{"id":"2e3ca06b-d918-40e2-a5f0-291a0ace093b","arxiv_id":"2412.15989","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Bayesian analysis of IPTA DR2 shows scalar-induced gravitational waves dominate and PBH merger backgrounds are strongly disfavored relative to an astrophysical SMBHB explanation.","lead":"The authors searched the International Pulsar Timing Array dataset for gravitational waves from merging primordial black holes, together with the gravitational-wave background induced by the same fluctuations that formed the black holes. They conclude that black hole merger signals cannot explain the detected nanohertz hum under the standard formation scenario, and that supermassive black hole binaries remain the preferred source.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Clustering factor Rcl is the load-bearing input; free-Rcl test saturates its prior, so the 'strongly favors SMBHB' claim rests on external constraints rather than PTA data alone.","rationale":"The reader's CONDITIONAL verdict is appropriate. The central claim that PBH mergers cannot explain the PTA signal is essentially a statement that the Poissonian clustering model predicts a small Rcl. The free-Rcl analysis does not settle this because (i) it multiplies the same merger-rate template and cannot correct errors in the mass/redshift dependence, and (ii) its posterior for log10 Rcl in the PBH-only model runs to the prior edge, leaving the model comparison inconclusive (log10 B = 0.05). The abstract's 'strongly favors' should therefore be read as applying to the fixed-clustering model, or to the model once external PBH constraints are folded in, not to the PTA likelihood alone. A direct N-body measurement of the late-time merger rate would test the analytic Rcl and either support or undermine the fixed-Rcl conclusion. This does not change the verdict: the paper is careful and transparent about its limitations, but the quantitative Bayes factors should be treated as order-of-magnitude estimates, exactly as the reader concluded.","tokens_in":36065,"tokens_out":28552,"duration_ms":235732,"concrete_test":"Run N-body simulations (e.g., Delos et al., arXiv:2410.01876) for a broad PBH mass function with f_PBH ≈ 10⁻⁴ and ns ≈ 0.97, and measure the late-time two-body capture merger rate. Extract the effective Rcl by dividing the simulated rate by f_PBH² ∫ ϕ(m₁)ϕ(m₂)(m₁+m₂)^{10/7}/(m₁m₂)^{5/7} dlnm₁ dlnm₂, and compare to the analytic Eq. (C15). If the simulated Rcl is more than an order of magnitude larger than the analytic value, the fixed-Rcl exclusion of PBH mergers weakens; if it agrees within a factor of a few, the conclusion is robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that PBH mergers cannot explain the IPTA DR2 signal rests on the analytic clustering factor Rcl computed in Appendix C from Poissonian clustering. This calculation is the weakest link: it combines a Press-Schechter halo mass function (Eq. 22) derived for monochromatic PBHs with extrapolations to broad mass functions (e.g., ⟨f²m⟩ in Eq. C6) and a dynamical-heating prescription (Eq. C11) with an assumed velocity distribution and NFW profiles. A specific red flag is the halo radius relation (Eq. C8), which scales as (⟨m f_PBH⟩)^{1/2} M_h^{5/6}; combining Eq. (C5) with ρ_c(z_h) ∝ (1+z_h)³ instead gives r_h ∝ (⟨m f_PBH⟩)^{−1/2} M_h^{5/6}, suggesting a possible sign error that could change Rcl by orders of magnitude for f_PBH ~ 10⁻⁴. The free-Rcl test (Sec. V B) does not resolve this because it only rescales the same template and cannot correct shape errors, and because the posterior for log10 Rcl in the PBH-only model saturates the prior bound of 10, with a Bayes factor vs. SMBHB of only 0.05. Thus the abstract's statement that IPTA DR2 'strongly favors' an astrophysical origin is either conditional on the Poissonian model or relies on external PBH constraints (LVK, CMB) rather than the PTA likelihood alone.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper performs a Bayesian search in IPTA DR2 for a stochastic gravitational-wave background sourced by primordial black holes, combining scalar-induced GWs from the curvature perturbations that form the PBHs with GWs from early-Universe PBH binaries and from late-time dynamical capture in Poisson-induced PBH halos. A broad PBH mass function is derived from a power-law primordial power spectrum with hard cutoffs, and constraints from CMB mu-distortions and fPBH<=1 are imposed. The analysis is run twice: once with the clustering factor Rcl computed from the analytic halo model, and once with Rcl as a free nuisance parameter. The authors find that