{"id":"870ad042-694d-4ef1-8433-bd2651c5e2af","arxiv_id":"2509.05957","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Quasi-periodic submillisecond microstructure was detected in the interpulses of PSRs J0627+0706 and J0953+0755, with properties matching or differing from the main pulse depending on the pulsar.","lead":"Using FAST telescope data, this paper reports submillisecond microstructure pulses inside the interpulse emission of two pulsars, the first such detections, and compares them with main-pulse microstructure. The result gives a new handle on whether main and interpulse emission come from the same or different magnetic regions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"First-detection claim lacks a statistical null test: manual selection among three ACF/FFT estimators could classify noise peaks as IP microstructure at the reported 5% rate.","rationale":"The reader's weakest_assumption is essentially the same as mine: the reliability of ACF/FFT identifications and manual selection for distinguishing genuine IP microstructure from noise and RFI. My concrete test makes the requirement explicit—a null test on surrogates and off-pulse noise. The paper's own §4.2 caveats about small samples and the time-resolution-limited MP-IP difference in J0953+0755 reinforce rather than weaken this concern. I do not see grounds for outright rejection: the analysis is standard, uses public FAST data, and shows representative examples. But the first-detection claim is conditional on demonstrating that the manual selection does not manufacture false positives. Hence the reader's CONDITIONAL verdict stands unchanged.","tokens_in":27730,"tokens_out":7439,"duration_ms":86509,"concrete_test":"Run a permutation null on the public FAST data: for each high-SNR IP pulse, take the pulse residual exactly as in §3, then generate 1000 surrogate residuals by randomizing the phases of the FFT (preserving amplitudes) or by shuffling the on-pulse phase bins. Apply the same ACF/FFT pipeline with a deterministic operationalization of the manual criteria (e.g., ACF first minimum < -0.2, FFT peak > 3 sigma above the median power, and consistency of at least two of the three P_mu estimators to within 30%). Count the fraction of surrogates that pass. If the false-positive rate is comparable to 5% (J0627+0706) or contributes substantially to 37% (J0953+0755), the first-detection claim is not supported. Independently, apply the same pipeline to off-pulse noise windows of the same duration.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Conclusion 1) is that quasi-periodic microstructure is detected in interpulses for the first time. The detection pipeline in Section 3, steps 4-8 combines (i) smoothing-spline denoising and residual extraction, (ii) reading tau_mu from the first ACF minimum, (iii) reading P_mu from three different ACF/FFT features, and (iv) manually selecting pulses that 'exhibit obvious quasi-periodic characteristics' and manually choosing among the three estimates, with RFI rejection also by eye. No false-alarm probability is computed against a null hypothesis of noise or residual RFI. For J0627+0706 the IP detection rate is 42/792 = 5.3%; for J0953+0755 it is 381/1021 = 37%. The IP components are much weaker than the MPs, and the residual extraction is effectively a high-pass filter that can imprint quasi-oscillatory structure on white noise. Because the headline is a first detection, the burden of proof is to show that this detection rate is not what the same pipeline would produce on pure noise. The paper explicitly acknowledges in §4.2 that sample sizes are small and J0953's MP-IP differences are at the time-resolution limit, but it never quantifies the null rate.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript analyzes FAST single-pulse observations of four interpulse pulsars (J0627+0706, J0826+2637, J0953+0755, J1946+1805) with the goal of detecting and measuring quasi-periodic microstructure in both main pulses (MP) and interpulses (IP). The analysis pipeline follows Mitra et al. (2015): it denoises single pulses, removes low-frequency power with a kernel smoother, computes ACFs and FFTs of the residuals, and extracts the characteristic timescale tau_mu and quasi-period P_mu, with manual selection of pulses judged to show quasi-periodic behavior. The paper reports the first detection of quasi-periodic microstructure in IP emission (for J0627+0706 and J0953+0755), claims that MP and IP microstructure are consistent for J0627+0706 but that IP values are smaller for J0953+0755, and presents a refreshed P_mu-P power-law fit using literature plus four new MP measurements.","tokens_in":27985,"tokens_out":5236,"duration_ms":61940,"significance":"If