Pith. sign in

REVIEW 4 major objections 4 minor 4 references

A LArTPC data-selection pipeline using only ionization charge can identify rare Michel-electron signals fast enough to run online, demonstrated on real MicroBooNE data.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 00:10 UTC pith:6VPONMJW

load-bearing objection A credible first charge-only LArTPC online-selection proof-of-principle, whose 'keeps up' claim leans on an average the paper's own tail contradicts—but the algorithm demonstration survives. the 4 major comments →

arxiv 2602.11138 v2 pith:6VPONMJW submitted 2026-02-11 hep-ex

Demonstration and performance of an online data selection algorithm for liquid argon time projection chambers using MicroBooNE

MicroBooNE collaboration: P. Abratenko , D. Andrade Aldana , L. Arellano , J. Asaadi , A. Ashkenazi , S. Balasubramanian , B. Baller , A. Barnard
show 185 more authors
G. Barr D. Barrow J. Barrow V. Basque J. Bateman B. Behera O. Benevides Rodrigues S. Berkman A. Bhat M. Bhattacharya V. Bhelande A. Binau M. Bishai A. Blake B. Bogart T. Bolton M.B. Brunetti L. Camilleri D. Caratelli F. Cavanna G. Cerati A. Chappell Y. Chen J.M. Conrad M. Convery L. Cooper-Troendle J.I. Crespo-Anadon R. Cross M. Del Tutto S.R. Dennis P. Detje R. Diurba Z. Djurcic K. Duffy S. Dytman B. Eberly P. Englezos A. Ereditato J.J. Evans C. Fang B.T. Fleming W. Foreman D. Franco A.P. Furmanski F. Gao D. Garcia-Gamez S. Gardiner G. Ge S. Gollapinni E. Gramellini P. Green H. Greenlee L. Gu W. Gu R. Guenette P. Guzowski L. Hagaman M. D. Handley O. Hen A. Hergenhan M. Harrison S. Hawkins C. Hilgenberg G.A. Horton-Smith A. Hussain B. Irwin M.S. Ismail C. James X. Ji J.H. Jo A. Johnson R.A. Johnson D. Kalra G. Karagiorgi W. Ketchum A. Kelly M. Kirby T. Kobilarcik K. Kumar N. Lane J.-Y. Li Y. Li K. Lin B.R. Littlejohn L. Liu S. Liu W.C. Louis X. Luo T. Mahmud N. Majeed C. Mariani J. Marshall N. Martinez D.A. Martinez Caicedo F. Martinez Lopez M. G. Manuel Alves S. Martynenko A. Mastbaum I. Mawby N. McConkey B. McConnell L. Mellet J. Mendez J. Micallef T. Mohayai A. Mogan M. Mooney A.F. Moor C.D. Moore L. Mora Lepin M. A. Hernandez Morquecho M.M. Moudgalya S. Mulleria Babu D. Naples A. Navrer-Agasson N. Nayak M. Nebot-Guinot C. Nguyen L. Nguyen J. Nowak N. Oza O. Palamara N. Pallat V. Paolone A. Papadopoulou V. Papavassiliou H. Parkinson S.F. Pate N. Patel Z. Pavlovic E. Piasetzky K. Pletcher I. Pophale X. Qian J.L. Raaf V. Radeka A. Rafique M. Reggiani-Guzzo J. Rodriguez Rondon M. Rosenberg M. Ross-Lonergan I. Safa D.W. Schmitz A. Schukraft W. Seligman M.H. Shaevitz R. Sharankova J. Shi L. Silva E.L. Snider S. Soldner-Rembold J. Spitz M. Stancari J. St. John T. Strauss A.M. Szelc N. Taniuchi K. Terao C.Thorpe D. Torbunov D. Totani M. Toups A. Trettin Y.-T. Tsai J. Tyler M.A. Uchida T. Usher B. Viren J. Wang L. Wang M. Weber H. Wei A.J. White S. Wolbers T. Wongjirad K. Wresilo W. Wu E. Yandel T. Yang L.E. Yates H.W. Yu G.P. Zeller J. Zennamo C. Zhang Y. Zhang
This is my paper
classification hep-ex
keywords LArTPConline data selectiontrigger primitiveMichel electrondata acquisitionMicroBooNEreal-time processingcharge-only readout
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper aims to prove that online data selection for liquid argon time projection chambers (LArTPCs) is feasible using only ionization charge information, without relying on scintillation light. It demonstrates an emulated three-stage software pipeline on pre-recorded MicroBooNE data that selects Michel electrons from stopping cosmic-ray muons—a rare, low-energy topology that resembles supernova neutrino signals. The authors show the pipeline runs quickly enough, on average, to keep pace with the detector's streaming data, and that its selection efficiency on real data matches simulation. This matters because future large LArTPCs like DUNE will generate terabytes per second and need fast, intelligent triggers to preserve rare physics without drowning in data.

