pith. sign in

arxiv: 2606.26228 · v1 · pith:GJ3B3OCTnew · submitted 2026-06-24 · ⚛️ physics.data-an · astro-ph.GA· cs.LG· hep-ph

Interpreting "Interpretability" and Explaining "Explainability" in Machine Learning in Physics

Rikab Gambhir , Luisa Lucie-Smith , Jesse Thaler This is my paper

Pith reviewed 2026-06-26 00:51 UTC · model grok-4.3

classification ⚛️ physics.data-an astro-ph.GAcs.LGhep-ph
keywords interpretabilityexplainabilitymachine learningphysicsmodel transparencydomain knowledgemodel design
0
0 comments X

The pith

Machine-learned models in physics face the same scientific questions as classical models, with interpretability and explainability treated as deliberate design choices.

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

The paper defines interpretability as structural transparency of a model and explainability as the ability to map its outputs onto domain knowledge in physics. It argues these are not fixed traits but choices made during model design, alongside trade-offs such as transparency versus expressivity. The authors stress that machine-learned models differ from classical ones only in scale and must meet the same standards of scientific scrutiny. Task specification and plans for intervention form a core part of that design process. A sympathetic reader would care because this framing moves the discussion from seeking inherent properties toward practical decisions about when and how much transparency or mapping is needed for a given physics problem.

Core claim

Interpretability concerns the structural transparency of a model while explainability concerns its scientific content through mapping onto domain knowledge; both are deliberate modeling choices rather than inherent properties, and machine-learned models remain subject to the same scientific questions as classical models, differing only in scale.

What carries the argument

The distinction between structural transparency (interpretability) and domain-knowledge mapping (explainability), together with their respective trade-offs against expressivity and adaptability.

If this is right

  • Model design must include explicit choices about the level of structural transparency required for a given task.
  • Post-hoc tools become necessary when intrinsic structure alone cannot deliver the needed domain mapping.
  • Task specification determines whether interpretability, explainability, or neither is prioritized.
  • Intervention plans must be part of the initial model architecture rather than added afterward.

Where Pith is reading between the lines

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

  • The same distinction could guide model choices in other data-intensive sciences where domain mapping is central.
  • Reframing the properties as choices suggests new architectures that optimize one dimension while accepting limits on the other.
  • It implies that discussions of black-box models can shift from prohibition to calibrated trade-off decisions based on the physics task.

Load-bearing premise

The proposed distinction between structural transparency and domain-knowledge mapping captures the primary practical and conceptual needs when applying machine learning to physics problems.

What would settle it

An example of a machine-learned physics model that cannot be made scientifically usable without simultaneously achieving both high structural transparency and high domain-knowledge mapping, or a case in which a machine-learned model evades the standard scientific questions applied to classical models.

Figures

Figures reproduced from arXiv: 2606.26228 by Jesse Thaler, Luisa Lucie-Smith, Rikab Gambhir.

Figure 1
Figure 1. Figure 1: Examples of scientific goals and existing models in HEP, astrophysics, [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The trade-off between interpretability and expressivity. Models that [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The trade-off between explainability and adaptability: the ability to model [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
read the original abstract

We review the concepts of interpretability and explainability as they apply to machine learning in physics. We define interpretability as concerning the structural transparency of a model (the ability to understand or approximate its inner workings) and explainability as concerning the scientific content of a model (the ability to map it onto domain knowledge). We discuss the trade-offs each entails (interpretability vs. expressivity; explainability vs. adaptability), the contexts in which each is needed, and the intrinsic and post-hoc tools available for achieving them. Throughout, we emphasize that machine-learned models are subject to the same scientific questions as classical models, differing only in scale, and that interpretability and explainability are best understood as deliberate modeling choices rather than inherent properties. We also emphasize the importance of task specification and intervention plans as a core aspect of model design.

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

0 major / 2 minor

Summary. The manuscript reviews interpretability and explainability in machine learning applied to physics. It defines interpretability as structural transparency (understanding or approximating inner workings) and explainability as the ability to map the model onto domain knowledge. The paper discusses associated trade-offs (interpretability vs. expressivity; explainability vs. adaptability), the contexts in which each is relevant, and available tools. It argues that machine-learned models face the same scientific questions as classical models (differing only in scale), positions interpretability and explainability as deliberate modeling choices rather than inherent properties, and stresses the importance of task specification and intervention plans as core aspects of model design.

