pith. sign in

arxiv: 2606.30489 · v1 · pith:PQONAHTOnew · submitted 2026-06-29 · 📊 stat.ML · cs.LG· hep-ex· hep-th· physics.data-an

Factorizable Normalizing Flows for parameter-dependent density morphing

Davide Valsecchi , Mauro Doneg\`a , Rainer Wallny This is my paper

Pith reviewed 2026-06-30 03:38 UTC · model grok-4.3

classification 📊 stat.ML cs.LGhep-exhep-thphysics.data-an
keywords normalizing flowsdensity estimationparameter-dependent modelingdensity morphinghigh energy physicsunbinned likelihoodfactorized transformationscontinuous parameters
0
0 comments X

The pith

Factorizable Normalizing Flows represent a density that changes with continuous parameters by composing a fixed reference flow with a learnable polynomial transformation that factors over each parameter.

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

The paper shows how to model densities that deform continuously with multiple parameters without training a new flow for every joint setting. A single high-fidelity flow is learned at a reference point and then composed with a transformation that is polynomial in the parameters and independent across them. Each parameter's isolated effect is trained on samples that vary only that parameter, after which any combination is recovered by adding the individual transformations at inference time. This structure is tested on a controlled case with two known deformations applied together, where the model recovers the true changes and the optimal likelihood; optional cross terms handle extra correlations when parameters vary strongly at once. The result is a model whose cost grows linearly with the number of parameters while the likelihood stays tractable.

Core claim

Factorizable Normalizing Flows model the parameter-dependent density as a fixed high-fidelity flow for a reference configuration composed with a learnable transformation that is polynomial in the parameters and factorized over them. Each parameter's effect is learned independently from samples in which that parameter alone is varied. The combined response of many parameters is recovered by summation at inference without ever sampling their combinatorially large joint space. On a controlled problem with two interpretable deformations applied jointly, the learned transformation reproduces the true deformations and matches the optimal likelihood, while optional interaction terms capture residua

What carries the argument

The learnable transformation that is polynomial in the parameters and factorized over them, composed with a fixed reference flow.

If this is right

  • Each parameter's deformation is learned from samples that vary only that parameter.
  • The response for any combination of parameters is recovered by summing the individual transformations.
  • The overall model scales linearly with the number of parameters rather than exponentially.
  • The likelihood remains tractable for downstream inference.
  • Optional interaction terms can be added to capture residual correlations under strong joint variations.

Where Pith is reading between the lines

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

  • The same structure could be applied to any simulation or measurement campaign where densities must be evaluated across continuous parameter grids.
  • Because individual effects are learned separately, the method may reduce the number of full joint samples needed in large-scale physics analyses.
  • If the polynomial order is increased systematically, the same factorization could test whether higher-order parameter effects remain separable.
  • The explicit factorization makes it straightforward to inspect which single parameter drives a given change in the density shape.

Load-bearing premise

The way a density deforms with parameters can be captured by a polynomial that factors over the parameters and composes with one fixed reference flow.

What would settle it

On the controlled test case with two known joint deformations, train the model and check whether the recovered transformation exactly matches the true deformations and reaches the optimal likelihood value.

Figures

Figures reproduced from arXiv: 2606.30489 by Davide Valsecchi, Mauro Doneg\`a, Rainer Wallny.

Figure 1
Figure 1. Figure 1: Schematic of the Factorizable Normalizing Flow. A systematic [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Schematic representation of the Factorizable Normalizing Flow layer. The scale [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Nominal dataset at ν = 0: the kinematic density p(x) and the marginal score density p(y | c), for class A (blue) and class B (orange). Both vary nontrivially between the two classes, illustrating the structure that the nominal flows must capture [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Effect of the two nuisance parameters, comparing the nominal distribution ( [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Validation of the factorizable residual for the kinematic density [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Validation of the score residual for p(y | x, c, ν) at the conditioning point x = (1, 1). (a) Learned affine response of the first component y1: the scale factor exp(s) and the shift t versus νshift (left pair) and νsqueeze (right pair), at four representative points in y and for both classes; the component y2 behaves analogously and is omitted. (b) Induced displacement field ∆y = Tν(y | x, ν)−y for +1σ va… view at source ↗
Figure 7
Figure 7. Figure 7: Goodness of fit across the nuisance plane, measured by the per-event [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Direct view of the learned cross-term for the score residual ( [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
read the original abstract

