{"id":"e48f43a6-d38d-4f0e-aa38-cfdee161c8f4","arxiv_id":"2412.09956","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The angular power spectra of the anisotropic gravitational wave background from local compact binary mergers show a Poisson-like slope and are dominated by the radial distribution of sources within 50 Mpc/h.","lead":"This paper uses computer simulations of galaxy formation and binary stars to predict how the gravitational wave background from merging black holes and neutron stars varies across the sky. It finds that the variation is driven by very nearby sources, and that subtracting those nearby sources makes predictions from different simulation versions much more consistent.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The slope-2 scaling and cross-type factor-of-2 alignment may be dominated by Poisson shot noise, so the residual spectra do not yet demonstrate LSS tracing.","rationale":"The reader identifies the realism of the mock-lightcone construction (interpolation, GABE-COSMIC rates, periodic replication) as the weakest assumption. That concern is legitimate, but the shot-noise issue is more fundamental because it affects even a perfect mock: the headline slope-2 scaling is the expected signature of a Poisson-weighted point-source catalogue, and the paper itself notes that the high-l shape resembles Poisson noise from Cusin et al. (2019). The central claim that remains load-bearing is the interpretation of the residual spectra, after subtracting <50 Mpc/h sources, as tracing large-scale structure and providing a theoretical upper limit. Without an explicit shot-noise decomposition, the observed slope, the factor-of-2 alignment across source types, and the reduced scatter across observer positions could all be produced by the common discreteness of the host-galaxy catalogue rather than by the clustering of the underlying LSS. I therefore recommend keeping the reader's CONDITIONAL verdict: the analysis is internally coherent and the qualitative anisotropy signal is plausible, but the paper should add a shot-noise floor calculation and a residual-clustering comparison before the quantitative claims are used as benchmarks. The proposed test directly settles which interpretation is correct.","tokens_in":17345,"tokens_out":11496,"duration_ms":146974,"concrete_test":"For each of the seven catalogues and each source type, compute the analytic shot-noise spectrum C_l^shot = 4π (Σ_i w_i^2)/(Σ_i w_i)^2 using the weights w_i = R_k(zk,m1,m2) (dE/dfr) / [(1+z_k) d_c,k^2] from Eq. (6), both for the full d_c<250 Mpc/h catalogue and for the residual catalogue with d_c>50 Mpc/h, applying the same HEALPix pixel window as the measured C_l. If C_l^measured is consistent with C_l^shot at the 10% level for l≳30, the slope-2 scaling and cross-type alignment are shot-noise dominated. If C_l^measured − C_l^shot remains positive, l-dependent, and similar across BNS/NSBH/BBH and across r1–r7, the LSS-tracing interpretation survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing gap is that the paper never separates the measured angular power spectrum into a clustered LSS term and a Poisson shot-noise term. For a weighted catalogue of discrete sources, with w_i the contribution of galaxy i in Eq. (6), the shot-noise part of the overdensity field is C_l^shot = 4π Σ_i w_i^2 / (Σ_i w_i)^2, independent of l; hence log10[l(l+1)C_l/(2π)] rises as β≈2 in log10 l. This is exactly the 'characteristic slope' quoted in the abstract and Sec. 5 for l≳30–300. The authors acknowledge in Sec. 4.2 that the 1-bin approximation retains the typical shape and that this shape matches the Poisson-noise predictions of Cusin et al. (2019), but they do not compute the shot-noise floor or subtract it. Since BNS, NSBH and BBH sources share the same host galaxies, their weights are strongly correlated, so the factor-of-2 agreement after removing <50 Mpc/h sources, and the reduction of scatter across r1–r7, could be a common shot-noise floor rather than evidence that the residual anisotropy traces LSS. If shot noise dominates over most of the quoted l range, the 'theoretical upper limit' and the LSS-probe interpretation are not established.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs seven all-sky mock lightcone gravitational-wave event catalogues from the Millennium simulation combined with the GABE semi-analytic galaxy formation model and the COSMIC binary population synthesis code, restricted to comoving distances below 250 Mpc/h (z <~ 0.085). For each catalogue, the authors compute HEALPix maps of the SGWB overdensity delta_GW from merging BBHs, NSBHs, and BNSs, and then evaluate angular power spectra C_l expressed as log10[ell(ell+1)C_l/(2pi)]. The central claims are that (i) at high ell (approximately ell > 30-300) all spectra rise with a characteristic slope beta ~ 2 in log10 ell; (ii) the large scatter among the seven catalogues is mainly caused by the radial distribution of nearby sources within ~50 Mpc/h; (iii) after subtracting those nearby sources, the spectra for the three source types become aligned to within a factor of about 2 over ell = 1-1000; and (iv) the residual anisotropy can therefore serve as a probe of large-scale structure, while the local-universe catalogues set a theoretical upper limit on the SGWB anisotropy from compact binary mergers.","tokens_in":17549,"tokens_out":4591,"duration_ms":52239,"significance":"If the conclusions are correct, the paper would strengthen the case that the anisotropic SGWB from compact binary coalescences can be used as an astrophysical probe of the cosmic web, and it would provide a concrete theoretical target for future searches. The construction of lightcone mock catalogues from a state-of-the-art galaxy formation plus binary population synthesis pipeline is a useful contribution, and the paper is honest