Pith. sign in

REVIEW 3 major objections 4 minor 7 references

GATE 10 Monte Carlo particle transport simulation -- Part I: development and new features

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read GATE 10 rebuilds the medical-physics Monte Carlo toolkit around a Python user interface, multithreading, library embedding, and a CI-driven development process, while keeping Geant4 as its transport engine.

desk verdict GATE 10 release paper is a solid, honest feature list for a major open-source MC update; the missing cross-version equivalence data is a real but not fatal gap for a Part I software description. read the letter →

arxiv 2507.09842 v2 pith:BDF5HXXT submitted 2025-07-14 physics.med-ph physics.comp-ph

David Sarrut , Nicolas Arbor , Thomas Baudier , Julien Bert , Konstantinos Chatzipapas , Martina Favaretto , Hermann Fuchs , Loïc Grevillot
show 25 more authors
Hussein Harb Gert Van Hoey Maxime Jacquet Sébastien Jan Yihan Jia George C. Kagadis Han Gyu Kang Paul Klever Olga Kochebina Wojciech Krzemien Lydia Maigne Philipp Mohr Guneet Mummaneni Valentina Paneta Panagiotis Papadimitroulas Alexis Pereda Axel Rannou Andreas F. Resch Emilie Roncali Maxime Toussaint Carlotta Trigila Charalampos Tsoumpas Jing Zhang Karl Ziemons Nils Krah
This is my paper
classification physics.med-phphysics.comp-ph
keywords GATE10MonteCarlosimulationmedicalphysicsGeant4PythoninterfaceparticletransportPET/SPECTimagingvariancereduction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper presents GATE 10, a major release of the open-source Monte Carlo simulation platform for medical physics. The central claim is that GATE has moved from a standalone application controlled by macro scripts to a Python-driven software platform: users install it with a single command, write simulations in Python, and can embed it as a library inside other programs. The release adds parallel worlds, tessellated and boolean solids, scene voxelization, new scoring actors, dynamic parameterization of runs, and new source models such as voxelized and time-activity-curve sources. The authors argue that this redesign makes Monte Carlo simulation easier to learn and share, supports industrial integration, and is maintained through a continuous-integration testing loop with more than 230 automated tests. A companion paper covers the internal architecture.

What carries the argument

The mechanism that carries the release is the new Python binding layer that wraps Geant4 and GATE's own C++ core, exposed through a manager-based object model in which every component derives from a GateObject base class. That base class is what allows the automatic JSON serialization of an entire simulation and the dynamic parameterization mechanism that reconfigures components such as CT images or source rotations run by run. The second key mechanism is the switch from the former System/sensitive-detector pipeline to Geant4 Primitive Scorers for hit collection, which is what makes any volume able to record hits. The continuous-integration loop, in which every feature must arrive with a test, documentation, and a pull request and stochastic outputs are compared statistically, is the supporting mechanism that is supposed to keep the platform reliable as it grows.

What would settle it

Re-simulate a benchmark case, such as a PET phantom or an external-beam dose distribution, with the same geometry, physics list, and random seed in GATE 9.4 and in GATE 10, then compare dose maps and list-mode output statistically; any difference beyond the tolerance of the stochastic tests would falsify the implicit equivalence claim.

Watch

Extended reading notes

Core claim

GATE 10 is a rebuild of the GATE Monte Carlo application around Python, with Geant4 still doing the particle transport. Its defining move is to treat the simulation description as Python code rather than as macro commands: geometry, sources, physics lists, scorers, and run timing are all configured through a Python interface, while the underlying C++ engine is wrapped by a new binding layer designed from scratch. Alongside this, the release replaces the old System/sensitive-detector mechanism for collecting hits with Geant4 Primitive Scorers, makes any volume able to record hits, and introduces a manager-based object model that can serialize the whole simulation to JSON. The paper reports that the result is cross-platform, installable via pip, and embeddable as a library, with multithreading and multiprocessing support; full Windows support and multithreaded ROOT output are still pending.

Load-bearing premise

GATE 10's physics results are the same as GATE 9's despite the new Python interface and the replacement of the old detector-hit mechanism, an equivalence the paper asserts but does not demonstrate with side-by-side regression comparisons.

Editorial extensions

If this is right

  • Users can install the package with pip install opengate and set up and run simulations entirely from Python, with no C++ compilation for standard use.
  • External software such as treatment-planning or image-reconstruction systems can embed GATE as a library rather than running it as a separate executable.
  • Scenarios that previously required multiple simulation runs, such as breathing motion, rotating gantries, or changing activity distributions, can be expressed in one script through run-based dynamic parameterization.
  • Medical physics teaching can use a single Python-based tool for Monte Carlo exercises, lowering the barrier compared with writing Geant4 C++ code.
  • Because every accepted feature must include an automated statistical test, the codebase is positioned to grow through external contributions without a central team writing all validations.