Significance. If the proposed distinction and framing hold, the paper offers a clear conceptual scaffold that could help physicists treat ML models with the same scrutiny applied to traditional models. Framing these properties as choices rather than fixed attributes encourages explicit design decisions around transparency and scientific utility. The emphasis on task specification aligns with standard scientific practice and may reduce misuse of black-box models in physics applications.

minor comments (2)
  1. [Abstract] Abstract: the trade-off discussion is stated at a high level; adding one or two brief physics-specific examples (e.g., in a regression or classification task) would make the distinction between structural transparency and domain mapping more concrete without altering the central argument.
  2. The manuscript would benefit from explicit cross-references to key prior reviews on interpretability in the physical sciences to situate the proposed definitions relative to existing usage in the literature.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive and accurate summary of the manuscript, as well as the recommendation for minor revision. The assessment that the proposed distinction offers a useful conceptual scaffold and that framing these properties as deliberate choices aligns with scientific practice is encouraging. No major comments were provided in the report.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The manuscript is a conceptual review paper that defines interpretability (structural transparency) and explainability (domain-knowledge mapping) as deliberate modeling choices. It contains no equations, no fitted parameters, no derivations, and no predictions that could reduce to inputs by construction. All claims are discursive and definitional; the central reframing (ML models face the same scientific questions as classical models, differing only in scale) is presented as a modeling perspective rather than a derived result. No self-citation load-bearing steps, uniqueness theorems, or ansatzes appear. The derivation chain is empty by design.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central contribution rests on the domain assumption that distinguishing structural transparency from scientific mapping is a useful modeling choice for physics applications; no free parameters or invented entities are introduced.

axioms (1)
  • domain assumption Interpretability and explainability are best treated as deliberate modeling choices rather than fixed properties of any given algorithm.
    Invoked in the abstract as the framing for the entire discussion.

pith-pipeline@v0.9.1-grok · 5684 in / 1198 out tokens · 20690 ms · 2026-06-26T00:51:54.036825+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

120 extracted references · 68 canonical work pages · 15 internal anchors

  1. [1]

    2019 , eprint=

    Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations , author=. 2019 , eprint=

    2019

  2. [2]

    arXiv e-prints , keywords =

    E(2) Equivariant Neural Networks for Robust Galaxy Morphology Classification. arXiv e-prints , keywords =. doi:10.48550/arXiv.2311.01500 , archivePrefix =. 2311.01500 , primaryClass =

    work page doi:10.48550/arxiv.2311.01500

  3. [3]

    Rotation-invariant convolutional neural networks for galaxy morphology prediction

    Rotation-invariant convolutional neural networks for galaxy morphology prediction. , keywords =. doi:10.1093/mnras/stv632 , archivePrefix =. 1503.07077 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stv632

  4. [4]

    KiDS-SBI: Simulation-based inference analysis of KiDS-1000 cosmic shear

    von Wietersheim-Kramsta, Maximilian and Lin, Kiyam and Tessore, Nicolas and Joachimi, Benjamin and Loureiro, Arthur and Reischke, Robert and Wright, Angus H. KiDS-SBI: Simulation-based inference analysis of KiDS-1000 cosmic shear. Astron. Astrophys. 2025. doi:10.1051/0004-6361/202450487. arXiv:2404.15402

    work page doi:10.1051/0004-6361/202450487 2025

  5. [5]

    Deep learning insights into cosmological structure formation , author =. Phys. Rev. D , volume =. 2024 , month =. doi:10.1103/PhysRevD.109.063524 , url =

    work page doi:10.1103/physrevd.109.063524 2024

  6. [6]

    DeepLSS: Breaking Parameter Degeneracies in Large-Scale Structure with Deep-Learning Analysis of Combined Probes , author =. Phys. Rev. X , volume =. 2022 , month =. doi:10.1103/PhysRevX.12.031029 , url =

    work page doi:10.1103/physrevx.12.031029 2022

  7. [7]

