{"id":"ee0b3679-ea75-48b2-ac0a-39e29d449be5","arxiv_id":"2608.10358","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A 3-sigma excess of fast radio bursts is found along galaxy cluster lines of sight in the CHIME/FRB second baseband catalog, about 1.4% of detections, attributed to member galaxies and gravitational lensing.","lead":"Using CHIME/FRB's second baseband catalog, the authors found 26 fast radio bursts aligned with galaxy clusters, where about 13 would be expected by chance, a 3-sigma rate enhancement. The result suggests galaxy clusters both host extra FRBs and magnify background ones, offering a new way to study FRB origins and the distant universe.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Null Monte Carlo rejects background FRBs (Appendix B step 5), yet §3.4 defines interlopers as unmagnified background FRBs; the expected interloper count is biased low, inflating the 3-sigma excess.","rationale":"The reader identifies P(z|DM) model error as the weakest assumption, but the more load-bearing problem is internal: even granting the z-DM model exactly, Appendix B step 5 conditions the null on z < z_clust for clusters within 3×r500. The paper's own definition of an interloper (§3.4) is an unmagnified background FRB, so the null should include such FRBs. Their exclusion suppresses the high-score tail of the null and artificially boosts both the Anderson-Darling rejection in Fig. 1 and the 26-vs-13.4 comparison that produces p=0.0008. This directly undermines the central claim, not through a contested population model but through a mismatch between the stated null and the simulated null. The paper's robustness tests vary population parameters but not this conditioning, and its 'same search on both samples' argument in §3.4 cannot fix a null that omits a physical population. The proposed test resolves the question: if including background FRBs raises the expected count to within Poisson noise of 26, the enhancement claim fails. I therefore recommend rejection of the central claim as written, though the analysis could be revived by correcting the null and re-running the pipeline.","tokens_in":34906,"tokens_out":8785,"duration_ms":84799,"concrete_test":"Rerun the interloper simulation with Appendix B step 5 modified to accept all simulated FRBs independent of cluster redshifts (i.e., sample z and DM from z-DM without any z < z_clust condition), then reapply the same P_cc threshold and count the expected associations for gamma = -1.0, -0.5, and 0.0. If the expected count becomes statistically consistent with 26 (e.g., p >= 0.05), the claimed 3-sigma rate enhancement is an artifact of the foreground-only null. Also recompute the P_cc distribution and the Fig. 1 Anderson-Darling p-value with this background-inclusive null.","verdict_should_be":"REJECT","load_bearing_attack":"Section 3.4 defines interlopers as 'unmagnified background FRBs that happen to be aligned with a foreground cluster,' and the central claim compares the 26 detected associations to mock catalogs intended to represent that null. Appendix B, step 5, however, rejects every simulated FRB 'with a redshift greater than that of a cluster within 3×r500,' i.e., it removes background FRBs from the null. Under the no-cluster-effect hypothesis, FRBs are distributed independently of clusters, so a substantial number of unmagnified FRBs behind the low-redshift DECaLS clusters (z_FRB > z_clust) should appear in the mock catalogs. Because those FRBs have low P(z<z_clust|DM), they receive high association scores and are exactly the events the search counts. Excluding them from the null lowers the expected interloper count (13.4±3.7 for gamma=-0.5) and inflates the observed excess (26; p=0.0008, claimed 3-sigma). The same foreground-only conditioning enters the P_cc Monte Carlo in §2 and the Anderson-Darling test in Fig. 1, so both the association sample and the significance are affected. This is not cured by the gamma and n_SFR grids or the robustness tests in §3.4, none of which alter the z > z_clust rejection. The argument in §3.4 that applying the same search to mock and real data cancels modeling error does not apply: the null distribution itself is missing a physical population.