Pith. sign in

REVIEW 4 major objections 5 minor 22 references

The Highly-Granular Time-of-Flight Neutron Detector for the BM@N experiment

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a newly designed highly granular time-of-flight neutron detector can identify neutrons and reconstruct their energies in heavy-ion collisions at BM@N up to 4A GeV, with a simulated yield of about one billion…

desk verdict Useful BM@N neutron detector simulations, but the headline yield is internally inconsistent with the stated flux and needs a correction. read the letter →

arxiv 2412.00455 v1 pith:4R5356ZI submitted 2024-11-30 hep-ex nucl-ex

S. Morozov , V. Bocharnikov , D. Finogeev , M. Golubeva , F. Guber , A. Izvestnyy , N. Karpushkin , A. Makhnev
show 5 more authors
P. Parfenov D. Serebryakov A. Shabanov V. Volkov A. Zubankov
This is my paper · ORCID
classification hep-exnucl-ex
keywords time-of-flightneutrondetectorheavy-ioncollisionsreconstructionazimuthalflowplasticscintillatorsiliconphotomultipliergraphneuralnetworkyields
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

The paper reports the design and simulated performance of a new highly granular time-of-flight neutron detector for the BM@N heavy-ion experiment. Its central claim is that the detector can identify primary neutrons and reconstruct their kinetic energies in the range from about 300 MeV to 4 GeV, using plastic scintillator cells read out by silicon photomultipliers and time-of-flight measurement. In simulations of Bi+Bi collisions at 3A GeV, the detector reaches roughly 50% detection efficiency, 10–30% energy resolution, and a reconstruction purity of 0.7 at that efficiency, yielding an estimated $1.2\times10^9$ reconstructed single-neutron events per month of running (about $1\times10^9$ in the summary's rounded figure). A first Xe+CsI run at 3.8A GeV is predicted to give about $0.9\times10^9$ single-neutron events per month. These numbers matter because neutron yields and azimuthal neutron flow are observables sensitive to the symmetry-energy term of the equation of state of dense nuclear matter, which is poorly constrained above 1 GeV per nucleon.

What carries the argument

The load-bearing object is the HGND module: an $11\times11$ matrix of $4\times4\times2.5\,\mathrm{cm}^3$ polystyrene scintillator cells read out by silicon photomultipliers on one side and LEDs for calibration on the other, with copper absorbers between layers and a veto first layer. Time-of-flight is the working identity: neutron velocity is inferred from the distance to the target divided by the hit time, and kinetic energy follows from $E = m_n(1/\sqrt{1-(v/c)^2}-1)$, supported by a 100 ps FPGA-based TDC and measured cell resolution of 130–150 ps. The copper absorber and layer stacking create a sampling calorimeter that converts neutron interactions into small clusters of fired cells, which are then grouped either by geometric and time clustering or by a graph neural network whose nodes are hit coordinates and energy deposits. The simulation chain of a heavy-ion event generator plus a detector-response simulation provides the efficiency, background, and yield numbers.

What would settle it

In the first Xe+CsI data run, compare the measured single-neutron event rate per $10^6$ beam ions with the simulated value of about 12.7% of events at the HGND surface; a sustained rate disagreement beyond a factor of two, after correcting for live time and trigger efficiency, would falsify the monthly yield estimate.

Watch

Extended reading notes

Core claim

The central claim is that a compact, highly segmented detector—16 alternating layers of $11\times11$ arrays of $4\times4\times2.5\,\mathrm{cm}^3$ plastic scintillator cells interleaved with 3 cm copper absorber plates, with the first layer acting as a charged-particle veto—can turn time-of-flight hits into a clean neutron sample in the fixed-target BM@N environment. Neutron kinetic energy is obtained from the fastest hit in a cluster through $E = m_n(1/\sqrt{1-(v/c)^2}-1)$, with per-cell time resolution measured at 130–150 ps and a TDC precision near 40 ps. With a 35 ns time cut that suppresses background by about a factor of 6 while retaining 92% of primary neutrons, the simulated detection efficiency is about 50% and the energy resolution is 10–30% over $0.3$–$4$ GeV. Two reconstruction strategies—a cluster method and a graph-neural-network method—both reach a reconstruction efficiency around 0.7 at fixed purity 0.7 in simplified single-neutron tests, with the machine-learning method correcting time-of-flight overestimates at 2–4 GeV.

