{"id":"7237add5-af33-44ed-ba7b-6a770697d17e","arxiv_id":"2507.21332","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"A log-normal primordial black hole population fitted to LIGO-Virgo-KAGRA catalogs gives a local merger rate of 23.5 to 30.3 per cubic gigaparsec per year and a future stochastic background signal.","lead":"Using gravitational wave catalogs, this paper fits a model in which black holes from the early universe merge and produce observed events. It obtains a local merger rate of about 25 per cubic gigaparsec per year, and predicts a gravitational wave hum that the Einstein Telescope might detect.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Likelihood Eq 3.2 omits GWTC selection effects and event-count information, so the quoted PBH posteriors and the resulting SGWB prediction are not supported by the data.","rationale":"I read the paper as claiming two things: (i) the GWTC data, under a log-normal PBH mass function, tightly constrain the population parameters and give a local PBH merger rate in 'excellent agreement' with LVK; (ii) the same population yields a SGWB detectable by ET-B. The load-bearing step for both is the Bayesian likelihood in Eq. 3.2. If that likelihood ignores the selection function, the inferred mass function and abundance are not measurements of the PBH population but fits to the detection-biased sample. The paper explicitly acknowledges the omission, so the concern is not speculative. The local-rate prior truncation then makes the 'agreement' claim circular. I agree with the reader's weakest_assumption; this is the main reason the paper's central inference is not sound. The SGWB normalization issue noted in the reader's rationale is severe as well, but it is downstream of the same inference; the selection-function flaw is the single most load-bearing defect. A selection-corrected re-analysis is a concrete, feasible check. If such a re-analysis still yields similar posteriors (which is unlikely given the known GWTC selection bias), the reader's REJECT could be revisited; as written, UNCHANGED is appropriate.","tokens_in":12872,"tokens_out":7296,"duration_ms":84301,"concrete_test":"Re-do the inference with a proper hierarchical likelihood using the GWTC-3 posterior samples for each event, a detection efficiency P_det(m1,m2,z) computed from the O3 sensitivity (e.g., from the LVK injection campaigns or the selection function used in Ref. [10]), and a Poisson count term exp(-N_exp) in addition to the per-event terms. Remove the local-rate prior truncation. If the resulting marginal posteriors for Mc, sigma, log10 fPBH, alpha (and R0) shift by more than roughly 1 sigma or their uncertainties grow by an order of magnitude, the paper's headline parameter estimates are artifacts of the incorrect likelihood.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The decisive weakness is the likelihood in Eq. 3.2. It treats each of the N_obs GWTC events as an independent draw from the model rate R_PBH(m1,m2,z|Lambda), normalized only by a geometric phase-space integral N(Lambda) over a fixed mass-redshift box (Eq. 3.3). It omits the LVK detection probability P_det(m1,m2,z), the Poisson term exp(-N_expected) that links the total number of observed events to the model rate, and the measurement uncertainty on each event's masses and redshift (GWTC provides posterior samples, not point values). The text states this is done 'without requiring ... detailed modelling of selection effects' (end of section 3). Because the GWTC catalog is detection-biased toward high masses and low redshifts, the likelihood is not the correct model of the observed data; consequently the quoted posterior values and error bars for Mc, sigma, log10 fPBH, and alpha are not credible population constraints. The reported 'agreement' of the local rate with LVK is also circular: section 3 first restricts the prior to models whose local rate lies in 17.9-44 Gpc^-3 yr^-1, and section 5 then reports the resulting 23.5-30.3 Gpc^-3 yr^-1 posterior as independent support.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes an inference of primordial black hole (PBH) population parameters from gravitational-wave transient catalog (GWTC) data, assuming a log-normal PBH mass function and a merger rate model that accounts for gravitational torques from surrounding PBHs. The authors perform a Bayesian analysis with a likelihood built from the model merger rate, impose a prior restriction that the local PBH merger rate lie within the LVK empirical range 17.9–44 Gpc^-3 yr^-1, and report posterior estimates M_c = 21.44 ± 0.79 M_sun, sigma = 0.84 ± 0.03, log10 fPBH = -2.67 ± 0.01, alpha = 2.19 ± 0.16, with a local rate of 23.5–30.3 Gpc^-3 yr^-1. They then compute the stochastic gravitational-wave