Pith. sign in

REVIEW 3 major objections 6 minor 23 references

Developments in NuWro Monte Carlo generator

T0 review · 3 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read NuWro's next release replaces simplified nuclear models with an argon spectral function, the Valencia 2020 meson-exchange-current model, and the Ghent hybrid pion model, and the paper reports improved agreement with MicroBooNE, MINERvA…

desk verdict A clear, honest status report on NuWro's planned upgrades; the physics is in the companion papers, and the one real soft spot—the two-parameter sampling approximation to Valencia 2020—is disclosed but not validated here. read the letter →

arxiv 2501.11470 v2 pith:XLMX5M7Y submitted 2025-01-20 hep-ph

Hemant Prasad , Jan T. Sobczyk , Artur M. Ankowski , J. Luis Bonilla , Rwik Dharmapal Banerjee , Krzysztof M. Graczyk , Beata E. Kowal This is my paper · ORCID
classification hep-ph
keywords NuWroneutrinoeventgeneratorneutrino-nucleusscatteringargonspectralfunctionmesonexchangecurrentsingle-pionproductionaxialformfactormachinelearningcrosssections
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

NuWro is a Monte Carlo generator, a computer program that simulates neutrino interactions with atomic nuclei by sampling random outcomes. This paper reports the physics changes planned for its next release: an argon spectral function for quasi-elastic scattering, the MINERvA axial form factor, the Valencia 2020 model for meson-exchange currents, and the Ghent hybrid model for single-pion production. The paper's central claim is that these replacements bring generator predictions into closer agreement with MicroBooNE, MINERvA, and T2K data, particularly for exclusive final-state observables such as pion kinematics and transverse kinematic imbalance. The paper also reports machine-learning fits to electron-carbon scattering data and their transfer to other nuclear targets. These upgrades matter because oscillation experiments such as DUNE and Hyper-Kamiokande need precise models of neutrino-nucleus scattering to control systematic uncertainties.

What carries the argument

The machinery is NuWro's modular event-generation structure, in which inclusive nuclear responses are stored as tables and exclusive final-state kinematics are modeled separately. The two pieces carrying the new physics are the argon spectral function, which supplies the probability of removing a proton with a given momentum and removal energy, and a two-parameter nucleon sampling function that mimics the correlated two-nucleon phase space of the Valencia 2020 model for 2p2h events. For single-pion production, the Ghent hybrid model replaces the previous delta-only resonance treatment with four resonances, interference with the nonresonant background, and a Regge description at high invariant mass. In the machine-learning work, the central object is a bootstrap-trained deep network used as an empirical cross-section function, extended to other nuclei through transfer learning.

What would settle it

Generate 2p2h events with the new NuWro implementation and compare the outgoing nucleon momentum, opening-angle, and isospin distributions directly with the published Valencia 2020 model; if the two-parameter sampling function cannot reproduce the correlated-nucleon phase space that drives the exclusive observables, the claimed improvements in transverse kinematic imbalance would not hold.

Watch

Extended reading notes

Core claim

The paper's core claim is that the next NuWro release moves the generator from simple Fermi-gas and delta-resonance descriptions to data-anchored nuclear models. The argon spectral function, derived from electron knockout measurements, describes quasi-elastic events; the Valencia 2020 model separates two-nucleon and three-nucleon knockout and predicts the momenta and isospin of outgoing nucleons, which NuWro approximates with a two-parameter sampling function; and the Ghent hybrid model covers pion production through the first and second resonance regions, nonresonant background interference, and a Regge high-energy tail. In comparisons reported from companion papers, the upgraded generator gives lower chi-squared per degree of freedom against MicroBooNE CC1p0$\pi$ and MINERvA exclusive data, with reconstructed neutron momentum remaining the one exception. On the machine-learning side, the paper claims that bootstrap-trained deep networks reproduce electron-carbon scattering at low and high angles, and that fine-tuning transfers those predictions to other nuclei with limited data.

Load-bearing premise

The load-bearing premise is that the external physics models, especially the Valencia 2020 exclusive kinematics, which NuWro approximates with a two-parameter nucleon sampling function, are correct and are implemented faithfully in the generator, because this paper reports validations from companion papers rather than an independent derivation or full reproduction.

