pith. sign in

arxiv: 2606.10960 · v1 · pith:TIBM7T7Pnew · submitted 2026-06-09 · ✦ hep-ex

Understanding Energy Dependent Hadronic Calorimeter Response from a Machine Learning Perspective

Shuai-Chun Wang , Huang-Ran Shen , Wan-Bing He , Wei-Hu Ma , Peng-Jie Li , De-Qing Fang , Yu-Gang Ma This is my paper

Pith reviewed 2026-06-27 10:52 UTC · model grok-4.3

classification ✦ hep-ex
keywords hadronic calorimetersmachine learningenergy reconstructionshower topologydual readoutintrinsic resolutionsampling fraction
0
0 comments X

The pith

Machine learning reconstruction using all signal channels and full 3D shower data reaches an intrinsic hadronic energy resolution of (10.8 ± 0.3)% / √(E/GeV).

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

The paper examines machine-learning methods for reconstructing the energy deposited by hadronic showers in calorimeters, testing the value of different signal channels and spatial information up to 10 GeV. It establishes that the intrinsic resolution limit is approximately (10.8 ± 0.3)% / √(E/GeV) when scintillation light, Cherenkov light, charged-particle signals, and the complete three-dimensional shower topology are all available. Traditional signal summing yields much worse resolution, but machine-learning models improve it substantially even when only 10% of the energy is sampled, moving the figure from (57.6 ± 3.7)% / √(E/GeV) to (34.1 ± 2.8)% / √(E/GeV). These gains are presented as motivation for pairing high-granularity dual-readout calorimeters with learned reconstruction in future experiments. All results rest on ideal simulations that omit detector effects.

Core claim

When every available signal channel and the full three-dimensional shower topology are supplied to a machine-learning model, the intrinsic energy resolution of hadronic showers reaches (10.8 ± 0.3)% / √(E/GeV). The same models raise resolution from (57.6 ± 3.7)% / √(E/GeV) to (34.1 ± 2.8)% / √(E/GeV) even at a 10% sampling fraction, demonstrating that multi-channel and spatially detailed inputs carry decisive information beyond simple summation.

What carries the argument

Machine-learning models trained on multi-channel signals (scintillation light, Cherenkov light, charged particles) together with the complete three-dimensional shower topology for hadronic energy reconstruction.

If this is right

  • High-granularity dual-readout calorimeter designs become more valuable when paired with machine-learning reconstruction.
  • Detailed spatial shower features contribute measurably to energy resolution beyond integrated signal sums.
  • Multi-channel information (scintillation, Cherenkov, and charged-particle channels) is required to approach the intrinsic resolution limit.
  • Energy resolution can be improved at fixed sampling fraction by replacing traditional summation with learned reconstruction.

Where Pith is reading between the lines

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

  • Calorimeter designs could trade some sampling fraction for finer granularity and more readout channels if machine learning is used downstream.
  • The reported numbers set a benchmark that full-detector simulations with realistic effects must be compared against before claims about future experiments can be assessed.
  • Similar machine-learning approaches may apply to other reconstruction tasks that currently rely on simple signal summation in particle detectors.

Load-bearing premise

The simulations used to train and test the machine-learning models contain no detector effects.

What would settle it

Running the same machine-learning models on simulated events that include realistic detector noise, inefficiencies, and response non-uniformities and comparing the resulting resolution numbers to the ideal-case values.

read the original abstract

To meet the precision requirements of future high-energy physics experiments, improving the energy resolution of hadronic calorimeters remains a critical challenge. This work presents a systematic investigation of hadronic energy reconstruction using machine learning, highlighting the roles of various signal channels, including scintillation light, Cherenkov light, charged particles, and the full three-dimensional topology of hadronic showers in the energy range up to 10 GeV. Throughout this study, detector effects are not taken into account. Under these conditions, the intrinsic resolution of hadronic showers reaches approximately $(10.8\pm0.3)\% / \sqrt{E/GeV}$ when all signal channels and the full 3D shower information are fully utilized. Compared with the traditional signal-summing approach, machine-learning-based reconstruction can significantly improve energy resolution, even under a limited sampling fraction of 10\%, enhancing it from $(57.6\pm3.7)\%/\sqrt{E/GeV}$ to $(34.1\pm2.8)\%/\sqrt{E/GeV}$. These results highlight the critical importance of both multi-channel information and detailed spatial shower features in hadronic energy reconstruction, and demonstrate the substantial potential of combining high-granularity and dual-readout calorimeter designs with machine-learning-based reconstruction techniques for future experiments.