Reading between the lines

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

  • If the binding layer proves behaviorally equivalent to GATE 9, the same architecture could be reused to expose other Geant4-based simulation tools to Python, not just medical-physics applications.
  • The lack of a published regression comparison means early GATE 10 adopters should re-run their own validated GATE 9 benchmarks before switching production workflows.
  • With Python as the interface, GATE 10 is likely to absorb more machine-learning-driven tools, since GAN-based sources and data analysis already run in the Python ecosystem.
  • The decision to keep ROOT for list-mode output while exposing images through ITK points toward a possible migration to standardized formats such as the ETSI emission-tomography format once it stabilizes.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This manuscript describes GATE 10, a major rewrite of the open-source Monte Carlo simulation toolkit GATE, replacing the legacy macro interface with a Python-based interface and adding features in geometry (parallel worlds, tessellated and boolean solids, voxelized geometry and voxelization of scenes), actors and scorers (phase space, digitizers, LET, RBE, TLE, free flight), sources (scanned pencil beams, voxelized, TAC, PHID, GAN, optiGAN), dynamic parameterization, JSON serialization, and a CI-driven development workflow. The paper is Part I of a two-part series; the internal architecture is largely deferred to Part II. It is primarily a software-release/feature description with illustrative examples, and it includes no quantitative validation against previous GATE versions, Geant4 reference results, or measured data.

Significance. If the software works as described, this is a potentially important contribution to open-source Monte Carlo simulation in medical physics: a pip-installable, cross-platform, Python-based interface to Geant4 with a visible public repository and a stated CI/test infrastructure of 230+ tests. These are concrete, checkable strengths that support the descriptive claim that GATE 10 exists and is developed openly. However, the medical-physics significance of the release depends on the preservation of validated GATE 9 physics and digitizer behavior through the new Python binding layer and the move to Geant4 Primitive Scorers; the manuscript does not yet provide evidence for that preservation. The paper is transparent about some limitations (e.g., Windows support, ROOT/MT incompatibility, offline coincidence sorter), but it does not address the equivalence gap, so the significance is conditional on a validation study or a clearly delimited claim.

major comments (3)
  1. [§2.4, §2.5] The replacement of the System sensitive-detector mechanism with Geant4 Primitive Scorers is a fundamental change to the hit-collection chain, and the new Python binding layer (§2.2) is designed from scratch, yet the paper presents no cross-version equivalence test: there is no comparison of GATE 10 against GATE 9, against Geant4 reference results, or against measured data for dose, energy spectra, or imaging/digitizer chains. The CI suite described in §2.5 contains feature-specific regression tests with fixed seeds, but those tests do not establish that the new hit-collection and binding chain reproduce validated GATE 9 behavior. Since the paper's central claim is that GATE 10 is a major, reliable successor for medical physics, this is a load-bearing gap. Please add a benchmark/validation section with representative quantitative comparisons and stated statistical tolerances, or explicitly restrict the paper's claims to software functionality and defer all physics-preservation claims to future work.
  2. [§2.2] The Python binding layer is the interface through which every GATE 10 simulation is configured and run, but its architecture is deferred to Part II and no validation of the binding layer itself is reported in Part I, such as evidence that a parameter set in Python produces the same Geant4 configuration as the equivalent legacy GATE macro. Because the paper's usability and reliability claims depend on this layer, please either provide a set of binding-level equivalence tests or state explicitly in Part I that such tests are still in progress and that users should not assume bitwise or statistical equivalence with GATE 9 until they are completed.
  3. [§3.2.5, §3.2.6, §3.3.6, §3.3.8] Several performance claims are presented as point values without experimental context, error bars, or, in one case, a citation: the TLE hybrid mode is said to be up to 100× faster (§3.2.5, citing earlier work), the free-flight actor reportedly achieves approximately 50-fold speedup (§3.2.6, no citation), PHID shows a 30× speedup (§3.3.6), and optiGAN reportedly halves simulation time with more than 92% similarity (§3.3.8). For a release paper, each of these should either be substantiated with the test conditions (geometry, physics list, statistics, hardware) or clearly attributed to prior publications and identified as not re-validated in GATE 10; otherwise readers cannot assess whether the claims hold in the new release.
minor comments (4)
  1. [§3.1.3] The function name 'substract_volumes(A, B)' appears with a typo; it should be 'subtract_volumes' if that is the intended API.
  2. [References] The bibliography lists the Arce et al. 2025 Medical Physics reference twice with identical content; please remove the duplicate entry.
  3. [§2.5] The statement that a randomly selected subset of tests is run on each platform per commit, while the complete suite runs only weekly, is important for interpreting the '230+ tests' claim; please make this limitation more prominent so readers do not infer that every commit is fully tested on every platform.
  4. [§3.3.2] The connection between MeanEnergyPerIonPair values (0.5 eV and 6.0 eV) and the annihilation-photon acollinearity angular distribution is not explained; a one-sentence description of the underlying model would help users understand why these particular values produce the stated FWHM.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is a software release description with no derivation whose inputs already contain its conclusions.