Load-bearing premise

The yield and performance numbers assume that the simulated neutron multiplicities and energy spectra from the event generator, and the simulated detector response, match the real Bi+Bi and Xe+CsI collisions at BM@N.

Editorial extensions

If this is right

  • HGND can deliver about $1\times10^9$ reconstructed single-neutron events per month of BM@N running, giving a dataset large enough for differential neutron yield and azimuthal-flow measurements at beam energies up to 4A GeV.
  • The 35 ns time cut removes roughly six times more background than signal while keeping 92% of primary neutrons, so primary neutrons with kinetic energy above about 300 MeV can be selected cleanly.
  • The two independent reconstruction methods—cluster analysis and graph neural networks—can cross-check each other, and the machine-learning route compensates the time-of-flight energy overestimation seen in the cluster method at 2–4 GeV.
  • If multiple-neutron events are also reconstructed, the monthly statistics rise from about $1.2\times10^9$ to about $1.5\times10^9$ reconstructed neutron events for Bi+Bi at 3A GeV.

Reading between the lines

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

  • If the simulated performance transfers to the real detector, this would extend neutron-flow constraints from sub-GeV heavy-ion measurements into the multi-GeV regime where the symmetry-energy term of the nuclear equation of state is least constrained.
  • The graph-neural-network energy-regression model is trained only on events identified as containing neutrons; retraining on events with multiple neutrons would likely be needed before the reported purity holds for high-multiplicity central collisions.
  • A natural testable extension is to use the two 7 m 'arms' to measure the reaction-plane dependence of neutron azimuthal flow differentially in rapidity, which the current single-neutron performance study does not yet demonstrate.
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

4 major / 5 minor

Summary. The paper describes the design and simulation studies of a highly granular time-of-flight neutron detector (HGND) for the BM@N fixed-target program at JINR. Two geometric options are presented (a single module at 5 m and a two-arm module at 7 m), along with the scintillator/SiPM cell design, the FPGA/TDC readout, and Geant4-based simulations using the DCM-QGSM-SMM event generator for Bi+Bi and Xe+CsI collisions. The authors report a neutron detection efficiency around 50%, an energy resolution of 10-30% from time-of-flight reconstruction with 150 ps time smearing, and compare a cluster method with a graph-neural-network approach, quoting reconstruction efficiency at a fixed purity of 0.7. They estimate a yield of about 1e9 reconstructed neutrons per month of BM@N operation.

Significance. If correct, the paper demonstrates a workable detector concept for neutron measurements in the 300 MeV to 4 GeV range at BM@N, with two independent reconstruction strategies and a transparent simulation setup. The strengths are the clearly specified thresholds, time cuts, time smearing, reconstruction definitions, and the held-out 50/50 train/test split for the machine-learning method. The performance numbers are internally plausible as preliminary simulation results. However, the central yield estimate contains an arithmetic inconsistency with the stated acceptance flux, and several simulation inputs are not fully documented; these issues must be resolved before the quantitative physics claims can be accepted.