Normalizing Flows excel at modeling a single fixed density, yet many problems across the sciences, such as high energy physics, instead require modeling how that density deforms as a function of continuous parameters: the strength of a physical effect, a calibration constant, or a source of systematic uncertainty. Learning a separate flow for every parameter configuration quickly becomes intractable, since the number of joint settings grows exponentially with the number of parameters. We introduce Factorizable Normalizing Flows (FNFs), which represent the parameter-dependent density as a fixed, high-fidelity flow for a reference configuration composed with a learnable transformation that is polynomial in the parameters and factorized over them. This structure has a practical consequence: each parameter's effect is learned in isolation, from samples in which that parameter alone is varied. The combined response of many parameters is then recovered by summation at inference, without ever sampling their combinatorially large joint space. On a controlled problem with two interpretable deformations applied jointly to the data, the learned transformation reproduces the true deformations and matches the optimal likelihood, while optional interaction terms capture residual correlations when several parameters vary strongly at once. The resulting model is interpretable, scales linearly with the number of parameters, and keeps the likelihood tractable. This provides a general tool for any inference workflow requiring continuous density morphing, and directly enables the next generation of unbinned likelihood fits in high energy physics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper introduces Factorizable Normalizing Flows (FNFs) to model how a density deforms continuously with multiple parameters. A fixed high-fidelity reference flow is composed with a learnable polynomial transformation that factorizes over parameters (with optional interaction terms). Each parameter's effect is learned separately from single-parameter variation samples; the joint effect is recovered by summation at inference time. On a controlled two-parameter example the learned map is reported to recover the true deformations and achieve optimal likelihood while remaining interpretable and linearly scaling.

Significance. If the empirical claims hold, the construction supplies a practical route to continuous density morphing that avoids the exponential cost of joint sampling, preserves tractable likelihoods, and yields interpretable per-parameter effects. This directly addresses a recurring need in high-energy physics for unbinned likelihood fits that incorporate systematic variations without retraining separate flows.

major comments (1)
  1. [Abstract] Abstract: the central empirical claim states that the learned transformation 'reproduces the true deformations and matches the optimal likelihood' on the controlled two-parameter problem, yet no numerical values (log-likelihoods, KL divergences, reconstruction errors), error bars, or baseline comparisons are supplied. This absence leaves the strength of the supporting evidence difficult to evaluate.
minor comments (2)
  1. The weakest-assumption paragraph in the reader's note correctly identifies that the method relies on the deformation being adequately captured by a polynomial factorized map; the manuscript should state this modeling assumption explicitly and discuss its domain of validity.
  2. Implementation details (network architectures, polynomial degree, training procedure, and how the reference flow is held fixed) are not mentioned in the abstract and should be added to the methods section for reproducibility.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the supportive summary and recommendation of minor revision. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central empirical claim states that the learned transformation 'reproduces the true deformations and matches the optimal likelihood' on the controlled two-parameter problem, yet no numerical values (log-likelihoods, KL divergences, reconstruction errors), error bars, or baseline comparisons are supplied. This absence leaves the strength of the supporting evidence difficult to evaluate.

    Authors: We agree that the abstract would benefit from explicit quantitative support for the central claim. The detailed log-likelihood values, KL divergences, reconstruction errors, and baseline comparisons are already reported with error bars in Section 4 (Experiments) and Table 1 of the manuscript. In the revised version we will insert a concise summary of the key numerical results (e.g., achieved log-likelihood matching the optimum within reported uncertainty) directly into the abstract while preserving its length constraints. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper defines FNFs explicitly as a fixed reference flow composed with a new learnable polynomial-in-parameters factorized transformation (with optional interactions). The central performance claim is validated on a controlled synthetic problem by direct comparison to the known true deformations and optimal likelihood; the factorization is learned from isolated single-parameter samples and combined by summation at inference. No equation reduces the reported reproduction of deformations or likelihood match to a quantity already fitted inside the same model by construction. The structure is presented as an ansatz chosen for tractability and linear scaling, not derived from a self-citation chain or uniqueness theorem. This is a self-contained construction against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated beyond the core modeling choice of a factorized polynomial transformation.

