arxiv: 2604.23177 · v1 · submitted 2026-04-25 · ⚛️ physics.ins-det · nucl-ex

Recognition: unknown

Readout and PID using AIML for SoLID High Background Cherenkov Detectors

Zhiwen Zhao , Bishnu Karki , Bo Yu , Andrew Smith , Gary Swift , Simon Gorbaty , Jingyi Zhou , Haiyan Gao

show 5 more authors

Benjamin Raydo Alexandre Camsonne Kishansingh Rajput Marco Contalbrigo Roberto Malaguti

Authors on Pith no claims yet

Pith reviewed 2026-05-08 07:02 UTC · model grok-4.3

classification ⚛️ physics.ins-det nucl-ex

keywords SoLIDCherenkov detectorparticle identificationMAROC readoutmultilayer perceptronpion kaon separationhigh backgroundGeant4 simulation

0 comments

The pith

Quad and pixel readouts combined with multilayer perceptrons achieve over 90% efficiency for pion and kaon identification in high-background Cherenkov detectors.

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

The paper presents a MAROC-based sum readout system for multianode photomultiplier tubes in the SoLID Cherenkov detectors, generating pixel, quadrant, and total signals. Bench tests confirm it handles the high rates expected at SoLID with good performance. Geant4 simulations show that simple photoelectron counting fails for pi/K separation amid background, but multilayer perceptron models using the quad and pixel data succeed with efficiencies above 90%, beating PMT-only approaches. This matters for experiments needing reliable particle identification in noisy, high-rate settings like SoLID at Jefferson Lab.

Core claim

The quad and pixel readout schemes with multilayer perceptron models trained on MAROC data achieve pion and kaon identification efficiencies above 90%, significantly better than PMT-only readout or simple photoelectron cuts under realistic beam-related background conditions simulated with Geant4 for the heavy-gas Cherenkov detector.

What carries the argument

The MAROC sum readout electronics that provide simultaneous pixel, quadrant-sum, and total-sum signals from multianode PMTs, paired with multilayer perceptron AI models for pattern-based particle identification.

If this is right

The MAROC system sustains rates at or above SoLID expectations while maintaining acceptable pedestal behavior and signal linearity.
Multilayer perceptron models perform substantially better than simple photoelectron-counting cuts for pi/K separation.
The combination of high-rate electronics and AIML-based recognition offers a practical path to robust PID in the SoLID environment.
Quad and pixel schemes clearly outperform the PMT-only readout scheme.

Where Pith is reading between the lines

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

If the simulated performance holds in real data, this readout and AI method could be adapted to other high-background Cherenkov or similar detectors.
Using AI pattern recognition might allow simpler hardware designs by compensating with software in future experiments.
Further training on more varied background conditions could improve robustness for varying beam intensities.

Load-bearing premise

The Geant4 simulations accurately capture the real detector response, background conditions, and signal patterns that will occur during actual SoLID running.

What would settle it

Measuring the actual pion and kaon identification efficiencies in the SoLID Cherenkov detectors during beam operation and comparing them to the simulated values above 90%.

Figures

Figures reproduced from arXiv: 2604.23177 by Alexandre Camsonne, Andrew Smith, Benjamin Raydo, Bishnu Karki, Bo Yu, Gary Swift, Haiyan Gao, Jingyi Zhou, Kishansingh Rajput, Marco Contalbrigo, Roberto Malaguti, Simon Gorbaty, Zhiwen Zhao.

**Figure 1.** Figure 1: SoLID open geometry setup in Geant4 simulation. view at source ↗

**Figure 2.** Figure 2: Engineering design of HGC. The top left CAD picture shows the front view without front view at source ↗

**Figure 3.** Figure 3: (Top) photos of the MAROC sum readout system. It includes an adapter board, an ASIC view at source ↗

**Figure 4.** Figure 4: (Left) The schematic representation of the summing mechanism of MAROC sum board view at source ↗

**Figure 5.** Figure 5: Saturation study of MAROC sum board with one pixel fired. The number of photoelec view at source ↗

**Figure 6.** Figure 6: Duration of hit distribution from pixel readout of the MAROC sum board at 3 different view at source ↗

**Figure 7.** Figure 7: A number of photoelectrons (NPE) distribution with TDC threshold +30 DAC unit above view at source ↗

**Figure 8.** Figure 8: Schematic layout of the bench test. The entire setup was placed inside the black box to view at source ↗

**Figure 9.** Figure 9: Comparison between the sum and pixel scaler rates. The agreement between the two view at source ↗

**Figure 10.** Figure 10: Left: TDC counts VS sum ADC integral. Right: TDC counts vs quad ADC integral. view at source ↗

**Figure 11.** Figure 11: Schematic layout for the LED ring test. For signals, lights from the pulsed LED was view at source ↗

**Figure 12.** Figure 12: Top: Number of photoelectron distribution for LED ring with pixel readout. The green view at source ↗

**Figure 13.** Figure 13: The accuracy for the Hough transformation algorithm at background of 370kHz/pixel. view at source ↗

**Figure 14.** Figure 14: Parameters of the recognized rings using the Hough transformation for pixel readout. view at source ↗