background (SGWB) from this population and find a mean total energy density of 1.67e-6 peaking near 158 Hz, with an Einstein Telescope (ET-B) SNR of about 3.24, while current detectors would not see it.","tokens_in":13132,"tokens_out":12242,"duration_ms":136842,"significance":"The topic is timely and the paper addresses a question of current interest: whether PBH mergers can contribute to LVK observations and to the SGWB. The model is explicitly formulated and the SGWB computation follows standard methods. If the inference were sound, the result would provide an interesting constraint on the PBH abundance and merger history. However, as detailed below, the likelihood is not a correct model of the GWTC data, the fPBH constraint is effectively imposed by a prior box, and the consistency with LVK rates is circular. These problems are load-bearing, so the specific numerical claims are not supported.","major_comments":[{"comment":"The likelihood in Eq. (3.2) is not a correct model of the GWTC data. It evaluates the model rate at point estimates of masses and redshift, omits the detection probability P_det(m1,m2,z), and omits the Poisson event-count term exp(-N_expected) that links the total number of observed events to the model rate. Because the GWTC catalog is strongly selection-biased toward high masses and low redshifts, these omissions bias all inferred parameters. The statement at the end of Section 3 that the analysis proceeds 'without requiring ... detailed modelling of selection effects' is precisely the problem, not a simplification. Additionally, the normalization integral in Eq. (3.3) integrates R_PBH over z without the proper comoving-volume and (1+z) weighting that converts a comoving merger rate density into a detector-frame event count. The quoted posteriors in Table 1 and all downstream results rest on this incorrect likelihood.","section":"Sec. 3, Eqs. (3.2)-(3.3)"},{"comment":"The use of the empirical local merger rate is circular. In Section 3, the parameter space is explicitly restricted to models whose local merger rate lies in the range 17.9–44 Gpc^-3 yr^-1 (the LVK empirical range), and in Section 5 the posterior local rate of 23.5–30.3 Gpc^-3 yr^-1 is presented as 'excellent agreement' with LVK. This is not an independent validation. Moreover, because the model rate in Eq. (2.6) depends on fPBH only through an overall multiplicative factor (fPBH^{53/37}), and the normalized likelihood in Eq. (3.2) divides by N(Λ), fPBH cancels exactly from the likelihood. The reported posterior log10 fPBH = -2.67 ± 0.01 is therefore entirely a consequence of the imposed local-rate prior box, not a constraint from the GWTC events.","section":"Sec. 3 prior restriction; Sec. 5 Fig. 6"},{"comment":"The caption of Figure 1 states that 'Orange vertical and horizontal lines indicate the injected parameter values.' In a Bayesian analysis of real observational data, there are no injected parameter values; such wording is only appropriate for a mock-injection study. This raises a serious question about whether the reported posteriors come from real GWTC data or from a simulation. The manuscript must clarify this point, because the central claim of the paper is the inference of PBH parameters from actual GWTC events. If the results are from an injection study, they do not constitute empirical constraints.","section":"Fig. 1 caption"},{"comment":"The predicted SGWB is never compared with the LVK O3 upper limit Omega_GW(25 Hz) <= 1.04e-9 quoted in the Introduction. Since the inferred local PBH merger rate (23.5–30.3 Gpc^-3 yr^-1) is comparable to the full observed BBH rate, the predicted background, with mean total energy density 1.67e-6, should be checked against this existing limit. As written, the spectrum appears to overshoot the limit unless the spectral shape is extremely unusual, and the absence of any such comparison leaves the consistency claim in Section 5 incomplete.","section":"Secs. 4-5"}],"minor_comments":[{"comment":"The factor 0.85 in the abundance formula is not derived or referenced; please clarify its origin (it appears to be Omega_cdm/(Omega_cdm+Omega_b)) or provide a citation.","section":"Sec. 2, Eq. (2.2)"},{"comment":"The hard restriction that the local merger rate lie in 17.9–44 Gpc^-3 yr^-1 is a prior constraint and should be listed explicitly as part of the prior, since it is the dominant constraint on fPBH.","section":"Sec. 3, Table 1"},{"comment":"The redshift-evolution power law (1+z)^alpha is introduced in Eq. (4.3), but Section 3's likelihood in Eq. (3.2) does not state how alpha enters the model rate used for inference; please specify the full rate model used in the Bayesian analysis.","section":"Sec. 4, Eq. (4.3)"},{"comment":"The text sets the overlap reduction