    The Importance of Being Interpretable: Toward an Understandable Machine Learning Encoder for Galaxy Cluster Cosmology , volume=

    Ntampaka, Michelle and Vikhlinin, Alexey , year=. The Importance of Being Interpretable: Toward an Understandable Machine Learning Encoder for Galaxy Cluster Cosmology , volume=. The Astrophysical Journal , publisher=. doi:10.3847/1538-4357/ac423e , number=

    work page doi:10.3847/1538-4357/ac423e

  8. [8]

    Interpreting deep learning models for weak lensing , author =. Phys. Rev. D , volume =. 2020 , month =. doi:10.1103/PhysRevD.102.123506 , url =

    work page doi:10.1103/physrevd.102.123506 2020

  9. [9]

    Real-Time Gravitational Wave Science with Neural Posterior Estimation , author =. Phys. Rev. Lett. , volume =. 2021 , month =. doi:10.1103/PhysRevLett.127.241103 , url =

    work page doi:10.1103/physrevlett.127.241103 2021

  10. [10]

    , volume =

    Z. Dark energy survey year 3 results: Cosmology with peaks using an emulator approach. Mon. Not. Roy. Astron. Soc. 2022. doi:10.1093/mnras/stac078. arXiv:2110.10135

    work page doi:10.1093/mnras/stac078 2022

  11. [11]

    2025, @doi [ ] 10.1093/mnras/stae2629 , https://ui.adsabs.harvard.edu/abs/2025MNRAS.536.1303J 536, 1303

    Jeffrey, N. and others. Dark energy survey year 3 results: likelihood-free, simulation-based wCDM inference with neural compression of weak-lensing map statistics. Mon. Not. Roy. Astron. Soc. 2024. doi:10.1093/mnras/stae2629. arXiv:2403.02314

    work page doi:10.1093/mnras/stae2629 2024

  12. [12]

    Nature Reviews Physics , keywords =

    Cosmological simulations of galaxy formation. Nature Reviews Physics , keywords =. doi:10.1038/s42254-019-0127-2 , archivePrefix =. 1909.07976 , primaryClass =

    work page doi:10.1038/s42254-019-0127-2 1909

  13. [13]

    Discovering Physical Concepts with Neural Networks

    Iten, Raban and Metger, Tony and Wilming, Henrik and del Rio, L \' dia and Renner, Renato. Discovering Physical Concepts with Neural Networks. Phys. Rev. Lett. 2020. doi:10.1103/PhysRevLett.124.010508. arXiv:1807.10300

    work page doi:10.1103/physrevlett.124.010508 2020

  14. [14]

    International Conference on Learning Representations , year =

    Higgins, Irina and Matthey, Loic and Pal, Arka and Burgess, Christopher and Glorot, Xavier and Botvinick, Matthew and Mohamed, Shakir and Lerchner, Alexander , title =. International Conference on Learning Representations , year =. 1804.03599 , archivePrefix =

    Pith/arXiv arXiv

  15. [15]

    Advances in Neural Information Processing Systems , volume =

    Chen, Xi and Duan, Yan and Houthooft, Rein and Schulman, John and Sutskever, Ilya and Abbeel, Pieter , title =. Advances in Neural Information Processing Systems , volume =. 2016 , eprint =

    2016

  16. [16]

    Machine Learning: Science and Technology , keywords =

    Predicting the thermal Sunyaev-Zel'dovich field using modular and equivariant set-based neural networks. Machine Learning: Science and Technology , keywords =. doi:10.1088/2632-2153/ac78c2 , archivePrefix =. 2203.00026 , primaryClass =

    work page doi:10.1088/2632-2153/ac78c2

  17. [17]

    , keywords =

    Fanaroff-Riley classification of radio galaxies using group-equivariant convolutional neural networks. , keywords =. doi:10.1093/mnras/stab530 , archivePrefix =. 2102.08252 , primaryClass =

    work page doi:10.1093/mnras/stab530

  18. [18]

    Physics-informed neural networks in the recreation of hydrodynamic simulations from dark matter.Mon