major comments (4)
  1. [Section 3.1] The yield estimate is not internally consistent with the quoted acceptance flux. The Introduction states that the neutron flux in the HGND acceptance is about 500 neutrons/s; over 30 days this gives 1.3e9 incoming neutrons, and multiplying by the quoted 50% mean efficiency gives at most 6.5e8 reconstructed neutrons. Since single-neutron events are a subset of all neutron events, the claimed 1.2e9 reconstructed single-neutron events is arithmetically impossible under these inputs. The estimate also cannot be reproduced because the listed inputs ('1e6 ions per spill, 50% duty factor and 70% efficiency of Nuclotron, 2% interaction length of target') omit the spill repetition rate or number of spills per month. Please supply a complete formula with all beam parameters and a corrected yield number, and harmonize it with the Summary's 'about 1e9'.
  2. [Section 3.1, Fig. 7] The 'mean efficiency of the HGND of 50%' used in the yield calculation is not defined. Figure 7 shows a strong energy dependence of the detection efficiency, so a single 50% value must be accompanied by the weighting spectrum. The yield should be computed by averaging the energy-dependent efficiency over the simulated primary-neutron spectrum at the chosen HGND position, and the resulting spectrum-weighted mean should be quoted instead of an ad-hoc 50% number.
  3. [Section 3] The paper does not provide any validation of the DCM-QGSM-SMM event generator for neutron production at BM@N energies, nor of the Geant4 hadronic transport models used in bmnroot. The quoted multiplicity, efficiency, purity, and yield all inherit these model assumptions. Please add a comparison with existing data (for example, the Xe+CsI run at 3.8A GeV already collected in 2023, or published neutron spectra from similar reactions) or at least state the generator dependence as a systematic uncertainty with a concrete cross-check; without this, the quantitative claims are conditional on an unvalidated input model.
  4. [Section 3.1, Figs. 9 and 10] The beam energy labels are inconsistent between the text and the figure captions. Section 3.1 states the Bi+Bi performance studies are done at 3.0A GeV, but the Fig. 9 caption reads 'Bi+Bi @ 3.8A GeV'; for Xe+CsI the text says 3.8A GeV while the Fig. 10 caption reads 3A GeV. Beam energy directly changes the neutron multiplicity and time-of-flight distributions used to derive the yield, so these inconsistencies must be resolved and the simulation results checked against the actual energy used.
minor comments (5)
  1. [Section 4.1] The velocity cut 'v < c' is not a meaningful restriction for massive particles, all of which satisfy v < c; the text should specify the actual threshold value used for gamma and charged-particle rejection.
  2. [Fig. 7] The sentence 'The difference for 1 GeV neurons' should read '1 GeV neutrons'.
  3. [Section 4.3, Fig. 13] The text refers to an energy-resolution comparison in Fig. 13 without stating numerical values or the definition of the systematic shift for the cluster method; a short quantitative summary would improve reproducibility.
  4. [Throughout] There are several typographical errors, including 'syimmetric' (Introduction), 'assemled' and 'matix' (Section 2.1), 'L VDS' (Sections 2.1 and 2.2), and 'yieds' (Summary); these should be corrected.
  5. [Section 3.1, Fig. 6] The right panel of Fig. 6 is described as 'background neutron on all surfaces of the HGND', but the text does not specify which surfaces are included or whether this background includes neutrons from hadronic interactions in the detector material; a sentence clarifying the definition would help.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the derivation chain is self-contained, though the headline yield estimate is arithmetically inconsistent with the stated 500 n/s flux.

full rationale

The paper's central derivations are self-contained rather than circular. Neutron energy reconstruction is obtained from the standard relativistic time-of-flight formula, with time resolution taken from external beam tests and then used as input to the simulation. The detection efficiency is computed from Geant4/bmnroot simulations and applied back to the same simulated spectra as a self-consistency correction, not as a fitted parameter that predetermines a prediction. The GNN classifier and energy regressor are trained on a 50% split and evaluated on the held-out half, so the reported efficiency and purity are not fitted to the test set. The yield estimate in Section 3.1 is a forward multiplication of simulated multiplicity, efficiency, and beam-operation parameters; no parameter is fitted to the quantity it later claims to predict. Self-citations to the HGND design concept, TDC development, and time-resolution measurements are supported by independent beam measurements and do not carry the argument by themselves. The only notable quantitative concern is an internal arithmetic inconsistency: with about 500 neutrons per second entering the acceptance and a 50% mean efficiency, one month of continuous operation yields at most about 6.5e8 reconstructed neutron events, whereas Section 3.1 quotes about 1.2e9 single-neutron events per month. This is an inconsistency in the headline number, not an equivalence between inputs and outputs, so it does not constitute circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper's central claim depends on the fidelity of two simulation pillars: the event generator DCM-QGSM-SMM and the Geant4-based bmnroot framework. Neither is validated against real HGND collision data in this paper, so the performance numbers are simulation-level estimates. The analysis cuts (35 ns, 3 MeV) are hand-chosen and affect all quoted metrics, and the monthly yield estimate relies on several operational assumptions. No new physical entities are introduced.