the first-detection claim is correct, the paper fills a genuine observational gap: microstructure has been studied extensively in main pulses and in other neutron-star classes, but not in interpulse emission. The simultaneous MP-IP comparison is also a promising diagnostic for emission geometry, and the paper uses publicly released FAST data. The main weakness is that the detection pipeline relies heavily on manual selection and visual RFI rejection, and no false-alarm rate is computed. The headline conclusion therefore remains plausible but not quantitatively established. The P_mu-P relation is a useful update, although it is a confirmation rather than a new result and depends on the reliability of the new MP measurements.","major_comments":[{"comment":"No null test is provided for the detection of quasi-periodic microstructure. The residual extraction is effectively a high-pass filter, which can imprint oscillatory structure on white noise and produce ACF minima and FFT peaks. The reported detection rates are 42/792 = 5.3% for the J0627 IP and 381/1021 = 37% for the J0953 IP, but the reader cannot tell whether these rates exceed what the same pipeline would produce on pure noise or residual RFI. The authors should process off-pulse noise windows (or simulated noise) through the identical smoothing, residual, ACF/FFT, and manual-selection protocol and report the resulting false-alarm rate. This is load-bearing for Conclusion 1.","section":"Section 3, steps 4-8; Table 3"},{"comment":"The text defines the first ACF minimum as tau_mu in step 6 and then again defines the first ACF minimum as P_mu in step 7(1). If taken literally, P_mu would equal tau_mu for every pulse, contradicting Table 3 (e.g., J0627+0706 MP: tau_mu = 0.46 ms, P_mu = 0.82 ms). The intended estimator must be specified precisely (e.g., first ACF maximum after zero, second minimum, or a harmonic relation from the FFT peak). Because P_mu enters the main conclusions and the P_mu-P fit, this ambiguity is not merely cosmetic.","section":"Section 3, step 7(1)"},{"comment":"The reported MP-IP differences for PSR J0953+0755 sit at the time-resolution limit. The paper states that the median Delta(tau_mu) = 0.06 ms equals the 49.152 microsecond bin width and median Delta(P_mu) = 0.11 ms equals twice that width. With such quantization, the observed \"smaller IP\" values could be a binning artifact. The authors acknowledge the resolution limit in the text, but they do not show that the distributions of Delta(tau_mu) and Delta(P_mu) are inconsistent with what would arise from a common underlying distribution after discretization. A bootstrap or simulated-bin test is needed before Conclusion 3 can be accepted.","section":"Section 4.1.3 and Section 4.2; Figure 13; Table 3"},{"comment":"Component boundaries are visually estimated for several pulsars and are used to define the phase windows from which single pulses are extracted. For J0953+0755 the MP/IP boundary is set by the minimum intensity after low-pass filtering, and for J0627+0706 and J1946+1805 the boundaries are \"visually estimated.\" The IP detection results depend directly on these windows: if the IP window includes a small amount of MP or bridge emission, the residual statistics could be contaminated. The authors should test at least a few neighboring boundary choices and show that the IP detection rates and P_mu/tau_mu values are stable.","section":"Section 3, step 1; Table 2"},{"comment":"The comparison between MP and IP (or PC) microstructure is stated qualitatively, with medians and interquartile ranges, but no statistical test is applied. For J0627+0706 the simultaneous-sample size is only 17 pulses, and for J0826+2637 the PC sample is 7 pulses; the paper itself flags the latter as not statistically significant. A two-sample test (e.g., KS or bootstrap) on the tau_mu and P_mu distributions, or at least a bootstrap confidence interval on the medians of Delta(tau_mu) and Delta(P_mu), is needed to support the claims of consistency in J0627+0706 and difference in J0953+0755.","section":"Section 4.2, Figures 10-13"}],"minor_comments":[{"comment":"Table 4 lists \"J1946+1905(MP)\" but the pulsar is J1946+1805 throughout the rest of the paper. The conclusion also contains typos such as \"mainpulse\" and \"exit\" instead of \"exist.