function to 1 for 'co-located detector pairs (such as Advanced LIGO and Virgo)', but Advanced LIGO and Virgo are not co-located; this choice is inappropriate and affects the reported LIGO/Virgo SNR values.","section":"Sec. 4, Eq. (4.6)"},{"comment":"The text reports an ET-B 'SNR of O(10^1)' while giving a mean SNR of about 3.24; 3.24 is O(1), not O(10), so please correct the order-of-magnitude statement.","section":"Sec. 5"},{"comment":"The description of the posterior predictive distribution as 'the probability of parameter values theta given the observed data d' is incorrect; a posterior predictive distribution describes predicted future observations, not parameter values.","section":"Sec. 3, Fig. 2"},{"comment":"The manuscript does not specify which GWTC events were used, how many, how the point estimates of masses and redshift were obtained, or the values of mmin, mmax, and zmax in Eq. (3.3); this information is needed to reproduce the analysis.","section":"Reproducibility"}],"recommendation":"major_revision","confidential_remarks":"The central inference is compromised by the likelihood in Eq. (3.2) and by the circular use of the local-rate prior. The caption of Fig. 1 mentioning 'injected parameter values' is particularly concerning: if the results are from a mock-injection study, the paper would need to be fundamentally reframed. I recommend major revision rather than immediate rejection because the underlying model and SGWB formalism are valid in principle and could be retained after a proper reanalysis with a selection-function-aware likelihood and event posterior samples. Please also verify the provenance of Fig. 1 during the revision process."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Name],\n\nThe paper is a PBH population fit to the GWTC black-hole merger catalog plus a stochastic-background forecast. The machinery is almost entirely assembled from existing work: merger rate with third-body torques from Liu et al. 2019/2023, redshift power-law from Mukherjee & Silk, SGWB integral from Christensen. That doesn't kill it, but it means the only new product is the specific posterior and the SNR numbers.\n\nWhat it does well: the analysis is transparent, uses Bilby/dynesty, reports PPDs, and the SGWB part is a clean textbook implementation. The paper also flags the clustering caveat (low fPBH < 3e-3 justifies ignoring clustering). So it's a plausible template for this type of study.\n\nThe soft spots are severe, though. The likelihood in Eq. 3.2 treats the observed events as random draws from the model rate normalized only by the phase-space integral in Eq. 3.3. There is no detection probability, no Poisson term linking the total number of events to the model, and no measurement uncertainty on masses and redshift. The text says this is done 'without requiring ... detailed modelling of selection effects,' but that phrase does not make the likelihood correct. GWTC is heavily biased to high masses and low redshift, so all four posterior parameters are suspect. The reported uncertainties, e.g. log10 fPBH = -2.67 ± 0.01, reflect the prior and the normalization, not the data.\n\nRelated to that is the circularity. Section 3 restricts the prior to models with local merger rate in 17.9–44 Gpc^-3 yr^-1, and Section 5 calls the resulting 23.5–30.3 range 'excellent agreement.' That is agreement with the input, not with the data.\n\nThere are also smaller internal inconsistencies. The text says ET-B would reach SNR of O(10^1) but the mean SNR is 3.24, which is O(1); and it quotes LIGO/Virgo as O(10^-2) presumably. The mean total energy density 1.67e-6 should be compared carefully to the LVK upper limit they cite; if it's an integrated quantity, say so.\n\nWho is this for? A reader who wants a worked exercise in PBH inference might look at it, but no one should cite the numbers. With a selection-corrected likelihood, inclusion of event counts, and corrected SNR wording, the pipeline could become a useful baseline. I would send it to a referee with a strong request to fix those issues, but as is, the central claims are not supported.","headline":"Standard PBH machinery, but the inference skips selection effects and event counts, the local-rate agreement is circular, and the SNR statement is inconsistent—so the quoted posteriors are not credible.","tokens_in":13718,"tokens_out":3931,"would_cite":false,"duration_ms":45489,"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":"This paper argues that a log-normal mass spectrum of primordial black holes, inferred from the published binary black hole catalogs, gives a local merger rate consistent with observations and predicts a stochastic gravitational-wave…","keywords":["primordial black holes","gravitational waves","stochastic gravitational-wave