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes the second CHIME/FRB baseband catalog against the DECaLS cluster catalog of Wen & Han (2024), using a score L = Σ_s / P(z < z_clust | DM) that combines a projected NFW density map with the probability that a burst is foreground to an aligned cluster. Monte Carlo simulations are used to assign each FRB a probability of chance coincidence P_cc, yielding a sample of 26 cluster-associated FRBs. The authors then simulate mock FRB catalogs to estimate the expected number of interloper (unmagnified background) associations, obtaining 13.4±3.7 for γ=−0.5, and compare with 26 observed associations to claim a 3σ rate enhancement that constitutes 1.4±0.4% of the second baseband catalog. The paper also studies cluster DM contributions, identifies a repeating FRB and two Coma-cluster intersections, and highlights FRB 20211113A aligned with Abell 2218 as a possible gravitationally lensed burst.","tokens_in":35313,"tokens_out":5943,"duration_ms":54671,"significance":"If the claimed enhancement is real, the result would demonstrate that cluster sightlines contain a measurable excess of FRBs, providing a new probe of FRB progenitors, cluster environments, and gravitational lensing of fast transients. The paper is clearly written, uses a well-defined masked sample, and is commendably transparent about the DM-based selection bias in the cluster sample (Section 3.1). It also tests a grid of population parameters (γ, n_SFR) and provides a concrete candidate for gravitational lensing with a path toward follow-up. However, the central statistical claim depends on a null simulation that appears to omit a physical population, and the significance estimate is therefore not currently supported.","major_comments":[{"comment":"The null Monte Carlo rejects every simulated FRB with a redshift greater than that of a cluster within 3×r500, i.e., it removes unmagnified background FRBs from the mock catalogs. Yet Section 3.4 defines interlopers as 'unmagnified background FRBs that happen to be aligned with a foreground cluster.' Under the no-cluster-effect null, these background FRBs are a physical population that must be present in the mock catalogs; because they have low P(z<z_clust|DM) and therefore high association scores, excluding them lowers the expected interloper count (13.4±3.7 for γ=−0.5) and inflates the claimed excess (26 associations; p=0.0008). The same foreground-only conditioning enters the P_cc Monte Carlo in Section 2 and the Anderson-Darling test in Figure 1. The robustness tests in Section 3.4 vary γ, n_SFR, host parameters, and completeness, but none of them alter the z>z_clust rejection in Appendix B. The argument that applying the same search to mock and real data cancels modeling error does not apply here, because the null distribution itself is missing a physical population.","section":"Appendix B, step 5; Section 3.4"},{"comment":"Both the association score L = Σ_s / P(z<z_clust|DM) and the null Monte Carlo are generated from the same z-DM population model, but the predicted DM distribution of this model is not validated against the second CHIME/FRB baseband catalog; Appendix B states that the default parameters are consistent only with the first CHIME catalog. If the true DM–redshift relation differs from the model in a way not spanned by the γ and n_SFR grids (for example, in the host-galaxy DM distribution or the local rate normalization), then the P_cc distribution and the interloper counts would be jointly biased, potentially creating a spurious excess. I recommend a direct two-sample comparison of the simulated and observed DM distributions (for example, a Kolmogorov–Smirnov or Anderson–Darling test) and a rerun of the null Monte Carlo with a model re-fit to the second catalog's DM distribution.","section":"Section 2, Table 1; Appendix B"}],"minor_comments":[{"comment":"Table 2 lists an entry as 'FRB Unknown' while Table 3 lists FRB 20191220C; the one-to-one correspondence between the cluster-association table and the FRB property table should be made explicit.","section":"Tables 2 and 3"},{"comment":"The label 'DESI' in Figure 14 appears to be a typo for 'DECaLS,' since the text and catalog refer to DECaLS; please correct the label.","section":"Figure 14"},{"comment":"For the highest-purity subsample (P(z<z_clust|DM)<0.01) the best-fit concentration is c=0.8±1.2, which is still below the expected c=2.6; the sentence claiming agreement with the 'characteristic expectation' should be qualified to reflect the