Editorial extensions

If this is right

  • The new NuWro release will be the first to include an argon spectral function derived from electron-scattering coincidence data in a neutrino event generator, directly serving MicroBooNE and SBND analyses on argon.
  • With the MINERvA axial form factor, the spectral-function model describes MicroBooNE CC1p0$\pi$ data with $\chi^2/\mathrm{d.o.f.}=0.7$, compared with 1.0 for the Fermi-gas model, indicating consistency between the two experiments.
  • The 2020 Valencia MEC implementation lowers $\chi^2/\mathrm{d.o.f.}$ for most MINERvA CC1p0$\pi$ exclusive variables, with reconstructed neutron momentum $|p_n|$ the reported exception.
  • The Ghent hybrid model removes the hadronization contribution from single-pion production and improves agreement with MINERvA and T2K in pion kinematics and transverse kinematic imbalance, while nuclear effects remain a challenge.
  • Bootstrap deep neural networks trained on electron-carbon scattering reproduce the data at low and high angles, and transfer learning extends these empirical cross sections to lithium, oxygen, aluminum, calcium, and iron with limited data.

Reading between the lines

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

  • Editorial inference: If the argon spectral function performs as well on other nuclear targets as it does on argon, spectral functions may replace Fermi-gas initial states as the default in generators, shifting one of the largest systematic uncertainties in long-baseline oscillation analyses.
  • Editorial inference: The two-parameter sampling approximation for the Valencia 2020 exclusive kinematics is the most fragile link in the new MEC implementation; a direct phase-space comparison against the original model would reveal whether the reported improvements come from the physics or from the sampling tune.
  • Editorial inference: The transfer-learning result suggests that empirical, data-driven cross sections could eventually bypass nuclear models altogether for many generator tasks, but only if the fine-tuned predictions are tested against neutrino data, not just electron scattering.
  • Editorial inference: The Ghent hybrid model's success may push other generators to abandon hadronization-based pion production below a few GeV; the remaining nuclear-effect discrepancies indicate the next bottleneck is the initial-state description rather than the hadron physics.
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

3 major / 6 minor

Summary. This conference proceedings contribution announces planned physics improvements for the NuWro Monte Carlo event generator. The upcoming release will add an argon spectral function for quasi-elastic scattering with a MINERvA axial form factor, implement the 2020 Valencia model for meson exchange currents with explicit treatment of outgoing-nucleon kinematics, replace the current Delta-resonance-only pion production model with the Ghent hybrid model, and apply deep neural networks (including transfer learning) to empirical fits of electron-nucleus cross sections. The paper presents algorithms and flowcharts for the new MEC implementation and cites companion papers for quantitative validation, including chi-squared comparisons against MicroBooNE and MINERvA data.

Significance. If the described developments are realized and validated, they would improve the accuracy of neutrino-nucleus event generators needed for DUNE and Hyper-Kamiokande. The paper is a concise status report with clear references to the underlying publications and to publicly available code. Its main strengths are the breadth of the planned improvements and the explicit discussion of algorithm details, such as the four-step MEC generation procedure. However, the central quantitative claims (chi-squared values and improved agreement with data) are not presented in the manuscript but only cited to companion papers, and the approximation used for the 2020 Valencia exclusive kinematics is not validated against the reference model. The manuscript is therefore more a summary of ongoing work than a self-contained research article.