full rationale

The manuscript describes a software release and its features; it does not contain a derivation chain that could reduce to its own inputs. The central claim, that GATE 10 exists and has Python-based configuration, multithreading/multiprocessing, library embedding, and a CI-driven development process, is externally checkable in the public repository and PyPI package. Feature descriptions such as parallel worlds, tessellated solids, boolean solids, digitizer actors, LET and RBE actors, and source models are presented as implemented functionality, not as results predicted from fitted parameters. Performance claims, such as the PHID 30x speedup and the TLE up-to-100x speedup, are attributed to prior published studies (Sarrut et al., 2024; Noblet et al., 2016) rather than being derived anew in this paper, so they are citations of independent evidence rather than self-referential validation. The paper's self-citations to earlier GATE reference papers and to the companion Part II paper (Krah et al., 2025) are contextual: they identify the project's history and defer architectural details, but they are not used to justify the paper's central release claim, nor do they forbid alternative interpretations of any result. The optiGAN 92% similarity statement cites Mummaneni et al. (2025) and Trigila et al. (2023); even though some authors overlap with this manuscript, the claim is about an external validation study, not a conclusion forced by the present definitions. The absence of a GATE 9-to-GATE 10 numerical equivalence or regression comparison is a potential validation gap, but that is a correctness or completeness concern, not circularity. No equation in the paper is defined in terms of the quantity it is said to predict, no fitted parameter is renamed as a prediction, and no load-bearing argument reduces to an unverified self-citation. The honest finding is therefore no significant circularity, with a score of 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central claim is a software release description, not a mathematical derivation. It relies on Geant4's physics validity, the correctness of a custom Python binding layer, and faithful manual translation of GATE 9 macros. No free parameters or invented physical entities are introduced.

assumptions (3)
  • domain assumption Geant4's particle transport engine and its physics lists are accurate enough for medical physics simulations.
    GATE 10 adds no native physics models; all transport is delegated to Geant4 (Section 1 and Section 3.3.1). The reliability of every GATE 10 result inherits Geant4's validation.
  • domain assumption The custom Python-to-Geant4 binding layer preserves C++ behavior and thread safety.
    Section 2.2 states that a binding layer was designed from scratch, with details deferred to Part II. The correctness of all Python-exposed features depends on this unverified-in-Part-I assumption.
  • domain assumption Users can translate GATE 9 macros into semantically equivalent Python scripts, and manual translation preserves the intended simulation logic.
    Section 2.4 notes that no automatic translation tool exists. The central migration story relies on users doing faithful manual transcription.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GATE 10 Monte Carlo particle transport simulation -- Part I: development and new features." pith.science (2026). https://pith.science/paper/BDF5HXXT

@misc{pith2026250709842,
  author       = {Pith},
  title        = {Pith review of: GATE 10 Monte Carlo particle transport simulation -- Part I: development and new features},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BDF5HXXT}},
  note         = {Machine review of arXiv:2507.09842}
}
read the original abstract

We present GATE version 10, a major evolution of the open-source Monte Carlo simulation application for medical physics, built on Geant4. This release marks a transformative evolution, featuring a modern Python-based user interface, enhanced multithreading and multiprocessing capabilities, the ability to be embedded as a library within other software, and a streamlined framework for collaborative development. In this Part 1 paper, we outline GATE's position among other Monte Carlo codes, the core principles driving this evolution, and the robust development cycle employed. We also detail the new features and improvements. Part 2 will detail the architectural innovations and technical challenges. By combining an open, collaborative framework with cutting-edge features, such a Monte Carlo platform supports a wide range of academic and industrial research, solidifying its role as a critical tool for innovation in medical physics.

Figures

Figures reproduced from arXiv: 2507.09842 by the authors.