**Figure 15.** Figure 15: Left: Schematic layout for the cosmic ring test. Two scintillator bars, one at the top view at source ↗

**Figure 16.** Figure 16: Top: Number of photoelectron distribution for cosmic data with pixel readout. The view at source ↗

**Figure 17.** Figure 17: Three sample pion and three sample kaon HGC events with beam background in three view at source ↗

**Figure 18.** Figure 18: HGC pion and kaon NPE counts (top) and NPE cut efficiency (bottom) at momentum view at source ↗

**Figure 19.** Figure 19: Tag–prediction difference distributions (left) and AUC ROC (right) for PMT (top), view at source ↗

read the original abstract

We present the development of readout electronics and artificial-intelligence-based particle-identification methods for the SoLID Cherenkov detectors at Jefferson Lab. To operate in the high-rate, high-background SoLID environment, we designed a MAROC sum readout system for multianode photomultiplier tubes that provides simultaneous pixel, quadrant-sum, and total-sum signals. Bench studies show that the system can sustain rates at or above those expected for SoLID while maintaining acceptable pedestal behavior and signal linearity. Using realistic Geant4 simulations for the heavy-gas Cherenkov detector, we then investigate $\pi/K$ separation with beam-related background. A simple photoelectron-counting cut is insufficient under these conditions, whereas multilayer perceptron models trained on PMT, quad, and pixel readout data perform substantially better. The quad and pixel readout schemes achieve pion and kaon efficiencies above 90\% and clearly outperform PMT-only readout. These results demonstrate that the combination of high-rate MAROC sum electronics and AIML-based pattern recognition provides a practical path toward robust SoLID Cherenkov PID.

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.

Desk Editor's Note private letter to a colleague

MAROC multi-granularity readout plus MLP on pixel/quad signals delivers >90% simulated pi/K efficiency for SoLID Cherenkov under background, with bench-backed electronics.

read the letter

The key point is that this paper shows a practical MAROC sum readout for multianode PMTs that feeds pixel and quadrant signals into simple multilayer perceptrons, yielding over 90% pion and kaon efficiencies in Geant4 simulations of the SoLID heavy-gas Cherenkov with realistic beam background, while outperforming basic PMT counting. Bench tests separately confirm the electronics handle the target rates with stable pedestals and linearity. That combination is the concrete advance here. The multi-level readout supplies the spatial pattern information the models use to separate signal from background, and the simulations quantify the gain clearly enough to see the effect. The hardware work is directly testable and the methods are described at a level that lets others replicate the setup. The main limitation is that the PID performance numbers rest on simulation alone. No beam data validates how well the trained models hold up once real detector response and background fluctuations are in play, so the claimed advantage could move once those conditions are measured. The paper does not overstate this; it frames the results as showing a workable path rather than a finished solution. This is aimed at the SoLID collaboration and other groups building high-rate Cherenkov detectors in nuclear physics. Detector physicists who need to extract PID from multi-anode tubes in noisy environments will find the readout design and the choice of MLP inputs useful. I would send it to peer review. The hardware claims are grounded in bench data, the simulation results are transparent, and the problem it addresses is real for the experiment.

Referee Report

0 major / 1 minor

Summary. The manuscript presents the design of a MAROC sum readout system for multianode PMTs in the SoLID Cherenkov detectors that simultaneously provides pixel, quadrant-sum, and total-sum signals. Bench tests confirm the system sustains rates at or above SoLID expectations with acceptable pedestal and linearity performance. Geant4 simulations of the heavy-gas Cherenkov detector with beam-related background then show that multilayer perceptron models trained on the quad and pixel readout data achieve pion and kaon identification efficiencies above 90 percent and substantially outperform both a simple photoelectron-counting cut and PMT-only readout.

Significance. If the simulated performance gains hold under real conditions, the work supplies a practical hardware-plus-ML solution for robust PID in high-rate, high-background Cherenkov detectors. The independent bench-test results strengthen the electronics claims, while the quantified comparison of readout schemes in simulation provides clear evidence of the advantage of the multi-level MAROC architecture.

minor comments (1)

[Geant4 simulation and PID results sections] The PID performance claims rest on Geant4 simulations; adding a short paragraph on how the simulation parameters (background spectra, detector response, etc.) were validated or tuned against existing data would help readers gauge how robust the reported >90 percent efficiencies are likely to remain when applied to actual SoLID running conditions.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the thorough review and the positive recommendation to accept the manuscript. The summary accurately captures the key contributions of the MAROC sum readout electronics and the AIML-based PID approach for the SoLID heavy-gas Cherenkov detector.