large uncertainty.","section":"Section 3.2"},{"comment":"Step 2 resamples each FRB's declination from a Normal distribution with standard deviation equal to the baseband localization uncertainty, but the text in Section 2 describes localization to approximately 1 arcminute; the exact uncertainty model and its validation should be stated.","section":"Appendix B, step 2"}],"recommendation":"major_revision","confidential_remarks":"The central statistical claim rests on a null simulation that is inconsistent with the paper's own definition of interlopers: background FRBs are excluded from the null Monte Carlo, which biases the expected interloper count and the resulting p-value. This is a load-bearing issue, but it is fixable by including background FRBs in the null and recomputing the significance. If the revised null no longer shows a 3σ excess, the paper's main conclusion would need to be substantially softened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is the first measurement claiming a 3-sigma excess of CHIME/FRB bursts toward DECaLS clusters, and the analysis is unusually transparent. But I think the headline significance is overstated, and the stress-test note identifies the reason. In Appendix B step 5, the Monte Carlo null rejects every simulated FRB with redshift beyond a cluster within 3 r500. That removes the very population—unmagnified background FRBs—that Section 3.4 defines as interlopers. Under the no-cluster-effect hypothesis, those background alignments are part of the null. Dropping them from the mock catalogs biases the expected interloper count low (13.4±3.7), so 26 looks like a 3-sigma excess. The same foreground-only conditioning enters the P_cc Monte Carlo and the Anderson-Darling test, so both the sample selection and the significance are affected. The gamma/nSFR grids don't address this, and the argument that running the same search on real and mock data cancels modeling error doesn't cancel a missing physical population.\n\nCredit where due: the paper is honest about the DM-selection bias in the cluster sample and avoids using the same DM to infer cluster DM profiles; the DM-ySZ temperature estimate and the case studies (Coma, Abell 2218, repeater) are useful and carefully discussed. The forecasting framework from Sammons et al. 2025 is applied sensibly, and the population-model sensitivity tests are a good-faith attempt. The citation pattern is appropriate.\n\nSecondary concern: P(z|DM) is derived from the same z-DM model that generates the null, and the model is not validated against the second catalog's DM distribution. That's a real circularity, but less damning than the background-rejection issue.\n\nWho is this for? FRB population theorists and CHIME users. It deserves a serious referee, but in its current form the central claim is not established. I'd send it out with a request for a revised interloper simulation that includes background FRBs.","headline":"A careful but flawed first measurement: the claimed 3-sigma cluster excess probably shrinks once you include the background FRBs the null simulation throws away.","tokens_in":35951,"tokens_out":4422,"would_cite":false,"duration_ms":42116,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Fast radio bursts arrive 3-sigma too often behind galaxy clusters","keywords":["fast radio bursts","galaxy clusters","gravitational lensing","dispersion measure","intracluster medium","CHIME/FRB","FRB population","rate enhancement"],"falsifier":"Locate host galaxies for all 26 candidate cluster FRBs with sub-arcsecond precision and measure their redshifts: if essentially every host lies foreground to its aligned cluster (z_host < z_cluster) and no excess dispersion remains in the cluster sample, the claimed rate enhancement would be ruled out. A second, independent check is to repeat the same Monte Carlo association test on a different arcminute-localized FRB catalog; if no low-Pcc excess appears there, the enhancement does not generalize.","tokens_in":34718,"feed_emoji":"🌌","tokens_out":6162,"duration_ms":52007,"temperature":0.7,"pith_summary":"Galaxy clusters sit in front of or around a small fraction of fast radio burst (FRB) lines of sight, and the paper argues that these special sight lines carry information that average-sky surveys wash out. It builds a statistical association score that combines how closely a burst lines up