scalar-induced GWs dominate the nHz band, that the PBH-merger contribution is subdominant under the Poissonian-clustering model, and that the IPTA DR2 signal favors an SMBHB interpretation over the PBH models considered. The central conclusion is stated as being conditional on standard Gaussian PBH formation and Poissonian clustering, with enhanced clustering identified as a possible loophole.","tokens_in":36449,"tokens_out":7773,"duration_ms":74812,"significance":"If the central conclusion holds, the paper is a useful contribution to the PBH interpretation of PTA data: it is one of the few analyses to include scalar-induced GWs and both early and late PBH merger channels for a broad, first-principles mass function, and it demonstrates quantitatively that mu-distortion cutoffs suppress the previously claimed late-time merger enhancement. The use of PTArcade, the explicit prior choices, the posterior predictive PBH mass functions, and the free-Rcl robustness test are strengths. The main result is, however, conditional on an analytic clustering model whose derivation contains a load-bearing scaling error (see major comment 1), and the free-Rcl analysis shows that the PTA likelihood alone does not strongly exclude PBH-only models. A careful revision of the clustering calculation and of the wording of the abstract's 'strongly favors' claim is needed.","major_comments":[{"comment":"The claimed scaling of the virial radius with the mass-averaged clustering quantity appears to have the wrong sign. Combining the spherical-collapse condition rho_h ~ 178 rho_c(z_h) with Eq. (C5), which gives (1+z_h) proportional to (langle m f_PBH rangle / M_h)^{1/2}, yields r_h proportional to (langle m f_PBH rangle)^{-1/2} M_h^{5/6}, whereas Eq. (C8) has the positive exponent +1/2. Because the clustering factor in Eq. (C15) contains v_vir^{-11/7} delta_cl^2 r_h^3 and delta_cl is proportional to M_h/r_h^3, this sign error changes R_cl by a factor of order (langle m f_PBH rangle)^{31/14}; for langle m f_PBH rangle ~ 10^{-3} M_sun, the printed relation overestimates R_cl by roughly four orders of magnitude or more. This affects the fixed-R_cl posteriors and the derived PBH merger contribution, and the derivation must be corrected and the fixed-R_cl analysis recomputed before the quantitative exclusion claim can be accepted.","section":"Appendix C, Eq. (C8)"},{"comment":"The free-Rcl analyses do not provide strong evidence against PBH-only models. The maximum posterior for log10 R_cl sits at the upper prior edge (log10 R_cl = 10) in both free-Rcl runs, and for the PBH-only case the Bayes factor is log10(B_SMBHB,PBH) = 0.05, which is inconclusive. The abstract's statement that IPTA DR2 'strongly favors' an astrophysical origin is therefore not supported by the PTA likelihood alone; it relies on external constraints (fPBH <= 1, CMB, LVK) applied in Sec. VI. Please qualify the central claim as conditional on the Poissonian-clustering model plus external PBH constraints, and state explicitly which Bayes factor supports the word 'strongly'.","section":"Sec. V B and Table I"},{"comment":"The late-time suppression factor S_late is taken from a monochromatic-mass computation and extrapolated to broad mass functions without a quantified error; the manuscript itself notes that this extension is 'non-trivial and still an open issue'. Because Eq. (18) multiplies the early-binary rate that enters all three models in Table I, the early-binary contribution carries an unquantified systematic that the free-Rcl parameter cannot absorb, since Rcl only rescales the late-binary template in Eq. (24). The authors should either demonstrate that the posteriors and Bayes factors are insensitive to this extrapolation or incorporate the extrapolation uncertainty into the model comparison.","section":"Sec. IV A, Eq. (18)"}],"minor_comments":[{"comment":"The statement 'The best fit values used for the GW spectrum shown in Fig. 5 are in violation of this L VK bound' conflicts with the use of the same values as the maximum-likelihood/maximum-posterior spectrum in the main text; please clarify whether the displayed spectrum is a posterior maximum or a best-fit that is excluded by LVK, and specify the status of the posterior sample used for Fig. 6.","section":"Appendix D 3"},{"comment":"The Bayes factors in Table I are quoted without an evidence scale; please state the Jeffreys-scale interpretation used for 'strong evidence' and note explicitly that the numerical values are specific to IPTA DR2 and to the prior choices in Table II, not to the newer NANOGrav/EPTA/PPTA datasets.","section":"Table I and