pith. sign in

arxiv: 2512.01612 · v1 · submitted 2025-12-01 · ⚛️ physics.ins-det

Neural network-based deconvolution for GeV-Scale Gamma-Ray Spectroscopy

Zhuofan Zhang , Mingxuan Wei , Kyle Fleck , Jun Liu , Xinjian Tan , Gianluca Sarri , Wenchao Yan This is my paper

Pith reviewed 2026-05-17 03:14 UTC · model grok-4.3

classification ⚛️ physics.ins-det
keywords gamma-ray spectroscopyneural networkdeconvolutionGeV scaleMonte CarloU-Netdenoising autoencoderpositron spectra
0
0 comments X

The pith

Neural network deconvolution reconstructs precise GeV-scale gamma-ray spectra from measured positrons.

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

This paper proposes a machine learning approach to overcome challenges in high-energy gamma-ray spectroscopy between multi-MeV and GeV energies. It optimizes a spectrometer design via Monte Carlo simulations to maximize positron yield and minimize noise. A denoising autoencoder first reduces statistical noise in the measured positron spectra, and then a U-Net architecture performs the deconvolution to recover the incident gamma-ray spectrum. This method offers a new way to diagnose gamma rays in experiments involving strong-field quantum electrodynamics and high-energy photon sources.

Core claim

The authors present a gamma-ray spectrometer design optimized through Monte Carlo simulations for maximum positron yield and minimal noise. They introduce a two-stage neural network framework consisting of a denoising autoencoder to suppress statistical noise in measured positron spectra and a U-Net to solve the ill-posed inverse problem of reconstructing the incident gamma spectra from those measurements.

What carries the argument

Two-stage neural network: a denoising autoencoder followed by a U-Net, structured around the spectrometer's positron output to handle noise suppression and spectral inversion.

If this is right

  • Improved precision in spectral reconstruction for multi-MeV to GeV gamma rays.
  • New diagnostic tool for strong-field QED experiments and high-energy photon sources.
  • Reduction of statistical noise effects in gamma-ray measurements.
  • Potential for application in laboratory astrophysics and high-energy-density science.

Where Pith is reading between the lines

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

  • This methodology could be adapted for other types of radiation spectroscopy where inverse problems arise.
  • Success would encourage integration of similar ML techniques in real-time experimental diagnostics.
  • Further testing on actual experimental data would validate generalization from simulations.

Load-bearing premise

Neural networks trained solely on Monte Carlo-simulated positron spectra will generalize accurately to real experimental data without major domain shifts or unaccounted systematics.

What would settle it

Direct comparison of the neural network-reconstructed gamma-ray spectra with those from conventional methods or known input spectra in a controlled experiment, checking for systematic deviations.

Figures

Figures reproduced from arXiv: 2512.01612 by Gianluca Sarri, Jun Liu, Kyle Fleck, Mingxuan Wei, Wenchao Yan, Xinjian Tan, Zhuofan Zhang.

Figure 1
Figure 1. Figure 1: FIG. 1. Top view schematic of the gamma-ray [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. (a) The normalized spectra of positrons produced by a 1 GeV monochromatic gamma-ray beam after [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Time-integrated fluence distribution for (a) photons, (b) positrons, and (c) electrons resulting from the [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Schematic of the overall reconstruction [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Multiple response functions were generated by simulating the monochromatic gamma-ray response using a [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Schematic of machine learning architecture: (a) Denoising autoencoder (DAE) with symmetric [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. The reconstructed photon spectrum obtained by applying machine learning algorithms to the dataset(For [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Performance of different models: (a) RMSE; (b) PSNR; (c) SSIM for the SIRT algorithm, FCNN+MLE [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
read the original abstract

High-energy gamma-ray spectroscopy is crucial for studying and advancing the application of high-energy photons in areas like strong-field physics, high-energy-density science, and laboratory astrophysics. However, high-energy gamma-ray spectroscopy in the multi-MeV to GeV range faces significant challenges in precise spectral reconstruction. This study presents a machine learning-based inversion approach that combines a spectrometer design with advanced deconvolution algorithms. We develop a gamma-ray spectrometer optimized through Monte Carlo simulations for maximum positron yield and minimal noise. A two-stage neural network framework is proposed based on the structure of the spectrometer: a denoising autoencoder suppresses statistical noise in measured positron spectra, while a U-Net architecture solves the ill-posed inverse problem to reconstruct incident gamma spectra. This approach establishes a new methodology for gamma-ray diagnostics in strong-field QED experiments and high-energy photon sources.