\"","section":"Table 4; Conclusion"},{"comment":"The notation is confusing: the same symbol P_mu is used for the candidate period from three methods, but the text does not explain how the three candidate values are reconciled when they differ. Also, figure axis labels read \"Frequency (KHz)\"; the unit should be kHz.","section":"Section 3, step 7 and Figure 2"},{"comment":"The caption of Figure 1 mentions vertical red lines in the top-right and bottom-left insets, but the figure description does not say what they mark. Table 2 gives phase ranges, but the relation between the table ranges and the inset rectangles is not always visually obvious.","section":"Section 3, step 1 and Figure 1"},{"comment":"The log-space least-squares fit appears to be unweighted, despite the data having asymmetric and heteroscedastic errors. Since some data points are duplicated (J2145-0750 and J1913+1330 appear twice) and one point (J0901-4046) is a rough estimate, a weighted fit or a bootstrap with an explicit treatment of the duplicate entries would be more robust. At minimum, the authors should state whether the shown uncertainties include the fit covariance.","section":"Section 4.3, Table 4"},{"comment":"The statement that in the IP of J0953+0755 \"almost all pulses exhibiting quasi-periodic microstructure are superimposed on low-frequency envelopes\" is made without a count or fraction. Please provide the number of such pulses out of the 381 detected IP pulses.","section":"Section 4.4"}],"recommendation":"major_revision","confidential_remarks":"The core problem is not circularity or an internally inconsistent derivation; it is the lack of a false-alarm test for a detection claim that rests on manual selection and high-pass residual filtering. If the authors can supply a null-signal control (off-pulse noise through the same pipeline) and clarify the P_mu estimator, the first-detection claim becomes much stronger. The J0953 MP-IP difference also needs a resolution-aware test. I would not reject, but the revision must address the null-test point before the paper can be considered publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe one thing you should know: this paper reports the first credible detection of quasi-periodic microstructure inside interpulses, for PSRs J0627+0706 and J0953+0755. If that holds, it is a genuinely new observational handle. The rest of the paper—MP/IP comparison, the P_mu-P fit—is mostly confirmation of known scaling.\n\nWhat is good: they use public FAST data, follow the established M15 ACF/FFT framework, and show illustrative pulses, histograms, and full tables. They are also honest about limitations: the J0826 PC sample is only 7 pulses, the J0953 MP-IP differences sit at the time-resolution boundary, and component boundaries are partly visual. The literature work is thorough, and the re-fit of P_mu-P with their new MP points is a useful meta-analysis even if it is not a new relation.\n\nSoft spots, in order:\n\n1. No null test for the detection. The pipeline does smoothing-spline residual extraction, ACF/FFT peak picking, then manual selection of pulses with \"obvious quasi-periodic characteristics\" and manual choice among three P_mu estimators. For J0627, only 42 of 792 high-SNR IP pulses (5.3%) were classified as quasi-periodic; for J0953, 381 of 1021 (37%). The stress-test concern is fair: a high-pass residual of pure noise can produce quasi-oscillatory ACF structure. The paper never quantifies the false-alarm rate of the same pipeline on noise, and the headline claim is exactly a first detection.\n\n2. The J0953 IP-smaller-than-MP result is at the resolution limit. The paper acknowledges that the median delta_tau equals the time resolution and delta_P equals twice that. So this conclusion is suggestive, not firm.\n\n3. Minor: no analysis code is provided, and the P_mu-P fit includes a rough estimate for J0901-4046.\n\nNone of this is fatal. The two IP detections are shown with enough examples that I would bet they are real, especially J0953 with its large sample. But the strongest wording would be much easier to defend with a null-hypothesis simulation or a blinded automated selection.\n\nWho it is for: pulsar observers working on microstructure, interpulse geometry, and the universal P_mu-P scaling. It deserves a serious referee, not a desk rejection. I would send it to review and ask for a null test and a clearer statement that the J0953 difference is resolution-limited. I would cite the IP detection if it survives refereeing.