    Physics-informed neural networks in the recreation of hydrodynamic simulations from dark matter. , keywords =. doi:10.1093/mnras/stad3394 , archivePrefix =. 2303.14090 , primaryClass =

    work page doi:10.1093/mnras/stad3394

  19. [19]

    2025 , eprint=

    SPINN: Advancing Cosmological Simulations of Fuzzy Dark Matter with Physics Informed Neural Networks , author=. 2025 , eprint=

    2025

  20. [20]

    The Mira-Titan Universe. III. Emulation of the Halo Mass Function. , keywords =. doi:10.3847/1538-4357/abac5c , archivePrefix =. 2003.12116 , primaryClass =

    work page doi:10.3847/1538-4357/abac5c 2003

  21. [21]

    , keywords =

    Aricò, Giovanni and Angulo, Raul E and Contreras, Sergio and Ondaro-Mallea, Lurdes and Pellejero-Ibañez, Marcos and Zennaro, Matteo , year=. The BACCO simulation project: a baryonification emulator with neural networks , volume=. Monthly Notices of the Royal Astronomical Society , publisher=. doi:10.1093/mnras/stab1911 , number=

    work page doi:10.1093/mnras/stab1911

  22. [22]

    Toward a halo mass function for precision cosmology: the limits of universality

    Tinker, Jeremy L. and Kravtsov, Andrey V. and Klypin, Anatoly and Abazajian, Kevork and Warren, Michael S. and Yepes, Gustavo and Gottlober, Stefan and Holz, Daniel E. Toward a halo mass function for precision cosmology: The Limits of universality. Astrophys. J. 2008. doi:10.1086/591439. arXiv:0803.2706

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/591439 2008

  23. [23]

    and Frenk, Carlos S

    The Structure of Cold Dark Matter Halos. , keywords =. doi:10.1086/177173 , archivePrefix =

    work page doi:10.1086/177173

  24. [24]

    Trudy Astrofizicheskogo Instituta Alma-Ata , year = 1965, month = jan, volume =

    On the Construction of a Composite Model for the Galaxy and on the Determination of the System of Galactic Parameters. Trudy Astrofizicheskogo Instituta Alma-Ata , year = 1965, month = jan, volume =

    1965

  25. [25]

    Spectral synthesis method

    Springel, Volker. The Cosmological simulation code GADGET-2. Mon. Not. Roy. Astron. Soc. 2005. doi:10.1111/j.1365-2966.2005.09655.x. arXiv:astro-ph/0505010

    work page doi:10.1111/j.1365-2966.2005.09655.x 2005

  26. [26]

    Efficient Computation of CMB anisotropies in closed FRW models

    Lewis, Antony and Challinor, Anthony and Lasenby, Anthony. Efficient computation of CMB anisotropies in closed FRW models. 2000. doi:10.1086/309179. arXiv:astro-ph/9911177

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/309179 2000

  27. [27]

    The EAGLE project: Simulating the evolution and assembly of galaxies and their environments

    The EAGLE project: simulating the evolution and assembly of galaxies and their environments. , keywords =. doi:10.1093/mnras/stu2058 , archivePrefix =. 1407.7040 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stu2058

  28. [28]

    2019, MNRAS, 490, 3196, doi: 10.1093/mnras/stz2338

    First results from the TNG50 simulation: the evolution of stellar and gaseous discs across cosmic time. , keywords =. doi:10.1093/mnras/stz2338 , archivePrefix =. 1902.05553 , primaryClass =

    work page doi:10.1093/mnras/stz2338 1902

  29. [29]

    Introducing the Illustris Project: Simulating the coevolution of dark and visible matter in the Universe

    Introducing the Illustris Project: simulating the coevolution of dark and visible matter in the Universe. , keywords =. doi:10.1093/mnras/stu1536 , archivePrefix =. 1405.2921 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stu1536

  30. [30]

    The Cosmic Linear Anisotropy Solving System (CLASS) II: Approximation schemes

    The Cosmic Linear Anisotropy Solving System (CLASS). Part II: Approximation schemes. , keywords =. doi:10.1088/1475-7516/2011/07/034 , archivePrefix =. 1104.2933 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/1475-7516/2011/07/034 2011

  31. [31]

    TPZ : Photometric redshift PDFs and ancillary information by using prediction trees and random forests