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

2 major / 2 minor

Summary. The paper claims to introduce a machine learning-based deconvolution method for reconstructing GeV-scale gamma-ray spectra from measured positron distributions. It combines a Monte Carlo-optimized spectrometer design with a two-stage neural network consisting of a denoising autoencoder to suppress noise in positron spectra followed by a U-Net to solve the ill-posed inverse problem for incident gamma spectra, targeting applications in strong-field QED experiments and high-energy photon sources.

Significance. If the approach generalizes beyond simulations, it could establish a practical new diagnostic tool for high-energy gamma-ray spectroscopy where traditional unfolding methods struggle with ill-posedness and noise. The simulation-driven spectrometer optimization represents a constructive element of the work.

major comments (2)
  1. [Abstract and methodology] Abstract and methodology description: The central claim that the two-stage NN solves the ill-posed inverse problem and establishes a new methodology is not supported by any quantitative validation, such as reconstruction accuracy metrics, direct comparisons to conventional unfolding techniques, error bars on reconstructed spectra, or robustness tests against realistic noise models.
  2. [Training and evaluation framework] Training and evaluation framework: All training data derive from Monte Carlo simulations assuming idealized detector response, perfect tracking, and known noise statistics, with no tests for domain shift, injected systematics (e.g., energy-dependent efficiency, scattering tails, pile-up, or calibration drift), or performance on experimental data; this directly undermines the generalization assumption required for the claimed diagnostic utility.
minor comments (2)
  1. [Neural network architecture] Provide explicit details on network hyperparameters, loss functions, and input/output dimensions to improve reproducibility of the denoising autoencoder and U-Net stages.
  2. [Spectrometer design] Clarify the exact definition of the spectrometer geometry parameters and how they were optimized in the Monte Carlo simulations.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive and detailed report. We agree that strengthening the quantitative validation and clarifying the simulation-based scope will improve the manuscript. We have revised the paper to incorporate additional metrics, comparisons, and robustness tests while maintaining the focus on the proposed methodology as a simulation-driven proof of concept.

read point-by-point responses

  1. Referee: [Abstract and methodology] Abstract and methodology description: The central claim that the two-stage NN solves the ill-posed inverse problem and establishes a new methodology is not supported by any quantitative validation, such as reconstruction accuracy metrics, direct comparisons to conventional unfolding techniques, error bars on reconstructed spectra, or robustness tests against realistic noise models.

    Authors: We acknowledge the need for explicit quantitative support. In the revised manuscript we have added a dedicated 'Performance Evaluation' subsection that reports reconstruction accuracy via mean squared error, spectral fidelity (overlap integral), and chi-squared metrics across a range of input spectra. Direct comparisons are now included against conventional unfolding methods (singular value decomposition and iterative Bayesian unfolding), with quantitative tables showing superior performance of the two-stage network in terms of bias and variance. Error bars are derived from 100 independent Monte Carlo realizations with different random seeds, and robustness is demonstrated by injecting Poisson and Gaussian noise at varying signal-to-noise ratios, reporting the resulting degradation in reconstruction quality. These additions directly support the central claims with numerical evidence. revision: yes

  2. Referee: [Training and evaluation framework] Training and evaluation framework: All training data derive from Monte Carlo simulations assuming idealized detector response, perfect tracking, and known noise statistics, with no tests for domain shift, injected systematics (e.g., energy-dependent efficiency, scattering tails, pile-up, or calibration drift), or performance on experimental data; this directly undermines the generalization assumption required for the claimed diagnostic utility.

