{"id":"29252947-5fcd-49fc-8713-24d40c4f0661","arxiv_id":"1908.03507","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"CHIME/FRB discovered eight new repeating fast radio burst sources and found, at about 4 sigma, that repeater bursts are intrinsically wider than bursts that have not repeated.","lead":"Astronomers report eight new repeating fast radio burst sources found with the CHIME telescope, bringing the known repeater population from two to ten. The wider sample reveals that repeater bursts tend to be broader than one-off bursts, hinting that the two types may have different origins.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Width-difference claim rests on cross-era samples: §4.4 admits the two samples came from different commissioning stages, yet the §4.5 width test never controls for that selection.","rationale":"The reader identified sub-burst independence as the weakest assumption, but the paper partly addresses this with a source-averaged comparison that still gives ~3.5σ significance. The less-mitigated vulnerability is that the repeater and non-repeater samples were collected under different commissioning configurations. The paper explicitly flags this selection concern in §4.4 for emission bandwidths but does not apply the same scrutiny to the width test in §4.5. This is a cross-cutting bias that would affect every burst in one sample, so it is more load-bearing than the within-source clustering issue. I would keep the reader's CONDITIONAL verdict: the discovery of eight new repeaters is solid, but the 4σ width claim should be re-examined with epoch- and pipeline-matched samples before being treated as settled.","tokens_in":36297,"tokens_out":6684,"duration_ms":74267,"concrete_test":"Reprocess the 12 non-repeating CHIME/FRB events from CHIME/FRB Collaboration et al. (2019a) with the same burst-fitting pipeline and the later commissioning configuration (including the multi-spectral-index search) used for the repeater bursts, then rerun the k-sample Anderson-Darling test on widths from matched epochs. If the significance drops below roughly 3σ, the width excess is a selection artifact rather than evidence for different emission mechanisms.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that repeater bursts are intrinsically wider than apparently non-repeating CHIME/FRB bursts, with 4σ significance (§4.5). For this to hold, the width distributions must be directly comparable. The paper's own §4.4 states that the two samples were detected at different stages of commissioning and were likely subject to different selection biases, giving as an example the real-time search moving from a flat spectral-index assumption to searches over −3 and +3. The §4.5 width comparison does not match epochs or pipeline configurations, and it only excludes repeater bursts with S/N<10; it does not quantify whether the earlier pipeline was equally sensitive to wide, low-S/N, or steep-spectrum bursts. A difference in detection efficiency as a function of width could produce exactly the observed excess without any difference in emission mechanism. The DM-distribution null check does not constrain width-dependent selection, and the source-averaged 3.5σ result still inherits the same cross-era bias. Thus the 4σ population claim is underdetermined by the presented analysis.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the discovery of eight new repeating fast radio burst sources with the CHIME telescope, providing sky localizations, dispersion measures, burst properties (widths, fluences, drift rates, scattering times), and detailed follow-up on two low-DM sources including a rotation measure and polarization measurement for Source 1. The authors also compare the DM distributions and burst width distributions of repeaters against the first twelve CHIME/FRB apparent non-repeaters, and report a ~4σ significance that repeater bursts are wider, suggesting different emission mechanisms. Additional results include repetition-rate estimates and a comparison of scattering times.","tokens_in":36436,"tokens_out":5487,"duration_ms":51879,"significance":"If the width-difference claim holds, it would be a substantive step toward distinguishing repeating from apparently non-repeating FRB populations. The paper is also valuable as a carefully documented catalog: localizations are calibrated with ~30,000 pulsar events, burst properties are measured with multiple independent methods, and the detailed analysis of Source 1 (DM, RM, polarization, persistent radio source limits) is a substantial contribution. The authors are appropriately cautious in several places, for example in declining to draw conclusions from the peak-flux--width correlation due to selection effects and correlated sub-bursts, and in stating the caveats on emission-bandwidth differences. The central width comparison, however, relies on assumptions about cross-era selection that are explicitly acknowledged for bandwidths but not for widths, and on independent-draw statistics for clustered sub-bursts.","major_comments":[{"comment":"The width comparison in Section 4.5 does not control for the different commissioning epochs of the repeater and non-repeater samples. Section 4.4 explicitly states that 'the two samples were detected at different stages of commissioning and were thus likely subject to different selection biases,' giving the evolution of the spectral-index search as an example. The Section 4.5 analysis excludes only repeat bursts with S/N < 10 and checks against DM smearing, but it does not quantify