free parameters (4)
  • Time-of-flight cut = 35 ns
    Chosen to reduce background by about 6x while retaining 92% of primary neutrons; affects efficiency and purity results.
  • Cell energy threshold = 3 MeV (about 0.5 MIP)
    Applied to fired cells before clustering and ML; affects efficiency and cluster size.
  • Mean HGND efficiency for yield estimate = 50%
    Used to convert simulated neutron multiplicity into monthly reconstructed event count; a rough average.
  • BM@N operation efficiency parameters = 1e6 ions/spill, 50% duty factor, 70% Nuclotron efficiency, 2% interaction length
    Operational assumptions in the statistics estimate.
assumptions (4)
  • domain assumption DCM-QGSM-SMM generator accurately reproduces neutron production and multiplicities in Bi+Bi at 3A GeV and Xe+CsI at 3.8A GeV
    Used in Section 3 and 4 to generate events; if wrong, the performance metrics, background ratio, and yield estimates change.
  • domain assumption Geant4 hadronic physics models accurately simulate neutron transport and detection in plastic scintillator with copper absorbers
    All efficiency and resolution results come from Geant4 via bmnroot; unvalidated against a real HGND prototype (only a single-cell time-resolution beam test is cited).
  • standard math The standard relativistic time-of-flight formula correctly maps measured time to kinetic energy
    Used in Section 4.1 for energy reconstruction; standard kinematics.
  • domain assumption The ML models trained on simulated data generalize to real detector response
    GNN classification and regression are trained and tested on the same simulation framework; no data/simulation domain shift is considered.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Highly-Granular Time-of-Flight Neutron Detector for the BM@N experiment." pith.science (2026). https://pith.science/paper/4R5356ZI

@misc{pith2026241200455,
  author       = {Pith},
  title        = {Pith review of: The Highly-Granular Time-of-Flight Neutron Detector for the BM@N experiment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4R5356ZI}},
  note         = {Machine review of arXiv:2412.00455}
}
read the original abstract

A new Highly-Granular time-of-flight Neutron Detector (HGND) is being developed and constructed to measure azimuthal neutron flow and neutron yields in nucleus-nucleus interactions in heavy-ion collisions with energies up to 4A GeV in the fixed target experiment BM@N at JINR. Details of the detector design and results of performance studies for neutron identification and reconstruction are shown. Comparison of simulations for different options of the HGND layout at the BM@N is presented. Several proposed methods of neutron reconstruction including machine learning and cluster methods are discussed.

Figures

Figures reproduced from arXiv: 2412.00455 by the authors.

Figure 1
Figure 1. Schematic view of the BM@N experiment. The legend has list of sub-detectors [15]. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Left (a): Top view of position options for the HGND at BM@N experimental area. Position [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Left: schematic view of HGND structure. Right: The HGND active layer view. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: The schematic view of HGND support structure: the view from the beam point (left), the side [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Left: neutron rapidity spectra for different positions of the HGND. Right: neutron transverse [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Left: primary neutrons at the HGND surface. Right: background neutron on all surfaces of [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: The result of simulation for neutron detection efficiency study. Red dots: single HGND option, [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Left: Mean value of energy reconstructed for different energy of incident neutrons for 7m [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Left: Neutron multiplicity for Bi+Bi @ 3.8A GeV for 2-arms HGND option. Right: single [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: The number of single neutron events for Xe+CsI @ 3.8A GeV is about 12.7% of [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 11
Figure 11. Figure 11: Example of an event with a neutron (kinetic energy of 1.58 GeV) hitting the HGND. Red [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]
Figure 12
Figure 12. Figure 12: Efficiency of neutron reconstruction at purity=0.7 [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]
Figure 13
Figure 13. Figure 13: Dependence of reconstructed kinetic energy of the neutron on its simulated energy using the [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 12 canonical work pages