background","binary black hole mergers","log-normal mass function","hierarchical Bayesian inference","merger rate","dark matter"],"falsifier":"Recompute the posterior including the published detection probability as a function of component masses and redshift. If the best-fit $\\alpha$ or $\\log_{10} f_{\\mathrm{PBH}}$ moves by more than the quoted error bars, the reported population is not robust; alternatively, a next-generation detector that saw no background above the predicted level near $158$ Hz, after subtracting resolved events, would rule out the steep-redshift-evolution scenario.","tokens_in":12640,"feed_emoji":"🕳️","tokens_out":12580,"duration_ms":121190,"temperature":0.7,"pith_summary":"The paper sets out to show that the merger history of primordial black holes (PBHs) can be inferred from current gravitational-wave observations, and that the same population leaves an imprint in the stochastic gravitational-wave background. Assuming a log-normal mass function, a Bayesian likelihood built from the predicted merger rate returns a central mass of about $21.4$ solar masses, a width of $0.84$, a PBH abundance of $\\log_{10} f_{\\mathrm{PBH}} \\approx -2.67$, and a redshift-evolution index of $2.19$. The resulting local merger rate, $23.5$–$30.3~\\mathrm{Gpc}^{-3}\\,\\mathrm{yr}^{-1}$, sits inside the range estimated from the observed binary black holes. Integrating the same rate over cosmic history with a standard waveform model yields a stochastic background with mean energy density $\\langle \\Omega_{\\mathrm{GW}}\\rangle = 1.67\\times10^{-6}$ peaking near $158$ Hz, within reach of a next-generation ground-based detector. If correct, the analysis connects a dark-matter candidate to both resolved merger statistics and an unresolved background that future instruments could measure.","feed_headline":"Primordial black holes fit merger data; background peaks at 158 Hz","feed_subtitle":"Inferred log-normal population near 21 solar masses matches observed rate and yields a detectable stochastic background.","key_machinery":"The central object is the differential merger rate density of PBH binaries, $R(t,m_i,m_j)$, built from the gravitational torque that a third PBH exerts on a newly forming pair. Together with the log-normal mass function $P(m)$ and a power-law redshift evolution $(1+z)^\\alpha$, it fixes both the event likelihood (3.2) and the stochastic background integral (4.2). The load-bearing identity is the rate formula, which scales roughly as $f_{\\mathrm{PBH}}^{53/37} m^{-21/37}$ and controls how the four population parameters map onto the observed events and the background spectrum.","core_discovery":"Under a log-normal PBH mass function, the paper's hierarchical Bayesian analysis of the published transient catalog yields $M_c = 21.44^{+0.79}_{-0.77}\\,M_\\odot$, $\\sigma = 0.84^{+0.03}_{-0.03}$, $\\log_{10} f_{\\mathrm{PBH}} = -2.67^{+0.01}_{-0.01}$, and $\\alpha = 2.19^{+0.16}_{-0.16}$. The model's local PBH binary merger rate, $23.5$–$30.3~\\mathrm{Gpc}^{-3}\\,\\mathrm{yr}^{-1}$, matches the empirically estimated range. Propagating this population to the unresolved background gives a mean total energy density of $1.67\\times10^{-6}$ with the spectrum peaking near $158$ Hz, and a predicted signal-to-noise ratio of about $3.24$ for a next-generation triangular detector with one year of observation, while current detectors fall well below threshold.","pith_inferences":["The likelihood omits detection probability; a selection-corrected re-analysis would likely widen the quoted uncertainties and could shift $\\alpha$ and $f_{\\mathrm{PBH}}$ toward different best-fit values, so the current error bars should be read as conditional on that omission.","The same rate machinery could be applied to subsolar-mass PBH mergers, where a stochastic background of overlapping events above a few hundred hertz might become a probe of lighter PBH populations.","If PBHs cluster in dark-matter halos even at $f_{\\mathrm{PBH}} \\sim 10^{-3}$, the background spectrum would develop a turnover at higher frequencies; the present paper's assumption of a Poisson spatial distribution is a testable simplification that future halo-occupation modelling could relax.","Because PBH mergers are rare and individually loud, the predicted background should be non-Gaussian and 'popcorn-like'; measuring the statistical distribution of the background could distinguish it from a smooth astrophysical component."],"forward_implications":["PBHs with $\\log_{10} f_{\\mathrm{PBH}} \\approx -2.67$ make up roughly $0.2\\%$ of the dark matter, below the threshold where PBH clustering would matter.","The steep redshift index $\\alpha \\approx 2.2$ implies most PBH mergers happen at high redshift, before star formation