Figure 1
Figure 1. Example of the parallel worlds feature through prostate brachytherapy treatment simulation. The isodoses, overlaid on the patient’s CT scan, are normalized with respect to the prescription dose [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Example of the tessellated geometry feature through skin dose exposure calculation in X-ray guided intervention, for different poses and positions of the physician around the patient. the GATE 10 simulation of a cardiologist’s exposure to scattering radiation during an X-ray-guided intervention in the operating room. The simulations were based on different poses of the ICRP 145 phantom created using a third party so… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

7 extracted references · 5 canonical work pages

  1. [1]

    2016Nuclear instruments and methods in physics research section A: Accelerators, Spectrometers, Detectors and Associated Equipment835, 186–225

    Allison J, Amako K, Apostolakis J, Arce P, Asai M, Aso T, Bagli E, Bagulya A, Banerjee S, Barrand G et al. 2016Nuclear instruments and methods in physics research section A: Accelerators, Spectrometers, Detectors and Associated Equipment835, 186–225. Arce P, Archer J W, Arsini L, Bagulya A, Bolst D, Brown J M C, Caccia B, Chacon A, Cirrone G A P, Cortés-G...

  2. [8]

    REFERENCES 21 Mercolli L, Steinberger W M, Läppchen T, Amon M, Bregenzer C, Conti M, Cardoso  R F, Mingels C, Moskal P, Rathod N, Sari H, Stępień E Ł, Weidner S, Rominger A, Shi K and Seifert R 2025medRxiv . URL: https://www.medrxiv.org/content/early/2025/05/29/2025.05.28.25328504.1 Moskal P, Baran J, Bass S, Choiński J, Chug N, Curceanu C, Czerwiński E,...

    2025

  3. [12]

    M, Gu C, Rübenach J, Peresano M and Turra R 2020 ‘scikit-hep/uproot: 3.12.0’

    Pivarski J, Das P, Burr C, Smirnov D, Feickert M, Gal T, Kreczko L, Smith N, Biederbeck N, Shadura O, Proffitt M, benkrikler, Dembinski H, Schreiner H, Rembser J, R. M, Gu C, Rübenach J, Peresano M and Turra R 2020 ‘scikit-hep/uproot: 3.12.0’. URL: https://zenodo.org/record/3952728#.X4lPLUL7TUJ Qi J and Huang B 2022IEEE Transactions on Medical Imaging41(1...

    work page arXiv 2020

  4. [76]

    Toussaint M, Loignon-Houle F, Auger É, Lapointe G, Dussault J P and Lecomte R 2024 Physics in Medicine & Biology

    Tashima H and Yamaya T 2024IEEE Transactions on Radiation and Plasma Medical Sciences 8(8), 853–866. Toussaint M, Loignon-Houle F, Auger É, Lapointe G, Dussault J P and Lecomte R 2024 Physics in Medicine & Biology. URL: http://iopscience.iop.org/article/10.1088/1361-6560/ad70f1 Trigila C, Srikanth A and Roncali E 2023Machine Learning: Science and Technolo...

    work page doi:10.1088/1361-6560/ad70f1 2024

  5. [86]

    URL: https://doi.org/10.1007/BF02748591 Cortés-Giraldo M A and Carabe A 2015Physics in Medicine and Biology60(7), 2645–

    Colombino P, Fiscella B and Trossi L 1965Il Nuovo Cimento (1955-1965)38(2), 707–723. URL: https://doi.org/10.1007/BF02748591 Cortés-Giraldo M A and Carabe A 2015Physics in Medicine and Biology60(7), 2645–

    work page doi:10.1007/bf02748591 1955

  6. [5203]

    Huang B and Qi J 2024Physics in Medicine & Biology69(24), 24NT01

    Harris C R, Millman K J, van der Walt S J, Gommers R, Virtanen P, Cournapeau D, Wieser E, Taylor J, Berg S, Smith N J, Kern R, Picus M, Hoyer S, van Kerkwijk M H, Brett M, Haldane A, Del Río J F, Wiebe M, Peterson P, Gérard-Marchant P, REFERENCES 20 Sheppard K, Reddy T, Weckesser W, Abbasi H, Gohlke C and Oliphant T E 2020 Nature 585(7825), 357–362. Huang...

    work page arXiv 2020

  7. [5249]

    URL: http://dx.doi.org/10.1088/0031-9155/59/24/7703 Srikanth A, Trigila C and Roncali E 2024Machine Learning: Science and Technology 5(2), 027001

    URL: https://dx.doi.org/10.1088/0031-9155/52/17/010 Smekens F, Létang J M, Noblet C, Chiavassa S, Delpon G, Freud N, Rit S and Sarrut D 2014 Phys Med Biol59(24), 7703–7715. URL: http://dx.doi.org/10.1088/0031-9155/59/24/7703 Srikanth A, Trigila C and Roncali E 2024Machine Learning: Science and Technology 5(2), 027001. Publisher: IOP Publishing. URL: https...

    work page doi:10.1088/0031-9155/52/17/010 2014

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.