences may be non–negligible when many bootstraps are required for classical uncertainty quantification. 10

    work page 2013

  7. [7]

    Balasubramanian, L

    R. Balasubramanian, L. Brenner, C. Burgard, G. Cowan, V. Croft, W. Verkerke, and P. Verschuuren, (2019), arXiv:1910.14654 [physics.data-an]

    Pith/arXiv arXiv 2019

  8. [8]

    Fully bayesian unfolding,

    G. Choudalakis, “Fully bayesian unfolding,” (2012), arXiv:1201.4612 [physics.data-an]

    Pith/arXiv arXiv 2012

  9. [9]

    Measurement of the top quark pair production charge asymmetry in proton-proton collisions at 7 TeV using the ATLAS detector

    G. Aadet al.(ATLAS), JHEP02, 107 (2014), arXiv:1311.6724 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  10. [10]

    Measurement of the charge asymmetry in dileptonic decays of top quark pairs in $pp$ collisions at $\sqrt{s}=7$ TeV using the ATLAS detector

    G. Aadet al.(ATLAS), JHEP05, 061 (2015), arXiv:1501.07383 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  11. [11]

    Adrianiet al.(LHCf), Phys

    O. Adrianiet al.(LHCf), Phys. Rev. D94, 032007 (2016), arXiv:1507.08764 [hep-ex]

    Pith/arXiv arXiv 2016

  12. [12]

    Measurement of the charge asymmetry in top-quark pair production in the lepton-plus-jets final state in $pp$ collision data at $\sqrt{s}=8$ TeV with the ATLAS detector

    G. Aadet al.(ATLAS), Eur. Phys. J. C76, 87 (2016), [Erratum: Eur.Phys.J.C 77, 564 (2017)], arXiv:1509.02358 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  13. [13]

    Measurement of the charge asymmetry in highly boosted top-quark pair production in $\sqrt{s} =$ 8 TeV $pp$ collision data collected by the ATLAS experiment

    G. Aadet al.(ATLAS), Phys. Lett. B756, 52 (2016), arXiv:1512.06092 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  14. [14]

    Measurements of the charge asymmetry in top-quark pair production in the dilepton final state at $\sqrt{s}=8$ TeV with the ATLAS detector

    G. Aadet al.(ATLAS), Phys. Rev. D94, 032006 (2016), arXiv:1604.05538 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  15. [15]

    Aadet al.(ATLAS), JHEP08, 159 (2016), arXiv:1606.00689 [hep-ex]

    G. Aadet al.(ATLAS), JHEP08, 159 (2016), arXiv:1606.00689 [hep-ex]

    Pith/arXiv arXiv 2016

  16. [16]

    Measurements of top quark spin observables in $t\bar{t}$ events using dilepton final states in $\sqrt{s} = 8$ TeV $pp$ collisions with the ATLAS detector

    M. Aaboudet al.(ATLAS), JHEP03, 113 (2017), arXiv:1612.07004 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  17. [17]

    Aidalaet al.(PHENIX), Phys

    C. Aidalaet al.(PHENIX), Phys. Rev. D99, 092003 (2019), arXiv:1901.08405 [hep-ex]

    work page arXiv 2019

  18. [18]

    Measurement of the energy asymmetry in $t\bar{t}j$ production at 13 TeV with the ATLAS experiment and interpretation in the SMEFT framework

    G. Aadet al.(ATLAS), Eur. Phys. J. C82, 374 (2022), arXiv:2110.05453 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  19. [19]

    Measurement of cross-sections for production of a $Z$ boson in association with a flavor-inclusive or doubly $b$-tagged large-radius jet in proton-proton collisions at $\sqrt{s} = 13$ TeV with the ATLAS experiment

    G. Aadet al.(ATLAS), Phys. Rev. D108, 012022 (2023), arXiv:2204.12355 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  20. [20]

    Evidence for the charge asymmetry in $pp \rightarrow t\bar{t}$ production at $\sqrt{s}= 13$ TeV with the ATLAS detector

    G. Aadet al.(ATLAS), JHEP08, 077 (2023), arXiv:2208.12095 [hep-ex]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  21. [21]

    Theano: A python framework for fast computation of mathematical expressions,

    T. T. D. Team, R. Al-Rfou, G. Alain, A. Almahairi, C. Angermueller, D. Bahdanau, N. Ballas, F. Bastien, J. Bayer, A. Be- likov, A. Belopolsky, Y. Bengio, A. Bergeron, J. Bergstra, V. Bisson, J. B. Snyder, N. Bouchard, N. Boulanger-Lewandowski, X. Bouthillier, A. de Br´ ebisson, O. Breuleux, P.-L. Carrier, K. Cho, J. Chorowski, P. Christiano, T. Cooijmans,...

    Pith/arXiv arXiv 2016

  22. [22]

    Variational inference with normalizing flows,

    D. J. Rezende and S. Mohamed, “Variational inference with normalizing flows,” (2016), arXiv:1505.05770 [stat.ML]

    Pith/arXiv arXiv 2016

  23. [24]

    D’Agostini, Nucl

    G. D’Agostini, Nucl. Instrum. Meth.A362, 487 (1995)

    work page 1995

  24. [25]

    W. H. Richardson, J. Opt. Soc. Am.62, 55 (1972)

    1972

  25. [26]

    L. B. Lucy, Astronomical Journal79, 745 (1974)

    work page 1974

  26. [27]

    L. A. Shepp and Y. Vardi, IEEE transactions on medical imaging1, 113 (1982)

    work page 1982

  27. [28]

    Efron, The Annals of Statistics7, 1 (1979)

    B. Efron, The Annals of Statistics7, 1 (1979)

    1979

  28. [29]

