{"id":"94273a07-cb1d-494c-af28-b7c16aa896cf","arxiv_id":"2610.31365","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"FRB 20210601A fired nine bursts in 106.5 seconds and none in roughly 320 hours of later monitoring, evidence that some repeating fast radio bursts are episodic storm emitters rather than steady repeaters.","lead":"Astronomers found a fast radio burst source that fired nine bright bursts in under two minutes and then stayed silent for years despite hundreds of hours of additional observing. The find suggests some repeaters may switch on in rare, intense storms that current surveys systematically miss, and predicts several more such sources in the next decade.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"Population forecast rests on Eq. (2)/(3); the exposure law may understate FRB 20210601A's own daily exposure by ~2x, and phase-locked storms would change pdet qualitatively.","rationale":"The observational core of the paper is convincing: nine bursts in a 106.5 s window, consistency with Poisson within the window, Monte Carlo KS testing, absence of sub-threshold excess, and a bootstrap check against side-lobe truncation all support the extreme-clustering claim. The outlier status relative to CHIME/FRB repeaters is also reasonably supported by the conservative p<1.6e-4 estimate. I therefore do not object to the main empirical finding. The load-bearing weakness is the population extrapolation. The reader identified the exposure model and storm-phase assumption as the weakest point, and I agree. My reading sharpens this: the paper's own exposure numbers for FRB 20210601A appear inconsistent with Eq. (2) at δ≈65° by about a factor of two, which directly scales the inferred population size and the decade forecast. The phase-randomness assumption in Eq. (3) is also not tested, and known periodic repeaters show that phase structure exists in at least some sources. The authors are transparent that Twait is unconstrained and that the forecast is anchored to the single detection, but the abstract presents the 1-4 forecast without that caveat, and the title claims evidence for a hidden class while Section 5 leaves the question uncertain. These issues justify the reader's CONDITIONAL verdict; I do not see a reason to move to reject, since the central clustering observation stands and the population model is explicitly framed as a simplified first framework.","tokens_in":38161,"tokens_out":11286,"duration_ms":96458,"concrete_test":"Recompute the Section 4.4 posterior using a Monte Carlo pdet built from the actual recorded CHIME/FRB transit epochs and exposure windows rather than Eq. (2), and repeat with storm phases locked to a grid of fixed sidereal-day-commensurate periods instead of uniform random daily phases. If the 68% interval on N or the 10-year forecast moves outside the quoted 9(+200/-8) or 1-4 ranges, the population claim must be presented as conditional on the exposure and phase model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The hidden-class population claim in Section 4.4 depends on the detectability model, not on the clustering evidence, which is strong. The load-bearing step is pdet(δ) built from Eq. (3) with the empirical exposure law d(δ)=E0/cos(δ), E0≈2.2 min (Eq. 2). For the detection bin δ≈65°, this gives d≈5.2 min/day. But Section 2 reports that FRB 20210601A's daily exposure at the detection epoch was approximately 10 minutes, and the quoted 320 hr over 2025 transits implies an average of about 9.5 min per non-zero transit. If the true daily exposure in the detection bin is larger by roughly a factor of two, then Λ(δ≈65°) for fixed N is underestimated by the same factor; since the one-detection likelihood anchors Λ≈1, the inferred N=9(+200/-8) and the 1-4 forecast scale nearly inversely with pdet and shift materially. Separately, Eq. (3) averages storm phase uniformly over the day. If episodic storms recur at a fixed physical phase (spin, orbit, or magnetospheric clock), the probability of overlap with CHIME's fixed sidereal dwell is not (d+Tstorm)/1440 per storm but can be near 0 or near 1 depending on phase, changing the N-Twait posterior qualitatively. The authors acknowledge that Twait is essentially unconstrained and that the forecast recovers the single detection by construction (Section 4.4.1), but the exposure/phase assumption remains the unstated load-bearing element of the population claim. The clustering and outlier claims are robust to this concern.