{"id":"3bd3d2e0-ba1d-4723-ae00-16682220b3f8","arxiv_id":"2505.14219","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Searching FAST arrival times of FRB 20201124A with Monte Carlo significance estimation yields no evidence for the about 1.7 second periodicity claimed by Du et al. (2025).","lead":"A reanalysis of 1,690 bursts from FRB 20201124A finds no significant second-scale periodicity, contradicting a recent claim of a 1.7-second period in the same FAST data. The disagreement comes down to how detection significance is estimated and whether closely spaced bursts are kept or removed.","discovery_kind":"replication","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The non-detection rests on cutting bursts with <0.4s waiting times; on MJD 59347 this removes the close triplets that drive the disputed ~1.7s signal, and no test proves such trimming preserves a real periodic signal.","rationale":"The paper is transparent and internally consistent: it defines test statistics, frequency spacing, and two null sampling schemes before applying them, and it checks compatibility with Du et al. The strongest claim, however, is a non-detection, and its validity is only as strong as the most permissive assumption that removes the signal. The reader identified exactly this weak point: the 0.4s threshold in Sec 3.1.1. I find no separate assumption that is more load-bearing. The log-normal null fit and the neglect of non-continuous windows (footnote 3) are secondary; even if the observation windows were continuous, the trimming issue would remain. The absence of code or injection tests makes it impossible to verify that the pipeline recovers a planted periodic signal after trimming, which is the minimal check needed to defend a null result against the filtering critique. I therefore agree with the reader's conditional verdict and recommend keeping it: the paper should be accepted only with the robustness analysis in the concrete test (or an equivalent injection-recovery study) made a condition. I do not see grounds to reject the paper: the statistical framework is sound, and the competing claim by Du et al. is not automatically vindicated by this critique. One minor completeness issue: Figure 1 cites 'Chen et al. 2025' but no matching reference appears in the reference list; this should be fixed but does not affect the central argument.","tokens_in":6794,"tokens_out":5283,"duration_ms":52800,"concrete_test":"Re-analyze MJD 59347 without the 0.4s pre-processing: compute the chi2 statistic on all bursts including the two close triplets, and estimate its p-value with a null generated by sampling arrival times from the full empirical waiting-time distribution of FRB 20201124A (including the short mode), using the same number of bursts, total duration, and frequency grid as the paper. If the un-trimmed chi2=5.8 remains below the detection threshold after this correction, the 0.4s cutoff is not the decisive factor; if it crosses the threshold, the non-detection is contingent on rejecting the very bursts that produce the claimed periodicity.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Sec 4, abstract) is a non-detection of strict periodicity. The load-bearing step is the pre-processing in Sec 3.1.1: any burst with a preceding burst <0.4s is discarded. On MJD 59347 this reduces chi2 from 5.8 (Du et al. 2025) to 4.3, converting an apparent periodicity into a non-detection. The paper's Appendix A shows that including short-waiting-time bursts inflates the null distribution of the H5 statistic, but this only says a null with such bursts has a broader tail; it does not show the specific triplets in MJD 59347 are engine substructure rather than genuine periodic engine activity. The proposed magnetar-substructure motivation is qualitative and is not backed by an injection/recovery test or a model showing that a periodic engine with clustered bursts would remain detectable after this trimming. Since the defended conclusion is exactly that no such periodicity is present, and the threshold removes the only bursts that produce the claimed periodicity, the argument risks circularity: the signal is filtered out and then reported as absent. This is the single most load-bearing concern; it does not mean the paper is wrong, but the non-detection is not yet established independently of the 0.4s choice.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reanalyzes the burst arrival times of FRB 20201124A observed by FAST between April and June 2021, searching for second-scale periodicity within individual observing days. The authors define a preprocessing step that discards bursts with a waiting time shorter than 0.4 s before a preceding burst, motivated by the bimodal waiting-time distribution, and then apply two test statistics (a truncated H-test with N=5 and a binned χ² test with 30 bins). Significance is estimated by Monte Carlo sampling of null TOA sequences using two models: uniform draws over the daily span and a log-normal waiting-time model with parameters fitted to the long-mode of the data. The central claim is that no significant periodicity is found on any day, specifically contradicting the ~1.7 s periodicities reported for MJDs 59310 and 