free parameters (1)
  • polynomial coefficients
    Coefficients of the parameter-dependent polynomial transformation are learned from data during training.
axioms (1)
  • domain assumption Density morphing can be represented by composition with a factorized polynomial transformation
    This is the central structural assumption enabling isolated per-parameter learning.

pith-pipeline@v0.9.1-grok · 5793 in / 1178 out tokens · 59669 ms · 2026-06-30T03:38:01.479732+00:00 · methodology

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

46 extracted references · 27 canonical work pages

  1. [1]

    The frontier of simulation-based inference

    Kyle Cranmer, Johann Brehmer, and Gilles Louppe. The frontier of simulation-based inference. Proc. Nat. Acad. Sci., 117(48):30055–30062, 2020. doi:10.1073/pnas.1912789117

    work page doi:10.1073/pnas.1912789117 2020

  2. [2]

    Ivan Kobyzev, Simon J. D. Prince, and Marcus A. Brubaker. Normalizing Flows: An Introduction and Review of Current Methods.IEEE Trans. Pattern Anal. Mach. Intell., 43(11): 3964–3979, 2021. doi:10.1109/TPAMI.2020.2992934

    work page doi:10.1109/tpami.2020.2992934 2021

  3. [3]

    Johnson, and Patrick M

    Michele Vallisneri, Marco Crisostomi, Aaron D. Johnson, and Patrick M. Meyers. Rapid parameter estimation for pulsar-timing-array datasets with variational inference and normalizing flows.Phys. Rev. Lett., 135:071401, Aug 2025. doi:10.1103/p3f7-rbmv. URL https://link.aps.org/doi/10.1103/p3f7-rbmv

    work page doi:10.1103/p3f7-rbmv 2025

  4. [4]

    Normalizing Flows for Probabilistic Modeling and Inference.J

    George Papamakarios, Eric Nalisnick, Danilo Jimenez Rezende, Shakir Mohamed, and Balaji Lakshminarayanan. Normalizing Flows for Probabilistic Modeling and Inference.J. Mach. Learn. Res., 22(57):1–64, 2021. URLhttps://jmlr.org/papers/v22/19-1028.html

    2021

  5. [5]

    tex.eprint: https://www.science.org/doi/pdf/10.1126/science.aaw1147

    Frank Noé, Simon Olsson, Jonas Köhler, and Hao Wu. Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning.Science, 365(6457):eaaw1147, 2019. doi:10.1126/science.aaw1147

    work page doi:10.1126/science.aaw1147 2019

  6. [6]

    M. S. Albergo, G. Kanwar, and P. E. Shanahan. Flow-based generative models for Markov chain Monte Carlo in lattice field theory.Phys. Rev. D, 100(3):034515, 2019. doi:10.1103/PhysRevD.100.034515. 14 Preprint Valsecchiet al

    work page doi:10.1103/physrevd.100.034515 2019

  7. [7]

    The CMS statistical analysis and combination tool: COMBINE.Comput

    CMS Collaboration. The CMS statistical analysis and combination tool: COMBINE.Comput. Softw. Big Sci., 8(1):19, 2024. doi:10.1007/s41781-024-00121-4

    work page doi:10.1007/s41781-024-00121-4 2024

  8. [8]

    HistFactory: A tool for creating statistical models for use with RooFit and RooStats

    Kyle Cranmer, George Lewis, Lorenzo Moneta, Akira Shibata, and Wouter Verkerke. HistFactory: A tool for creating statistical models for use with RooFit and RooStats. Technical report, New York U., 2012. URLhttps://cds.cern.ch/record/1456844

    arXiv 2012

  9. [9]

    Interpolation between multi-dimensional histograms using a new non-linear moment morphing method.Nucl

    Max Baak, Stefan Gadatsch, Robert Harrington, and Wouter Verkerke. Interpolation between multi-dimensional histograms using a new non-linear moment morphing method.Nucl. Instrum. Meth. A, 771:39–48, 2015. doi:10.1016/j.nima.2014.10.033

    work page doi:10.1016/j.nima.2014.10.033 2015

  10. [10]

    Springer, Cham, 3rd edition, 2023

    Luca Lista.Statistical Methods for Data Analysis: With Applications in Particle Physics, volume 1003 ofLecture Notes in Physics. Springer, Cham, 3rd edition, 2023. ISBN 978-3-031-19933-2. doi:10.1007/978-3-031-19934-9

    work page doi:10.1007/978-3-031-19934-9 2023

  11. [11]