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"This paper reports the discovery and detailed characterization of FRB 20210601A, a CHIME/FRB repeater that emitted nine bursts within 106.5 seconds during a single side-lobe transit and was not detected in roughly 320 hours of additional exposure spanning 2018-2025. The authors characterize the burst properties, polarization, localization, and environment; compare the source with other CHIME/FRB repeaters; show that the within-storm waiting times are consistent with a Poisson process while the long quiescence is incompatible with a stationary rate inferred from the active window; and construct a toy population model of episodic repeaters with declination-dependent exposure. The model yields a CHIME/FRB-detectable population of N = 9^{+200}_{-8} sources and a forecast of 1-4 additional discoveries in the next decade. The paper explicitly acknowledges several limitations, including the essentially unconstrained storm recurrence time T_wait and the fact that the forecast recovers the single detection by construction.","tokens_in":38567,"tokens_out":8807,"duration_ms":74619,"significance":"The observational core of this paper is strong and important. The contrast between the active-window rate of 0.075 bursts/s and the subsequent 188 hours of silence is enormous: if the active-window rate had persisted, tens of thousands of bursts would have been expected, so the change in repetition behavior is essentially model-independent. The paper is also careful in several respects: the burst definition is stated explicitly, the side-lobe nature of the detections is checked against the CHIME beam model, composite and baseband data are used to separate instrumental from intrinsic spectral structure, and the main limitations of the population model are acknowledged. The population-level claim of a hidden class of episodic repeaters is much more fragile: it rests on a simplified exposure law, on an assumed random phase between storms and CHIME dwell windows, and on a single detected event. The paper is honest about these uncertainties, but the headline population size and forecast are not yet robust enough to support the 'evidence for a hidden class' framing.","major_comments":[{"comment":"The exposure model used for the population inference appears inconsistent with the reported exposure of FRB 20210601A itself. Section 2 states that the daily exposure at the detection epoch was approximately 10 minutes, and the quoted 320 hours over 2025 non-zero transits implies roughly 9.5 minutes per usable transit, whereas Eq. (2) with E0 ≈ 2.2 minutes gives d(65°) ≈ 5.2 minutes/day, a factor of about two lower. This factor enters Λ(δ) linearly through Eqs. (3)-(4), and because the one-detection likelihood accumulates most posterior weight near Λ ≈ 1, the inferred N and the 1-4 forecast scale almost inversely with p_det. The authors should either adopt the measured per-source daily exposure for the detection bin or propagate the uncertainty in E0 explicitly and show how N and the forecast respond. As written, the population size and forecast are sensitive to a plausible calibration error in the exposure model.","section":"§2 and §4.4, Eq. (2)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Equation (3) assumes that storm phases are uniformly random relative to the telescope's daily dwell window and treats successive storms as independent trials. If episodic activity is phase-locked to a physical clock (spin, orbit, or magnetospheric cycle), the probability of overlap with CHIME's fixed sidereal dwell is not (d + T_storm)/1440 per storm but can be near zero or near unity depending on the phase. This would change the N-T_wait posterior qualitatively, not just by a numerical factor. The paper acknowledges that T_wait is essentially unconstrained, but the population forecast still depends on the phase-randomization assumption. A sensitivity analysis that contrasts random-phase storms with phase-locked storms, or that introduces and marginalizes over a phase parameter, is needed before the hidden-class population claim can be regarded as supported.","section":"§4.4, Eq. (3)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The one-detection likelihood in Eq. (6) is conditioned on the detection of FRB 20210601A, so, as the authors acknowledge, the mean prediction at the Catalog 2 epoch recovers the historical count by construction rather than by independent validation. The consequence is that the quoted interval N = 9^{+200}_{-8} and the 1-4 discovery