59347 by Du et al. (2025). The authors attribute the discrepancy to their more careful null-distribution estimation and, for MJD 59347, to the removal of two triplets of closely spaced bursts.","tokens_in":7053,"tokens_out":3360,"duration_ms":29860,"significance":"If the non-detection is correct, the paper provides an important cautionary result for FRB periodicity searches: it shows that the claimed second-scale periodicity in FRB 20201124A is not robust to a more principled treatment of the null distribution and burst pre-processing. The manuscript is methodologically transparent, explicitly defining the search procedure, frequency spacing, test statistics, and null-sampling strategies, and it includes a supplemental appendix demonstrating that short-waiting-time bursts inflate the tail of the H5 statistic. The two independent null sampling strategies and the use of both χ² and H statistics add robustness. However, the central claim depends critically on the 0.4 s threshold, which is not validated by an injection/recovery test, and the significance estimate for MJD 59310 relies on extrapolating a Gumbel fit beyond the 500 Monte Carlo samples; these are substantive gaps that a revision should address.","major_comments":[{"comment":"The 0.4 s waiting-time threshold is the load-bearing element of the non-detection claim, but the manuscript does not establish that this threshold preserves genuine periodic signals while removing only substructure. In §4.1 the authors state that rejecting three bursts from two triplets on MJD 59347 reduces χ² from 5.8 to 4.3, which is exactly what converts the Du et al. (2025) detection into a non-detection. Appendix A shows that including short-waiting-time bursts widens the null distribution of H5, but this only demonstrates that the null is broader when such bursts are present; it does not prove that the specific triplets in MJD 59347 are unrelated to a periodic engine. An injection/recovery test (or a dedicated model of sub-burst structure) is needed to show that a periodic signal with clustered bursts would remain detectable after the 0.4 s cut. Without such a test, the non-detection is not established independently of the threshold choice.","section":"§3.1.1, §4.1, Appendix A"},{"comment":"The significance estimate for MJD 59310 rests on a Gumbel distribution fitted to 500 Monte Carlo maxima and then extrapolated to p-values as small as 10⁻³, which is then compared to a Bonferroni-corrected threshold of about 10⁻⁹. The paper offers no validation that the Gumbel tail is accurate beyond the sampled range; for an extreme-value distribution, a slight mis-fit can change the extrapolated p-value by orders of magnitude. The authors should either increase the number of Monte Carlo samples enough to resolve the tail directly, or provide a diagnostic (e.g., a quantile-quantile plot on the largest order statistics) that justifies the extrapolation. In addition, the log-normal null parameters (4.5, 1.42) are fitted to the same dataset under analysis (§3.2, item 2), which mildly pulls the null toward the data; the paper should discuss or test the sensitivity of the conclusions to this fitting procedure.","section":"§3.2, §4"},{"comment":"The frequency spacing formula Δf ∼ Δφ/(max t_i − min t_i) uses the full observed span of arrival times, but footnote 3 admits that most daily observations were not continuous. If the actual observing windows are shorter than the span, the effective frequency resolution for a phase-coherent signal is coarser, and the search could miss periodicities whose frequencies fall between the sampled grid points. The authors dismiss this as a negligible effect without a quantitative argument. A quantitative estimate, or a test using only the continuous segments, would strengthen the claim that the search is complete over the 0.1–10 Hz range.","section":"§3.1.4, footnote 3"}],"minor_comments":[{"comment":"The manuscript contains several language slips: the abstract has 'F AST' with an extra space, and §6 says 'We searched' and 'We compared my search strategy' where 'our' is expected; these should be corrected in a revision.","section":"Abstract and §6"},{"comment":"The figure legend inside the panel reads 'Chen et al. 2025' while the caption and text refer to C. Du et al. (2025); this reference mismatch is confusing and should be fixed.","section":"Figure 1"},{"comment":"The displayed definition 'p-value59310 ≡ P(χ²_search ≤ χ²_Gumbel)' appears to have the inequality direction reversed; a p-value should express the probability that the null statistic exceeds the observed value. Please clarify the notation.","section":"§4, Eq. (p-value)"},{"comment":"The description of the log-normal null sampling says 'ensuring the total duration is similar to that of the detected bursts up to 10% using rejection sampling' without explaining how the rejection step works or how the 10% tolerance is enforced; a concise algorithmic description would improve reproducibility.