begins.","The predicted stochastic background peaks near $158$ Hz and is accessible to a next-generation triangular detector, while current detectors are predicted to be about two orders of magnitude short.","The local PBH merger rate of $23.5$–$30.3~\\mathrm{Gpc}^{-3}\\,\\mathrm{yr}^{-1}$ is consistent with the observed binary black hole rate, so a substantial fraction of the detected events could be primordial in origin.","The spectral peak at $158$ Hz provides a distinguishing feature that could separate PBH mergers from the smoother astrophysical background."],"supporting_citations":[{"why":"Provides the merger rate density formula with the third-body torque that sets the binary angular momentum.","marker":"[51]"},{"why":"Applies the merger-history rate model to gravitational-wave data, giving the expression the present analysis adopts.","marker":"[76]"},{"why":"Supplies the catalog of observed binary black hole mergers and the empirical local merger rate range used to restrict the parameter space.","marker":"[10]"},{"why":"Sets up the redshift-integrated formula for the stochastic gravitational-wave background energy density.","marker":"[19]"},{"why":"Provides the analytic phenomenological fit to the gravitational-wave energy spectrum emitted by each binary coalescence.","marker":"[80]"},{"why":"Determines the abundance threshold below which PBH clustering can be neglected, justifying the Poisson assumption at the inferred $f_{\\mathrm{PBH}}$.","marker":"[62]"}],"fun_headline_variants":["Stochastic background peaks at 158 Hz from PBH mergers","Primordial black hole mergers fingerprint the stochastic background","Bayesian analysis ties PBH mergers to observed GW background","Primordial black hole merger rate inferred from LIGO-Virgo data","PBH merger history written in the gravitational-wave background"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The inference treats the observed merger events as an unbiased random draw from the model's predicted rate, ignoring the selection bias of the detector network toward heavy, nearby binaries; if that selection matters, the four inferred population parameters and their error bars would shift.","fun_headline_variants_meta":{"raw":{"variants":["Stochastic background peaks at 158 Hz from PBH mergers","Primordial black hole mergers fingerprint the stochastic background","Bayesian analysis ties PBH mergers to observed GW background","Primordial black hole merger rate inferred from LIGO-Virgo data","PBH merger history written in the gravitational-wave background"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000689,"raw_usage":{"total_tokens":3168,"prompt_tokens":1036,"completion_tokens":2132,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":652,"completion_tokens_details":{"reasoning_tokens":2049}},"tokens_in":652,"tokens_out":2132,"duration_ms":17501,"temperature":1.0,"reasoning_tokens":2049,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T12:53:31.053013+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the posterior including the published detection probability as a function of component masses and redshift. If the best-fit $\\alpha$ or $\\log_{10} f_{\\mathrm{PBH}}$ moves by more than the quoted error bars, the reported population is not robust; alternatively, a next-generation detector that saw no background above the predicted level near $158$ Hz, after subtracting resolved events, would rule out the steep-redshift-evolution scenario.","supporting_citations":[{"cited_title":"Liu, Z.-Q","cited_arxiv_id":null,"evidence_quote":"Applies the merger-history rate model to gravitational-wave data, giving the expression the present analysis adopts."},{"cited_title":"Liu, Z.-K","cited_arxiv_id":null,"evidence_quote":"Provides the merger rate density formula with the third-body torque that sets the binary angular momentum."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the catalog of observed binary black hole mergers and the empirical local merger rate range used to restrict the parameter space."},{"cited_title":"Christensen,Measuring the stochastic gravitational-radiation background with laser-interferometric antennas, Physical Review D46 (1992) 5250","cited_arxiv_id":null,"evidence_quote":"Sets up the redshift-integrated formula for the stochastic gravitational-wave background energy density."},{"cited_title":"Ajith, S","cited_arxiv_id":null,"evidence_quote":"Provides the analytic phenomenological fit to the gravitational-wave energy spectrum emitted by each binary coalescence."},{"cited_title":"Hütsi, M","cited_arxiv_id":null,"evidence_quote":"Determines the abundance threshold below which PBH clustering can be neglected, justifying the Poisson assumption at the inferred $f_{\\mathrm{PBH}}$."}],"review_version":1}