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 / 0 minor

Summary. The manuscript systematically explores machine-learning-based reconstruction of hadronic shower energies up to 10 GeV using multiple signal channels (scintillation, Cherenkov, charged-particle) and full 3D topology information. Simulations are performed without detector effects. The work reports an intrinsic resolution of (10.8±0.3)%/√(E/GeV) when all information is utilized and shows that ML improves resolution relative to simple signal summing, for example from (57.6±3.7)%/√(E/GeV) to (34.1±2.8)%/√(E/GeV) at a 10% sampling fraction.

Significance. If the results hold, the paper supplies a useful benchmark for the ultimate limits of hadronic energy resolution and quantifies the gains available from combining multi-channel readout, high granularity, and ML techniques. This provides concrete guidance for the design of future precision calorimeters. The explicit statement that detector effects are omitted is noted, so the reported numbers serve as an idealized reference rather than a direct prediction for real detectors.

major comments (1)
  1. [Abstract] Abstract: The claim that the results 'demonstrate the substantial potential of combining high-granularity and dual-readout calorimeter designs with machine-learning-based reconstruction techniques for future experiments' is load-bearing for the paper's broader impact. This statement rests on simulations that explicitly exclude detector effects (as stated: 'Throughout this study, detector effects are not taken into account'). Because real effects such as noise, finite light yield, and position-dependent response would affect both the summing baseline and the ML performance, the manuscript should either qualify the applicability statements or include a discussion/sensitivity study showing how the quoted factor-of-~1.7 improvement would change under realistic conditions.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the positive assessment of the work as an idealized benchmark and for the constructive major comment. We address it below by qualifying the applicability statements.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The claim that the results 'demonstrate the substantial potential of combining high-granularity and dual-readout calorimeter designs with machine-learning-based reconstruction techniques for future experiments' is load-bearing for the paper's broader impact. This statement rests on simulations that explicitly exclude detector effects (as stated: 'Throughout this study, detector effects are not taken into account'). Because real effects such as noise, finite light yield, and position-dependent response would affect both the summing baseline and the ML performance, the manuscript should either qualify the applicability statements or include a discussion/sensitivity study showing how the quoted factor-of-~1.7 improvement would change under realistic conditions.

    Authors: We agree that the abstract claim should be qualified to reflect the idealized conditions. The results are presented as a benchmark for intrinsic limits and relative gains from multi-channel information plus ML versus summing, when detector effects are absent. We will revise the abstract (and conclusions) to state explicitly that the quoted resolutions and improvement factor apply under these idealized conditions and serve as an upper-bound reference for future designs. A sensitivity study to realistic effects (noise, finite yield, etc.) is outside the present scope, as it would require new, extensive simulations; the manuscript already states the omission of detector effects throughout. This qualification addresses the concern on applicability without overstating the results. revision: partial

Circularity Check

0 steps flagged

No significant circularity; results are direct ML outputs on idealized simulations

full rationale

The paper applies machine learning models to simulated hadronic shower data under explicitly idealized conditions (no detector effects). Reported resolutions such as (10.8±0.3)%/√E and ML improvements from (57.6±3.7)% to (34.1±2.8)% at 10% sampling are obtained by training and evaluating on these simulations. No self-definitional loops, fitted inputs renamed as predictions, load-bearing self-citations, uniqueness theorems, or ansatzes smuggled via citation appear in the abstract or described methodology. The derivation chain is self-contained as a standard simulation-plus-ML workflow without reduction to its own inputs by construction.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The paper's claims depend on the validity of the simulation model and the ML training process, which involve assumptions about data generation and model optimization not detailed here.

free parameters (1)
  • Machine learning model parameters
    ML models have many tunable parameters fitted during training, but specifics not provided in abstract.
axioms (1)
  • domain assumption The simulation without detector effects accurately represents the intrinsic hadronic shower response.
    Explicitly stated that detector effects are not taken into account.

pith-pipeline@v0.9.1-grok · 5782 in / 1368 out tokens · 39295 ms · 2026-06-27T10:52:14.107460+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

38 extracted references

  1. [1]

    Fernow,Introduction to Experimental Particle Physics, Cambridge University Press (2023)

    R.C. Fernow,Introduction to Experimental Particle Physics, Cambridge University Press (2023)

    2023

  2. [2]

    Fabjan and F

    C.W. Fabjan and F. Gianotti,Calorimetry for particle physics,Rev. Mod. Phys.75(2003) 1243

    2003

  3. [3]