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is essentially a direct rebuttal of Du et al. (2025), and the threshold issue is the crux of the controversy. The authors should be encouraged to add an injection/recovery test in the revision, as this is the most direct way to break the circularity concern. Also note the reference mismatch in Figure 1 (Chen vs Du) and the reversed p-value inequality; these are minor but suggest a final proofreading pass is needed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a careful, transparent reanalysis that makes a credible case against the Du et al. 1.7 s periodicity claim for FRB 20201124A, but the case is not airtight. The new content is mostly methodological: two test statistics (H5 and binned chi2), two null sampling strategies (uniform and log-normal waiting times), and Monte Carlo significance estimation on the public FAST data. The paper shows that for MJD 59310 the claimed signal vanishes once the null is sampled properly, and for MJD 59347 the chi2 drops from 5.8 to 4.3 after rejecting bursts with waiting times under 0.4 s. The appendix demonstrating that short-waiting-time bursts inflate the H5 null tail is a useful addition.\n\nThe soft spot is exactly that 0.4 s cut. On MJD 59347, removing two triplets of close bursts is what converts the claimed periodicity into a non-detection, and the paper does not prove that those triplets are engine substructure rather than genuine periodic activity. The magnetar-substructure motivation is qualitative, and there is no injection/recovery test showing a periodic engine with clustered emission would survive this pre-processing. Appendix A shows the null broadens when short waits are included, but it is not the same as computing the p-value for the original chi2 of 5.8 against the appropriate null. I do not think the argument is circular in a damning sense--the cut is motivated by the observed bimodal waiting-time distribution and by the physics of burst substructure, not by chasing the null--but the threshold choice needs explicit robustness analysis before the non-detection is established independently of it.\n\nOther concerns are minor: the log-normal null is fitted to the same dataset, which mildly pulls the null toward the data, but the uniform sampling provides a cross-check; the neglect of non-continuous observation windows is acknowledged and likely minor; no code or injection tests are released, which would help. The frequency-spacing and H-test harmonic choices are justified clearly.\n\nThis paper is for anyone working on periodicity searches in repeating FRBs. It deserves a serious referee: the referee should ask for an ablation of the 0.4 s threshold and an injection test. I would accept it conditionally; the central refutation is likely correct but not yet fully established.","headline":"A credible but threshold-dependent refutation of the 1.7s periodicity claim; the 0.4s waiting-time cut needs an injection test before the non-detection is fully established.","tokens_in":7597,"tokens_out":3043,"would_cite":true,"duration_ms":28749,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A reanalysis of FAST observations of FRB 20201124A finds no significant second-scale periodicity in any of 45 observing days, contradicting the claimed 1.7 s period.","keywords":["fast radio bursts","FRB 20201124A","periodicity search","waiting time distribution","null distribution","H-test","chi-square test","repeating FRB"],"falsifier":"If a re-analysis of the same FAST data that keeps the close sub-0.4 s bursts, models their correlation explicitly in the null, and applies a look-elsewhere correction still returns a periodicity peak above the 5-sigma threshold at about 1.7 s on MJD 59347 or MJD 59310, then the paper's non-detection claim would be wrong.","tokens_in":6534,"feed_emoji":"📡","tokens_out":6563,"duration_ms":62336,"temperature":0.7,"pith_summary":"The paper aims to show that the about 1.7 second periodicity claimed for the repeating fast radio burst FRB 20201124A is not statistically significant when the null distribution is sampled carefully. It defines a full periodicity-search pipeline, applies it to the 1863 FAST-detected bursts from April to June 2021, and finds that no individual observing day shows significant constant-frequency periodicity in the 0.1 to 10 Hz range. The two days on which a companion detection was claimed, MJD 59310 and MJD 59347, do not survive the corrected significance tests. If correct, the result removes the strongest current evidence for a second-scale periodic clock in this repeater and shifts the burden onto properly constructed null hypotheses rather than textbook chi-square approximations.","feed_headline":"No second-scale periodicity in FRB 20201124A's FAST bursts","feed_subtitle":"A reanalysis that samples the null distribution finds the claimed 1.7 s period is not significant on any day.","key_machinery":"The load-bearing object is the empirical null distribution of the maximal periodicity statistic. Instead of relying on theoretical chi-square or H-test distributions, the paper samples 500 mock arrival-time sequences per day under two models (uniform in time, or log-normal waiting times matching the long mode), forces the same 0.4 s pre-processing threshold and same burst count, searches each mock set over frequency, and records the best statistic. A Gumbel extreme-value fit to those maxima sets the p-value for the observed