Sec. V"},{"comment":"The averaged quantity in Eq. (C7), defined as f_PBH^2 times the mass-weighted integral of phi(m), is central to both Eq. (23) and Eq. (C15); please define it once in the main text (or in a single appendix location) and use a notation that distinguishes it clearly from langle f_PBH m rangle used in Eq. (C3), since the two enter different parts of the halo model.","section":"Eq. (C7) and Eq. (23)"},{"comment":"The right-hand panels of Figs. 4 and 5 show several spectral components but do not label a 'total' curve explicitly; adding the total PBH+SIGW spectrum and the SMBHB reference model in the same panel would make the model comparison easier to interpret.","section":"Figs. 4 and 5"}],"recommendation":"major_revision","confidential_remarks":"The sign-error concern in Appendix C is genuine and should be fixed before publication; the corrected direction likely strengthens the exclusion of the Poissonian late-binary channel, but the printed derivation is not self-consistent as it stands. The free-Rcl saturation and the weak Bayes factor for the PBH-only model mean the abstract's 'strongly favors' claim needs substantial qualification. The paper is within scope for a cosmology/astro-ph journal and is publishable after the clustering calculation is corrected and the conditional language is made precise."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a careful read and a serious referee, but the quantitative results are shakier than the tone suggests. There is an apparent sign error in the halo radius relation, Eq. (C8), which changes the clustering factor by orders of magnitude, and the free-Rcl analysis saturates its prior, so the abstract's 'strongly favors SMBHB' is too strong for the PBH-only model.\n\nWhat is actually new: they combine scalar-induced GWs with early and late PBH merger channels for a broad, QCD-featured mass function, and show that CMB mu-distortion constraints remove the asymmetric-binary boost predicted in Ref. [55]. That is a clean and useful negative result. The Bayesian analysis in PTArcade is careful, the external constraints are applied consistently, and the authors are honest about many merger-rate uncertainties.\n\nThe soft spots are real but mostly fixable. First, Eq. (C8): using their own Eq. (C5) and virialization gives r_h proportional to (⟨m f_PBH⟩)^{-1/2} M_h^{5/6}, not the positive power written. The sign error does not undermine the main exclusion; if anything, their Rcl is too large, so the conclusion survives, but the absolute rates and Rcl values cannot be trusted until corrected. Second, in the free-Rcl analysis with only PBH binaries, log10 Rcl sits at the prior boundary 10 and the Bayes factor versus SMBHB is 0.05, i.e., odds of about 1.1:1. That is not strong evidence; the strong preference comes from the model that also includes SIGWs. The abstract blurs this. Third, the best-fit spectrum in Fig. 5 violates the LVK bound, as the authors admit in Appendix D3; that does not affect the Bayes factors, but it is sloppy and should be fixed. Fourth, the late-time suppression factor Slate is extrapolated from a monochromatic mass distribution to broad mass functions without validation, and no code or data release is provided.\n\nWho this is for: PTA cosmology and PBH phenomenology readers, especially those trying to interpret the nHz common-spectrum process. It deserves a serious referee; the central negative claim is likely correct, but the paper needs a revision to fix the sign error, temper the abstract, and ideally release the code. I would take it to peer review.","headline":"Solid and useful negative result for PBH-PTA interpretations, but a likely sign error in the halo radius relation and a prior-saturating free parameter make the quantitative claims weaker than the abstract suggests.","tokens_in":36932,"tokens_out":6362,"would_cite":true,"duration_ms":50882,"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":"PBH mergers alone cannot explain the pulsar timing background","keywords":["primordial black holes","pulsar timing arrays","stochastic gravitational wave background","scalar-induced gravitational waves","PBH clustering","IPTA DR2","model comparison","CMB mu-distortion constraints"],"falsifier":"A computation of $R_{\\rm cl}$ from N-body simulations of PBH cluster formation with enhanced (non-Poissonian) initial clustering that yields $R_{\\rm cl}\\gtrsim 10^4$ Gpc$^{-3}$ yr$^{-1}$ while satisfying $f_{\\rm PBH}\\le 1$ and CMB $\\mu$-distortion limits would disprove the central exclusion. Equivalently, a PTA detection whose high-frequency tail scales as $\\Omega_{\\rm gw}\\propto