{"id":"a2628c30-56a3-4951-b929-9850d1691bd3","arxiv_id":"2608.11296","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"The FRB DM-galaxy cross-correlation is detected at 6.5 sigma and disfavors a no-feedback baryon distribution at about 9 sigma.","lead":"Using 130 localized fast radio bursts and 27 million DESI galaxies, this paper measures a 6.5-sigma correlation between FRB dispersion measure and galaxy density, and uses it to infer the clustering of cosmic electrons. It is the strongest such detection to date and includes a first joint comparison with the kinematic Sunyaev-Zeldovich effect.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Jackknife covariance from 150 patches around 130 shared FRBs is the load-bearing weak point; the paper's own Npatch test shows ±1σ shifts in the headline SNR.","rationale":"The reader's CONDITIONAL verdict identifies exactly the same weak assumption, and I agree with it. The central measurement is plausible and has useful independent support: the real-space and harmonic-space measurements agree, the comparison with Wang et al. (2025) is consistent at 2.2σ, the fitted FLAMINGO large-scale amplitude is close to unity, and the FRB-kSZ comparison is consistent within the quoted uncertainties. None of these cross-checks, however, calibrates the error on the headline significance: the reported SNR is a direct function of the jackknife covariance. Since the paper's own Npatch test yields ±1σ shifts in the fiducial SNR, the load-bearing concern is not whether a signal exists but whether the reported significance levels are stable. A bootstrap or mock-based covariance is the direct arbiter of that question. I did not find an independent internal inconsistency in the halo-model or kSZ formalism that would change the verdict; the kSZ joint constraint is explicitly framed as a first demonstration. The verdict therefore remains CONDITIONAL: the paper should be accepted only after the significance robustness is demonstrated with a covariance estimator that is less sensitive to the jackknife patch choice.","tokens_in":36359,"tokens_out":7139,"duration_ms":76126,"concrete_test":"Bootstrap-resample the 130 FRBs with replacement, recompute the 15-bin real-space correlation and the full jackknife covariance for the log10 M*>10 bin with Npatch=150, and re-derive the SNR in Eq. 8. Repeat this over 1000 resamples. If the median bootstrap SNR falls below 5.5, or if the bootstrap covariance exceeds the patch-jackknife covariance by more than 30% on the diagonal, the reported 6.5σ is inflated by the covariance estimator and the ~9σ NFW-disagreement should be re-quoted with the bootstrap or a mock-based covariance.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline significances (6.5σ detection; ~9σ against a no-feedback NFW profile) are computed with a jackknife covariance built from Npatch=150 sky patches for only 130 FRBs, with the same FRBs reused in every radial bin (Section 4.1). The paper itself reports that the fiducial SNR for the log10 M*>10 bin moves by approximately ±1σ when Npatch is changed by a factor of two, and the harmonic-space SNR uses an analytic covariance that explicitly neglects cross-redshift-slice covariance (Section 4.2). That patch-count sensitivity means the covariance estimate is not in the asymptotic regime; the Hartlap correction in Eq. 5 only corrects the inversion bias of a Wishart-distributed covariance and cannot fix an estimator that does not yet have stable variance. If the jackknife underestimates the variance — for example, because patches containing a single FRB dominate the residuals or because FRB shot noise is strongly non-Gaussian — both quoted significances shrink. This is the load-bearing assumption: the central claim is a significance statement, and the covariance estimator sets that significance.