{"id":"c2d5de45-15bd-4f3c-976c-53c924a179c6","arxiv_id":"2506.14887","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"PSR J1823-3021A emits over 37,000 giant pulses per hour in the UHF band, with a first measured scattering time of 5.5 microseconds at 1 GHz.","lead":"Astronomers detected more than 37,000 giant radio pulses per hour from a millisecond pulsar in a globular cluster using MeerKAT's UHF receiver, about 13 times the rate seen at higher frequencies. The new data also provide the first measurement of how interstellar scattering smears this pulsar's pulses.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 37,000/hr rate depends on treating all 358 off-pulse 7<S/N<10 candidates as false positives and extrapolating their rate into C1/C2 via phase uniformity; the known C3 component from Abbate et al. (2020) and the paper's own 83 vs 100/hr inconsistency make this correction under-tested.","rationale":"I focused on the false-positive correction because it is the only systematic that can change the headline number by more than its stated uncertainty. The rest of the analysis—power-law fits, multi-peak morphologies, polarisation, scattering—is secondary and does not feed back into the rate. The spectral-index discrepancy in the abstract (≈−3 vs −1.6) is real but does not change the detection claim. The scattering measurement is first-of-its-kind but separate. The raw data are not yet public, so the proposed check cannot be independently repeated by readers, strengthening the case for CONDITIONAL rather than ACCEPT. I agree with the reader that the core result is credible; the condition is to settle the false-positive correction with a phase-resolved test. The reader's weakest assumption is the same correction; I sharpen it by pointing to the definitional circularity and the C3 prior, hence partial agreement.","tokens_in":16510,"tokens_out":12912,"duration_ms":124959,"concrete_test":"Make the C1/C2 phase boundaries and the Abbate et al. (2020) C3 phase explicit, then re-run the S/N 7–10 selection on the existing single-pulse archives, tallying candidates in independent off-pulse sub-bins (and in the C3 window) rather than one aggregate 0.838-wide bin. Test the off-pulse counts for Poisson uniformity (e.g., chi-square or KS) and recompute the expected in-window false positives from the measured per-bin rates. Also recompute the corrected rate under a bootstrap that perturbs the phase-window definition by ±1 bin. If the corrected rate stays ≥37,000/hr under all variants, the headline is robust; if not, the rate should be quoted without the 'over' qualifier.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3 estimates the number of false positives inside the C1/C2 windows as 358/0.838 × (1−0.838) ≈ 69, and bases the S/N 7–10 correction on this. Two assumptions carry the calculation: (i) every one of the 358 off-pulse candidates is a false positive, and (ii) the false-positive rate per unit pulse phase is constant across the whole period. Assumption (i) is close to definitional: candidates outside C1/C2 are called false positives, so the statement 'we found 358 false positives' cannot independently validate the absence of real off-pulse emission. The paper itself notes that Abbate et al. (2020) saw a C3 component that is not significantly detected here; if any weak C3-type GPs are present in the 7<S/N<10 band, they are silently absorbed into the 358 and the extrapolated 69 is biased. Assumption (ii) is untested and is not just a theoretical worry: xprof maximises S/N over a range of boxcar widths, and the effective trials factor can differ between the broad off-pulse baseline and the structured C1/C2 windows, changing the false-positive probability with phase. The paper's own numbers are inconsistent on this correction: Section 3 gives ≈83 false positives/hr, while the Conclusions give ≈100/hr. The corrected total of 31,134 events is only ~470 events above the value that would round to 'over 37,000/hr'; a few hundred additional real false positives inside C1/C2 would move the headline. Thus the correction is the most load-bearing step for the central claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents MeerKAT UHF-band baseband observations of the globular-cluster millisecond pulsar PSR J1823−3021A. In 2,984 s of tracking data the authors detect 9,366 giant pulses with S/N > 10 and, after a phase-based false-positive correction for the 7 < S/N < 10 interval, estimate a corrected rate of 37,000 ± 200 GPs/hr, which they state is 13.5 times the L-band rate of Abbate et al. (2020). They report that the GPs are clustered in two phase components, follow steep power-law energy distributions, are mostly unpolarised, and include double- and triple-peaked morphologies; they also report a first scattering-time measurement, τ = 5.5 ± 0.6 μs at 1 GHz, and discuss implications for the connection between giant pulses and fast radio bursts.","tokens_in":16862,"tokens_out":6541,"duration_ms":56257,"significance":"The observation is a direct and valuable measurement: it is the largest GP sample from a globular-cluster MSP at UHF, with baseband time resolution that reveals subpulse structure and