    TPZ: photometric redshift PDFs and ancillary information by using prediction trees and random forests. , keywords =. doi:10.1093/mnras/stt574 , archivePrefix =. 1303.7269 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stt574

  32. [32]

    ANNz: estimating photometric redshifts using artificial neural networks

    ANNz: Estimating Photometric Redshifts Using Artificial Neural Networks. , keywords =. doi:10.1086/383254 , archivePrefix =. astro-ph/0311058 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/383254

  33. [33]

    CosmoPower: emulating cosmological power spectra for accelerated Bayesian inference from next-generation surveys , volume=

    Spurio Mancini, Alessio and Piras, Davide and Alsing, Justin and Joachimi, Benjamin and Hobson, Michael P , year=. CosmoPower: emulating cosmological power spectra for accelerated Bayesian inference from next-generation surveys , volume=. Monthly Notices of the Royal Astronomical Society , publisher=. doi:10.1093/mnras/stac064 , number=

    work page doi:10.1093/mnras/stac064

  34. [34]

    Advances in Neural Information Processing Systems 30 , editor =

    A Unified Approach to Interpreting Model Predictions , author =. Advances in Neural Information Processing Systems 30 , editor =. 2017 , publisher =

    2017

  35. [35]

    2017 , eprint=

    SmoothGrad: removing noise by adding noise , author=. 2017 , eprint=

    2017

  36. [36]

    International Journal of Computer Vision , author =

    Selvaraju, Ramprasaath R. and Cogswell, Michael and Das, Abhishek and Vedantam, Ramakrishna and Parikh, Devi and Batra, Dhruv , year=. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization , volume=. International Journal of Computer Vision , publisher=. doi:10.1007/s11263-019-01228-7 , number=

    work page doi:10.1007/s11263-019-01228-7

  37. [37]

    2015 , eprint=

    Learning Deep Features for Discriminative Localization , author=. 2015 , eprint=

    2015

  38. [38]

    2014 , eprint=

    Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps , author=. 2014 , eprint=

    2014

  39. [39]

    Why should i trust you?

    " Why should i trust you?" Explaining the predictions of any classifier , author=. Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining , pages=

  40. [40]

    CDM and early dark energy in latent space: A data-driven parametrization of the CMB temperature power spectrum

    Piras, Davide and Herold, Laura and Lucie-Smith, Luisa and Komatsu, Eiichiro. CDM and early dark energy in latent space: A data-driven parametrization of the CMB temperature power spectrum. Phys. Rev. D. 2025. doi:10.1103/PhysRevD.111.083537. arXiv:2502.09810

    work page doi:10.1103/physrevd.111.083537 2025

  41. [41]

    and Pontzen, Andrew

    Lucie-Smith, Luisa and Peiris, Hiranya V. and Pontzen, Andrew. Explaining Dark Matter Halo Density Profiles with Neural Networks. Phys. Rev. Lett. 2024. doi:10.1103/PhysRevLett.132.031001. arXiv:2305.03077

    work page doi:10.1103/physrevlett.132.031001 2024

  42. [42]

    A deep-learning model for the density profiles of subhaloes in IllustrisTNG

    Lucie-Smith, Luisa and Despali, Giulia and Springel, Volker. A deep-learning model for the density profiles of subhaloes in IllustrisTNG. Mon. Not. Roy. Astron. Soc. 2024. doi:10.1093/mnras/stae1487. arXiv:2403.12125

    work page doi:10.1093/mnras/stae1487 2024

  43. [43]

    and Pontzen, Andrew and Piras, Davide

    Guo, Ningyuan and Lucie-Smith, Luisa and Peiris, Hiranya V. and Pontzen, Andrew and Piras, Davide. Deep learning insights into non-universality in the halo mass function. Mon. Not. Roy. Astron. Soc. 2024. doi:10.1093/mnras/stae1696. arXiv:2405.15850

    work page doi:10.1093/mnras/stae1696 2024

  44. [44]

    and Pontzen, Andrew

    Lucie-Smith, Luisa and Peiris, Hiranya V. and Pontzen, Andrew. An interpretable machine learning framework for dark matter halo formation. Mon. Not. Roy. Astron. Soc. 2019. doi:10.1093/mnras/stz2599. arXiv:1906.06339

    work page doi:10.1093/mnras/stz2599 2019

  45. [45]