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper measures the cross-correlation between FRB dispersion-measure fluctuations (δDM) and the DESI Legacy Survey Bright Galaxy Sample using 130 localized FRBs, including 11 new DSA-110 sources. It presents the measurement in configuration space and in harmonic space, reports a 6.5σ detection, compares the signal to FLAMINGO simulations and to a baryonification-style halo model, and interprets the results as disfavoring a no-feedback NFW scenario at roughly 9σ. It also converts the DM signal to optical depth, compares it to recent kSZ measurements of BGS galaxies, and makes a first attempt at joint FRB-kSZ constraints on the kSZ optical-depth degeneracy, expressed as constraints on r or fσ8 and on the velocity bias factor b_v.","tokens_in":36630,"tokens_out":3824,"duration_ms":64374,"significance":"If the reported significance survives scrutiny, this is a landmark measurement: it would be the strongest detection of the DM-galaxy cross-correlation with localized FRBs, the first mass-binned measurement of this statistic, and a genuinely new handle on the clustering of cosmic electrons on scales from sub-Mpc to tens of Mpc. The joint FRB-kSZ analysis is a novel and promising proof of concept for breaking the optical-depth degeneracy. The paper is also notable for adding new DSA-110 FRB redshifts and for making the FLAMINGO comparison in a way that keeps the two-halo amplitude close to unity. The main caveat is that the headline significance and the NFW-disagreement both rest on a jackknife covariance whose own convergence test shows roughly ±1σ shifts, and the harmonic-space cross-check explicitly neglects cross-redshift-slice covariance. These are fixable in revision, and the underlying data are valuable.","major_comments":[{"comment":"The jackknife covariance is the load-bearing element for the headline 6.5σ and ~9σ claims, and the paper itself reports that the fiducial SNR for the log10 M*>10 bin changes by approximately ±1σ when Npatch is changed by a factor of two. With 130 FRBs shared across all radial bins and 150 patches, the covariance estimator is not in the asymptotic regime, and the Hartlap correction in Eq. (5) only corrects inversion bias of a Wishart-distributed covariance; it cannot fix an estimator whose variance is still changing with patch count. Please provide a convergence test, a bootstrap over FRBs (or FRB jackknife), and/or an analytic covariance estimate with a shot-noise model, and propagate the resulting covariance uncertainty into the quoted SNR and the NFW-disagreement significance.","section":"Section 4.1, Eq. (5) and the Npatch discussion"},{"comment":"The harmonic-space SNR is computed with an analytic covariance that neglects covariance between the five redshift slices, justified by the assumptions that FRB noise dominates and that galaxy slices are uncorrelated. Since the same FRBs are reused in every slice and photometric redshift errors can scatter galaxies between slices, the neglected off-diagonal term should be quantified rather than assumed small. Please estimate this term (e.g., with mocks or a hybrid jackknife) or state clearly that the harmonic-space detection significance is an approximation; otherwise the harmonic-space result cannot serve as an independent validation of the configuration-space claim.","section":"Section 4.2, Eq. (11) and the following paragraph"},{"comment":"Two ad hoc cuts are applied after inspecting the data: the exclusion of FRB 20240304B because of its high redshift, and the δDM>300 pc/cm3 cut applied to seven FRBs in the southern portion of DESI. The paper asserts that these cuts are not expected to bias the measurement significantly, but for the redshift cut no quantitative test is shown, and for the southern cut only an amplitude/noise statement is given. Because the second cut is applied directly on δDM, it can in principle correlate with foreground structure and affect the cross-correlation. Please add a robustness table showing the fiducial SNR, best-fit parameters, and NFW-disagreement significance with and without each cut, and consider presenting a blinded version of the analysis.","section":"Section 3.1 and Section 6.4"},{"comment":"The ~9σ disfavoring of the no-feedback NFW scenario and the matter power spectrum suppression curve in Figure 7 are both outputs of the same three-parameter BCEmu halo model fitted to the same data vector used for the detection. Figure 7's hatched curve is therefore a fit-derived extrapolation through the baryonification framework, not an independent measurement, and the text should label it as such. Please add a posterior predictive check or a scale-resolved goodness-of-fit diagnostic to show that the halo model actually describes the data, and state explicitly that the quoted NFW-disagreement significance inherits the covariance uncertainty discussed in the first major comment.","section":"Section 5.2 and Figure 7"}],"minor_comments":[{"comment":"The abstract claims a 6.5σ detection 'in both configuration and harmonic