    Authors: We agree that the study relies exclusively on Monte Carlo simulations with idealized assumptions, which is standard for developing new diagnostic concepts prior to hardware realization. The revised manuscript now includes explicit tests for domain shift by training on nominal GEANT4 parameters and evaluating on perturbed sets with varied energy-dependent efficiencies and added scattering tails. Injected systematics such as pile-up and calibration drift have been simulated and the network's degradation quantified. However, because the spectrometer design is optimized but not yet constructed, no experimental data exist for direct validation. We have expanded the 'Limitations and Outlook' section to discuss these points and outline planned beam-test validation. revision: partial

standing simulated objections not resolved
  • Direct performance validation on experimental data, as no measured data are available for the proposed Monte Carlo-optimized spectrometer design.

Circularity Check

0 steps flagged

No circularity: proposed simulation-trained NN deconvolution is self-contained

full rationale

The paper proposes a spectrometer design optimized via Monte Carlo simulations followed by a two-stage neural network (denoising autoencoder plus U-Net) to invert measured positron spectra into incident gamma spectra. No load-bearing step reduces by construction to a fitted parameter, self-definition, or self-citation chain; the architecture and training procedure are presented as an independent methodology whose performance is evaluated on held-out simulations rather than being forced by the target result itself.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim depends on the fidelity of Monte Carlo simulations for both spectrometer optimization and generation of training labels, plus the assumption that the chosen NN architectures can solve the ill-posed inverse problem without overfitting to simulation artifacts.

free parameters (2)
  • Neural network architecture hyperparameters
    Layer counts, filter sizes, learning rates, and training epochs for the autoencoder and U-Net are not reported and must be chosen or fitted during development.
  • Spectrometer geometry parameters
    Dimensions and materials were optimized via Monte Carlo to maximize positron yield; these choices directly affect the response matrix used for training.
axioms (2)
  • domain assumption Monte Carlo simulations accurately capture the spectrometer response and noise statistics for training data generation
    Invoked to justify both design optimization and supervised training of the networks.
  • domain assumption The inverse problem of recovering gamma spectra from positron spectra is solvable by the chosen U-Net architecture
    Central to the claim that the second stage reconstructs incident spectra.

pith-pipeline@v0.9.0 · 5455 in / 1386 out tokens · 34307 ms · 2026-05-17T03:14:25.501661+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

54 extracted references · 54 canonical work pages · 4 internal anchors

  1. [1]

    double-differential

    is a key factor in optimizing spectrometers. MC sim- ulations can be used to analyze the relationship between the total positron yield and tungsten target thickness under different energy pencil-shaped gamma-ray inci- dence conditions. In our simulation, a 10 8 pencil-shaped monochromatic gamma ray beam with an energy of 1 GeV was simulated penetrating tu...

    work page 2025

  2. [2]

    Betatron Radia- tion and Bremsstrahlung in the Interaction of Intense Laser Pulse with Solid Target,

    L. Li, R. Li, L. Ju, K. Jiang, M. Yu, T. Huang, H. Zhang, S. Wu, B. Qiao, C. Zhou, and X. He, “Betatron Radia- tion and Bremsstrahlung in the Interaction of Intense Laser Pulse with Solid Target,”Applied Sciences, vol. 13, p. 6632, May 2023

    work page 2023

  3. [3]

    Bremsstrahlung hard x-ray source driven by an electron beam from a self-modulated laser wakefield accelerator,

    N. Lemos, F. Albert, J. L. Shaw, D. Papp, R. Polanek, P. King, A. L. Milder, K. A. Marsh, B. B. Pollock, B. M. Hegelich, J. D. Moody, J. Park, R. Tommasini, G. J. Williams, and C. Joshi, “Bremsstrahlung hard x-ray source driven by an electron beam from a self-modulated laser wakefield accelerator,”Plasma Physics and Con- trolled Fusion, vol. 60, no. 5, p....

    work page 2018

  4. [4]

    Charge-neutral, GeV-scale electron- positron pair beams produced using bremsstrahlung gamma rays,