    Density estimation using Real NVP

    Laurent Dinh, Jascha Sohl-Dickstein, and Samy Bengio. Density estimation using Real NVP. In International Conference on Learning Representations (ICLR), 2017

    2017

  12. [12]

    Kingma, Tim Salimans, Rafal Jozefowicz, Xi Chen, Ilya Sutskever, and Max Welling

    Diederik P. Kingma, Tim Salimans, Rafal Jozefowicz, Xi Chen, Ilya Sutskever, and Max Welling. Improved Variational Inference with Inverse Autoregressive Flow. InAdvances in Neural Information Processing Systems, volume 29, pages 4743–4751, 2016

    2016

  13. [13]

    Masked autoregressive flow for density estimation, 2018

    George Papamakarios, Theo Pavlakou, and Iain Murray. Masked autoregressive flow for density estimation, 2018. URLhttps://arxiv.org/abs/1705.07057

    Pith/arXiv arXiv 2018

  14. [14]

    MADE: Masked Autoencoder for Distribution Estimation

    Mathieu Germain, Karol Gregor, Iain Murray, and Hugo Larochelle. MADE: Masked Autoencoder for Distribution Estimation. InProceedings of the 32nd International Conference on Machine Learning (ICML), volume 37 ofPMLR, pages 881–889, 2015

    2015

  15. [15]

    Kingma and Prafulla Dhariwal

    Diederik P. Kingma and Prafulla Dhariwal. Glow: Generative Flow with Invertible 1x1 Convolutions. InAdvances in Neural Information Processing Systems, volume 31, pages 10215–10224, 2018

    2018

  16. [16]

    Neural spline flows,

    Conor Durkan, Artur Bekasov, Iain Murray, and George Papamakarios. Neural spline flows,

  17. [17]

    URLhttps://arxiv.org/abs/1906.04032

    arXiv 1906

  18. [18]

    PyTorch: An Imperative Style, High-Performance Deep Learning Library

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Kopf, Edward Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, Junjie Bai, and Soumith Chintala. PyTorch: An Imperative Style, High-Perfo...

    2019

  19. [19]

    Zuko: Normalizing flows in pytorch, 2022

    François Rozet et al. Zuko: Normalizing flows in pytorch, 2022. URL https://pypi.org/project/zuko

    2022

  20. [20]

    valsdav/factorizable-normalizing-flow: v1.0, 2026

    Davide Valsecchi. valsdav/factorizable-normalizing-flow: v1.0, 2026. URL https://doi.org/10.5281/zenodo.21011625

    work page doi:10.5281/zenodo.21011625 2026

  21. [21]

    Decoupled weight decay regularization, 2019

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization, 2019. URL https://arxiv.org/abs/1711.05101

    Pith/arXiv arXiv 2019

  22. [22]

    Computational Optimal Transport.Found

    Gabriel Peyré and Marco Cuturi. Computational Optimal Transport.Found. Trends Mach. Learn., 11(5-6):355–607, 2019. doi:10.1561/2200000073

    work page doi:10.1561/2200000073 2019

  23. [23]

    OT-Flow: Fast and Accurate Continuous Normalizing Flows via Optimal Transport

    Derek Onken, Samy Wu Fung, Xingjian Li, and Lars Ruthotto. OT-Flow: Fast and Accurate Continuous Normalizing Flows via Optimal Transport. InProceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 9223–9232, 2021. doi:10.1609/aaai.v35i10.17113

    work page doi:10.1609/aaai.v35i10.17113 2021

  24. [24]

    Chin-Wei Huang, Ricky T. Q. Chen, Christos Tsirigotis, and Aaron Courville. Convex potential flows: Universal probability distributions with optimal transport and convex optimization, 2021. URLhttps://arxiv.org/abs/2012.05942

    arXiv 2021

  25. [25]

    Zico Kolter

    Brandon Amos, Lei Xu, and J. Zico Kolter. Input convex neural networks, 2017. URL https://arxiv.org/abs/1609.07152. 15 Preprint Valsecchiet al

    Pith/arXiv arXiv 2017

  26. [26]

    Improving and Generalizing Flow-Based Generative Models with Minibatch Optimal Transport.Trans