space,' but Table 1 lists SNRCℓ = 6.2 for the harmonic-space measurement; please harmonize the wording or the numbers.","section":"Abstract and Table 1"},{"comment":"The text contains the typo 'armcimute-scale'; it should read 'arcminute-scale'.","section":"Section 1"},{"comment":"Equation (12) includes an explicit (1+z_g) weight in the FRB-optical-depth estimator, while the conversion in Eq. (C18) already contains a (1+z) factor; please clarify whether this double counting is intended and how the random-pair term is weighted.","section":"Section 4.3, Eq. (12)"},{"comment":"The caption should define the hatched region and state whether it includes parameter uncertainty; as presented, the hatched curve has no error band despite being derived from a posterior.","section":"Figure 7"},{"comment":"The statement that the 100 cMpc foreground-background separation requirement 'eliminates approximately half of the possible FRB galaxy pairs' should specify which mass bin and which redshift cuts it refers to, since the fraction will vary with the galaxy sample.","section":"Section 4.1"},{"comment":"The quantity SNR_DM is central to the NFW-disagreement claim, but the table caption does not define the exact model vector and data bins used to compute it; please spell this out in the text or caption.","section":"Table 1 footnote"}],"recommendation":"major_revision","confidential_remarks":"The paper's own Npatch sensitivity test and the two ad hoc cuts are the main obstacles to accepting the headline significance. The central detection is plausible and the new DSA-110 data are a useful contribution, but the covariance estimate needs a robustness layer before the 6.5σ and ~9σ claims can be considered established. I saw no evidence of inappropriate behavior; the authors are appropriately candid about their limitations. I would recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper does something genuinely new: it pushes the FRB DM-galaxy cross-correlation to 130 localized FRBs, adds 11 new DSA-110 hosts, measures the signal in stellar-mass bins, and does the first joint FRB-kSZ analysis. The central detection is real: configuration space and harmonic space both give roughly 6σ, and the agreement with earlier CHIME work is reassuring. The mass-binned trend, the FLAMINGO comparison, and the honest discussion of systematics are all solid. The CAP-filter estimator for FRBs is a nice bridge to kSZ.\n\nThe soft spots are real but not fatal. The headline significances rest on a jackknife covariance with Npatch=150 for 130 FRBs, and the paper itself reports the SNR for the M*>10 bin shifts by ~±1σ when Npatch changes by a factor of two. That is a genuine weakness; the Hartlap correction fixes inversion bias, not estimator variance. The harmonic-space SNR uses an analytic covariance that neglects inter-slice correlations. Having said that, the harmonic-space measurement independently gives 6.2σ, so the detection is not a one-estimator fluke. I would guess the true significance is somewhat below 6.5σ but not catastrophically. The ~9σ no-feedback disfavor is more fragile, because it depends on the fit-derived halo model and the same covariance; Figure 7 is an extrapolation through the baryonification model, not a direct measurement, and the paper is appropriately cautious about that.\n\nThe two ad hoc cuts, especially the δDM>300 pc/cm³ cut in the south, are post-hoc but checked: the paper says including those FRBs does not change the amplitude significantly, only the noise. That mitigation is reasonably convincing.\n\nBottom line: this is a serious step for FRB two-point statistics, not a breakthrough. The paper deserves full peer review, but the covariance estimation and the independence of the ~9σ claim need to be addressed before the numbers are quoted. For anyone working on FRB cosmology or baryon feedback, it is worth reading and citing.","headline":"A real detection and a promising first FRB-kSZ joint step, with a jackknife covariance that wobbles by ~1σ under patch choice—worth refereeing, but treat the headline significances as provisional.","tokens_in":37179,"tokens_out":1736,"would_cite":true,"duration_ms":26098,"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":"With 130 localized fast radio bursts and 27 million galaxies, the FRB DM-galaxy cross-correlation is detected at 6.5 sigma and disfavors gas tracing dark matter at about 9 sigma.","keywords":["fast radio bursts","dispersion