whether the earlier detection pipeline had the same sensitivity to wide, low-S/N, or steep-spectrum bursts. A width-dependent detection efficiency difference between epochs could produce exactly the observed excess without any intrinsic population difference. The DM-distribution null check does not constrain width-dependent selection, and the source-averaged result inherits the same cross-era bias. The abstract's 4σ claim is therefore underdetermined by the presented analysis. I request an epoch-matched comparison, injection-based sensitivity simulations as a function of width, or a clear statement that this selection effect is unmodeled and a corresponding softening of the claim.","section":"Section 4.4 and Section 4.5"},{"comment":"The Anderson-Darling and Kolmogorov-Smirnov tests on individual Gaussian components treat every sub-burst as an independent random draw. Yet multiple components from the same burst share a single detection event, propagation path, and source activity state, and bursts from the same repeating source are clustered. The paper itself acknowledges correlated sub-bursts in the peak-flux versus width analysis ('sub-burst measurements in our sample – which likely possess correlated noise properties for sub-bursts grouped in the same detection event') but the width distribution test does not correct for this clustering. The headline 4σ significance is from the per-component test, while the source-averaged result is ~3.5σ. The analysis should either use a hierarchical or bootstrap test that resamples at the burst/source level, or the paper should report the source-averaged significance as the primary result and qualify the abstract accordingly.","section":"Section 4.5, width-distribution tests"},{"comment":"The paper states: 'For the sources which do not have a significant width measurement, we assume the corresponding 95% confidence upper limit to be the measured value.' If a substantial fraction of the apparent non-repeaters have only upper limits, replacing the true (unknown) width with the upper-limit value truncates the distribution and could bias the comparison toward narrower widths for that sample. Please report how many bursts in each sample are treated this way, and test sensitivity by repeating the comparison with alternative treatments (e.g., excluding upper limits, or drawing values from the measured distribution).","section":"Section 4.5, treatment of upper limits"}],"minor_comments":[{"comment":"The abstract reports 'with 4σ significance' without noting that this is the per-component test; the source-averaged significance is ~3.5σ. Please specify which test the headline number refers to, given the clustering caveat.","section":"Abstract and Section 4.5"},{"comment":"In Section 4.7 the text says 'Figure 8 shows the observed and scaled Poisson repetition rates,' but the repetition-rate figure is Figure 9 (Figure 8 is the peak-flux versus width plot). Please correct the cross-reference.","section":"Section 4.7 and Figure 9"},{"comment":"The sentence 'six sources were detected twice, another three times, and one ten times' is grammatically ambiguous; it would be clearer as 'six sources were detected twice, one source three times, and one source ten times.'","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The paper is a valuable catalog paper with careful calibration and detailed source characterization, and the width-difference result is likely of interest to the FRB community. However, the central width claim in the abstract rests on cross-era comparisons and independent-draw statistics that the authors themselves flag as problematic in adjacent analyses. I believe the result can be made robust with re-analysis (epoch matching, source-level resampling, quantifying upper-limit effects), so major revision is appropriate rather than rejection. I would also encourage the collaboration to consider posting the burst-fitting code or width measurements as machine-readable tables if not already done."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this is the paper that turns the repeater sample from a curiosity into a population. Eight new repeating FRBs, careful localizations, low-DM targets, an RM measurement, and a genuine attempt at comparing repeaters to apparent non-repeaters. Most of the paper is solid, and the width claim is the one place I'd push back.\n\nThe detection work is genuinely careful. The localizations use ~30k pulsar events to set empirical confidence regions, which is a step up from earlier CHIME/FRB papers. The baseband confirmation for Sources 1, 5, and 8 is a nice extra. Source 1's low DM excess and RM of -114.6 rad/m² are interesting; their discussion of whether it could be Galactic is honest and reasonably thorough. The repetition-rate section was also upfront about the selection bias in per-object rates.\n\nThe soft spot is the 4σ width comparison in §4.5. The paper itself, in §4.4, says the repeater and non-repeater samples were collected during different commissioning stages with different search assumptions (e.g., flat spectral index early, later searching indices from -3 to +3). That means the two samples could have different selection in width, fluence, or spectral shape. The width test does not match epochs or pipeline configurations, so the observed width excess could be at least partly instrumental. They try to argue against a specific bandwidth effect by refitting non-repeater widths on 100-MHz sub-bands, but that does not address the broader selection difference. Also, the test treats each Gaussian sub-burst as an independent sample, even though sub-bursts from the same burst share propagation paths and detection conditions; §4.5 notes this problem for the peak-flux-width correlation but does not correct for it in the width distribution test. The per-source average reduces but does not eliminate the clustering issue, and it lowers the significance to ~3.5σ. So I read the width result as a plausible hint, not a settled 4σ population difference. The abstract's 'suggesting different emission mechanisms' overreaches.