enables the first scattering measurement for this pulsar. The comparison with Abbate et al. (2020) is an external benchmark, and the power-law and scattering fits are outputs rather than inputs to the main rate claim. The paper also provides useful single-pulse morphology statistics and a clear statement of the FRB-motivation context. However, the headline rate depends on an estimated false-positive subtraction whose uniformity assumption is not independently tested, and there are internal inconsistencies in the reported false-positive rate and in the spectral-index statement. These issues are fixable and should be addressed before publication.","major_comments":[{"comment":"The headline 37,000 ± 200 GPs/hr is obtained by assuming that the 358 candidates found outside the C1/C2 phase ranges (83.8% of the period) are all false positives and are uniformly distributed in phase, yielding 69 expected false positives inside C1/C2. This uniformity assumption is load-bearing and is not tested: xprof maximizes S/N over a sequence of boxcar widths, so the effective number of trials, and hence the false-positive rate, can differ between the broad off-pulse baseline and the structured C1/C2 windows. In addition, the paper itself notes that the C3 component seen by Abbate et al. (2020) is not significantly detected here; real weak GPs in the 7 < S/N < 10 band outside C1/C2 would be silently absorbed into the 358 and would bias the extrapolation. The quoted ±200 error also does not include any systematic from this correction. Please test the uniformity assumption (for example, with injected pulses or a phase-dependent trials-factor estimate), report the resulting systematic uncertainty, and reconcile the 83/hr rate quoted in Section 3 with the ≈100/hr rate quoted in Section 5.","section":"Section 3, false-positive correction"},{"comment":"The abstract states that the GPs have steep spectral indices of ≈ −3, but Section 3.2 reports that the 400 brightest GPs have a spectral-index distribution centered at −1.6 with standard deviation 0.32 and median −1.57, while the integrated profile has −3.30. This is a direct internal contradiction in a property used to motivate the UHF/L-band rate comparison. The authors should state clearly whether the steep index refers to the integrated profile or to the GPs, and adjust the abstract and Section 3.2 accordingly.","section":"Section 3.2 and the abstract"},{"comment":"The first scattering measurement is based on the 50 brightest GPs (49 after one failed fit), which are selected before the scattering fit. If the scattering time correlates with intrinsic brightness or profile width, the mean τ = 5.5 ± 0.6 μs could be biased; the paper reports only the mean and the range of α, not the distribution of τ across the 49 GPs. I request the distribution of individual τ values and a brief discussion of whether the bright-GP selection affects the mean.","section":"Section 3.6, scattering measurement"}],"minor_comments":[{"comment":"The phrase 'We found 358 false positives in a phase range of 0.838' should read 'in the 83.8% of the pulse period outside C1 and C2' to avoid ambiguity about whether 0.838 is a phase offset.","section":"Section 3"},{"comment":"The reference to 'Manch-ester et al. 2005' contains a line-break typo; please use the standard citation 'Manchester et al. (2005)'.","section":"Section 2"},{"comment":"The text says the average error on the GP spectral indices is 0.29, 'significantly larger than the measurement errors in Abbate et al. (2020) (≈0.09)'; please specify whether these are formal fit errors or the scatter of the distribution, since the two are not directly comparable.","section":"Section 3.2"},{"comment":"The sentence 'Examples of all these events were seen in Abbate et al. (2020)' is confusing because the preceding sentence quantifies double- and triple-peak GPs; please clarify which events were previously seen and which are new to this work.","section":"Section 3.4"},{"comment":"The sentence 'However, it is still hard to say if a significant periodicity can be seen from our work if we only considered the brightest GPs' is unclear; please rephrase to specify what test would be applied and what the result is.","section":"Section 4.2"}],"recommendation":"major_revision","confidential_remarks":"The paper presents a strong observational dataset and the main measurement is likely sound for S/N > 10. My recommendation of major revision is driven by the load-bearing false-positive correction and by the internal inconsistencies in the false-positive rate and the spectral-index statement; these are fixable within the scope of the manuscript. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this paper. The S/N > 10 measurement is rock solid — 9,366 GPs in 50 minutes, 11,300/hr, a clean detection. The headline 37,000/hr for S/N > 7 is a reasonable extrapolation, but the false-positive correction that gets you there is the one place I'd want more work before trusting the exact number.