major comments (3)
  1. [Section II.B] The implementation of the 2020 Valencia MEC model uses a nucleon sampling function with two adjustable parameters to approximate the correlations between outgoing nucleons. However, the manuscript does not provide any quantitative comparison between this sampling function and the correlated two-nucleon phase space of the original 2020 Valencia model (Ref. [11]). Since the distinguishing feature of that model is its prediction of correlated outgoing-nucleon momenta and isospin decomposition, the fidelity of this approximation is load-bearing for the claimed improvements in exclusive observables. Please include a comparison (e.g., a plot or a numerical metric) of the sampling function's phase-space distributions against those of Ref. [11], or explicitly state that the validation is entirely delegated to the companion paper Ref. [12].
  2. [Section II.B, final paragraph] The sentence stating that NuWro with the new hadronic model 'produces lower χ2/d.o.f except in the case of |pn|' is a quantitative claim that is not supported by any table, figure, or chi-squared value in this manuscript. Since this is the paper's primary evidence that the new MEC model improves agreement with MINERvA CC1p0π data, the relevant numbers or a summarizing plot should be included, or the claim should be unambiguously attributed to Ref. [12] with a note that the full comparison appears there.
  3. [Section III] The phrase 'model-independent reconstruction of lepton-nucleus interactions' overstates what the DNN framework accomplishes. The described work produces empirical fits to inclusive electron-carbon cross-section data and uses transfer learning to other nuclei; this is a data-driven parameterization, not a reconstruction of interaction dynamics in a model-independent sense. Please clarify the intended meaning and adjust the wording to avoid the implication that the DNN recovers underlying interaction mechanisms without any theoretical input.
minor comments (6)
  1. [Throughout] The name 'Wroc law' appears with a space and should be written as 'Wroclaw' or 'Wrocław' consistently.
  2. [Section II.B] The phrase 'squared-four momentum' should be 'four-momentum squared' for standard terminology.
  3. [Section I] The sentence 'The current version of NuWro available on GitHub ... is 21.09.2' would read more smoothly as 'The current version of NuWro, available on GitHub ..., is 21.09.2.'
  4. [Section III] The phrase 'extending from the quasi-elastic peak' should be capitalized ('Quasi-elastic') or rephrased, and the sentence beginning 'extending' is a fragment.
  5. [Figures 2 and 3] The flowchart in Fig. 3 is referenced but not described in the text; a brief explanation of the boxed region and the meaning of f(cos θ*) would improve readability.
  6. [Section II.C] The sentence 'The transition from the low energy regime to the high energy regime is described in [19]' is vague; it would be clearer to specify the kinematic variable (e.g., invariant mass W or Q2) that defines the transition.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all load-bearing validation is against external data (MicroBooNE, MINERvA, T2K), so self-citations do not reduce the claims to the paper's inputs.

full rationale

This paper is a proceedings-style summary of NuWro generator developments; it contains no derivation whose output is equivalent to its input. The quasi-elastic improvements are validated by comparing the argon spectral-function implementation to MicroBooNE CC1p0π data (Ref. [9]), with the χ2/d.o.f. taken from the authors' companion paper [3]; the MINERvA axial form factor is an external parametrization, and the electron-scattering spectral functions originate from JLab Hall A, not from this paper. The Valencia-2020 MEC section explicitly delegates implementation and validation to companion papers [11,12]; the two-parameter nucleon sampling function of Ref. [12] is an approximation to the Valencia exclusive kinematics, but the comparison to MINERvA CC1p0π is an external benchmark, and the reported worsening of |pn| is the opposite of what would be expected if the parameters had been tuned to that data. The Ghent hybrid model [19] is likewise tested against MINERvA and T2K data. The machine-learning portion is explicitly described as 'empirical fits' and transfer learning with test datasets excluded from training, so no fitted quantity is relabeled as a prediction. Self-citations are present and numerous, but in each case the cited companion paper provides the algorithm or fit and confronts it with independent experimental data; therefore the self-citations are real evidence rather than circular support. No equation in this paper is defined in terms of the quantity it is used to predict.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The central claims rest on external models and on the authors' companion papers. No new physical entities are introduced. The free parameters listed are implementation or training parameters, not fitted constants introduced to make a derivation work.

free parameters (2)
  • Nucleon sampling function parameters (two) = not stated in this paper
    Section II.B and Fig. 3 state the new MEC implementation uses a nucleon sampling function with two adjustable parameters to approximate Valencia 2020 phase space; values and fitting procedure are deferred to Ref [12].
  • DNN weights for bootstrap and MC dropout fits = trained on electron-carbon inclusive data
    Section III and Ref [21] describe empirical DNN fits; network weights are free parameters fitted to training data, though the models are tested on held-out data.
assumptions (6)
  • domain assumption Argon spectral function extracted from JLab Hall A (e,e'p) data can be used for neutrino-argon charged-current quasielastic scattering.
    Invoked in Section II.A; it carries electron-scattering observables into the weak interaction sector without direct weak-process validation.
  • domain assumption The MINERvA parametrization of the axial form factor correctly describes the nucleon axial current in the quasi-elastic channel.
    Section II.A adopts this parametrization as an input; it is a fit to prior data, not derived in this paper.
  • domain assumption The Valencia 2020 model correctly predicts exclusive 2p2h and 3p3h kinematics for meson exchange current events.
    Section II.B imports Ref [11]; the manuscript does not re-derive or independently test the model.
  • domain assumption The Ghent hybrid model correctly describes single-pion production across the resonance and Regge regions.
    Section II.C replaces the existing Delta-only model with Ref [19]; correctness is assumed from the companion paper.
  • domain assumption DNN fits trained on electron-carbon scattering transfer to other nuclear targets after fine-tuning.
    Section III summarizes Ref [22]; the transferability claim is supported only by the authors' own work.
  • standard math Standard quantum-field-theory and nuclear-many-body formalism underlying all cited models.
    Background assumptions of relativistic kinematics, response functions, and final-state interactions are used throughout without proof.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Developments in NuWro Monte Carlo generator." pith.science (2026). https://pith.science/paper/XLMX5M7Y