  1. [1]

    G. F. Burgio, I. Vidana, The Equation of State of Nuclear Matter : from Finite Nuclei to Neutron Stars, Universe 6 (8) (2020) 119. arXiv:2007.04427, doi:10.3390/universe6080119

    arXiv 2020

  2. [2]

    Elliptic Flow: Transition from out-of-plane to in-plane Emission in Au + Au Collisions

    C. Pinkenburg, et al., Elliptic flow: Transition from out-of-plane to in-plane emission in Au + Au col- lisions, Phys. Rev. Lett. 83 (1999) 1295–1298. arXiv:nucl-ex/9903010, doi:10.1103/PhysRevLett. 83.1295

    work page Pith review arXiv 1999

  3. [3]

    Liu, et al., Sideward flow in Au + Au collisions between 2-A-GeV and 8-A-GeV, Phys

    H. Liu, et al., Sideward flow in Au + Au collisions between 2-A-GeV and 8-A-GeV, Phys. Rev. Lett. 84 (2000) 5488–5492. arXiv:nucl-ex/0005005, doi:10.1103/PhysRevLett.84.5488

    arXiv 2000

  4. [4]

    Differential Elliptic Flow in 2 - 6 AGeV Au+Au Collisions: A New Constraint for the Nuclear Equation of State

    P. Chung, et al., Differential elliptic flow in 2-A-GeV - 6-A-GeV Au+Au collisions: A New constraint for the nuclear equation of state, Phys. Rev. C 66 (2002) 021901. arXiv:nucl-ex/0112002, doi: 10.1103/PhysRevC.66.021901

    work page Pith review arXiv 2002

  5. [5]

    Sorensen, et al., Dense nuclear matter equation of state from heavy-ion collisions, Prog

    A. Sorensen, et al., Dense nuclear matter equation of state from heavy-ion collisions, Prog. Part. Nucl. Phys. 134 (2024) 104080. arXiv:2301.13253, doi:10.1016/j.ppnp.2023.104080

    arXiv 2024

  6. [6]

    Leifels, et al., Exclusive studies of neutron and charged particle emission in collisions of Au-197 + Au-197 at 400-MeV/nucleon, Phys

    Y. Leifels, et al., Exclusive studies of neutron and charged particle emission in collisions of Au-197 + Au-197 at 400-MeV/nucleon, Phys. Rev. Lett. 71 (1993) 963–966. doi:10.1103/PhysRevLett. 71.963

    doi:10.1103/physrevlett 1993

  7. [7]

    Lambrecht, et al., Energy dependence of collective flow of neutrons and protons in Au-197 + Au-197 collisions, Z

    D. Lambrecht, et al., Energy dependence of collective flow of neutrons and protons in Au-197 + Au-197 collisions, Z. Phys. A 350 (1994) 115–120. doi:10.1007/BF01290679

    doi:10.1007/bf01290679 1994

  8. [8]

    Long, G.-F

    X.-X. Long, G.-F. Wei, Effects of incompressibility K0 in heavy-ion collisions at intermediate ener- gies, Phys. Rev. C 109 (5) (2024) 054619. arXiv:2402.12912, doi:10.1103/PhysRevC.109.054619

    arXiv 2024

Show all 22 references
  1. [9]

    Senger, The heavy-ion program at the upgraded Baryonic Matter@Nuclotron Experiment at NICA, PoS CPOD2021 (2022) 033

    P. Senger, The heavy-ion program at the upgraded Baryonic Matter@Nuclotron Experiment at NICA, PoS CPOD2021 (2022) 033. doi:10.22323/1.400.0033

    2022 doi

  2. [10]