with known clusters and how likely its dispersion measure places it behind the cluster, then runs Monte Carlo simulations of a cluster-blind sky. Applying this to the arcminute-localized subset of the second CHIME/FRB baseband catalog yields 26 cluster-associated bursts where about 13 to 14 are expected by chance, a 3-sigma excess corresponding to 1.4% ± 0.4% of the catalog. The authors propose that the extra bursts come about equally from FRBs emitted by cluster member galaxies and from gravitational lensing magnification of background sources, with the member-galaxy channel sensitive to non-star-forming progenitor channels and the lensing channel sensitive to high-redshift evolution. If right, cluster-aligned FRBs become a way to constrain parts of the FRB population that are otherwise invisible to current surveys.","feed_headline":"Fast radio bursts arrive 3-sigma too often behind galaxy clusters","feed_subtitle":"26 cluster-aligned bursts make up 1.4% of CHIME/FRB detections, pointing to member galaxies and lensed background sources.","key_machinery":"The load-bearing object is a per-burst association score L = Σs(RA,Dec)/P(z < z_clust|DM). Σs is an all-sky map of projected cluster density built from Navarro-Frenk-White profiles with concentration c = 6, truncated and normalized by each cluster's M500 mass; the denominator uses a redshift-dispersion population model to estimate how likely a burst with a given extragalactic dispersion measure lies foreground to the most distant aligned cluster. The score is large for high-dispersion bursts closely aligned with nearby clusters. Converting L into a probability of chance coincidence Pcc through Monte Carlo simulations of cluster-blind skies, and then comparing the 26 observed low-Pcc bursts against the interloper count from mock catalogs, is what turns a set of alignments into a rate excess.","core_discovery":"The paper's central claim is that the rate of FRBs is enhanced toward massive galaxy clusters at the 3-sigma level. Using a cluster catalog from DECaLS and a score L = Σs(RA,Dec)/P(z < z_clust|DM), where Σs is a projected Navarro-Frenk-White density map of cluster matter and P(z < z_clust|DM) is the model-dependent probability that a burst lies foreground to its most distant aligned cluster, the authors identify 26 FRBs with chance-coincidence probability below 1/892. Monte Carlo realizations of a cluster-blind sky predict 13.4 ± 3.7 such associations for the fiducial energy-function slope γ = −0.5 (14.1 ± 3.7 for γ = 0, and 11.6 ± 3.4 for γ = −1.0), so the probability that all 26 are interlopers is p = 0.0008, 0.0016, and ≤0.0001, respectively. They conclude that clusters cause 1.4% ± 0.4% of detections in the second CHIME/FRB baseband catalog, split roughly evenly between member-galaxy emission and lensing magnification, and they flag FRB 20211113A behind the strong lens Abell 2218 as a candidate lensed burst.","pith_inferences":["We infer that the same score could be applied to future arcsecond-localized FRB samples to test the member-versus-lensing decomposition directly: member bursts should sit at the cluster redshift with small excess DM, while lensed bursts should be high-DM, background, and magnified.","The claimed 1.4% excess predicts a specific cumulative signal: in a larger catalog of several thousand arcminute-localized bursts, the number of cluster-associated FRBs should grow faster than the interloper expectation, and the excess fraction should stay near 1%; a null result at that scale would falsify the enhancement.","The DM–y_sz temperature measurement, if extended to individual well-localized cluster FRBs, offers a per-sight-line ICM gas temperature probe independent of X-ray or SZ data alone, once host redshifts are available.","The authors' own simulation implies that an FRB aligned within r500 of a M500 ≥ 10^14 M_sun cluster is about equally likely to be a member burst or a background interloper, a prediction that differs from earlier claims and is testable with higher-mass cluster samples."],"forward_implications":["If the excess is real, cluster-aligned FRBs constrain the fraction of FRBs from non-star-forming progenitor channels, because cluster member galaxies have suppressed star formation while still producing some bursts.","The lensing component makes cluster sight lines sensitive to high-redshift evolution of the FRB energy