    Y. Noh, J. Song, M. Mirzaie, C. I. Hojbota, H.-i. Kim, S. Lee, J. Won, H. Song, C. Song, C.-M. Ryu, C. H. Nam, and W. Bang, “Charge-neutral, GeV-scale electron- positron pair beams produced using bremsstrahlung gamma rays,”Commun Phys, vol. 7, pp. 1–10, Jan. 2024

    work page 2024

  5. [5]

    Ultrabroad-band x-ray source using a picosecond, laser-driven plasma accelerator,

    N. Lemos, P. M. King, D. Rusby, I. Pagano, M. Sinclair, K. A. Marsh, J. L. Shaw, A. L. Milder, A. Pak, B. B. Pollock, M. Aufderheide, F. V. Hartemann, S. Q. Wu, Y. Hwang, B. M. Hegelich, J. Moody, P. Michel, C. Joshi, and F. Albert, “Ultrabroad-band x-ray source using a picosecond, laser-driven plasma accelerator,”Phys. Rev. Research, vol. 6, p. L032022, ...

    work page 2024

  6. [6]

    Ultrahigh Brilliance Multi-MeVγ-Ray Beams from Nonlinear Relativistic Thomson Scattering,

    G. Sarri, D. J. Corvan, W. Schumaker, J. M. Cole, A. Di Piazza, H. Ahmed, C. Harvey, C. H. Kei- tel, K. Krushelnick, S. P. D. Mangles, Z. Najmudin, D. Symes, A. G. R. Thomas, M. Yeung, Z. Zhao, and M. Zepf, “Ultrahigh Brilliance Multi-MeVγ-Ray Beams from Nonlinear Relativistic Thomson Scattering,”Phys. 12 Rev. Lett., vol. 113, Nov. 2014

    work page 2014

  7. [7]

    High-order multiphoton Thomson scattering,

    W. Yan, C. Fruhling, G. Golovin, D. Haden, J. Luo, P. Zhang, B. Zhao, J. Zhang, C. Liu, M. Chen, S. Chen, S. Banerjee, and D. Umstadter, “High-order multiphoton Thomson scattering,”Nature Photon, vol. 11, pp. 514– 520, Aug. 2017

    work page 2017

  8. [8]

    Gamma-ray vortex burst in nonlinear thomson scattering with refo- cusing spiral plasma mirror,

    W. Zhou, W. Yan, J. Wang, and L. Chen, “Gamma-ray vortex burst in nonlinear thomson scattering with refo- cusing spiral plasma mirror,”Ultrafast Science, vol. 3, p. 0005, 2023

    work page 2023

  9. [9]

    All-optical nonlinear Compton scattering performed with a multi-petawatt laser,

    M. Mirzaie, C. I. Hojbota, D. Y. Kim, V. B. Pathak, T. G. Pak, C. M. Kim, H. W. Lee, J. W. Yoon, S. K. Lee, Y. J. Rhee, M. Vranic, ´O. Amaro, K. Y. Kim, J. H. Sung, and C. H. Nam, “All-optical nonlinear Compton scattering performed with a multi-petawatt laser,”Nat. Photon., Oct. 2024

    work page 2024

  10. [10]

    A tld-based few-channel spectrometer for x ray fields with high fluence rates,

    R. Nolte, R. Behrens, M. Schn¨ urer, A. Rousse, and P. Ambrosi, “A tld-based few-channel spectrometer for x ray fields with high fluence rates,”Radiation protec- tion dosimetry, vol. 84, no. 1-4, pp. 367–370, 1999

    work page 1999

  11. [11]

    A tld-based few-channel spectrometer for mixed photon, electron, and ion fields with high fluence rates,

    R. Behrens and P. Ambrosi, “A tld-based few-channel spectrometer for mixed photon, electron, and ion fields with high fluence rates,”Radiation protection dosimetry, vol. 101, no. 1-4, pp. 73–76, 2002

    work page 2002

  12. [12]

    A bremsstrahlung spectrometer using k-edge and dif- ferential filters with image plate dosimeters,

    C. Chen, J. King, M. Key, K. Akli, F. Beg, H. Chen, R. Freeman, A. Mackinnon, A. MacPhee, P. Patel,et al., “A bremsstrahlung spectrometer using k-edge and dif- ferential filters with image plate dosimeters,”Review of Scientific Instruments, vol. 79, no. 10, 2008

    work page 2008

  13. [13]