Core claim

The central claim is that a physics-topology-sensitive, TPC-based data selection algorithm can operate online using charge information exclusively, and that this is the first demonstration of such a system on real LArTPC data. The algorithm identifies stopping muon decays by detecting two topological features: a Bragg peak—the localized energy spike where the muon stops—and a kink angle between the muon track and the emerging Michel electron. Validated on 202 seconds of continuously streamed MicroBooNE supernova data, the emulated pipeline retains 21% of drift regions as Michel candidates, consistent with CORSIKA cosmic-ray simulation, and achieves an average total processing time of 1315 mi

What carries the argument

The load-bearing mechanism is a three-stage software pipeline: (1) Trigger Primitive generation, which converts per-channel regions-of-interest into compact summaries (charge, amplitude, width) and stitches frames into 2.3 ms drift regions; (2) Trigger Candidate generation, which subdivides each drift region into slices, selects a candidate Bragg peak by charge and amplitude cuts, searches for track segments on either side, and requires a kink angle greater than 30 degrees; and (3) a High-Level Trigger that retrieves the full ROI data for selected frames. The Bragg-peak-plus-kink-angle selection is the physics discriminator that distinguishes rare stopping-muon decays from abundant crossing

Load-bearing premise

The central claim collapses if the assumption is wrong that keeping the average processing time below the 2.3 ms per-drift-region deadline guarantees the pipeline never falls behind; the measured processing-time tail extends past 6 ms, which would cause buffer buildup in a real streaming system.

What would settle it

Run the pipeline on a live data stream (not replayed from memory) and count the fraction of drift regions whose total processing time exceeds 2.3 ms; if that fraction is non-negligible, data buffers will fill and the no-backlog claim is contradicted. Alternatively, instrument the SEB buffer occupancy during a true streaming test and show it grows without bound.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Future LArTPCs such as SBND and DUNE can adopt charge-only online selection to drastically reduce data rates while preserving rare low-energy signatures.
  • The three-stage architecture—primitives, candidates, high-level trigger—mirrors DUNE's proposed trigger design, making this a direct proof-of-principle for that strategy.
  • Because Michel electrons mimic supernova-neutrino electron signatures, demonstrated online selection of this topology implies similar pipelines could target supernova bursts in off-beam data.
  • The framework's ability to run within latency limits on a commodity server suggests that CPU-based intelligent triggering is a realistic alternative to pure hardware triggers for large LArTPCs.
  • The same algorithm can serve as a low-energy calibration source, enabling high-statistics Michel-electron samples for energy calibration in future detectors.