    Insights into the origin of halo mass profiles from machine learning

    Lucie-Smith, Luisa and Adhikari, Susmita and Wechsler, Risa H. Insights into the origin of halo mass profiles from machine learning. Mon. Not. Roy. Astron. Soc. 2022. doi:10.1093/mnras/stac1833. arXiv:2205.04474

    work page doi:10.1093/mnras/stac1833 2022

  46. [46]

    and Pontzen, Andrew and Nord, Brian and Thiyagalingam, Jeyan and Piras, Davide

    Lucie-Smith, Luisa and Peiris, Hiranya V. and Pontzen, Andrew and Nord, Brian and Thiyagalingam, Jeyan and Piras, Davide. Discovering the building blocks of dark matter halo density profiles with neural networks. Phys. Rev. D. 2022. doi:10.1103/PhysRevD.105.103533. arXiv:2203.08827

    work page doi:10.1103/physrevd.105.103533 2022

  47. [47]

    Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl

    Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl. arXiv e-prints , keywords =. doi:10.48550/arXiv.2305.01582 , archivePrefix =. 2305.01582 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2305.01582

  48. [48]

    2026 , eprint=

    Why do we do astrophysics? , author=. 2026 , eprint=

    2026

  49. [49]

    Accelerating Large-Scale-Structure data analyses by emulating Boltzmann solvers and Lagrangian Perturbation Theory , volume=

    Arico’, Giovanni and Angulo, Raul and Zennaro, Matteo , year=. Accelerating Large-Scale-Structure data analyses by emulating Boltzmann solvers and Lagrangian Perturbation Theory , volume=. doi:10.12688/openreseurope.14310.2 , journal=

    work page doi:10.12688/openreseurope.14310.2

  50. [50]

    and Dolag, Klaus , year=

    Lehman, Kai and Schuster, Nico and Lucie-Smith, Luisa and Hamaus, Nico and Davies, Christopher T. and Dolag, Klaus , year=. Cosmological inference with cosmic voids and neural network emulators , volume=. doi:10.1051/0004-6361/202554592 , journal=

    work page doi:10.1051/0004-6361/202554592

  51. [51]

    , keywords =

    SPECULATOR: Emulating Stellar Population Synthesis for Fast and Accurate Galaxy Spectra and Photometry. , keywords =. doi:10.3847/1538-4365/ab917f , archivePrefix =. 1911.11778 , primaryClass =

    work page doi:10.3847/1538-4365/ab917f 1911

  52. [52]

    The universality of the virial halo mass function and models for non-universality of other halo definitions

    Despali, Giulia and Giocoli, Carlo and Angulo, Raul E. and Tormen, Giuseppe and Sheth, Ravi K. and Baso, Giacomo and Moscardini, Lauro. The universality of the virial halo mass function and models for non-universality of other halo definitions. Mon. Not. Roy. Astron. Soc. 2016. doi:10.1093/mnras/stv2842. arXiv:1507.05627

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stv2842 2016

  53. [53]

    , title =

    Marcinkevičs, Ričards and Vogt, Julia E. , title =. WIREs Data Mining and Knowledge Discovery , volume =. doi:https://doi.org/10.1002/widm.1493 , url =. https://wires.onlinelibrary.wiley.com/doi/pdf/10.1002/widm.1493 , abstract =

    work page doi:10.1002/widm.1493

  54. [54]

    A comprehensive guide to the physics and usage of PYTHIA 8.3

    Bierlich, Christian and others. A comprehensive guide to the physics and usage of PYTHIA 8.3. SciPost Phys. Codeb. 2022. doi:10.21468/SciPostPhysCodeb.8. arXiv:2203.11601

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.21468/scipostphyscodeb.8 2022

  55. [55]

    MadGraph 5 : Going Beyond

    Alwall, Johan and Herquet, Michel and Maltoni, Fabio and Mattelaer, Olivier and Stelzer, Tim. MadGraph 5 : Going Beyond. JHEP. 2011. doi:10.1007/JHEP06(2011)128. arXiv:1106.0522