\n\nBest,\n\n[You]","headline":"First IP microstructure detection is plausible but rests on manual selection without a null test; the real value is the new observational result, not the re-fitted P_mu-P relation.","tokens_in":28578,"tokens_out":2307,"would_cite":true,"duration_ms":26841,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["97.60.Gb","95.85.Bh"],"model":"deepseek-v4-flash","headline":"The paper reports the first detection of quasi-periodic microstructure, sub-millisecond intensity flicker, in the interpulse emission of pulsars, and confirms that the flicker's characteristic period tracks neutron-star spin across all clas","keywords":["pulsar microstructure","interpulse pulsars","quasi-periodic substructure","autocorrelation function","fast Fourier transform","rotation-period scaling","FAST observations","single-pulse morphology"],"falsifier":"Re-observe PSRs J0627+0706 and J0953+0755 at roughly five times finer time resolution than the current 49 µs, and require the same interpulse quasi-periodic peaks (Pµ ≈ 0.94 ms and ≈ 0.38 ms) and ACF first minima to reappear in independently cleaned residuals; if the peaks vanish or shift by more than the quoted errors, or if matching peaks appear in the ACF/FFT of pure noise processed the same way, the manual peak-picking was not seeing true interpulse microstructure.","tokens_in":27575,"feed_emoji":"📡","tokens_out":13391,"duration_ms":125887,"temperature":0.7,"pith_summary":"This paper asks a question no one had answered: do the faint secondary radio pulses of interpulse pulsars carry the same sub-millisecond intensity flicker ('microstructure') that main pulses are known to carry? Using single-pulse FAST data on four interpulse pulsars, it reports the first detection of quasi-periodic microstructure in interpulse emission (PSRs J0627+0706 and J0953+0755), plus a first detection in the post-cursor of PSR J0826+2637. It then compares main-pulse and interpulse flicker: statistically indistinguishable in J0627+0706, but measurably finer in the interpulse of J0953+0755. Finally it combines the new measurements with published data across normal pulsars, millisecond pulsars, magnetars, RRATs, and a 76-second pulsar, reconfirming Pµ(ms) ≈ 1.34 × P(s)^1.06. If correct, the results open interpulse emission to microstructure studies and sharpen the evidence that quasi-periodic flicker across all radio-emitting neutron stars is set by the rotation period itself.","feed_headline":"Sub-millisecond flicker found in pulsars' interpulses","feed_subtitle":"First detection links flicker period to spin from millisecond pulsars to magnetars.","key_machinery":"The load-bearing analysis chain: (1) fold FAST single pulses, excise RFI, denoise with smoothing-spline regression, and strip low-frequency power via Nadaraya-Watson kernel smoothing at bandwidth 0.075×N_on, leaving residuals that hold only the fast flicker; (2) read the ACF's first minimum as the timescale τµ, and take the quasi-period Pµ from three estimators — ACF first minimum, FFT peak of the residual, FFT peak of the residual's ACF — reconciled by manual visual selection; (3) fit the power law Pµ = A·P^α in log space across ~40 objects spanning five decades in spin period. The fixed point of the study is the empirical P–Pµ relation, Pµ(ms) ≈ 1.34·P(s)^1.06, which the new interpulse mea","core_discovery":"On its own terms, the discovery is that quasi-periodic microstructure is not confined to main pulses. From FAST single pulses, using autocorrelation-function (ACF) and fast-Fourier-transform (FFT) analysis of denoised pulse residuals, the authors identify sub-millisecond periodic modulation in the interpulses of PSR J0627+0706 (42 of 792 high-SNR pulses; Pµ = 0.94 ms) and PSR J0953+0755 (381 of 1021; Pµ = 0.38 ms), and in the post-cursor of PSR J0826+2637 (7 of 9 pulses, flagged as statistically insignificant). In J0627+0706 the interpulse flicker matches the main pulse within errors; in J0953+0755 it is finer. Re-fitting the population relation between Pµ and rotation period recovers Pµ(ms)","pith_inferences":["If the IP–MP flicker comparison is a genuine geometric diagnostic, it can be pushed further than the paper goes: a modest survey of high-SNR interpulse pulsars could map IP/MP Pµ ratios against MP–IP longitude separation, testing whether the ratio tracks the angular distance between sampled field lines — a prediction the paper's two objects cannot decide.","The paper's own resolution caveats (shortest τµ at 1–2× the sampling time) imply the fitted Pµ values may be upper limits; re-observations at finer time resolution could push τµ down and steepen the slope, so the near-unity exponent should be treated as provisional at the short-period end.","The same reasoning that predicts FRB host periods can be inverted: for repeating FRBs with quasi-periodic substructure, independent evidence of periodicity (burst clustering or an identified associated source) should appear near the predicted spin period — a falsifiable link between FRB microstructure and neutron-star rotation.","If the scaling is truly universal, the 6.45-hour coherent transient ASKAP J183950.5−075635.0 becomes a stress test: its substructure period should land near the power-law extension at P ≈ 23,000 s, which can be checked against current limits."],"forward_implications":["Interpulse emission is no longer smooth by default: quasi-periodic sub-millisecond flicker occurs in IP components, so IP radiation is structured on the same rapid timescales as MP radiation.","Within one pulsar, IP and MP flicker rates can differ (J0953+0755: Pµ = 0.38 ms vs 0.50 ms), giving a quantitative observable for how emission changes between the two sight-line cuts.","The reconfirmed near-linear scaling, Pµ(ms) = (1.337±0.114)·P(s)^(1.063±0.038), across normal pulsars, MSPs, magnetars, RRATs, and long-period pulsars implies the flicker clock is set by rotation period itself, not by age or field.","Applying that scaling to the quasi-periods measured in six FRBs predicts host spin periods from ~2 ms to ~122 s (Table 5), giving a testable handle on FRB central engines.","For interpulse pulsars, microstructure similarity between MP and IP can flag whether the two components come from the same pole and even the same flux tube, complementing polarization-based geometry arguments."],"supporting_citations":[{"why":"Supplies the ACF/FFT analysis method, the smoothing-spline and Nadaraya-Watson preprocessing, and most of the normal-pulsar Pµ data used in the population fit.","marker":"(Mitra et al. 2015)"},{"why":"Defines the P–Pµ and P–τµ relations across source classes that this paper reconfirms, and supplies the four magnetar data points.","marker":"(Kramer et al. 2024)"},{"why":"First detection of MSP microstructure; supplies two millisecond-pulsar data points for the fit.","marker":"(De et al. 2016)"},{"why":"Additional MSP microstructure measurements that extend and check the fit.","marker":"(Liu et al. 2022)"},{"why":"Supplies the 76-s pulsar J0901−4046 data point and the single-pulse morphology classification scheme used in Section 4.4.","marker":"(Caleb et al. 2022a)"},{"why":"First microstructure and quasi-periodic Pµ measurement in PSR B0950+08, the historical baseline the MP comparison must reproduce.","marker":"(Hankins 1971b)"},{"why":"Origin of the Δτµ ≈ 10^-3 P scaling that motivates the rotation-period dependence.","marker":"(Cordes 1976)"},{"why":"The rotating-vector-model polarization analysis used to assign MP and IP to the same or different magnetic poles and flux tubes.","marker":"(Sun et al. 2025)"},{"why":"Independent τµ/Pµ values for PSRs B0823+26 and B0950+08 used for cross-checks, plus the beaming Lorentz-factor formula.","marker":"(Lange et al. 1998)"}],"fun_headline_variants":["First interpulse microstructure seen by FAST","Pulsar interpulses show sub-ms flicker for first time","FAST reveals interpulse flicker tied to pulsar spin","Interpulse microstructure detected: links to spin period"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The first-detection claim stands on the assumption that the ACF dips and FFT peaks — picked partly by eye, on as few as 7 to 42 usable interpulse pulses — really trace periodic flicker in the emission, and are not noise or residual radio interference that survived the cleaning.","fun_headline_variants_meta":{"raw":{"variants":["First interpulse microstructure seen by FAST","Pulsar interpulses show sub-ms flicker for first time","FAST reveals interpulse flicker tied to pulsar spin","Interpulse microstructure detected: links to spin period"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000517,"raw_usage":{"total_tokens":2400,"prompt_tokens":854,"completion_tokens":1546,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":598,"completion_tokens_details":{"reasoning_tokens":1476}},"tokens_in":598,"tokens_out":1546,"duration_ms":11162,"temperature":1.0,"reasoning_tokens":1476,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T04:44:19.432539+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-observe PSRs J0627+0706 and J0953+0755 at roughly five times finer time resolution than the current 49 µs, and require the same interpulse quasi-periodic peaks (Pµ ≈ 0.94 ms and ≈ 0.38 ms) and ACF first minima to reappear in independently cleaned residuals; if the peaks vanish or shift by more than the quoted errors, or if matching peaks appear in the ACF/FFT of pure noise processed the same way, the manual peak-picking was not seeing true interpulse microstructure.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Origin of the Δτµ ≈ 10^-3 P scaling that motivates the rotation-period dependence."},{"cited_title":"The emission geometry of pulsars with interpulses","cited_arxiv_id":"2503.13824","evidence_quote":"The rotating-vector-model polarization analysis used to assign MP and IP to the same or different magnetic poles and flux tubes."},{"cited_title":"1998, A&A, 332, 111","cited_arxiv_id":null,"evidence_quote":"Independent τµ/Pµ values for PSRs B0823+26 and B0950+08 used for cross-checks, plus the beaming Lorentz-factor formula."}],"review_version":1}