forecast are statements conditional on the toy detectability model of Eqs. (2)-(3), not empirical constraints on the FRB population. I recommend adding a quantitative sensitivity analysis over E0, T_storm, and the phase assumption, and labeling the forecast as conditional on those modeling choices in the abstract and conclusions.","section":"§4.4.1 and Eq. (6)","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The reported probability p < 1.6e-4 is an empirical survival-product computed within the observed CHIME/FRB repeater sample; please state explicitly that it is a within-sample percentile statement rather than a model-based false-alarm probability for the underlying population, so that it is not misinterpreted as a population-level significance.","section":"§4.1, Fig. 2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The cumulative fluence index quoted in the text is -0.6(0.2), while the fit restricted to bursts above the 95% completeness threshold gives -1.1(0.9); the paper should discuss this difference explicitly in the main text, since the main-lobe burst-count correction of roughly 2.3 uses the shallower index.","section":"Figure A.6 and §4.1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The title's phrase 'Evidence for a Hidden Class of Episodic Repeaters' is stronger than the analysis supports, given the acknowledged sensitivity of the population inference to the detectability model and the single detected storm; I recommend either tempering the title and abstract or strengthening the population analysis with the sensitivities requested above.","section":"Title and Abstract","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The comparison with FAST sources is appropriately hedged about differing bands, fluence thresholds, and exposure strategies, but the claim that the 106.5 s window is 'the highest instantaneous burst rate yet reported' could be made more precise by stating the explicit conditions under which it holds, since the comparison sources are observed in a different band with different cadence.","section":"§4.2","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision","confidential_remarks":"The observational result is solid and suitable for publication, but the population-level claims need substantial revision or reframing. The exposure-model inconsistency in Eq. (2) and the phase assumption in Eq. (3) are fixable by sensitivity analysis, so I do not recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"Short version: the observational core is strong and should be published; the \"hidden class\" population inference is a toy model calibrated on one detection, and the abstract and title oversell it. Nine bursts in 106.5 s at fluences of about 2 Jy ms or more, then nothing in about 188 hr of subsequent exposure and 132 hr before, is overwhelmingly inconsistent with the rate measured in the active window. The paper earns that claim, and it does careful ancillary work: a sub-threshold search, a Monte Carlo KS test, side-lobe sensitivity scaling, and a candid audit that eight of the nine bursts were already in Catalog 2 plus one recovered faint burst. The comparison against FAST bursts from FRB 20201124A and FRB 20240114A at comparable fluence is a fair way to establish the \"highest instantaneous rate at this fluence\" claim, with band and cadence caveats stated.\n\nThe soft spots are in the population forecast. Section 4.4 uses a one-detection Poisson likelihood, and Section 4.4.1 admits the mean prediction recovers the single detection by construction. The numbers N = 9(+200/-8) and the 1-4 source forecast are therefore a consistency statement, not an independent prediction, and the abstract's unqualified \"forecast\" is stronger than the body supports. There is also a specific exposure-model problem. Equation (2) gives about 5.2 min/day at declination 65 degrees, while Section 2 says the source's daily exposure at detection was about 10 minutes, and the quoted 320 hr over 2025 transits averages about 9.5 minutes per transit. If the detection-bin exposure is understated by roughly a factor of two, then the expected detection rate at fixed N is understated by the same factor, and because the one-detection likelihood forces the sky-integrated rate near unity, the inferred N shifts roughly inversely. The authors do flag T_wait as essentially unconstrained and are transparent about the by-construction nature of the forecast, but they do not flag this exposure-law inconsistency. Equation (3) also averages storm phase uniformly over the day; a phase-locked storm clock would change p_det qualitatively. The clustering and outlier claims do not depend on any of this and are robust.