    Alexander Tong, Nikolay Malkin, Guillaume Huguet, Yanlei Zhang, Jarrid Rector-Brooks, Kilian Fatras, Guy Wolf, and Yoshua Bengio. Improving and Generalizing Flow-Based Generative Models with Minibatch Optimal Transport.Trans. Mach. Learn. Res., 2024, 2024. URLhttps://openreview.net/forum?id=HgDwiZrpVq. Introduces OT-CFM

    2024

  27. [27]

    Komiske, Eric M

    Patrick T. Komiske, Eric M. Metodiev, and Jesse Thaler. The Metric Space of Collider Events. Phys. Rev. Lett., 123(4):041801, 2019. doi:10.1103/PhysRevLett.123.041801

    work page doi:10.1103/physrevlett.123.041801 2019

  28. [28]

    A continuous calibration of the ATLAS flavour-tagging classifiers via optimal transportation maps, 2025

    ATLAS Collaboration. A continuous calibration of the ATLAS flavour-tagging classifiers via optimal transportation maps, 2025. Preprint

    2025

  29. [29]

    Mind the Gap: Navigating Inference with Optimal Transport Maps, 2025

    Malte Algren, Tobias Golling, Francesco Armando Di Bello, and Christopher Pollard. Mind the Gap: Navigating Inference with Optimal Transport Maps, 2025

    2025

  30. [30]

    Advancing Tools for Simulation-Based Inference.SciPost Phys

    Henning Bahl, Víctor Bresó-Pla, Giovanni De Crescenzo, and Tilman Plehn. Advancing Tools for Simulation-Based Inference.SciPost Phys. Core, 8:060, 2025. doi:10.21468/SciPostPhysCore.8.3.060

    work page doi:10.21468/scipostphyscore.8.3.060 2025

  31. [31]

    A Guide to Constraining Effective Field Theories with Machine Learning.Phys

    Johann Brehmer, Kyle Cranmer, Gilles Louppe, and Juan Pavez. A Guide to Constraining Effective Field Theories with Machine Learning.Phys. Rev. D, 98(5):052004, 2018. doi:10.1103/PhysRevD.98.052004

    work page doi:10.1103/physrevd.98.052004 2018

  32. [32]

    Unifying Simulation and Inference with Normalizing Flows.Phys

    Claudius Krause et al. Unifying Simulation and Inference with Normalizing Flows.Phys. Rev. D, 111:076004, 2025. doi:10.1103/PhysRevD.111.076004

    work page doi:10.1103/physrevd.111.076004 2025

  33. [33]

    Data-Driven High-Dimensional Statistical Inference with Generative Models.JHEP, 11:129, 2025

    Oz Amram and Manuel Szewc. Data-Driven High-Dimensional Statistical Inference with Generative Models.JHEP, 11:129, 2025. doi:10.1007/JHEP11(2025)129

    work page doi:10.1007/jhep11(2025)129 2025

  34. [34]

    Unbinned inclusive cross-section measurements with machine-learned systematic uncertainties.Phys

    Lisa Benato, Cristina Giordano, Claudius Krause, Ang Li, Robert Schöfbeck, Dennis Schwarz, Maryam Shooshtari, and Daohan Wang. Unbinned inclusive cross-section measurements with machine-learned systematic uncertainties.Phys. Rev. D, 112:052006, Sep 2025. doi:10.1103/zwzt-1rrw. URLhttps://link.aps.org/doi/10.1103/zwzt-1rrw

    work page doi:10.1103/zwzt-1rrw 2025

  35. [35]

    Unbinned multivariate observables for global SMEFT analyses from machine learning.JHEP, 03:033, 2023

    Raquel Gomez Ambrosio, Jaco ter Hoeve, Maeve Madigan, Juan Rojo, and Veronica Sanz. Unbinned multivariate observables for global SMEFT analyses from machine learning.JHEP, 03:033, 2023. doi:10.1007/JHEP03(2023)033

    work page doi:10.1007/jhep03(2023)033 2023

  36. [36]

    An implementation of neural simulation-based inference for parameter estimation in ATLAS.Rept

    ATLAS Collaboration. An implementation of neural simulation-based inference for parameter estimation in ATLAS.Rept. Prog. Phys., 2025. doi:10.1088/1361-6633/add370

    work page doi:10.1088/1361-6633/add370 2025

  37. [37]