    A spectrometer for pulsed and continuous photon radiation,

    R. Behrens, “A spectrometer for pulsed and continuous photon radiation,”Journal of Instrumentation, vol. 4, no. 03, p. P03027, 2009

    work page 2009

  14. [14]

    Measurements of high-energy radia- tion generation from laser-wakefield accelerated electron beams,

    W. Schumaker, G. Sarri, M. Vargas, Z. Zhao, K. Behm, V. Chvykov, B. Dromey, B. Hou, A. Maksimchuk, J. Nees,et al., “Measurements of high-energy radia- tion generation from laser-wakefield accelerated electron beams,”Physics of Plasmas, vol. 21, no. 5, 2014

    work page 2014

  15. [15]

    Pair spectrometer hodoscope for hall d at jefferson lab,

    F. Barbosa, C. Hutton, A. Sitnikov, A. Somov, S. Somov, and I. Tolstukhin, “Pair spectrometer hodoscope for hall d at jefferson lab,”Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 795, pp. 376– 380, 2015

    work page 2015

  16. [16]

    Compact high energy x-ray spectrometer based on forward Compton scattering for high intensity laser plasma experiments,

    S. Singh, R. Versaci, A. L. Garcia, L. Morejon, A. Fer- rari, M. Molodtsova, R. Schwengner, D. Kumar, and T. Cowan, “Compact high energy x-ray spectrometer based on forward Compton scattering for high intensity laser plasma experiments,”Rev. Sci. Instrum., vol. 89, no. 8, 2018

    work page 2018

  17. [17]

    Design of a com- pact spectrometer for high-flux MeV gamma-ray beams,

    D. J. Corvan, G. Sarri, and M. Zepf, “Design of a com- pact spectrometer for high-flux MeV gamma-ray beams,” Rev. Sci. Instrum., vol. 85, no. 6, 2014

    work page 2014

  18. [18]

    Compact broadband high- resolution Compton spectroscopy for laser-driven high- flux gamma rays,

    Z.-C. Zhang, T. Yang, G.-Y. Hu, M.-T. Li, W. Luo, N. An, and J. Zheng, “Compact broadband high- resolution Compton spectroscopy for laser-driven high- flux gamma rays,”Matter and Radiation at Extremes, vol. 6, no. 1, 2021

    work page 2021

  19. [19]

    High energy X- ray Compton spectroscopy via iterative reconstruction,

    D. Haden, G. Golovin, W. Yan, C. Fruhling, P. Zhang, B. Zhao, S. Banerjee, and D. Umstadter, “High energy X- ray Compton spectroscopy via iterative reconstruction,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 951, p. 163032, Jan. 2020

    work page 2020

  20. [20]

    Development of Compton X- ray spectrometer for high energy resolution single-shot high-flux hard X-ray spectroscopy,

    S. Kojima, T. Ikenouchi, Y. Arikawa, S. Sakata, Z. Zhang, Y. Abe, M. Nakai, H. Nishimura, H. Shiraga, T. Ozaki, S. Miyamoto, M. Yamaguchi, A. Takemoto, S. Fujioka, and H. Azechi, “Development of Compton X- ray spectrometer for high energy resolution single-shot high-flux hard X-ray spectroscopy,”Review of Scientific Instruments, vol. 87, p. 043502, Apr. 2016

    work page 2016

  21. [21]

    Compton Spectrometer for FACET-II,

    B. Naranjo, G. Andonian, A. Fukasawa, W. Lynn, N. Ma- jernik, J. Rosenzweig, Y. Sakai, O. Williams, M. Yadav, and Y. Zhuang, “Compton Spectrometer for FACET-II,” Proceedings of the 12th International Particle Accelerator Conference, vol. IPAC2021, pp. 4 pages, 3.981 MB, 2021

    work page 2021

  22. [22]