\n\nThat said, this is still a paper worth a serious referee. The discovery content is solid, the methods are mostly transparent, and the caveats they do include suggest clear thinking. The fix is to reframe the width claim as tentative, or to redo the comparison on matched epochs once the pipeline is stable.\n\nI'd cite this for the repeater sample and bring it to reading group. Send it to review; the referee should spend time on §4.5.","headline":"Eight new repeaters is a real step change, but the 4-sigma width claim is shakier than the abstract suggests.","tokens_in":37288,"tokens_out":2495,"would_cite":true,"duration_ms":25705,"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":"This paper reports eight new repeating fast radio burst sources from CHIME and shows that their bursts are intrinsically wider than those of apparently non-repeating bursts at about 4$\\sigma$ significance, evidence that the two classes…","keywords":["fast radio bursts","repeating FRBs","burst width distribution","dispersion measure","rotation measure","downward frequency drift","radio transient surveys","CHIME telescope"],"falsifier":"Repeat the Anderson\\,--\\,Darling comparison using only the first detected burst from each repeating source (or one width per burst envelope), so sub-bursts from the same detection are not counted as independent; if the separation falls below 3$\\sigma$, the claimed 4$\\sigma$ finding is inflated by correlated measurements.","tokens_in":36110,"feed_emoji":"📡","tokens_out":8384,"duration_ms":80012,"temperature":0.7,"pith_summary":"Eight new repeating fast radio burst sources, discovered in CHIME data, grow the known repeater population from two to ten. The paper establishes that the repeater sample spans dispersion measures from 103.5 to 1281 $\\mathrm{pc\\,cm^{-3}}$, including two nearby, low-excess-DM sources that are strong targets for multi-wavelength follow-up. Its central claim is a statistical one: at about 4$\\sigma$, bursts from repeating sources are intrinsically wider than bursts that have not been seen to repeat, which the authors read as evidence for different emission mechanisms. It also shows that repeater and non-repeater dispersion-measure distributions are indistinguishable, and that repeater morphology often includes downward-drifting sub-bursts, though not always. If the width difference holds, repetition status becomes a probe of underlying FRB physics rather than a mere observational accident.","feed_headline":"Repeater bursts are wider than one-off FRBs at 4-sigma significance","feed_subtitle":"CHIME's eight new repeating sources point to distinct emission physics behind bursts that repeat.","key_machinery":"The load-bearing comparison is the intrinsic temporal width of each fitted Gaussian burst component, measured in milliseconds after dedispersion and scattering fits. The paper's statistical engine is a pair of non-parametric two-sample tests (k-sample Anderson\\,--\\,Darling and Kolmogorov\\,--\\,Smirnov) applied to the width distributions, first per component and then per source using inverse-variance weighted averages; excluding S/N$<10$ repeat bursts controls for the lower trigger threshold for repeat candidates. Supporting this, a coherence-spectrum dedispersion analysis that maximizes sub-burst sharpness assigns structure-optimizing dispersion measures, and a 2-D auto-correlation fit with Monte Carlo resampling characterizes the downward-drifting sub-burst morphology seen in a subset of bursts.","core_discovery":"The paper's central claim is that bursts from repeating FRB sources are intrinsically wider than bursts from sources that have not yet been seen to repeat. Restricting both samples to CHIME detections in the 400\\,--\\,800 MHz band, and excluding repeat bursts with S/N below the new-source threshold to guard against a pipeline bias, the authors compare the widths of Gaussian burst components and find the two distributions differ at roughly 4$\\sigma$ (k-sample Anderson\\,--\\,Darling), with source-averaged widths differing at about 3.5$\\sigma$ and Kolmogorov\\,--\\,Smirnov tests giving 4\\,--\\,5$\\sigma$. They also find no statistically significant difference in the DM distributions of repeaters and apparent non-repeaters, and they measure downward frequency drifts in eleven bursts, complex sub-burst structure, and, for one source, a modest rotation measure of $-114.6 \\pm 0.6$ rad m$^{-2}$ that is far below FRB 121102's. The width difference, if it holds under selection corrections, is their main evidence that repeaters form a distinct population with different emission physics.","pith_inferences":["A concrete next test: pool published widths for all CHIME repeaters plus FRB 121102 and FRB 180814.J0422+73, but weight each burst envelope as a single draw; the expected significance drop will tell whether the 4$\\sigma$ result is dominated by a few multi-component bursts.","If narrow-band (100\\,--\\,150 MHz) repeater spectra are confirmed with forward beam models, spectral shape could join