\n\nThe title says \"Evidence for a Hidden Class\"; Section 5 says whether this is a statistical outlier or the first member of an under-sampled class remains uncertain. That mismatch should be fixed before publication.\n\nBottom line: send it to review. A serious referee should push on Equations (2)-(3) and the exposure bookkeeping, and the authors should soften the abstract and title. I would cite the observational result, not the population numbers.","headline":"A genuinely extreme repeating FRB with a solid clustering claim, wrapped in a population forecast that is too under-determined to carry the title.","tokens_in":40065,"tokens_out":809,"would_cite":true,"duration_ms":29003,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"FRB 20210601A fired nine bursts in a 106.5-second storm, the highest instantaneous burst rate yet reported from a repeating FRB at fluences near 2 Jy ms, then stayed silent through seven years of monitoring — evidence for a hidden class…","keywords":["fast radio bursts","repeating FRB sources","burst storms","episodic repeaters","CHIME/FRB survey","temporal clustering","radio transient sources","compact objects"],"falsifier":"Computing CHIME/FRB's true declination-dependent exposure from the beam model and telescope pointing logs would test the population claim directly, because the inferred $N=9^{+200}_{-8}$ scales with the assumed law $d(\\delta)=E_0/\\cos(\\delta)$; if the real daily dwell at the source's declination deviates substantially from about 2.2 minutes divided by $\\cos(\\delta)$, the forecast of one to four discoveries changes accordingly. Observationally, a second burst storm from FRB 20210601A within a few years, or one to four sibling storm-then-silence sources in the ongoing survey, would confirm the episodic class, whereas continued decades-long silence would push the interpretation toward a one-off or transient-lensing event.","tokens_in":39521,"feed_emoji":"📡","tokens_out":16597,"duration_ms":109858,"temperature":0.7,"pith_summary":"This paper reports a repeating fast radio burst source, FRB 20210601A, whose detectable activity is confined to a single 106.5-second storm: at least nine bursts at fluences of roughly 2 Jy ms or brighter, the highest instantaneous burst rate yet reported from a repeater at comparable sensitivity, followed by about seven years and roughly 320 hours of monitoring with no further bursts. The paper shows that this storm-then-silence pattern is statistically inconsistent with the stationary, single-rate repetition models usually used for repeaters, estimating a chance probability at most $1.6\\times10^{-4}$ within the CHIME/FRB repeater sample. Because the storm arrived through a telescope side-lobe, the true rate was at least 2.3 times higher, implying about 21 bursts at main-lobe sensitivity. The authors argue the source is the first clear example of a hidden class of highly episodic repeaters — sources dormant most of the time that erupt in short burst storms — and forecast a CHIME/FRB-detectable population of $9^{+200}_{-8}$ such sources, with one to four additional discoveries in the next decade. If the claim is right, FRB repetition statistics must treat state-switching as a real, selection-biased behavior rather than fitting a single average rate.","feed_headline":"An FRB fired 9 bursts in 106 seconds, then fell silent for years","feed_subtitle":"The pattern defies steady-rate models of FRB repetition and points to a hidden population of episodic repeaters.","key_machinery":"The central object is the burst storm: a short episode of dense emission, here fixed to the observed 106.5 s, treated as the unit of activity for sources that are otherwise dormant. The argument runs through three pieces of machinery. First, a comparison space of detections per active transit against waiting time between active transits, which locates FRB 20210601A as an extreme outlier among CHIME/FRB repeaters. Second, a side-lobe sensitivity correction: all nine bursts arrived through a beam side-lobe whose effective response was about 0.25 of the main lobe, so the storm rate was roughly four times higher than observed, implying about 21 bursts (and possibly about 70 under a steeper fluence index) within 106.5 s. Third, a two-parameter Poisson population model in which each source produces storms of duration $T_{\\rm storm}=106.5$ s every $T_{\\rm wait}$ days, with the chance of catching a storm set by the declination-dependent daily exposure $d(\\delta)=E_0/\\cos(\\delta)$ with $E_0\\approx2.2$ minutes; Bayesian nested sampling over the one observed detection yields the population size $N=9^{+200}_{-8}$ and the next-decade forecast of one to four discoveries. The paper explicitly acknowledges that $T_{\\rm wait}$ remains essentially unconstrained by the data.","core_discovery":"The central discovery is that a single repeating FRB can pack essentially all of its detectable activity into a 106.5-second window. FRB 20210601A emitted nine bursts on 2021 June 1, internally consistent with a Poisson process at 0.075 bursts per second, after roughly 132 hours of prior exposure and before 188 hours of subsequent exposure accumulated in daily 10-minute dwells over about seven years, with zero detections outside the window. The paper argues the sharp cutoff is real rather than a sensitivity artifact: mock transits show the side-lobe would have caught bursts brighter than about 10 Jy ms had activity continued, and silence beyond 146.4 seconds is already inconsistent with the storm's own rate. Within the CHIME/FRB repeater sample, the combination of nine intra-transit detections and a waiting time to the next transit of at least 1674 days places the source alone in its corner of the detections-per-transit versus waiting-time plane, with a chance occurrence probability at most $1.6\\times10^{-4}$ under conservative independence assumptions. On this basis the authors propose that FRB 20210601A is an extreme manifestation of a hidden population of episodic repeaters, and they use a two-parameter Poisson model with declination-dependent exposure to estimate a CHIME/FRB-detectable population of $9^{+200}_{-8}$ sources and a forecast of one to four additional discoveries over the next decade.","pith_inferences":["If episodic storms are as prevalent as the model suggests, many apparent one-off FRBs could be storms whose 106-second active windows fell entirely between the telescope's roughly 10-minute daily dwells; the paper gestures at this possibility for one-off sources but does not quantify the overlap correction.","The random-phase exposure model is the hinge of the population claim: if storm recurrence is phase-locked to a spin, orbit, or magnetospheric cycle, catchability is no longer simply $T_{\\rm storm}/T_{\\rm dwell}$, and the inferred population of $9^{+200}_{-8}$ sources could shift by orders of magnitude in either direction.","A faster test of the episodic hypothesis would be to re-mine archival baseband data at the positions of apparent one-off FRBs for sub-threshold mini-storms; even a two-burst cluster in a single transit would raise the episodic fraction above what the catalog currently reports.","The Galactic analogy with rare, roughly 90-second bright-mode windows in some rotating radio transients suggests a luminosity-scaled continuum of storm behavior; monitoring magnetars during X-ray burst activity for clustered radio emission on roughly 100-second timescales would test whether the same storm phenomenology operates locally."],"forward_implications":["Repetition statistics for FRBs must move beyond single time-averaged rates: if episodic behavior exists, steady-rate models will systematically misclassify such sources as weak repeaters or apparent one-offs.","If the population estimate holds, CHIME/FRB should find one to four additional storm-then-silence sources over the next decade, preferentially in sky regions with the highest combined sensitivity and daily exposure.","Because the storm sat in a side-lobe, its true rate was at least 2.3 times the nine detected bursts, about 21 bursts in under two minutes, so fixed-beam surveys can undercount extreme burst rates by factors of several.","The source's otherwise ordinary environment — excess dispersion measure near 360 pc cm$^{-3}$, rotation measure near $-150$ to $-170$ rad m$^{-2}$, temporal scattering below 0.37 ms — means episodic behavior can hide among normal repeaters and does not require an exotic host.","Targeted high-sensitivity monitoring during apparent quiescence can directly test whether bursts continue below CHIME/FRB's detection threshold or whether the source is genuinely off between storms."],"supporting_citations":[{"why":"The second CHIME/FRB Catalog: supplies eight of the nine bursts, the exposure statistics per declination, and the source's cataloged