search result. Appendix A shows the entire procedure matters: adding bursts from the short waiting-time mode shifts the tail of the maximal-H5 distribution to larger values, demonstrating that close bursts, if included, inflate significance under the null.","core_discovery":"The central claim, stated on the paper's own terms, is a non-detection: within the 45 observing days of FAST data, no constant-frequency timing model with frequency between 0.1 and 10 Hz produces a test statistic exceeding what is expected under an empirically sampled null. The analysis pre-processes arrival times by discarding bursts with a preceding burst less than 0.4 s away, then searches with two statistics, the binned chi-square (to match the prior claim) and the H5 harmonic test. Significance is estimated by Monte Carlo: 500 mock burst sets per day, drawn either uniformly or from the long log-normal waiting-time mode, are searched and the maximum statistic per set forms the null. For MJD 59310 the best chi-square value has a Gumbel-fitted p-value of $10^{-3}$, far above the Bonferroni-corrected 5-$\\sigma$ threshold around $10^{-9}$. For MJD 59347 the chi-square value drops from 5.8 to 4.3 once two triplets of bursts with sub-0.4 s waiting times are reduced to single representative bursts, placing the result well inside the null distribution. The paper therefore concludes that the claimed 1.7 s periodicity is an artifact of an improper null distribution and of including short-waiting-time substructure.","pith_inferences":["If accepted, the result suggests that published periodicity detections elsewhere that used theoretical null distributions for small burst counts should be re-examined with sampled nulls before being invoked as engine evidence.","The choice of 0.4 s is physically motivated but not derived; a decisive test would search for periodic structure among the discarded close bursts themselves, asking whether their arrival times prefer a common clock.","Pure constant-frequency periodicity is not the only possible clock; a quasi-periodic engine with phase wandering or frequency drift would be missed, and the full daily observation windows are short enough that longer continuous monitoring could still reveal a period.","The significance pipeline itself is reusable: any repeater with tens of bursts per session can be tested with the same Monte Carlo null, making the paper's contribution primarily a statistical template for repeating-FRB periodicity searches."],"forward_implications":["The claimed about 1.7 s period in FRB 20201124A would not be evidence for a rotating or periodically modulated engine; strict second-scale clocking is absent in this 2021 FAST campaign.","Future periodicity claims for repeating FRBs need an empirical null that includes waiting-time correlations and the look-elsewhere effect from scanning frequency, not the textbook chi-square distribution.","A pre-processing threshold that strips sub-0.4 s pairs is part of the analysis; if adopted, it removes substructure that otherwise biases null searches.","Applying the same pipeline to other GHz-ranged repeaters with dense burst counts can test whether second-scale periodicity exists in any currently known source.","The non-detection constrains constant-frequency periodic modulation at 0.1-10 Hz over 2-4 hour windows, independent of the day-scale periodicity seen in other repeaters."],"supporting_citations":[{"why":"Supplies the 1863 FAST barycentric burst arrival times from April-June 2021 that the periodicity search is run on.","marker":"H. Xu et al. (2022)"},{"why":"The companion detection of about 1.7 s periodicity in MJDs 59310 and 59347 that this paper reanalyzes and finds insignificant.","marker":"C. Du et al. (2025)"},{"why":"Earlier second-scale periodicity search in the same source whose methods and test statistic are compared with the present procedure.","marker":"C. Du et al. (2024)"},{"why":"Defines the H-test statistic that the paper adapts to small burst counts by limiting harmonics to N=5.","marker":"O. C. de Jager et al. (1989)"}],"fun_headline_variants":[],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that bursts separated by less than 0.4 s are substructure rather than engine periodicity, so they can be removed; if those close bursts are true periodic engine output, the analysis deletes the signal it claims to test.","fun_headline_variants_meta":{"error":"Client error '402 Payment Required' for url 'https://api.deepseek.com/chat/completions'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/402"},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:38:17.114787+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"If a re-analysis of the same FAST data that keeps the close sub-0.4 s bursts, models their correlation explicitly in the null, and applies a look-elsewhere correction still returns a periodicity peak above the 5-sigma threshold at about 1.7 s on MJD 59347 or MJD 59310, then the paper's non-detection claim would be wrong.","supporting_citations":[{"cited_title":"C., Raubenheimer, B","cited_arxiv_id":null,"evidence_quote":"Defines the H-test statistic that the paper adapts to small burst counts by limiting harmonics to N=5."}],"review_version":1}