f^{2/3}$ rather than the steeper SMBHB power law would reopen the PBH-binary channel.","tokens_in":35879,"feed_emoji":"🕳️","tokens_out":14323,"duration_ms":111008,"temperature":0.7,"pith_summary":"This paper asks whether the nanohertz gravitational-wave background seen by pulsar timing arrays could be produced by primordial black holes (PBHs), either through the gravitational waves inevitably induced by the curvature perturbations that form PBHs, or through the mergers of PBH binaries, or both. Using the International Pulsar Timing Array data release 2, the authors fit a combined model with scalar-induced gravitational waves and two PBH binary channels (early-universe binaries and binaries formed dynamically in late-time PBH clusters), for a broad PBH mass function derived from a power-law primordial curvature spectrum. They find that the merger contribution is subdominant: under the standard Gaussian-perturbation formation scenario with only Poissonian clustering, PBH mergers alone cannot reproduce the signal, and the nHz band is instead dominated by scalar-induced gravitational waves. A phenomenological boost to the clustering factor can make PBH binaries fit the data, but only at values roughly four orders of magnitude above the Poissonian expectation, and the resulting PBH abundances clash with independent constraints.","feed_headline":"PBH mergers alone cannot explain the pulsar timing background","feed_subtitle":"A Bayesian fit to IPTA DR2 favors supermassive black hole binaries over the modeled primordial black hole scenarios.","key_machinery":"The load-bearing object is the combined gravitational-wave spectrum $h^2\\Omega_{\\rm gw}(f)=h^2\\Omega_{\\rm gw}^{\\rm SIGW}(f)+h^2\\Omega_{\\rm gw}^{\\rm PBHB}(f)$, built from the curvature power spectrum $P_\\zeta(k)=A_\\zeta (k/k_\\star)^{n_s-1}$ with cut-offs $k_{\\min}, k_{\\max}$. The SIGW term is the second-order scalar-induced spectrum of Eq. (11); the PBHB term sums the early-universe binary merger rate of Eq. (14) and the late-time halo merger rate of Eq. (24), the latter controlled by the clustering factor $R_{\\rm cl}(z)$, a dimensionful factor in Gpc$^{-3}$ yr$^{-1}$ that packages the Poisson-induced halo mass function, cuspy density profiles, and dynamical heating. The argument turns on $R_{\\rm cl}$: with fixed parameters it stays at values $R_{\\rm cl}\\sim 1$--$10^2$ Gpc$^{-3}$ yr$^{-1}$ (or up to $\\sim 10^6$ for $n_s>1$, where average masses are tiny), far below the $\\gtrsim 10^4$ Gpc$^{-3}$ yr$^{-1}$ needed for late PBH binaries to dominate the signal.","core_discovery":"The central claim is that the common-spectrum process in IPTA DR2 does not select a PBH-merger origin. When the curvature power spectrum is constrained by CMB $\\mu$-distortion limits and the $f_{\\rm PBH}\\le 1$ overproduction limit, the large-scale cutoff removes PBHs heavier than roughly $10^3$ solar masses, which are precisely the asymmetric binaries that earlier work had identified as the dominant merger contribution. The fixed-clustering analysis then yields a spectrum dominated by scalar-induced GWs, with PBH binaries contributing at a much lower amplitude, and a log evidence ratio $\\log_{10} B \\simeq 1.9$ in favor of the astrophysical supermassive-black-hole-binary model. In the free-clustering analysis, the clustering factor $R_{\\rm cl}$ is driven to the prior boundary ($\\log_{10} R_{\\rm cl}=10$), and the PBH-only fit produces a PBH mass function that is excluded by microlensing, CMB, and ground-based interferometer merger-rate constraints. The authors conclude that, within the modeled scenarios, the IPTA DR2 signal strongly favors an astrophysical origin.","pith_inferences":["Editorial inference: because the inferred value of $R_{\\rm cl}$ lands at the prior boundary, IPTA DR2 data alone do not fix the merger-rate enhancement; they only set a lower limit, and a reanalysis with a wider prior or independent cluster constraints would sharpen the bound.","Editorial inference: the exclusion is limited to Gaussian-perturbation formation with Poissonian clustering; rerunning the same pipeline with non-Gaussian curvature perturbations that enhance PBH clustering at formation, a caveat the authors name, is the natural next test.","Editorial inference: applied to the upcoming combined IPTA DR3 dataset, the same template should either tighten the supermassive-black-hole-binary interpretation or reveal a residual component; the $\\Omega_{\\rm gw}\\propto f^{2/3}$ high-frequency scaling is a distinctive PBH-binary signature that separates the two.","Editorial inference: the $\\mu$-distortion bound at $k\\lesssim 