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1007/jhep06(2011)128 2011

  56. [56]

    GEANT4—a simulation toolkit

    Agostinelli, S. and others. GEANT4 - A Simulation Toolkit. Nucl. Instrum. Meth. A. 2003. doi:10.1016/S0168-9002(03)01368-8

    work page doi:10.1016/s0168-9002(03)01368-8 2003

  57. [57]

    Menzo, Tony and Roman, Alexander and Gleyzer, Sergei and Matchev, Konstantin and Fleming, George T. and H. HEPTAPOD: Orchestrating High Energy Physics Workflows Towards Autonomous Agency. 2025. arXiv:2512.15867

    arXiv 2025

  58. [58]

    Bakshi, S. D. and others. ArgoLOOM: agentic AI for fundamental physics from quarks to cosmos. 2025. arXiv:2510.02426

    arXiv 2025

  59. [59]

    MadAgents

    Plehn, Tilman and Schiller, Daniel and Schmal, Nikita. MadAgents. 2026. arXiv:2601.21015

    Pith/arXiv arXiv 2026

  60. [60]

    2015 , eprint=

    Distilling the Knowledge in a Neural Network , author=. 2015 , eprint=

    2015

  61. [61]

    2019 , eprint=

    On the Spectral Bias of Neural Networks , author=. 2019 , eprint=

    2019

  62. [62]

    The Pareto Frontier of Resilient Jet Tagging

    Gambhir, Rikab and LeBlanc, Matt and Zhou, Yuanchen. The Pareto Frontier of Resilient Jet Tagging. 39th Annual Conference on Neural Information Processing Systems : Includes Machine Learning and the Physical Sciences (ML4PS). 2025. arXiv:2509.19431

    arXiv 2025

  63. [63]

    Streamlining latent spaces in machine learning using moment pooling

    Gambhir, Rikab and Osathapan, Athis and Thaler, Jesse. Streamlining latent spaces in machine learning using moment pooling. Phys. Rev. D. 2024. doi:10.1103/PhysRevD.110.074020. arXiv:2403.08854

    work page doi:10.1103/physrevd.110.074020 2024

  64. [64]

    2023 , eprint=

    STL: Surprisingly Tricky Logic (for System Validation) , author=. 2023 , eprint=

    2023

  65. [65]

    2019 , eprint=

    Attention is not Explanation , author=. 2019 , eprint=

    2019

  66. [66]

    Proceedings of the National Academy of Science , keywords =

    Cranmer, Kyle and Brehmer, Johann and Louppe, Gilles , year=. The frontier of simulation-based inference , volume=. Proceedings of the National Academy of Sciences , publisher=. doi:10.1073/pnas.1912789117 , number=

    work page doi:10.1073/pnas.1912789117

  67. [67]

    CaloChallenge 2022: a community challenge for fast calorimeter simulation

    Amram, Oz and others. CaloChallenge 2022: a community challenge for fast calorimeter simulation. Rept. Prog. Phys. 2025. doi:10.1088/1361-6633/ae1304. arXiv:2410.21611

    work page doi:10.1088/1361-6633/ae1304 2022

  68. [68]

    and Kanwar, Gurtej and Romero-L \'o pez, Fernando and Shanahan, Phiala E

    Abbott, Ryan and Boyda, Denis and Fu, Yang and Hackett, Daniel C. and Kanwar, Gurtej and Romero-L \'o pez, Fernando and Shanahan, Phiala E. and Urban, Julian M. Variance reduction in lattice QCD observables via normalizing flows. 2026. arXiv:2603.02984

    arXiv 2026

  69. [69]

    and Shanahan, Phiala E

    Abbott, Ryan and Boyda, Denis and Kanwar, Gurtej and Romero-L \'o pez, Fernando and Hackett, Daniel C. and Shanahan, Phiala E. and Urban, Julian M. Progress in Normalizing Flows for 4d Gauge Theories. PoS. 2025. doi:10.22323/1.466.0066. arXiv:2502.00263

    work page doi:10.22323/1.466.0066 2025

  70. [70]