measure","galaxy cross-correlation","baryonic feedback","cosmic baryons","kinetic Sunyaev-Zeldovich effect","growth rate","halo model"],"falsifier":"Split the same 130 FRBs into two independent halves and recompute xi_{g,deltaDM} and its covariance; if the two halves disagree more than the jackknife error, or if the combined significance drops below about 3 sigma, the quoted 6.5 and 9 sigma results overstate the evidence. A cleaner test is to repeat the measurement with a much larger localized sample and an analytic covariance that includes shot noise and redshift-slice correlations.","tokens_in":36186,"feed_emoji":"📡","tokens_out":8553,"duration_ms":68573,"temperature":0.7,"pith_summary":"The paper is trying to establish that the two-point statistic correlating FRB dispersion measures with foreground galaxy positions is now a working probe of cosmic baryons, not just a theoretical proposal. Using 130 localized FRBs with host redshifts and the DESI Bright Galaxy Sample, the authors report a 6.5 sigma detection in both configuration and harmonic space, and find that models in which gas follows dark matter (no feedback) are disfavored at roughly 9 sigma. This matters because the same observable can map where the universe's diffuse gas sits around galaxies, constrain feedback that biases weak-lensing cosmology, and, combined with the kinematic Sunyaev-Zeldovich effect, break the optical-depth degeneracy to reach the growth rate f sigma8. The paper thus positions FRB two-point statistics as a complementary large-scale baryon probe rather than a small-sample curiosity.","feed_headline":"130 FRBs map cosmic gas around galaxies at 6.5 sigma","feed_subtitle":"Electron clustering traced by dispersion measures rules out gas following dark matter at 9 sigma.","key_machinery":"The central object is the DM contrast deltaDM, the per-FRB dispersion-measure excess defined by subtracting Milky Way, mean cosmic, and mean host contributions, which integrates the electron overdensity delta_e along the line of sight. Cross-correlating this line-of-sight electron column with foreground galaxy positions gives the projected galaxy-electron correlation function xi_{g,deltaDM}(r); the halo model then splits this into one-halo and two-halo terms, with the electron profile described by the baryonification (BCEmu) parameters log10 M_c, theta_ej, and eta. FLAMINGO simulations supply fiducial and strong-feedback predictions, and an NFW dark-matter-tracing model provides the no-feedback null. Converting DM to optical depth via delta_tau = sigma_T (1+z) deltaDM allows direct comparison with kSZ measurements, and the CAP-filter estimator makes the comparison aperture-matched.","core_discovery":"The paper's central claim is that the FRB DM-galaxy cross-correlation, measured with 130 localized FRBs with host redshifts and 27.3 million DESI BGS galaxies, is detected at 6.5 sigma in both configuration and harmonic space, the strongest such detection to date. The measured radial signal is incompatible with a no-feedback scenario in which gas traces the underlying dark matter at roughly 9 sigma. In halo-model and FLAMINGO comparisons, the lower-mass bins prefer strong feedback, and the electron clustering on scales beyond a few Mpc is presented as among the first direct measurements of large-scale electron clustering. The paper further claims that the FRB signal agrees with a recent kSZ measurement of the same galaxies and that, combined with kSZ, it yields a first estimate of the growth rate f sigma8 (or the velocity reconstruction parameter r) and of the kSZ velocity bias factor b_v.","pith_inferences":["If the electron clustering measured here is confirmed with larger samples, the same cross-correlation could serve as a direct calibration of baryonic feedback in weak-lensing analyses, potentially reducing a leading systematic in cosmological parameter constraints.","The FRB-kSZ combination, applied to future CMB surveys and thousands of localized FRBs, could yield an independent measurement of f sigma8 that does not rely on galaxy-velocity assumptions.","A testable extension is to split the FRB sample by DM selection cuts or host-galaxy properties; if the signal depends strongly on the high-DM tail, selection effects larger than assumed would show up in the cross-correlation amplitude.","Because FRBs can access sub-arcminute scales, the technique could eventually map gas profiles in individual halos and the