properties.","marker":"CHIME/FRB Collaboration et al. 2026a"},{"why":"The repeater sample and repetition framework against which FRB 20210601A is measured as an outlier, and the source of the exposure model's calibration.","marker":"A. M. Cook et al. 2026a"},{"why":"The instrument paper: beam model, real-time detection pipeline, and baseband recording used to derive the side-lobe sensitivity correction.","marker":"CHIME/FRB Collaboration et al. 2018"},{"why":"A representative stationary single-rate repetition model whose consistency with the repeater sample is the baseline the clustered source must break.","marker":"C. W. James 2023"},{"why":"The FAST burst catalog of FRB 20240114A used to benchmark the 106.5 s peak rate at comparable fluence and to supply the energy-budget scaling method.","marker":"J.-S. Zhang et al. 2025"},{"why":"FAST observations of FRB 20201124A providing the second benchmark distribution of burst counts per 106.5 s window at CHIME-comparable fluences.","marker":"D. J. Zhou et al. 2022"},{"why":"Tracking-beam monitoring of FRB 20220912A showing static CHIME beams miss about 70% of bursts, the evidence that the side-lobe storm was substantially more intense than the nine detections.","marker":"T. C. Abbott et al. 2026"},{"why":"Documentation of FRB 20201124A's transition from long quiescence to intense activity, the closest previously known episodic analog.","marker":"A. E. Lanman et al. 2022"},{"why":"The nested-sampling implementation used to evaluate the population posterior and produce the $N=9^{+200}_{-8}$ estimate and discovery forecast.","marker":"J. Buchner 2021"}],"fun_headline_variants":["FRB packs 9 bursts into 106 seconds, then goes quiet for years","Nine FRB bursts in 106 seconds, none in 320 hours of other exposure","A repeating FRB fires a 106-second storm, then seven years of silence","FRB 20210601A: 9 bursts in 106 seconds, 320 hours of silence","Episodic FRB repeater erupts for 106 seconds, then stays hidden for years"],"cache_read_input_tokens":0,"weakest_assumption_plain":"Everything beyond the clustering claim — the inferred hidden population of $9^{+200}_{-8}$ sources and the forecast of one to four more discoveries — rests on the assumption that burst storms occur at random phases relative to the telescope's roughly 10-minute daily look-in, so a storm's catchability is governed by the ratio of storm length to daily exposure as encoded in $d(\\delta)=E_0/\\cos(\\delta)$; if storm timing is instead locked to a physical clock such as a spin, orbit, or magnetospheric cycle, or if that exposure law is inaccurate, the population numbers change.","fun_headline_variants_meta":{"raw":{"variants":["FRB packs 9 bursts into 106 seconds, then goes quiet for years","Nine FRB bursts in 106 seconds, none in 320 hours of other exposure","A repeating FRB fires a 106-second storm, then seven years of silence","FRB 20210601A: 9 bursts in 106 seconds, 320 hours of silence","Episodic FRB repeater erupts for 106 seconds, then stays hidden for years"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":0.000149,"raw_usage":{"cost":0.0001490216,"is_byok":false,"cost_details":{"upstream_inference_cost":0.0001490216,"upstream_inference_prompt_cost":3.65456e-05,"upstream_inference_completions_cost":0.000112476},"total_tokens":1974,"prompt_tokens":1356,"completion_tokens":618,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":1193,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":516}},"tokens_in":1356,"tokens_out":618,"cost_source":"openrouter_usage","duration_ms":6348,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":516,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-23T08:14:21.138008+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"falsifier":"Computing CHIME/FRB's true declination-dependent exposure from the beam model and telescope pointing logs would test the population claim directly, because the inferred $N=9^{+200}_{-8}$ scales with the assumed law $d(\\delta)=E_0/\\cos(\\delta)$; if the real daily dwell at the source's declination deviates substantially from about 2.2 minutes divided by $\\cos(\\delta)$, the forecast of one to four discoveries changes accordingly. Observationally, a second burst storm from FRB 20210601A within a few years, or one to four sibling storm-then-silence sources in the ongoing survey, would confirm the episodic class, whereas continued decades-long silence would push the interpretation toward a one-off or transient-lensing event.","supporting_citations":[],"review_version":1}