    Accardi, J.L

    A. Accardi, J.L. Albacete, M. Anselmino, N. Armesto, E.C. Aschenauer, A. Bacchetta et al., Electron-ion collider: The next qcd frontier,The European Physical Journal A52(2016) 268

    2016

  4. [4]

    M.Benedikt,F.Zimmermann,B.Auchmann,W.Bartmann,J.P.Burnet,C.Carlietal.,Futurecircular collider feasibility study report,The European Physical Journal Special Topics234(2025) 5113

    2025

  5. [5]

    Brau, J.A

    J.E. Brau, J.A. Jaros and H. Ma,Advances in calorimetry,Ann. Rev. Nucl. Part. Sci.60(2010) 615

    2010

  6. [6]

    Wang, S.-C

    X.-Y. Wang, S.-C. Wang, W.-B. He and Y.-G. Ma,Cosmic ray test of shashlik electromagnetic calorimeter modules for nica-mpd,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment1069(2024) 169857

    2024

  7. [7]

    Wellisch,Hadronic shower models in geant4 — the frameworks,Computer Physics Communications140(2001) 65

    J. Wellisch,Hadronic shower models in geant4 — the frameworks,Computer Physics Communications140(2001) 65

    2001

  8. [8]

    D.E. Groom,Energy flow in a hadronic cascade: Application to hadron calorimetry,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment572(2007) 633

    2007

  9. [9]

    Z. Wang, R. Yuan, H.-Q. Liu, J. Chen, X. Chen, K.S. Khaw et al.,Design of hadronic calorimeter for darkshine experiment,Nuclear Science and Techniques35(2024) 148

    2024

  10. [10]

    Zhao, Q.-B

    Z.-Y. Zhao, Q.-B. Liu, J.-Y. Chen, J. Chen, J.-F. Chen, X. Chen et al.,Design of a lyso crystal electromagnetic calorimeter for dark photon detection in the darkshine experiment,Nuclear Science and Techniques36(2025) 41

    2025

  11. [11]

    E. Hilger,The zeus uranium-scintillator calorimeter for hera,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment257 (1987) 488

    1987

  12. [12]

    Celletti, A

    F. Celletti, A. Marchionni, P. Spillantini, Y. Kamyshkov, V. Pojidaev, M. Cerrada et al.,Test of prototype hadron calorimeter for the l3 experiment,Nuclear Instruments and Methods in Physics Research225(1984) 493

    1984

  13. [13]

    S. Lee, M. Livan and R. Wigmans,Dual-readout calorimetry,Rev. Mod. Phys.90(2018) 025002

    2018

  14. [14]

    Wigmans,25 years of dual-readout calorimetry,Instruments6(2022)

    R. Wigmans,25 years of dual-readout calorimetry,Instruments6(2022)

    2022

  15. [15]

    A. Pareti and on behalf of the IDEA dual-readout calorimeter group,Status of dual-readout calorimetry for future high-energy physics experiments,Journal of Instrumentation19(2024) C04029. – 25 –

    2024

  16. [16]

    R.Wigmans,Thedual-readoutapproachtocalorimetry,NuclearInstrumentsandMethodsinPhysics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment732(2013) 475

    2013

  17. [17]

    Amendola and on behalf of the CMS collaboration,An overview of the cms high-granularity calorimeter,Journal of Instrumentation20(2025) C01031

    C. Amendola and on behalf of the CMS collaboration,An overview of the cms high-granularity calorimeter,Journal of Instrumentation20(2025) C01031

    2025

  18. [18]

    Zubankov, S

    A. Zubankov, S. Afanasiev, M. Golubeva, F. Guber, A. Ivashkin, N. Karpushkin et al.,Performance study of the highly granular neutron detector prototype in the bm@n experiment,Nuclear Science and Techniques36(2025) 226

    2025

  19. [19]

    Thomson,Particle flow calorimetry,Journal of Physics: Conference Series293(2011) 012021

    M.A. Thomson,Particle flow calorimetry,Journal of Physics: Conference Series293(2011) 012021

    2011

  20. [20]

    M. Thomson,Particle flow calorimetry and the pandorapfa algorithm,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment611(2009) 25

    2009

  21. [21]

    Yu, X.-Y

    X.-Z. Yu, X.-Y. Wang, W.-H. Ma, S.-H. Fu, P.-F. Sun, J.-X. Song et al.,Production and test of sphenix w/scifiber electromagnetic calorimeter blocks in china,Nuclear Science and Techniques35(2024) 145

    2024

  22. [22]