CGM-IGM transition, a regime kSZ cannot reach."],"forward_implications":["Electron clustering around BGS galaxies is now measured on scales from roughly 0.3 to 30 comoving Mpc, establishing FRB two-point statistics as a direct tracer of large-scale electron clustering.","The no-feedback null, in which gas traces dark matter, is disfavored at about 9 sigma, indicating that baryons are redistributed relative to dark matter around these galaxies.","FRB and kSZ signals agree within uncertainties and are complementary: FRBs constrain scales above a few Mpc while kSZ is strongest near 1 Mpc.","A first joint FRB-kSZ analysis demonstrates how the kSZ optical-depth degeneracy can be broken, yielding constraints on f sigma8 and the kSZ velocity bias factor b_v.","With future samples, the same statistic is expected to provide precision constraints on cosmic baryons and to complement weak-lensing, X-ray, and CMB probes of feedback."],"supporting_citations":[{"why":"Provides the basis for excluding 'low-DM' localizations whose host identification depended on DM, protecting the sample from selection bias.","marker":"C. Leung et al. (2025b)"},{"why":"Supplies the harmonic-space pseudo-C_l procedure and the unlocalized-FRB measurement used as a cross-check of the localized-FRB result.","marker":"H. Wang et al. (2025)"},{"why":"Earlier measurement of the DM-galaxy correlation on a similar galaxy selection, used as a consistency check in configuration space.","marker":"M. Hussaini et al. (2025)"},{"why":"The kSZ measurement around DESI BGS galaxies that the FRB optical-depth signal is compared to and combined with in the joint f sigma8 analysis.","marker":"B. Hadzhiyska et al. (2026)"},{"why":"Proposed that FRBs can break the kSZ optical-depth degeneracy and defined the velocity bias factor b_v used here.","marker":"M. S. Madhavacheril et al. (2019)"},{"why":"Defines the FLAMINGO simulation suite whose DM and subhalo lightcones generate the fiducial and strong-feedback predictions.","marker":"J. Schaye et al. (2023)"},{"why":"The baryonification emulator (BCEmu) that maps halo electron profiles with parameters log10 M_c, theta_ej, and eta and translates fits into matter power-spectrum suppression.","marker":"S. K. Giri & A. Schneider (2021)"},{"why":"Supplies the NaMaster code used to compute the masked pseudo-C_l angular cross-spectrum and its covariance.","marker":"D. Alonso et al. (2019)"},{"why":"Provides the covariance-inversion correction used when converting jackknife covariance estimates into signal-to-noise ratios.","marker":"J. Hartlap et al. (2007)"}],"fun_headline_variants":["FRB signals trace cosmic gas distribution at 6.5σ","Gas follows dark matter? FRBs say no at 9σ","First joint FRB-kSZ probe breaks baryon degeneracy","130 FRBs reveal electron clustering beyond galaxy halos","FRB-galaxy cross-correlation maps cosmic baryons at 6.5σ"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline significances rest on a leave-one-patch-out error estimate built from only 130 FRBs shared across all separation bins, so if that estimate understates the true variance, the 6.5 sigma detection and 9 sigma no-feedback rejection would both shrink.","fun_headline_variants_meta":{"raw":{"variants":["FRB signals trace cosmic gas distribution at 6.5σ","Gas follows dark matter? FRBs say no at 9σ","First joint FRB-kSZ probe breaks baryon degeneracy","130 FRBs reveal electron clustering beyond galaxy halos","FRB-galaxy cross-correlation maps cosmic baryons at 6.5σ"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000299,"raw_usage":{"total_tokens":1756,"prompt_tokens":997,"completion_tokens":759,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":613,"completion_tokens_details":{"reasoning_tokens":668}},"tokens_in":613,"tokens_out":759,"duration_ms":7238,"temperature":1.0,"reasoning_tokens":668,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:13:20.217168+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Split the same 130 FRBs into two independent halves and recompute xi_{g,deltaDM} and its covariance; if the two halves disagree more than the jackknife error, or if the combined significance drops below about 3 sigma, the quoted 6.5 and 9 sigma results overstate the evidence. A cleaner test is to repeat the measurement with a much larger localized sample and an analytic covariance that includes shot noise and redshift-slice correlations.","supporting_citations":[],"review_version":1}