@misc{pith2026250111470,
  author       = {Pith},
  title        = {Pith review of: Developments in NuWro Monte Carlo generator},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XLMX5M7Y}},
  note         = {Machine review of arXiv:2501.11470}
}
abstract

In this article, we highlight physics improvements in the NuWro Monte Carlo event generator. The upcoming version of NuWro will incorporate the integration of the argon spectral function for quasi-elastic scattering, along with the MINER$\nu$A parametrization of the axial form factor. Additionally, the new release will feature the implementation of the Valencia 2020 model for meson exchange current. The previously used simplistic delta resonance model for single-pion production will be replaced by a more accurate Ghent hybrid model in the upcoming version of NuWro. We also discuss the recent advancements made by the Wroclaw Neutrino Group in applying machine-learning techniques to achieve model-independent reconstruction of lepton-nucleus interactions

Figures

Figures reproduced from arXiv: 2501.11470 by the authors.

Figure 1
Figure 1. FIG. 1. Double differential cross section [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Algorithm of the new MC MEC model in the context of NuWro implementation of 2020 Valencia model. The “boxed” [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Algorithm corresponding to “Generate outgoing nucleons” given in Fig. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 13 canonical work pages

  1. [3]

    Golan, C

    T. Golan, C. Juszczak, and J. T. Sobczyk, Final State Interactions Effects in Neutrino-Nucleus Interactions, Phys. Rev. C 86, 015505 (2012), arXiv:1202.4197 [nucl- th]

    arXiv 2012

  2. [12]

    J. E. Sobczyk, J. Nieves, and F. S´ anchez, Exclusive-final- state hadron observables from neutrino-nucleus multi- nucleon knockout, Phys. Rev. C 102, 024601 (2020), arXiv:2002.08302 [nucl-th]

    work page Pith review arXiv 2020

  3. [19]

    Bodek and U

    A. Bodek and U. K. Yang, Modeling neutrino and elec- tron scattering inelastic cross sections (2003), arXiv:hep- ex/0308007 [hep-ex]

    work page arXiv 2003

  4. [11]

    J. T. Sobczyk, Multinucleon Ejection Model for Meson Exchange Current Neutrino Interactions, Phys. Rev. C 86, 015504 (2012), arXiv:1201.3673 [hep-ph]

    arXiv 2012

  5. [1]

    Juszczak, J

    C. Juszczak, J. A. Nowak, and J. T. Sobczyk, Simulations from a new neutrino event generator, Nucl. Phys. B Proc. Suppl. 159, 211 (2006), arXiv:hep-ph/0512365

    arXiv 2006

  6. [2]

    boxed” region is the inclusive part of the MEC model. FIG. 3. Algorithm corresponding to “Generate outgoing nucleons

    Lepton kinematics. Input: Generate outgoing nucleons Compute event weight Store the event Produce double differential cross-sections Four-sets of hadronic tables FIG. 2. Algorithm of the new MC MEC model in the context of NuWro implementation of 2020 Valencia model. The “boxed” region is the inclusive part of the MEC model. FIG. 3. Algorithm corresponding...

    work page 2020

  7. [4]

    R. D. Banerjee, A. M. Ankowski, K. M. Graczyk, B. E. Kowal, H. Prasad, and J. T. Sobczyk, JLab spec- tral functions of argon in nuwro and their implications for MicroBooNE, Phys. Rev. D 109, 073004 (2024), arXiv:2312.13369 [hep-ph]

    arXiv 2024

  8. [5]

    Jiang, A

    L. Jiang, A. M. Ankowski, D. Abrams, et al. (Jefferson Lab Hall A Collaboration), Determination of the argon spectral function from ( e, e ′ p) data, Phys. Rev. D 105, 112002 (2022)

    work page 2022

Show all 23 references
  1. [6]