10^5$ Mpc$^{-1}$ is the main lever suppressing heavy PBHs, so a future spectral-distortion measurement at these scales would decide whether the late-time merger channel can ever be competitive."],"forward_implications":["If the paper is right, the nHz common-spectrum process in IPTA DR2 is not a signature of PBH mergers under Gaussian primordial perturbations, so explaining it cosmologically requires either non-Gaussian clustering or a different PBH formation channel.","The scalar-induced GW component remains a viable PBH-associated signal and constrains the primordial power-spectrum parameters $(A_\\zeta, n_s, k_{\\min}, k_{\\max})$ in the window where $f_{\\rm PBH}\\le 1$ and CMB $\\mu$-distortion limits are respected.","Models that force the signal to come only from late-time PBH binaries need $R_{\\rm cl}\\gtrsim 10^4$ Gpc$^{-3}$ yr$^{-1}$, which conflicts with Poissonian clustering and yields PBH abundances excluded by microlensing, CMB anisotropies, and ground-based interferometer merger-rate measurements.","The reported preference for an astrophysical origin is quantified by a log evidence ratio $\\log_{10} B_{\\rm SMBHB,PBH} \\simeq 1.9$ for the models including SIGWs, and $\\simeq 0.05$ for a PBH-merger-only model, favoring supermassive black hole binaries.","The analysis provides a reusable merger-rate template for broad PBH mass functions that can be applied to other gravitational-wave searches."],"supporting_citations":[{"why":"Earlier claim that broad-mass late-time PBH binaries produce a PTA-detectable GWB; the main baseline this paper revises.","marker":"[55]"},{"why":"CMB mu-distortion constraints that impose the small-k cutoff and suppress PBHs above roughly 10^3 solar masses.","marker":"[56]"},{"why":"IPTA DR2 pulsar timing dataset used for the search.","marker":"[57]"},{"why":"IPTA DR2 common-spectrum analysis prescription, including noise modeling and the 13 frequency bins.","marker":"[33]"},{"why":"Earlier SIGW search in IPTA DR2 that motivates the window-function and abundance treatment.","marker":"[48]"},{"why":"Semi-analytic scalar-induced GW spectrum used to compute the SIGW contribution.","marker":"[26]"},{"why":"Early-universe PBH binary formation and merger-rate formalism.","marker":"[89]"},{"why":"Clustering-factor framework for late-time PBH binaries including dynamical heating.","marker":"[54]"},{"why":"Halo mass function from Poisson fluctuations used in the late merger rate.","marker":"[93]"},{"why":"Halo merger-rate calculation with cuspy density profiles and velocity distribution.","marker":"[102]"}],"fun_headline_variants":["PBH mergers fail to explain pulsar timing background","PTA data favor astrophysical origin over PBH mergers","Primordial black holes cannot account for PTA signal","IPTA DR2 rejects PBH mergers for common spectrum signal","Pulsar timing background likely astrophysical, not primordial black holes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that late-time PBH halos form only from Poisson fluctuations in the PBH number density, with the paper's specific halo mass function from the standard collapse formalism, cuspy density profiles, and dynamical-heating prescription; if real clustering is stronger, the PBH merger background grows and the central exclusion weakens.","fun_headline_variants_meta":{"raw":{"variants":["PBH mergers fail to explain pulsar timing background","PTA data favor astrophysical origin over PBH mergers","Primordial black holes cannot account for PTA signal","IPTA DR2 rejects PBH mergers for common spectrum signal","Pulsar timing background likely astrophysical, not primordial black holes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000878,"raw_usage":{"total_tokens":3819,"prompt_tokens":988,"completion_tokens":2831,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":604,"completion_tokens_details":{"reasoning_tokens":2749}},"tokens_in":604,"tokens_out":2831,"duration_ms":19667,"temperature":1.0,"reasoning_tokens":2749,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T10:53:40.269708+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A computation of $R_{\\rm cl}$ from N-body simulations of PBH cluster formation with enhanced (non-Poissonian) initial clustering that yields $R_{\\rm cl}\\gtrsim 10^4$ Gpc$^{-3}$ yr$^{-1}$ while satisfying $f_{\\rm PBH}\\le 1$ and CMB $\\mu$-distortion limits would disprove the central exclusion. Equivalently, a PTA detection whose high-frequency tail scales as $\\Omega_{\\rm gw}\\propto f^{2/3}$ rather than the steeper SMBHB power law would reopen the PBH-binary channel.","supporting_citations":[],"review_version":1}