function; a steeper high-z slope (γ′ ≈ −2.5) nearly reconciles simulations with the observed 26 associations.","A repeating FRB (FRB 20200929C / FRB 20201125B) aligned with a foreground cluster could, with long-term monitoring, probe intracluster-medium variations and, if strongly lensed, provide time-delay cosmology.","Even unlocalized bursts carry statistical information about cluster dispersion and ICM temperature: the paper measures a mean ICM temperature of 3.25 ± 1.44 × 10^7 K from the DM–y_sz relation in the highest-purity subsample.","The cluster DM contribution, though small per burst, is detectable at the population level and should be accounted for in precision cosmological uses of FRB dispersion measures."],"supporting_citations":[{"why":"Supplies the ~1.6 million cluster positions, masses, and redshifts used to build the association map and define the sample footprint.","marker":"Z. L. Wen & J. L. Han (2024)"},{"why":"Provides the z-DM population-model machinery that produces P(z|DM) for the foreground-probability denominator in the score.","marker":"C. W. James et al. (2022a)"},{"why":"Sets the baseline FRB population parameters (energy function, host DM, rate) used in the mock catalogs and interloper simulations.","marker":"C. W. James et al. (2022b)"},{"why":"Provides the forecasting framework for lensing and DM contributions from massive foreground clusters, used to model the rate enhancement.","marker":"M. W. Sammons et al. (2025)"},{"why":"Provides independent CHIME/FRB population constraints and the alternative γ and n_SFR values tested in the analysis.","marker":"K. Shin et al. (2023)"},{"why":"Gives the earlier expectation that cluster-aligned FRBs are likely member-galaxy sources, which the paper's simulations compare against.","marker":"L. Connor et al. (2023)"},{"why":"Releases the second CHIME/FRB catalog with the position, DM, and morphology data used for the cluster association search.","marker":"CHIME/FRB Collaboration et al. (2026)"},{"why":"Describes the baseband localization pipeline that gives the ~arcminute positions needed to align bursts with clusters.","marker":"D. Michilli et al. (2021)"},{"why":"Constrains CHIME/FRB detection completeness and the 5 Jy ms fluence threshold used in the rate simulations.","marker":"M. Merryfield et al. (2023)"}],"fun_headline_variants":["3-sigma excess of fast radio bursts toward galaxy clusters","Cluster lensing and galaxies explain 3-sigma FRB excess","Fast radio bursts cluster around galaxy clusters","Galaxy clusters boost fast radio burst rates 3-sigma","FRBs behind clusters: 26 bursts, 1.4% extra"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole analysis leans on the assumed redshift-versus-dispersion relation of the FRB population: if the model's P(z|DM) is biased for the actual second CHIME/FRB baseband catalog in ways not covered by the tested energy-function and star-formation slopes, the low chance-coincidence scores and the 3-sigma excess could be an artifact of that bias rather than a real cluster effect.","fun_headline_variants_meta":{"raw":{"variants":["3-sigma excess of fast radio bursts toward galaxy clusters","Cluster lensing and galaxies explain 3-sigma FRB excess","Fast radio bursts cluster around galaxy clusters","Galaxy clusters boost fast radio burst rates 3-sigma","FRBs behind clusters: 26 bursts, 1.4% extra"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000852,"raw_usage":{"total_tokens":3792,"prompt_tokens":1119,"completion_tokens":2673,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":735,"completion_tokens_details":{"reasoning_tokens":2589}},"tokens_in":735,"tokens_out":2673,"duration_ms":19775,"temperature":1.0,"reasoning_tokens":2589,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:21:59.046822+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Locate host galaxies for all 26 candidate cluster FRBs with sub-arcsecond precision and measure their redshifts: if essentially every host lies foreground to its aligned cluster (z_host < z_cluster) and no excess dispersion remains in the cluster sample, the claimed rate enhancement would be ruled out. A second, independent check is to repeat the same Monte Carlo association test on a different arcminute-localized FRB catalog; if no low-Pcc excess appears there, the enhancement does not generalize.","supporting_citations":[],"review_version":1}