Where Pith is reading between the lines

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

  • The paper's 'keeps up without backing up' conclusion depends on the mean processing time (1315 μs) being under the 2.3 ms deadline, but the measured distribution extends to roughly 6000 μs; a real streaming system must also control the tail, not just the average, to avoid buffer overflow.
  • The emulation reads ROI data directly from memory after network transfer, so the network and disk I/O path—a likely bottleneck in genuine real-time operation—is never stressed; a hardware-in-the-loop test would be a stronger validation.
  • The bit-flip correction and short-waveform removal are data-quality steps specific to MicroBooNE's SN stream; porting the pipeline elsewhere will require equivalent preprocessing, which could be non-trivial in firmware.
  • If tail-latency is brought under control, the same three-stage design could be extended to other rare topologies—neutron-antineutron transitions, proton decay, or beam neutrinos—by replacing the Bragg/kink selection with appropriate discriminators.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper describes a multi-stage data selection algorithm for LArTPC detectors, targeting Michel electrons from stopping cosmic-ray muons using only TPC ionization charge information. The algorithm (trigger primitive generation, trigger candidate generation, and high-level trigger) is implemented on an 'emulated online' stream of pre-recorded MicroBooNE supernova data and validated with Monte Carlo samples and data-MC comparisons. The authors report a selection efficiency of about 20% for stopping muons, a processing time distribution with a mean of 1315 μs, and claim that this demonstrates, for the first time, online TPC-based data selection in a LArTPC using real data and charge information exclusively.

Significance. If the claims are substantiated, this work is an important proof-of-principle for online data selection in future large LArTPC experiments such as SBND and DUNE. The paper uses real detector data, provides a detailed description of the algorithm and its implementation, and includes data-MC validation. It also transparently discusses several data-quality issues (bit-flips, short waveforms) and their mitigation. However, the central claims of 'online' feasibility and 'keeping up without backing up' are weakened by the issues raised below, and the reported efficiency may be optimistic because the algorithm is tuned on the same Monte Carlo sample used for the efficiency measurement.

major comments (4)
  1. [Abstract and Sec. 2.2] The abstract claims the 'first demonstration of online data selection in a LArTPC using real data and charge information exclusively,' but Sec. 2.2 explicitly describes an 'emulated online' system in which pre-recorded ROI data are read directly from memory after network transfer. The actual network streaming path and finite DAQ buffers are not exercised. The claim should be qualified as 'emulated online' in the abstract and conclusions, or the demonstration should include the real-time network/streaming interface.
  2. [Sec. 4.2.3, Fig. 17] The statement 'Since the average processing time is below 2.3 ms, the system processes events faster than they arrive and therefore keeps up without backing up' is not correct. For a deterministic arrival process with service times that sometimes exceed the interarrival time (Fig. 17 shows a tail extending to ~6000 μs), a queue can build up even if the mean service time is below 2.3 ms. The mean being below the deadline guarantees only long-run stability, not bounded latency or absence of buffer overflow. The paper needs to report the tail probability, simulate the actual queue with finite buffers, or otherwise demonstrate that no data loss occurs.
  3. [Sec. 4.1 and Table 2] The algorithm thresholds (Bragg peak charge window 850–2150 ADC, maximum amplitude 100–200 ADC, kink angle >30°) were tuned using the same stopping-muon Monte Carlo sample on which the efficiency is reported. This is an in-sample estimate and may be optimistic. The paper should validate the algorithm on an independent sample (e.g., a withheld portion of the stopping-muon sample or a dedicated validation sample) and report the true-positive efficiency on the CORSIKA sample, which is currently only used to quote the fraction of all drift regions selected.
  4. [Sec. 4.2.2 and 4.2.3] The paper never quantifies the final data reduction achieved by the full pipeline. Process 2 retains 21% of drift regions (Sec. 4.2.2), but the output rate of Process 3 and the resulting fraction of data stored are not reported. Since the stated motivation is reducing data rates and storage, the paper should report the number of TCs passing the high-level trigger, the false-positive rate in real data, and the data-volume reduction relative to storing all SN data.
minor comments (4)
  1. [Sec. 3.3, Fig. 7] The data-MC comparison is described as showing 'good shape agreement,' but the distributions show visible shifts in integrated charge and width. Consider adding a quantitative comparison (e.g., chi-square or Kolmogorov-Smirnov) to support this statement.
  2. [Sec. 4.1, Table 2] The 'selection efficiency' is defined as the fraction of drift regions passing the selection, not the efficiency per Michel electron. This should be clarified in the text to avoid ambiguity, especially since a drift region can contain multiple muons.
  3. [Sec. 3.1] The bit-flip correction threshold of 50 ADC is described as empirical. It would be useful to state the sensitivity of the final selection to this threshold, or at least note that it was checked.
  4. [Sec. 2.1] Minor wording: 'This lossy-compressed data stream'—the stream is first zero-suppressed (lossy) and then Huffman-compressed (lossless); consider rewording to 'zero-suppressed and compressed data stream.'