    Kapishin, Studies of baryonic matter at the BM@N experiment (JINR), Nucl

    M. Kapishin, Studies of baryonic matter at the BM@N experiment (JINR), Nucl. Phys. A 982 (2019) 967–970. doi:10.1016/j.nuclphysa.2018.07.014

    2019 doi

  3. [11]

    Randrup, J

    J. Randrup, J. Cleymans, Maximum freeze-out baryon density in nuclear collisions, Phys. Rev. C 74 (2006) 047901. arXiv:hep-ph/0607065, doi:10.1103/PhysRevC.74.047901

    2006 arXiv

  4. [12]

    Friman, W

    B. Friman, W. Norenberg, V. D. Toneev, The Quark condensate in relativistic nucleus-nucleus collisions, Eur. Phys. J. A 3 (1998) 165–170. arXiv:nucl-th/9711065, doi:10.1007/s100500050163

    1998 arXiv

  5. [13]

    V. D. Kekelidze, R. Lednicky, V. A. Matveev, I. N. Meshkov, A. S. Sorin, G. V. Trubnikov, Three stages of the NICA accelerator complex, Eur. Phys. J. A 52 (8) (2016) 211. doi:10.1140/epja/ i2016-16211-2

    2016 doi

  6. [14]

    URL http://nica.jinr.ru/files/BM@N/BMN_CDR.pdf 13

    BM@N Conceptual Design Report. URL http://nica.jinr.ru/files/BM@N/BMN_CDR.pdf 13

  7. [15]

    Afanasiev, et al., The bm@n spectrometer at the nica accelerator complex (2024)

    S. Afanasiev, et al., The bm@n spectrometer at the nica accelerator complex (2024). arXiv: 2312.17573. URL https://arxiv.org/abs/2312.17573

    2024

  8. [16]

    Guber, et al., Development of high granular neutron time-of-flight detector for the bm@n exper- iment (2023)

    F. Guber, et al., Development of high granular neutron time-of-flight detector for the bm@n exper- iment (2023). arXiv:2309.09610. URL https://arxiv.org/abs/2309.09610

    2023 arXiv

  9. [17]

    Finogeev, F

    D. Finogeev, F. Guber, A. Izvestnyy, N. Karpushkin, A. Makhnev, S. Morozov, D. Serebryakov, Development of a 100 ps tdc based on a kintex 7 fpga for the high granular neutron time-of- flight detector for the bm@n experiment, Nuclear Instruments and Methods in Physics Research ...

    2024

  10. [18]

    R. Brun, F. Bruyant, F. Carminati, S. Giani, M. Maire, A. McPherson, G. Patrick, L. Urban, GEANT: Detector Description and Simulation Tool; Oct 1994, CERN Program Library, CERN, Geneva, 1993, long Writeup W5013. doi:10.17181/CERN.MUHF.DMJ1. URL https://cds.cern.ch/record/1082634

    1994

  11. [19]

    URL https://git.jinr.ru/nica/bmnroot

    Simulation and analysis framework for the bm@n experiment of the nica project. URL https://git.jinr.ru/nica/bmnroot

  12. [20]

    Baznat, A

    M. Baznat, A. Botvina, G. Musulmanbekov, V. Toneev, V. Zhezher, Monte-carlo generator of heavy ion collisions dcm-smm, Physics of Particles and Nuclei Letters 17 (3) (2020) 303–324. doi: 10.1134/s1547477120030024. URL http://dx.doi.org/10.1134/S1547477120030024

    2020 doi

  13. [21]

    Guber, A

    F. Guber, A. Ivashkin, N. Karpushkin, A. Makhnev, S. Morozov, D. Serebryakov, V. Baskov, V. Polyansky, Measurement of time resolution of scintillation detectors with eqr-15 silicon photode- tectors for the time-of-flight neutron detector of the bm@n experiment (2023). arXiv:23...

    2023 arXiv

  14. [22]

    T. Kipf, M. Welling, Semi-supervised classification with graph convolutional networks (09 2016). doi:10.48550/arXiv.1609.02907. 14

    2016 doi arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.