    Conceptual De- sign of a Highflux Multi-GeV Gamma-ray Spectrometer,

    K. Fleck, N. Cavanagh, and G. Sarri, “Conceptual De- sign of a Highflux Multi-GeV Gamma-ray Spectrometer,” Scientific Reports, vol. 10, no. 1, p. 9894, 2020

    work page 2020

  23. [23]

    Experimental characterization of a single- shot spectrometer for high-flux, GeV-scale gamma-ray beams,

    N. Cavanagh, K. Fleck, M. J. V. Streeter, E. Gerst- mayr, L. T. Dickson, C. Ballage, R. Cadas, L. Calvin, S. Dobosz Dufr´ enoy, I. Moulanier, L. Romagnani, O. Vasilovici, A. Whitehead, A. Specka, B. Cros, and G. Sarri, “Experimental characterization of a single- shot spectrometer for high-flux, GeV-scale gamma-ray beams,”Phys. Rev. Research, vol. 5, p. 04...

    work page 2023

  24. [24]

    Tech- nical design report for the luxe experiment,

    L. collaboration, H. Abramowicz, M. Almanza Soto, M. Altarelli, R. Aßmann, A. Athanassiadis, G. Avoni, T. Behnke, M. Benettoni, Y. Benhammou,et al., “Tech- nical design report for the luxe experiment,”Eur. Phys. J. Spec. Top., vol. 233, no. 10, pp. 1709–1974, 2024

    work page 1974

  25. [25]

    Compact low-noise GeV-scale electron-positron pair spectrome- ter,

    J. Song, Y. Noh, M. Mirzaie, C. I. Hojbota, H.-i. Kim, S. Lee, J. Won, C. Song, and W. Bang, “Compact low-noise GeV-scale electron-positron pair spectrome- ter,”Results in Physics, vol. 68, p. 108091, Jan. 2025

    work page 2025

  26. [26]

    A comprehensive review on Compton camera image reconstruction: From principles to AI innovations,

    S. M. Kim and J. S. Lee, “A comprehensive review on Compton camera image reconstruction: From principles to AI innovations,”Biomed. Eng. Lett., vol. 14, pp. 1175– 1193, Nov. 2024

    work page 2024

  27. [27]

    Maximum Likelihood Compton Polarimetry with the Compton Spectrometer and Im- ager,

    A. W. Lowell, S. E. Boggs, C. L. Chiu, C. A. Kierans, C. Sleator, J. A. Tomsick, A. C. Zoglauer, H.-K. Chang, C.-H. Tseng, C.-Y. Yang, P. Jean, P. V. Ballmoos, C.-H. Lin, and M. Amman, “Maximum Likelihood Compton Polarimetry with the Compton Spectrometer and Im- ager,”ApJ, vol. 848, p. 120, Oct. 2017

    work page 2017

  28. [28]

    Improved iterative Bayesian unfolding

    G. D’Agostini, “Improved iterative Bayesian unfolding,” arXiv:1010.0632, Oct. 2010

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  29. [29]

    Exploring Fully Bayesian Unfolding forγ-ray Spectra,

    V. M. Valsdottir, “Exploring Fully Bayesian Unfolding forγ-ray Spectra,” 2020

    work page 2020

  30. [30]

    Re- constructing Gamma-ray Energy Distributions from PE- DRO Pair Spectrometer Data,

    M. Yadav, M. H. Oruganti, B. Naranjo, G. Andonian, ¨O. Apsimon, C. P. Welsch, and J. B. Rosenzweig, “Re- constructing Gamma-ray Energy Distributions from PE- DRO Pair Spectrometer Data,”arXiv:2409.02113, Aug. 2024

    work page arXiv 2024

  31. [31]

    Strong-field quantum electrodynamics in the field of ultra-intense lasers,

    K. J. Fleck, “Strong-field quantum electrodynamics in the field of ultra-intense lasers,” 2025

    work page 2025

  32. [32]

    V. B. Berestetskii, L. P. Pitaevskii, and E. M. Lifshitz, Quantum Electrodynamics: Volume 4. Elsevier, Dec. 2012

    work page 2012

  33. [33]