Circularity Check

0 steps flagged

No significant circularity: the paper reports an empirical demonstration with MC-validated selection and measured timing; no derived result is equivalent to a fitted input.

full rationale

The paper's central outputs are (1) a measured performance characterization of a TPC-based data selection algorithm and (2) a demonstrated timing profile under an emulated online workflow. Neither is derived from the algorithm's inputs in a way that reduces to those inputs by construction. The selection thresholds in Sec. 4.1 were manually optimized on a subset of stopping-muon Monte Carlo event displays and then applied to independent MC samples and to real MicroBooNE SN data; the reported efficiencies in Table 2 are empirical characterizations, not fitted parameters renamed as predictions. The data-to-MC comparison in Sec. 3.3 is explicitly a validation/consistency check and is not used to fit the central selection result. The online-feasibility statement in Sec. 4.2.3 ('Since the average processing time is below 2.3 ms, the system processes events faster than they arrive and therefore keeps up without backing up') is an inference from the measured processing-time distribution, and the presence of a tail above 2.3 ms in Fig. 17 is a legitimate correctness or queueing-model concern, but it is not circularity: the conclusion is not equivalent to its input by construction. Self-citations, such as Ref. [2] for the continuous readout stream and Ref. [12] for trigger development, are contextual descriptions of the MicroBooNE detector and prior trigger work; they are not load-bearing in the sense of being the sole justification for a derived result. No uniqueness theorem, no ansatz smuggled in via citation, and no renaming of a known result were found. The paper is best assessed as a self-contained engineering demonstration whose central claims rest on measured and simulated performance, not on a circular derivation chain.

Axiom & Free-Parameter Ledger

8 free parameters · 5 axioms · 0 invented entities

The central claim of online feasibility rests on hardware/queueing assumptions (average processing time vs tail) and on MC realism; the selection thresholds are tuned parameters rather than derived constants. No new physical entities are introduced.

free parameters (8)
  • Bragg peak integrated charge window = 850-2150 ADC counts
    Manually optimized on stopping-muon MC (Sec. 4.1 Step 2, Fig. 9).
  • Bragg peak max amplitude window = 100-200 ADC counts
    Chosen in the same manual scan over 300 event displays (Sec. 4.1 Step 2).
  • Kink angle threshold = >30 degrees
    Chosen from MC efficiency vs false-positive trade-off (Table 2).
  • Bit-flip correction threshold = 50 ADC counts
    Empirically chosen to balance sensitivity vs noise (Sec. 3.1).
  • Short waveform removal length = <20 ADC words
    Based on presample/postsample (14) plus shaping-time requirement (Sec. 3.1).
  • Slice size = 36 channels x 575 time ticks
    Algorithmic sub-division to limit search space (Sec. 4.1 Step 2).
  • Track search contiguous channels = 13 channels
    Requirement for track segments in each direction (Sec. 4.1 Step 3).
  • TP start-time tolerance = +/-5 time ticks
    Used to suppress delta-ray false positives (Sec. 4.1 Step 3).
axioms (5)
  • domain assumption CORSIKA Monte Carlo plus MicroBooNE detector simulation accurately reproduces the distribution of trigger primitives in the real SN data stream after pre-processing.
    Used as a validation check (Sec. 3.3, Fig. 7) before applying the algorithm; agreement is judged 'good' visually but no quantitative goodness-of-fit is given.
  • domain assumption The kink angle in channel-time space distinguishes true stopping-muon decays from crossing-muon/delta-ray backgrounds.
    The separation shown in Fig. 11 is from simulation; the paper assumes this separation transfers to real data.
  • domain assumption The emulated online system's omission of network transfer and buffering does not change the throughput conclusion.
    Sec. 2.2 states ROI data is read directly from memory 'once already transferred over the network'; the transfer step is not emulated, so real-time latency of that path is unmeasured.
  • domain assumption Three 1.6 ms frames can be stitched into two 2.3 ms drift regions without loss of causal correlations.
    Sec. 4.2.1 describes stitching; any timing misalignment could affect track-finding in Process 2.
  • domain assumption The bit-flip correction algorithm (nearest power of two) removes most artifacts without distorting signals.
    Sec. 3.1 states the 50-ADC threshold was chosen empirically; residual bit-flips are neglected.