    Machine-learning approaches to accelerating lattice simulations

    Lawrence, Scott. Machine-learning approaches to accelerating lattice simulations. PoS. 2025. doi:10.22323/1.466.0010. arXiv:2502.02670

    work page doi:10.22323/1.466.0010 2025

  71. [71]

    Fast inference of deep neural networks in FPGAs for particle physics

    Duarte, Javier and others. Fast inference of deep neural networks in FPGAs for particle physics. JINST. 2018. doi:10.1088/1748-0221/13/07/P07027. arXiv:1804.06913

    work page doi:10.1088/1748-0221/13/07/p07027 2018

  72. [72]

    and Brandes, T

    Bal, A. and Brandes, T. and Iemmi, F. and Klute, M. and Maier, B. and Mikuni, V. and Arrestad, T. K. Distilling particle knowledge for fast reconstruction at high-energy physics experiments. Mach. Learn. Sci. Tech. 2024. doi:10.1088/2632-2153/ad43b1. arXiv:2311.12551

    work page doi:10.1088/2632-2153/ad43b1 2024

  73. [73]

    hls4ml: A Flexible, Open-Source Platform for Deep Learning Acceleration on Reconfigurable Hardware

    Schulte, Jan-Frederik and others. hls4ml: A Flexible, Open-Source Platform for Deep Learning Acceleration on Reconfigurable Hardware. 2025. arXiv:2512.01463

    arXiv 2025

  74. [74]

    (Eds.), Contributions to the Theory of Games, Volume II

    Shapley, Lloyd S. A Value for n -Person Games. Contributions to the Theory of Games. 1953. doi:10.1515/9781400881970-018

    work page doi:10.1515/9781400881970-018 1953

  75. [75]

    An operational definition of quark and gluon jets

    Komiske, Patrick T. and Metodiev, Eric M. and Thaler, Jesse. An operational definition of quark and gluon jets. JHEP. 2018. doi:10.1007/JHEP11(2018)059. arXiv:1809.01140

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1007/jhep11(2018)059 2018

  76. [76]

    Journal of High Energy Physics , year =

    Komiske, Patrick T. and Metodiev, Eric M. and Thaler, Jesse. Energy Flow Networks: Deep Sets for Particle Jets. JHEP. 2019. doi:10.1007/JHEP01(2019)121. arXiv:1810.05165

    work page doi:10.1007/jhep01(2019)121 2019

  77. [77]

    and Gambhir, Rikab and Tasissa, Abiy and Thaler, Jesse

    Ba, Demba and Dogra, Akshunna S. and Gambhir, Rikab and Tasissa, Abiy and Thaler, Jesse. SHAPER: can you hear the shape of a jet?. JHEP. 2023. doi:10.1007/JHEP06(2023)195. arXiv:2302.12266

    work page doi:10.1007/jhep06(2023)195 2023

  78. [78]

    and Youssef, Ahmed and Zupan, Jure

    Bierlich, Christian and Ilten, Phil and Menzo, Tony and Mrenna, Stephen and Szewc, Manuel and Wilkinson, Michael K. and Youssef, Ahmed and Zupan, Jure. Describing hadronization via histories and observables for Monte-Carlo event reweighting. SciPost Phys. 2025. doi:10.21468/SciPostPhys.18.2.054. arXiv:2410.06342

    work page doi:10.21468/scipostphys.18.2.054 2025

  79. [79]

    Lorentz-Equivariant Geometric Algebra Transformers for High-Energy Physics

    Spinner, Jonas and Bres \'o , Victor and de Haan, Pim and Plehn, Tilman and Thaler, Jesse and Brehmer, Johann. Lorentz-Equivariant Geometric Algebra Transformers for High-Energy Physics. 38th conference on Neural Information Processing Systems. 2024. arXiv:2405.14806

    arXiv 2024

  80. [80]

    SEAL - A Symmetry EncourAging Loss for High Energy Physics

    Hebbar, Pradyun and Madula, Thandikire and Mikuni, Vinicius and Nachman, Benjamin and Outmezguine, Nadav and Savoray, Inbar. SEAL - A Symmetry EncourAging Loss for High Energy Physics. 2025. arXiv:2511.01982

    arXiv 2025

Showing first 80 references.