    Geant4—a simulation toolkit,

    S. Agostinelli, J. Allison, K. a. Amako, J. Apostolakis, H. Araujo, P. Arce, M. Asai, D. Axen, S. Banerjee, G. Barrand,et al., “Geant4—a simulation toolkit,”Nu- clear Instruments and Methods in Physics Research Sec- tion A: Accelerators, Spectrometers, Detectors and Asso- ciated Equipment, vol. 506, pp. 250–303, July 2003. 13

    work page 2003

  34. [34]

    Characterization of relativistic electron–positron beams produced with laser- accelerated gev electrons,

    H. Song, C. M. Kim, J. Won, J. Song, S. Lee, C.-M. Ryu, W. Bang, and C. H. Nam, “Characterization of relativistic electron–positron beams produced with laser- accelerated gev electrons,”Scientific Reports, vol. 13, no. 1, p. 310, 2023

    work page 2023

  35. [35]

    Design of radiation conversion target for Compton gamma magnetic spectrometer,

    X. Weng, X. Tan, D. Hei, X. Zhang, B. Sun, K. Wei, and X. Liu, “Design of radiation conversion target for Compton gamma magnetic spectrometer,”AIP Ad- vances, vol. 11, p. 055011, May 2021. [35]https://pdg.lbl.gov/2022/AtomicNuclearProperties /

    work page 2021

  36. [36]

    The FLUKA code: Description and benchmarking,

    G. Battistoni, F. Cerutti, A. Fass` o, A. Ferrari, S. Muraro, J. Ranft, S. Roesler, and P. R. Sala, “The FLUKA code: Description and benchmarking,” inAIP Conference Pro- ceedings, vol. 896, (Batavia, Illinois (USA)), pp. 31–49, AIP, 2007

    work page 2007

  37. [37]

    Study of imaging plate detector sensitivity to 5-18 MeV electrons,

    G. Boutoux, N. Rabhi, D. Batani, A. Binet, J.-E. Ducret, K. Jakubowska, J.-P. N` egre, C. Reverdin, and I. Thfoin, “Study of imaging plate detector sensitivity to 5-18 MeV electrons,”Review of Scientific Instruments, vol. 86, p. 113304, Nov. 2015

    work page 2015

  38. [38]

    Response functions of imaging plates to photons, elec- trons and 4He particles,

    T. Bonnet, M. Comet, D. Denis-Petit, F. Gobet, F. Han- nachi, M. Tarisien, M. Versteegen, and M. M. Al´ eonard, “Response functions of imaging plates to photons, elec- trons and 4He particles,”Review of Scientific Instru- ments, vol. 84, p. 103510, Oct. 2013

    work page 2013

  39. [39]

    Overview of spectral unfolding techniques and uncertainty estimation,

    M. Reginatto, “Overview of spectral unfolding techniques and uncertainty estimation,”Radiation Measurements, vol. 45, pp. 1323–1329, Dec. 2010

    work page 2010

  40. [40]

    On the Solution of Fredholm Integral Equations of the First Kind,

    I. G. Burova and V. M. Ryabov, “On the Solution of Fredholm Integral Equations of the First Kind,”Wseas transactions on mathematics, vol. 19, pp. 699–708, Feb. 2021

    work page 2021

  41. [41]

    Denoising Autoencoder Aided Spectrum Reconstruction for Colloidal Quantum Dot Spectrometers,

    J. Zhang, X. Zhu, and J. Bao, “Denoising Autoencoder Aided Spectrum Reconstruction for Colloidal Quantum Dot Spectrometers,”IEEE Sensors Journal, vol. 21, pp. 6450–6458, Mar. 2021

    work page 2021

  42. [42]

    The Richardson-Lucy deconvolution method to extract LAMOST 1D spectra,

    M. Li, G.-W. Li, K. Lv, F.-Q. Duan, H. Haerken, and Y.- H. Zhao, “The Richardson-Lucy deconvolution method to extract LAMOST 1D spectra,”Res. Astron. Astro- phys., vol. 19, p. 145, Oct. 2019

    work page 2019

  43. [43]