pith-pipeline@v1.3.0-alltime-deepseek · 15907 in / 11066 out tokens · 96189 ms · 2026-08-03T00:10:24.573390+00:00 · methodology

0 comments
read the original abstract

The MicroBooNE detector is a liquid argon time projection chamber (LArTPC) that produces three-dimensional images of particle interactions using ionization charge collected by anode wire plane arrays and scintillation light collected by a light detection system. In addition to testing long-standing experimental neutrino anomalies and performing measurements of neutrino interactions with argon nuclei using the Fermilab Booster Neutrino Beam, MicroBooNE aims to develop methodologies for rare beyond the Standard Model and off-beam physics searches. Looking ahead to the upcoming Deep Underground Neutrino Experiment (DUNE), with MicroBooNE serving as a valuable testbed, achieving high sensitivity and livetime for off-beam physics while satisfying data processing and storage constraints will require data-driven, intelligent, and online or real-time data selection techniques. These techniques are essential for reducing data rates and preserving rare signals with high accuracy. In this paper, we describe a fast data selection algorithm suitable for online execution to identify electrons from stopping cosmic ray muons in the MicroBooNE detector utilizing ionization charge information, and present its performance. This represents the first demonstration of online data selection in a LArTPC using real data and charge information exclusively and provides an important proof-of-principle for applying such techniques to other LArTPC experiments such as the Short-Baseline Near Detector and DUNE.

discussion (0)

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

Reference graph

Works this paper leans on

4 extracted references · 3 linked inside Pith

  1. [14]

    Heck et al.,CORSIKA: A Monte Carlo code to simulate extensive air showers, Tech

    D. Heck et al.,CORSIKA: A Monte Carlo code to simulate extensive air showers, Tech. Rep. FZKA-6019, Forschungszentrum Karlsruhe (1998). – 22 –

    1998

  2. [15]

    ZeroMQ – The Guide

    P. Hintjens, “ZeroMQ – The Guide.”http://zguide.zeromq.org/page:all, 2010. [16]ICARUScollaboration,ICARUS at the Fermilab Short-Baseline Neutrino program: initial operation, Eur. Phys. J. C83(2023) 467 [2301.08634]

    Pith/arXiv arXiv 2010

  3. [17]

    Chung et al.,Real-time Anomaly Detection for Liquid Argon Time Projection Chambers, [2509.21817]

    S. Chung et al.,Real-time Anomaly Detection for Liquid Argon Time Projection Chambers, [2509.21817]. [18]ArgoNeuTcollaboration,A deep-learning based raw waveform region-of-interest finder for the liquid argon time projection chamber,J. Instrum.17(2022) P01018 [2103.06391]

    Pith/arXiv arXiv 2022

  4. [19]

    Y.-J. Jwa, G. Di Guglielmo, L. Arnold, L. Carloni and G. Karagiorgi,Real-Time Inference With 2D Convolutional Neural Networks on Field Programmable Gate Arrays for High-Rate Particle Imaging Detectors,Front. Artif. Intell.5(2022) 855184 [2201.05638]. – 23 –

    Pith/arXiv arXiv 2022