    Wei and H

    Y. Wei and H. Wang,Review of the performance of the cms hadron calorimeter,Particles9(2026)

    2026

  23. [23]

    Solovyanov and (on behalf ofAtlas Tile Community),The atlas tile calorimeter: Commissioning and preparation for collisions,Journal of Physics: Conference Series160(2009) 012053

    O. Solovyanov and (on behalf ofAtlas Tile Community),The atlas tile calorimeter: Commissioning and preparation for collisions,Journal of Physics: Conference Series160(2009) 012053

    2009

  24. [24]

    N. van der Kolk,Focal: A highly granular digital calorimeter,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment958 (2020) 162059

    2020

  25. [25]

    Abdul Khalek, A

    R. Abdul Khalek, A. Accardi, J. Adam, D. Adamiak, W. Akers, M. Albaladejo et al.,Science requirements and detector concepts for the electron-ion collider: Eic yellow report,Nuclear Physics A1026(2022) 122447

    2022

  26. [26]

    Achasov, X.C

    M. Achasov, X.C. Ai, L.P. An, R. Aliberti, Q. An, X.Z. Bai et al.,Stcf conceptual design report (volume 1): Physics & detector,Frontiers of Physics19(2023)

    2023

  27. [27]

    S. Lai, J. Utehs, A. Wilhahn, O. Bach, E. Brianne, A. Ebrahimi et al.,Software compensation for highly granular calorimeters using machine learning,Journal of Instrumentation19(2024) P04037

    2024

  28. [28]

    S. Lai, J. Utehs, A. Wilhahn, M. Fouz, O. Bach, E. Brianne et al.,Shower separation in five dimensions for highly granular calorimeters using machine learning,Journal of Instrumentation19 (2024) P10027

    2024

  29. [29]

    Aamir, G

    M. Aamir, G. Adamov, T. Adams, C. Adloff, S. Afanasiev, C. Agrawal et al.,Using graph neural networks to reconstruct charged pion showers in the cms high granularity calorimeter,Journal of Instrumentation19(2024) P11025

    2024

  30. [30]

    J.-L. Fei, A. Yuan, K. Wei, L. Sun and J.-K. Wang,Layered reconstruction framework for longitudinal segmented electromagnetic calorimeter,Nuclear Science and Techniques36(2025) 166

    2025

  31. [31]

    He, Y.-G

    W.-B. He, Y.-G. Ma, L.-G. Pang, H.-C. Song and K. Zhou,High-energy nuclear physics meets machine learning,Nuclear Science and Techniques34(2023) 88

    2023

  32. [32]

    C.L. Liao, Z. Quan, Y.W. Dong, M. Xu., C. Zhang, J.J. Wang et al.,Application of machine learning method for energy reconstruction on space based high granularity calorimeter,Experimental Astronomy58(2024) 12. – 26 –

    2024

  33. [33]

    Giannelli and R

    M.F. Giannelli and R. Zhang,Caloshowergan, a generative adversarial network model for fast calorimeter shower simulation,The European Physical Journal Plus139(2024) 597

    2024

  34. [34]

    Wu, X.-Y

    H.-K. Wu, X.-Y. Wang, Y.-M. Wang, Y.-J. Wang, D.-Q. Fang, W.-B. He et al.,Fudan multi-purpose active target time projection chamber (fmeta-tpc) for photonuclear reaction experiments,Nuclear Science and Techniques35(2024) 200

    2024

  35. [35]

    Akchurin, L

    N. Akchurin, L. Berntzon, A. Cardini, R. Ferrari, G. Gaudio, J. Hauptman et al.,Dual-readout calorimetry with lead tungstate crystals,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment584(2008) 273

    2008

  36. [36]

    Melikyan, I

    Y.A. Melikyan, I. Bearden, V. Buchakchiev, S. Jia, V. Kozhuharov and I. Møller,Characterization of the large-size ndl eqr20 silicon photomultipliers,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment1078(2025) 170604

    2025

  37. [37]

    Collaboration, G

    T.A. Collaboration, G. Aad, E. Abat, J. Abdallah, A.A. Abdelalim, A. Abdesselam et al.,The atlas experiment at the cern large hadron collider,Journal of Instrumentation3(2008) S08003

    2008

  38. [38]

    Bilki, J

    B. Bilki, J. Repond, J. Schlereth, L. Xia, Z. Deng, Y. Li et al.,Testing hadronic interaction models using a highly granular silicon–tungsten calorimeter,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment794(2015) 240. – 27 –

    2015