\n\nWhat's actually new: the first scattering measurement for this pulsar (tau = 5.5 ± 0.6 μs at 1 GHz), triple-peak GPs, and the UHF rate comparison with L-band. The paper is careful with RFI, uses a sensible GP definition, and compares properly with Knight (2007) and Abbate et al. (2020). The spectral index analysis of the 400 brightest GPs (Gaussian centered at −1.6) is interesting and contradicts the abstract's \"≈−3\" — that needs fixing.\n\nSoft spots. The false-positive correction assumes the rate per unit phase is uniform. The 358 off-pulse candidates are, by definition, false positives, so the 69 expected inside C1/C2 follows only if that rate holds. The trials factor in xprof could vary with phase, and the paper provides no check. There's also an internal inconsistency: Section 3 says 83/hr, the Conclusions say 100/hr. The margin above the \"over 37,000\" threshold is only a few hundred events, so this isn't just cosmetic. If the false-positive rate inside C1/C2 were several times higher, the headline would drop below 37,000. That said, the S/N > 10 rate and the scattering measurement don't depend on this correction. And note: if weak C3 GPs exist in the 7–10 range (Abbate et al. 2020 saw C3), they'd be absorbed into the 358, which would make the correction conservative, not threatening.\n\nBottom line: this is a solid observational paper for the giant-pulse/FRB-connection crowd. It deserves refereeing. I'd ask the authors to propagate the systematic uncertainty from the false-positive estimate, reconcile the two rates, and fix the abstract. The core results — the UHF rate at high S/N and the scattering time — will stand.","headline":"Solid UHF giant-pulse study with a new scattering measurement; the 37,000/hr headline is credible but rests on a false-positive correction that needs a proper systematic treatment.","tokens_in":17407,"tokens_out":4794,"would_cite":true,"duration_ms":58999,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Observing a globular-cluster millisecond pulsar at 544–1088 MHz with MeerKAT detects giant pulses at 37,000 per hour, 13.5 times the rate previously seen at L-band, and yields the pulsar's first scattering-time measurement.","keywords":["giant pulses","millisecond pulsars","globular clusters","radio astronomy","MeerKAT","pulsar scattering","fast radio bursts","PSR J1823-3021A"],"falsifier":"Compare candidate $S/N$ 7–10 events inside the C1 and C2 phase windows with the per-phase rate in equal off-pulse windows; a significant difference would invalidate the 69 expected false positives and change the 37,000 per hour claim.","tokens_in":16322,"feed_emoji":"📡","tokens_out":10631,"duration_ms":87762,"temperature":0.7,"pith_summary":"PSR J1823−3021A, a millisecond pulsar in the globular cluster NGC 6624, emits giant pulses at a high rate when observed at low radio frequencies. Using MeerKAT's UHF band (544–1088 MHz) for about 50 minutes, the authors detect 9,366 pulses with $S/N > 10$ and, after correcting for false positives in the $S/N$ 7–10 range, estimate a rate of 37,000 ± 200 giant pulses per hour, 13.5 times the rate seen in earlier L-band observations. The same baseband data give the first scattering-time measurement for this pulsar, $\\tau = 5.5 \\pm 0.6\\,\\mu\\mathrm{s}$ at 1 GHz. The authors also find that giant pulses cluster in two rotation phases, follow a single power law in energy with slopes near $-3$, and occasionally split into multiple peaks, including triple-peaked pulses seen for the first time. These properties bear on whether giant pulses from globular-cluster pulsars could underlie the fast radio bursts observed in other globular clusters.","feed_headline":"37,000 giant pulses per hour from one pulsar","feed_subtitle":"MeerKAT UHF observations beat the L-band rate by 13.5 times and measure interstellar scattering for the first time.","key_machinery":"The central objects are giant pulses — brief radio flashes far exceeding the pulsar's mean pulse flux — and the machinery that resolves them is the PTUSE baseband recording system on MeerKAT, which captured full-voltage data across 544 MHz of UHF bandwidth with 5.3 $\\mu\\mathrm{s}$ time resolution (2.65 $\\mu\\mathrm{s}$ for the 50 brightest pulses). Coherent dedispersion at the known dispersion measure, single-pulse $S/N$ measurement with a boxcar search, and a phase-based false-positive correction convert raw detections into the headline rate. Scattering times are extracted by fitting a Gaussian pulse convolved with an isotropic-screen scattering tail using the SCAMP-I code with MCMC sampling, applied to 49 bright, narrow pulses across four frequency sub-bands.","core_discovery":"The central claim is that PSR J1823−3021A is a far more prolific giant-pulse emitter at UHF frequencies than previously appreciated: the paper measures a detection rate of $37{,}000 \\pm 200$ pulses per hour with $S/N > 7$, a 13.5-fold increase over the L-band rate of about 3,000 per hour and higher than the 8.5-fold increase predicted by extrapolating the pulsar's steep spectrum. The paper also reports the first scattering-time measurement for this pulsar, $\\tau = 5.5 \\pm 0.6\\,\\mu\\mathrm{s}$ at 1 GHz, obtained by fitting an isotropic-screen scattering model to 49 bright, narrow giant pulses, with a mean scattering index $\\alpha = -2.5 \\pm 0.3$. In addition, the giant pulses are