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper reports bench measurements of MAROC electronics and Geant4-based simulations of Cherenkov detector response under beam-related background. PID performance is obtained by training and evaluating multilayer perceptron models on the simulated PMT/quad/pixel signals; this constitutes standard supervised learning on independent Monte Carlo outputs rather than any self-referential fit or definition. No equations reduce a claimed prediction to its own inputs by construction, no load-bearing self-citations are invoked for uniqueness or ansatz, and no known empirical patterns are merely renamed. The derivation chain is therefore self-contained against external simulation benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on the domain assumption that Geant4 faithfully models the detector and background; no explicit free parameters or new entities are introduced in the abstract.

axioms (1)

domain assumption Geant4 Monte Carlo simulation accurately reproduces the physical processes, light production, and detector responses in the heavy-gas Cherenkov detector under beam-related background
All PID performance numbers are derived from data generated by these simulations.

pith-pipeline@v0.9.0 · 5539 in / 1273 out tokens · 92180 ms · 2026-05-08T07:02:55.241366+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

12 extracted references · 1 canonical work pages

[1]

SoLID Project, Jefferson Lab.https://solid.jlab.org/
[2]

The solenoidal large intensity device (solid) for jlab 12 gev.Journal of Physics G: Nuclear and Particle Physics, 50(11):110501, sep 2023

J Arrington, J Benesch, A Camsonne, J Caylor, J-P Chen, S Covrig Dusa, A Emmert, G Evans, H Gao, J-O Hansen, G M Huber, S Joosten, V Khachatryan, N Liyanage, Z-E Meziani, M Nycz, C Peng, M Paolone, W Seay, P A Souder, N Sparveris, H Spiesberger, Y Tian, E Voutier, J Xie, W Xiong, Z-Y Ye, Z Ye, J Zhang, Z-W Zhao, X Zheng, and For the Jefferson Lab SoLID Co...

2023
[3]

C. Peng, J. Xie, S. Joosten, Z.-E. Meziani, M. Jones, E. Kaczanowicz, M. Rehfuss, N. Sparveris, M. Paolone, M. Foley, M. Minot, and M. Popecki. Performance of photosensors in a high-rate environment for gas cherenkov detectors.Journal of Instrumentation, 17(08):P08022, aug 2022

2022
[4]

M. Contalbrigoet al.The clas12 ring imaging cherenkov detector.Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 964:163791, 2020

2020
[5]

MAROC, a generic photomultiplier readout chip.IEEE Nucl

Sylvie Blin, Pierre Barrillon, and Christophe de La Taille. MAROC, a generic photomultiplier readout chip.IEEE Nucl. Sci. Symp. Conf. Rec., 2010:1690–1693, 2010

2010
[6]

Paulo Bedaque, Amber Boehnlein, Mario Cromaz, Markus Diefenthaler, Latifa Elouadrhiri, Tanja Horn, Michelle Kuchera, David Lawrence, Dean Lee, Steven Lidia, Robert McKeown, Wally Melnitchouk, Witold Nazarewicz, Kostas Orginos, Yves Roblin, Michael Scott Smith, Malachi Schram, and Xin-Nian Wang. A.I. for nuclear physics.Eur. Phys. J. A, 57(3):100, 2021

2021
[7]

Kuchera, Dean Lee, Witold Nazarewicz, Peter Ostroumov, Kostas Orginos, Alan Poon, Xin-Nian Wang, Alexander Scheinker, Michael S

Amber Boehnlein, Markus Diefenthaler, Nobuo Sato, Malachi Schram, Veronique Ziegler, Cris- tiano Fanelli, Morten Hjorth-Jensen, Tanja Horn, Michelle P. Kuchera, Dean Lee, Witold Nazarewicz, Peter Ostroumov, Kostas Orginos, Alan Poon, Xin-Nian Wang, Alexander Scheinker, Michael S. Smith, and Long-Gang Pang. Colloquium: Machine learning in nu- clear physics...

2022
[8]

Machine learning for imaging Cherenkov detectors.J

Cristiano Fanelli. Machine learning for imaging Cherenkov detectors.J. Instrum., 15(02):C02012, 2020

2020
[9]

Cisbani, A

E. Cisbani, A. Del Dotto, C. Fanelli, M. Williams, M. Alfred, F. Barbosa, L. Barion, V. Berd- nikov, W. Brooks, T. Cao, M. Contalbrigo, S. Danagoulian, A. Datta, M. Demarteau, A. Denisov, M. Diefenthaler, A. Durum, D. Fields, Y. Furletova, C. Gleason, M. Grosse- Perdekamp, M. Hattawy, X. He, H. van Hecke, D. Higinbotham, T. Horn, C. Hyde, Y. Ilieva, G. Ka...

2020
[10]

D. H. Dongwi, C.-J. Na¨ ım, L. Rhode, and A. Deshpande. Hybrid machine-learning particle identification for the epic proximity-focusing rich.arXiv:2512.14598, 2025

work page arXiv 2025
[11]

DeepRICH: Learning Deeply Cherenkov Detectors.Mach

Cristiano Fanelli and Jary Pomponi. DeepRICH: Learning Deeply Cherenkov Detectors.Mach. Learn.: Sci. Technol., 1:015010, 11 2020. 22

2020
[12]

FastDIRC: a fast Monte Carlo and reconstruction algorithm for DIRC detectors.J

John Hardin and Mike Williams. FastDIRC: a fast Monte Carlo and reconstruction algorithm for DIRC detectors.J. Instrum., 11(10):P10007, 2016. 23

2016