width as a classification feature, allowing rapid triage of single bursts for repeated-emission follow-up.","The low RM and low DM excess of Source 1, if it proves extragalactic, would place it in a region of DM\\,--\\,RM phase space occupied by older neutron stars, suggesting repeaters may span an age sequence from FRB 121102-like young nebulae to older, cleaner environments.","Source 2's potential host NGC 3403 could be tested with a targeted VLBI localization: a burst in the disk outskirts or halo would bound the host DM contribution and discriminate halo versus disk progenitor scenarios."],"forward_implications":["If the width gap is real, apparently non-repeating FRBs and repeating FRBs are not drawn from one homogeneous population, so models of FRB emission must explain why repeaters are systematically broader.","Low-DM Sources 1 and 2 become priority targets for interferometric localization; Source 1's low rotation measure and absence of a bright persistent radio source disfavor a young, FRB 121102-like magnetized wind nebula, at least at current sensitivity.","Repeater burst rates estimated from detections of repeat bursts are biased high because they select the active tail; the mild inconsistency between CHIME repeater rates and upper limits from a different survey can be resolved with a larger sample.","Downward-drifting sub-bursts, detected in nine repeaters and in none of the apparent non-repeaters, may be a useful (though not decisive) marker of repetition at about 1-ms resolution.","Source-averaged widths show a slightly weaker but still significant difference, meaning the population-level claim does not depend on any single hyperactive source."],"supporting_citations":[{"why":"It supplies the 12 apparent non-repeaters whose widths and DMs are the comparison sample for the central width claim.","marker":"CHIME/FRB Collaboration et al. 2019a"},{"why":"It provides the second known repeater FRB 180814.J0422+73, included in the repeater width and morphology sample.","marker":"CHIME/FRB Collaboration et al. 2019b"},{"why":"It contributes CHIME-band widths, drift rates, and completeness methodology for FRB 121102 bursts used in the comparison.","marker":"Josephy et al. 2019"},{"why":"It first reported that FRB 121102 bursts are wider than non-repeating Parkes bursts, the precedent this paper extends to CHIME.","marker":"Scholz et al. 2016"},{"why":"It established the downward-drifting sub-burst morphology and autocorrelation method used to characterize repeater structure.","marker":"Hessels et al. 2018"},{"why":"It characterized FRB 121102's repeating bursts, the archetype defining repeat-burst morphology and activity.","marker":"Spitler et al. 2016"},{"why":"It provides the NE2001 Galactic electron-density model used to estimate Milky Way DM contributions for the low-DM sources.","marker":"Cordes & Lazio 2002"},{"why":"It provides the YMW16 Galactic electron-density model used alongside NE2001 to constrain DM excess and source distances.","marker":"Yao et al. 2017"}],"fun_headline_variants":["Repeater bursts are wider than one-offs: 4-sigma evidence","8 new repeaters: bursts 4-sigma wider than non-repeaters","Wider bursts from repeaters hint at different FRB physics","CHIME finds repeaters emit wider bursts, 4-sigma difference","Repeat bursts are wider: new CHIME sources at 4-sigma"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The width comparison treats each measured sub-burst as an independent draw and assumes the repeater and non-repeater samples suffer no selection differences, though sub-bursts from the same burst or source share a common detection envelope and the two samples were gathered under different commissioning conditions.","fun_headline_variants_meta":{"raw":{"variants":["Repeater bursts are wider than one-offs: 4-sigma evidence","8 new repeaters: bursts 4-sigma wider than non-repeaters","Wider bursts from repeaters hint at different FRB physics","CHIME finds repeaters emit wider bursts, 4-sigma difference","Repeat bursts are wider: new CHIME sources at 4-sigma"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000812,"raw_usage":{"total_tokens":3662,"prompt_tokens":1150,"completion_tokens":2512,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":766,"completion_tokens_details":{"reasoning_tokens":2415}},"tokens_in":766,"tokens_out":2512,"duration_ms":18902,"temperature":1.0,"reasoning_tokens":2415,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:11:25.427168+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the Anderson\\,--\\,Darling comparison using only the first detected burst from each repeating source (or one width per burst envelope), so sub-bursts from the same detection are not counted as independent; if the separation falls below 3$\\sigma$, the claimed 4$\\sigma$ finding is inflated by correlated measurements.","supporting_citations":[{"cited_title":"CHIME/FRB Detection of the Original Repeating Fast Radio Burst Source FRB 121102","cited_arxiv_id":"1906.11305","evidence_quote":"It contributes CHIME-band widths, drift rates, and completeness methodology for FRB 121102 bursts used in the comparison."},{"cited_title":"G., Hessels , J","cited_arxiv_id":null,"evidence_quote":"It first reported that FRB 121102 bursts are wider than non-repeating Parkes bursts, the precedent this paper extends to CHIME."},{"cited_title":"G., Scholz , P., Hessels , J","cited_arxiv_id":null,"evidence_quote":"It characterized FRB 121102's repeating bursts, the archetype defining repeat-burst morphology and activity."}],"review_version":1}