    Measurement of off-shell Higgs boson production in theH→ZZ→4ℓ decay channel using a neural simulation-based inference technique in 13 TeV pp collisions with the ATLAS detector.Rept

    ATLAS Collaboration. Measurement of off-shell Higgs boson production in theH→ZZ→4ℓ decay channel using a neural simulation-based inference technique in 13 TeV pp collisions with the ATLAS detector.Rept. Prog. Phys., 2025. doi:10.1088/1361-6633/adcd9a

    work page doi:10.1088/1361-6633/adcd9a 2025

  38. [38]

    Refinable modeling for unbinned SMEFT analyses.Mach

    Robert Schöfbeck. Refinable modeling for unbinned SMEFT analyses.Mach. Learn. Sci. Tech., 6:015007, 2025. doi:10.1088/2632-2153/ad9fd1

    work page doi:10.1088/2632-2153/ad9fd1 2025

  39. [39]

    Multiscale Flow for Robust and Optimal Cosmological Analysis

    Biwei Dai and Uroš Seljak. Multiscale Flow for Robust and Optimal Cosmological Analysis. Proc. Nat. Acad. Sci., 121(9):e2309624121, 2024. doi:10.1073/pnas.2309624121

    work page doi:10.1073/pnas.2309624121 2024

  40. [40]

    olkopf, Bernhard , title =

    Maximilian Dax, Stephen R. Green, Jonathan Gair, Jakob H. Macke, Alessandra Buonanno, and Bernhard Schölkopf. Real-Time Gravitational Wave Science with Neural Posterior Estimation. Phys. Rev. Lett., 127(24):241103, 2021. doi:10.1103/PhysRevLett.127.241103

    work page doi:10.1103/physrevlett.127.241103 2021

  41. [41]

    One flow to correct them all: Improving simulations in high-energy physics with a single normalising flow and a switch.Comput

    Caio Daumann, Mauro Donegà, Johannes Erdmann, Massimiliano Galli, Jan Lukas Späh, and Davide Valsecchi. One flow to correct them all: Improving simulations in high-energy physics with a single normalising flow and a switch.Comput. Softw. Big Sci., 8(1):23, 2024. doi:10.1007/s41781-024-00125-0

    work page doi:10.1007/s41781-024-00125-0 2024

  42. [42]

    Generative Unfolding with Distribution Mapping

    Anja Butter, Sascha Diefenbacher, Nathan Huetsch, Vinicius Mikuni, Benjamin Nachman, Sofia Palacios Schweitzer, and Tilman Plehn. Generative Unfolding with Distribution Mapping. SciPost Phys., 18:200, 2025. doi:10.21468/SciPostPhys.18.6.200

    work page doi:10.21468/scipostphys.18.6.200 2025

  43. [43]

    Analysis-ready generative unfolding, 2025

    Anja Butter, Nathan Huetsch, Vinicius Mikuni, Benjamin Nachman, and Sofia Palacios Schweitzer. Analysis-ready generative unfolding, 2025. URL https://arxiv.org/abs/2509.02708. 16 Preprint Valsecchiet al

    arXiv 2025

  44. [44]

    Simulation-prior independent neural unfolding procedure, 2025

    Anja Butter, Theo Heimel, Nathan Huetsch, Michael Kagan, and Tilman Plehn. Simulation-prior independent neural unfolding procedure, 2025. URL https://arxiv.org/abs/2507.15084

    arXiv 2025

  45. [45]

    Machine Learning-Assisted Unfolding for Neutrino Cross-section Measurements with the OmniFold Technique.Phys

    T2K Collaboration. Machine Learning-Assisted Unfolding for Neutrino Cross-section Measurements with the OmniFold Technique.Phys. Rev. D, 112:012008, 2025. doi:10.1103/PhysRevD.112.012008

    work page doi:10.1103/physrevd.112.012008 2025

  46. [46]

    Neural networks for full phase-space reweighting and parameter tuning.Physical Review D, 101(9), May 2020

    Anders Andreassen and Benjamin Nachman. Neural networks for full phase-space reweighting and parameter tuning.Physical Review D, 101(9), May 2020. ISSN 2470-0029. doi:10.1103/physrevd.101.091901. URLhttp://dx.doi.org/10.1103/PhysRevD.101.091901. 17

    work page doi:10.1103/physrevd.101.091901 2020