    TensorFlow: Large- scale machine learning on heterogeneous systems,

    M. Abadi, A. Agarwal, P. Barham, E. Brevdo, Z. Chen, C. Citro, G. S. Corrado, A. Davis, J. Dean, M. Devin, S. Ghemawat, I. Goodfellow, A. Harp, G. Irving, M. Isard, Y. Jia, R. Jozefowicz, L. Kaiser, M. Kudlur, J. Levenberg, D. Man´ e, R. Monga, S. Moore, D. Murray, C. Olah, M. Schuster, J. Shlens, B. Steiner, I. Sutskever, K. Talwar, P. Tucker, V. Vanhouc...

    work page 2015

  29. [30]

    J. V. Dillon, I. Langmore, D. Tran, E. Brevdo, S. Vasudevan, D. Moore, B. Patton, A. Alemi, M. D. Hoffman, and R. A. Saurous, CoRRabs/1711.10604(2017), 1711.10604

    Pith/arXiv arXiv 2017

  30. [31]

    Germain, K

    M. Germain, K. Gregor, I. Murray, and H. Larochelle, (2015), arXiv:1502.03509 [cs.LG]

    Pith/arXiv arXiv 2015

  31. [32]

    Papamakarios, T

    G. Papamakarios, T. Pavlakou, and I. Murray, (2018), arXiv:1705.07057 [stat.ML]

    Pith/arXiv arXiv 2018

  32. [33]

    Ramachandran, B

    P. Ramachandran, B. Zoph, and Q. V. Le, arXiv preprint arXiv:1710.05941 (2017)

    Pith/arXiv arXiv 2017

  33. [34]

    Salvatier, T

    J. Salvatier, T. V. Wiecki, and C. Fonnesbeck, PeerJ Computer Science2, e55 (2016)

    work page 2016

  34. [35]

    M. D. Hoffman and A. Gelman, Journal of Machine Learning Research15, 1593 (2014)

    2014

  35. [36]

    Andreassen, P

    A. Andreassen, P. T. Komiske, E. M. Metodiev, B. Nachman, and J. Thaler, Phys. Rev. Lett.124, 182001 (2020), arXiv:1911.09107 [hep-ph]

    Pith/arXiv arXiv 2020

  36. [37]

    Pythia/Herwig + Delphes Jet Datasets for OmniFold Unfolding,

    A. Andreassen, P. Komiske, E. Metodiev, B. Nachman, and J. Thaler, “Pythia/Herwig + Delphes Jet Datasets for OmniFold Unfolding,” (2019)

    work page 2019

  37. [38]

    Bellmet al., (2017), arXiv:1705.06919 [hep-ph]

    J. Bellmet al., (2017), arXiv:1705.06919 [hep-ph]

    Pith/arXiv arXiv 2017

  38. [39]

    Sj¨ ostrand, S

    T. Sj¨ ostrand, S. Ask, J. R. Christiansen, R. Corke, N. Desai, P. Ilten, S. Mrenna, S. Prestel, C. O. Rasmussen, and P. Z. Skands, Comput. Phys. Commun.191, 159 (2015), arXiv:1410.3012 [hep-ph]

    Pith/arXiv arXiv 2015

  39. [40]

    ATLAS Collaboration, (2014),https://cds.cern.ch/record/1966419

    arXiv 2014

  40. [41]

    de Favereau, C

    J. de Favereau, C. Delaere, P. Demin, A. Giammanco, V. Lema ˆ ıtre, A. Mertens, and M. Selvaggi (DELPHES 3), JHEP 02, 057 (2014), arXiv:1307.6346 [hep-ex]

    Pith/arXiv arXiv 2014

  41. [42]

    Chatrchyanet al.(CMS), JINST3, S08004 (2008)

    S. Chatrchyanet al.(CMS), JINST3, S08004 (2008)

    work page 2008

  42. [43]

    Cacciari, G

    M. Cacciari, G. P. Salam, and G. Soyez, JHEP04, 063 (2008), arXiv:0802.1189 [hep-ph]

    Pith/arXiv arXiv 2008

  43. [44]

    Cacciari, G

    M. Cacciari, G. P. Salam, and G. Soyez, Eur. Phys. J.C72, 1896 (2012), arXiv:1111.6097 [hep-ph]

    Pith/arXiv arXiv 2012

  44. [45]

    A. J. Larkoski, I. Moult, and B. Nachman, Physics Reports841, 1–63 (2020)

    work page 2020

  45. [46]

    Thaler and K

    J. Thaler and K. Van Tilburg, Journal of High Energy Physics2011(2011), 10.1007/jhep03(2011)015

    doi:10.1007/jhep03(2011)015 2011

  46. [47]

    Thaler and K

    J. Thaler and K. Van Tilburg, Journal of High Energy Physics2012(2012), 10.1007/jhep02(2012)093

    doi:10.1007/jhep02(2012)093 2012

  47. [48]

    A. J. Larkoski, S. Marzani, G. Soyez, and J. Thaler, Journal of High Energy Physics2014(2014), 10.1007/jhep05(2014)146. 11

    doi:10.1007/jhep05(2014)146 2014

  48. [49]

    Dasgupta, A

    M. Dasgupta, A. Fregoso, S. Marzani, and G. P. Salam, Journal of High Energy Physics2013(2013), 10.1007/jhep09(2013)029

    doi:10.1007/jhep09(2013)029 2013

  49. [50]

    Huetsch, J

    N. Huetsch, J. Mari˜ no Villadamigo, A. Shmakov, S. Diefenbacher, V. Mikuni, T. Heimel, M. J. Fenton, K. T. Greif, B. Nachman, D. Whiteson, A. Butter, and T. Plehn, SciPost Physics18(2025), 10.21468/scipostphys.18.2.070

    doi:10.21468/scipostphys.18.2.070 2025