    Jiang, A

    L. Jiang, A. M. Ankowski, D. Abrams, et al. (The Jef- ferson Lab Hall A Collaboration), Determination of the titanium spectral function from (e, e ′ p) data, Phys. Rev. D 107, 012005 (2023)

    2023

  2. [7]

    A. M. Ankowski, O. Benhar, and M. Sakuda, Improv- ing the accuracy of neutrino energy reconstruction in charged-current quasielastic scattering off nuclear tar- gets, Phys. Rev. D 91, 033005 (2015)

    2015

  3. [8]

    Abratenko, R

    P. Abratenko, R. Acciarri, C. Adams, et al., Scintillation light in sbnd: simulation, reconstruction, and expected performance of the photon detection system, The Euro- pean Physical Journal C 84, 10.1140/epjc/s10052-024- 13306-3 (2024)

    2024 doi

  4. [9]

    Aguilar, L

    A. Aguilar, L. B. Auerbach, R. L. Burman, et al. (LSND Collaboration), Evidence for neutrino oscillations from the observation of νe appearance in a νµ beam, Phys. Rev. D 64, 112007 (2001)

    2001

  5. [10]

    Abratenko, M

    P. Abratenko, M. Alrashed, R. An, et al. (MicroBooNE Collaboration), First measurement of differential charged current quasielasticlike νµ-argon scattering cross sections with the microboone detector, Phys. Rev. Lett. 125, 201803 (2020)

    2020

  6. [13]

    Prasad, J

    H. Prasad, J. T. Sobczyk, A. M. Ankowski, J. L. Bonilla, R. D. Banerjee, K. M. Graczyk, and B. E. Kowal, New multinucleon knockout model in nuwro monte carlo gen- erator (2024), arXiv:2411.11523 [hep-ph]

    2024 arXiv

  7. [14]

    Abe et al

    K. Abe et al. (Hyper-Kamiokande Proto-), Physics po- tential of a long-baseline neutrino oscillation experiment using a J-PARC neutrino beam and Hyper-Kamiokande, PTEP 2015, 053C02 (2015), arXiv:1502.05199 [hep-ex]

    2015 arXiv

  8. [15]

    Acciarri et al

    R. Acciarri et al. (DUNE), Long-Baseline Neutrino Facil- ity (LBNF) and Deep Underground Neutrino Experiment (DUNE): Conceptual Design Report, Volume 1: The LBNF and DUNE Projects, (2016), arXiv:1601.05471 [physics.ins-det]

    2016

  9. [16]

    Alvarez-Ruso et al

    L. Alvarez-Ruso et al. (NuSTEC), NuSTEC White Paper: Status and challenges of neutrino–nucleus scattering, Prog. Part. Nucl. Phys. 100, 1 (2018), arXiv:1706.03621 [hep-ph]

    2018 arXiv

  10. [17]

    K. M. Graczyk, D. Kie lczewska, P. Przew locki, and J. T. Sobczyk, C A 5 axial form factor from bubble chamber ex- periments, Phys. Rev. D 80, 093001 (2009)

    2009

  11. [18]

    Sj¨ ostrand, S

    T. Sj¨ ostrand, S. Mrenna, and P. Skands, Pythia 6.4 physics and manual, Journal of High Energy Physics 2006, 026 (2006)

    2006

  12. [20]

    Q. Yan, K. Niewczas, A. Nikolakopoulos, R. Gonz´ alez- Jim´ enez, N. Jachowicz, X. Lu, J. Sobczyk, and Y. Zheng, The ghent hybrid model in nuwro: a new neutrino single-pion production model in the gev regime (2024), arXiv:2405.05212 [hep-ph]

    2024 arXiv

  13. [21]

    Hern´ andez, J

    E. Hern´ andez, J. Nieves, and M. Valverde, Weak pion production off the nucleon, Phys. Rev. D 76, 033005 (2007)

    2007

  14. [22]

    B. E. Kowal, K. M. Graczyk, A. M. Ankowski, R. D. Banerjee, H. Prasad, and J. T. Sobczyk, Empirical fits to inclusive electron-carbon scattering data obtained by deep-learning methods, Phys. Rev. C110, 025501 (2024)

    2024

  15. [23]

    K. M. Graczyk, B. E. Kowal, A. M. Ankowski, R. D. Banerjee, J. L. Bonilla, H. Prasad, and J. T. Sobczyk, Electron-nucleus cross sections from transfer learning (2024), arXiv:2408.09936 [hep-ph]

    2024 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.