    A platform for all-optical thomson/compton scattering with versatile parameters,

    S. Chen, W. Yan, M. Zhu, Y. Li, X. Hu, H. Xu, W. Zhou, G. Lu, M. Wei, L. Lu,et al., “A platform for all-optical thomson/compton scattering with versatile parameters,” High Power Laser Science and Engineering, pp. 1–16, 2024

    work page 2024

  44. [44]

    High- intensity lasers and research activities in China,

    Y. Li, L. Chen, M. Chen, F. Liu, Y. Gu, B. Guo, J. Hua, T. Huang, Y. Leng, F. Li, L. Li, R. Li, C. Lin, W. Lu, Z. Lyu, W. Ma, X. Ning, Y. Peng, Y. Wan, J. Wang, Z. Wang, Z. Wei, X. Yan, J. Zhang, B. Zhao, Z. Zhao, C. Zhou, K. Zhou, W. Zhou, J. Zhu, and P. Zhu, “High- intensity lasers and research activities in China,”High Power Laser Science and Engineeri...

    work page 2025

  45. [45]

    Nonlinear single Compton scattering of an electron wave packet,

    A. Angioi, F. Mackenroth, and A. Di Piazza, “Nonlinear single Compton scattering of an electron wave packet,” Phys. Rev. A, vol. 93, May 2016

    work page 2016

  46. [46]

    Locally monochromatic approximation to QED in intense laser fields,

    T. Heinzl, B. King, and A. J. MacLeod, “Locally monochromatic approximation to QED in intense laser fields,”Phys. Rev. A, vol. 102, Dec. 2020

    work page 2020

  47. [47]

    P. C. Hansen,Discrete Inverse Problems: Insight and Al- gorithms. No. 7 in Fundamentals of Algorithms, Philadel- phia: Society for Industrial and Applied Mathematics, 2010

    work page 2010

  48. [48]

    Deep learning as- sisted Shack–Hartmann wavefront sensor for direct wave- front detection,

    L. Hu, S. Hu, W. Gong, and K. Si, “Deep learning as- sisted Shack–Hartmann wavefront sensor for direct wave- front detection,”Opt. Lett., vol. 45, p. 3741, July 2020

    work page 2020

  49. [49]

    Data-driven science and machine learning methods in laser–plasma physics,

    A. D¨ opp, C. Eberle, S. Howard, F. Irshad, J. Lin, and M. Streeter, “Data-driven science and machine learning methods in laser–plasma physics,”High Power Laser Sci- ence and Engineering, vol. 11, p. e55, Jan. 2023

    work page 2023

  50. [50]

    U-Net: Convolutional Networks for Biomedical Image Segmen- tation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmen- tation,” inMedical Image Computing and Computer- Assisted Intervention – MICCAI 2015(N. Navab, J. Hornegger, W. M. Wells, and A. F. Frangi, eds.), (Cham), pp. 234–241, Springer International Publishing, 2015

    work page 2015

  51. [51]

    TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems

    M. Abadi, A. Agarwal, P. Barham, E. Brevdo, Z. Chen, C. Citro, G. S. Corrado, A. Davis, J. Dean, M. Devin, et al., “Tensorflow: Large-scale machine learning on heterogeneous distributed systems,”arXiv preprint arXiv:1603.04467, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  52. [52]

    Adam: A Method for Stochastic Optimization

    D. P. Kingma and J. Ba, “Adam: A method for stochas- tic optimization,”arXiv preprint arXiv:1412.6980, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  53. [53]

    Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

    Y. Gal and Z. Ghahramani, “Dropout as a bayesian ap- proximation: Representing model uncertainty in deep learning,”arXiv:1506.02142, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  54. [54]

    Feasibility of Deep Learning- Guided Attenuation and Scatter Correction of Whole- Body 68Ga-PSMA PET Studies in the Image Domain,

    S. Mostafapour, F. Gholamiankhah, H. Dadgar, H. Arabi, and H. Zaidi, “Feasibility of Deep Learning- Guided Attenuation and Scatter Correction of Whole- Body 68Ga-PSMA PET Studies in the Image Domain,” Clin Nucl Med, vol. 46, pp. 609–615, Aug. 2021

    work page 2021