strongly clustered in two rotation phases (C1 and C2), their energies follow a single power law (slopes $-3.02 \\pm 0.01$ and $-2.96 \\pm 0.03$ for C1 and C2), and the data reveal multi-peak morphologies — 119 double-peaked, 4 triple-peaked, and 18 events with both C1 and C2 in one rotation — as well as one event with quasi-periodic substructure resembling FRB 20201028.","pith_inferences":["Beyond the paper: going below 544 MHz could push the giant-pulse rate even higher if the steep spectrum continues, but stronger scattering and RFI at those frequencies could erase the gain; splitting the existing UHF band into sub-bands would test this directly.","Beyond the paper: if the measured scattering is entirely interstellar, coherently descattering the baseband voltages should reveal narrower intrinsic pulses and may convert some single-peaked giant pulses into multi-peaked ones.","Beyond the paper: the observed excess of same-rotation C1 and C2 events (18 found versus 8 expected, at about 3 sigma) hints at a weak correlation between the two emission components; a longer observation could confirm or rule it out.","Beyond the paper: the low polarisation of these giant pulses compared with the high linear polarisation of the M81 repeating FRB suggests that, if giant pulses power globular-cluster FRBs, the mechanism must accommodate a wide range of polarisation fractions."],"forward_implications":["The UHF-band rate of $37{,}000 \\pm 200$ GPs/hr implies the pulsar emits a giant pulse every 0.1 seconds on average, making it the most active globular-cluster giant-pulse emitter known.","Because the observed rate exceeds the 8.5× scaling prediction, the pulsar's giant-pulse spectrum must be even steeper, or the low-frequency emission mechanism more efficient, than the integrated-profile extrapolation suggests.","The first scattering measurement, $\\tau = 5.5 \\pm 0.6\\,\\mu\\mathrm{s}$ at 1 GHz, is consistent with interstellar-medium models, so the intrinsic UHF pulse widths are plausibly shorter than the observed widths.","The excess of multi-peak pulses over chance coincidence (119 double and 4 triple observed) points to nanoshot-like substructure within individual giant pulses rather than independent overlapping pulses.","The low polarisation and the strong 5.44 ms periodicity of arrival times separate this source from typical fast radio bursts, and the brightest detected pulse would be visible with MeerKAT only to about 40 kpc, whereas an M81-like FRB would require a pulse about $10^4$ times more luminous."],"supporting_citations":[{"why":"Provides the L-band detection-rate baseline of about 3,000 GPs/hr that this work's 13.5-fold increase and the 8.5-fold prediction are measured against.","marker":"Abbate et al. (2020)"},{"why":"Earlier UHF detection of 120 giant pulses with Parkes; supplies the previous power-law index and GP-rate comparison.","marker":"Knight (2007)"},{"why":"Describes the PTUSE backend and MeerKAT UHF system used for the baseband observations, including the RFI environment.","marker":"Bailes et al. (2020)"},{"why":"Provides the scattering transfer function model used in the SCAMP-I fits.","marker":"Geyer & Karastergiou (2016)"},{"why":"The SCAMP-I package applied to fit scattering timescales and scattering indices.","marker":"Oswald et al. (2021)"},{"why":"Discovery of the repeating FRB in the M81 globular cluster; it motivates the observation and is the comparison target for the FRB connection.","marker":"Kirsten et al. (2022)"}],"fun_headline_variants":["37,000 giant pulses per hour from one pulsar","Record giant-pulse rate seen from millisecond pulsar","MeerKAT UHF detects 13.5 times more giant pulses","First scattering measurement for pulsar J1823−3021A","UHF baseband boosts giant pulse count to 37,000/hr"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline rate assumes that the 358 false-positive candidates seen outside the C1 and C2 phase windows are uniformly distributed in rotation phase, so that the same per-phase rate applies inside those windows; if the in-window false-positive rate differs, the $37{,}000 \\pm 200$ per hour claim changes.","fun_headline_variants_meta":{"raw":{"variants":["37,000 giant pulses per hour from one pulsar","Record giant-pulse rate seen from millisecond pulsar","MeerKAT UHF detects 13.5 times more giant pulses","First scattering measurement for pulsar J1823−3021A","UHF baseband boosts giant pulse count to 37,000/hr"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000278,"raw_usage":{"total_tokens":1789,"prompt_tokens":1214,"completion_tokens":575,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":830,"completion_tokens_details":{"reasoning_tokens":485}},"tokens_in":830,"tokens_out":575,"duration_ms":5784,"temperature":1.0,"reasoning_tokens":485,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:09:43.028781+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare candidate $S/N$ 7–10 events inside the C1 and C2 phase windows with the per-phase rate in equal off-pulse windows; a significant difference would invalidate the 69 expected false positives and change the 37,000 per hour claim.","supporting_citations":[],"review_version":1}