about several limitations, such as the restricted redshift range and the reliance on interpolation between snapshots. The numerical pipeline (HEALPix sky maps, anafast angular power spectra, seven observer positions) is standard and appears internally consistent. However, the main interpretation hinges on separating the clustered, large-scale-structure-induced anisotropy from the Poisson shot-noise contribution. The paper does not compute or subtract the shot-noise floor, and the qualitative slope claim is not accompanied by any fit with uncertainties, which limits the strength of the headline conclusions.","major_comments":[{"comment":"The shot-noise floor is never computed or subtracted. For the weighted discrete-source overdensity field implicit in Eq. (6) and Eq. (8), shot noise contributes an approximately ell-independent term to C_l, so the plotted quantity log10[ell(ell+1)C_l/(2pi)] rises with slope beta ~ 2 in log10 ell. This is exactly the 'characteristic slope' quoted in the abstract and Section 5. The text already notes in Section 4.2 that even the 1-bin approximation retains the typical shape and that this shape matches the Poisson-noise predictions of Cusin et al. (2019), but it never evaluates the shot-noise term or subtracts it. Because BNS, NSBH, and BBH events share the same host galaxies, their weights are strongly correlated; therefore the factor-of-2 alignment after removing sources below 50 Mpc/h, and the reduced scatter across the r1-r7 catalogues, could be a common shot-noise floor rather than evidence that the residual anisotropy traces large-scale structure. This gap directly affects the LSS-probe and upper-limit claims and should be addressed by adding an explicit shot-noise calculation and, ideally, a decomposition of C_l into clustered and shot-noise components.","section":"Sec. 4.2, Fig. 3"},{"comment":"The claimed characteristic slope beta ~ 2 is not supported by a quantitative fit. The paper states that the spectra increase approximately linearly with log10 ell at higher ell, but no slope, intercept, uncertainty, or goodness-of-fit is reported for any catalogue or source type. The gray dashed reference lines in Figs. 3 and 4 have an arbitrary intercept ('different alpha'), so the agreement is purely visual. A fit of the slope over a specified ell range (for instance ell = 30-1000) with uncertainties is needed to support the central claim and to distinguish beta ~ 2 from other power-law behavior.","section":"Abstract and Sec. 5"},{"comment":"The 'theoretical upper limit' claim is not quantified. Fig. 5 presents ratios C_ell / C_ell,50Mpc/h for BBHs in the r2 catalogue only, with no error bars or variance across the seven observer positions, and the text summarizes reductions with broad ranges (e.g., 'approximately 1.7-2.5 times', 'roughly 1.3-1.6 times') without specifying how these ranges were derived. Since the paper states in Section 4.3 that the findings establish an upper limit on the anisotropy from compact binaries, the analysis should either quantify the limit with uncertainties and catalogue-to-catalogue scatter, or the claim should be softened to a statement about the trend in these particular mocks.","section":"Sec. 4.3, Fig. 5"},{"comment":"The decomposition into 'nearby' (<50 Mpc/h) and 'farther' sources uses a single distance cutoff, and the central subtraction result in the lower panel of Fig. 3 depends on this choice. No sensitivity test with other cutoffs (e.g., 30 or 100 Mpc/h) is presented. Since the paper's interpretation attributes the r1-r7 scatter primarily to the radial distribution of nearby sources, the robustness of this attribution to the chosen cutoff should be demonstrated; otherwise the 50 Mpc/h boundary appears ad hoc.","section":"Sec. 3.2/4.1, Fig. 3"}],"minor_comments":[{"comment":"The gray dashed 'Typical shape (beta = 2)' lines have arbitrary intercepts; the figure captions or legends should state explicitly that these are not fits to the data and that alpha is a free offset.","section":"Fig. 3 and Fig. 4 captions"},{"comment":"The notation uses R_k both for the merger rate and for its mean value; please introduce a distinct symbol, such as an overbar, to avoid confusing the rate with its binned average.","section":"Eq. (6)"},{"comment":"The sentence 'Consequently, other factors might contribute to the Poisson noise' does not logically follow from the preceding remark that the 1-bin approximation retains the typical shape; please rephrase to explain the intended chain of reasoning.","section":"Sec. 4.2"},{"comment":"The statement that the spectra 'exhibit negligible variation' over the LIGO frequency range 20-100 Hz is not shown in any figure; since this is a robustness claim, a figure or a quantitative bound would be helpful.","section":"Sec. 5"},{"comment":"The color-bar label '-1 21164.7' is unclear; the units of delta_GW and the meaning of the color scale should be specified.","section":"Fig. 2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the mock-catalogue construction is a useful contribution. I do not see a circularity problem: the GABE-COSMIC catalogue is an input model, not a target of the analysis. The main risk is that the headline claims currently overstate what the analysis establishes because the shot-noise contribution is not separated from the clustered anisotropy. I recommend major revision, with the shot-noise decomposition and a quantitative slope fit as the essential additions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new things here are the seven GABE-COSMIC all-sky lightcone catalogs, the observer-position variation study, and the empirical finding that subtracting sources within 50 Mpc/h reduces the scatter across catalogs and aligns the three source types within a factor of ~2. That last result is interesting and not in the earlier Jenkins/Cusin papers. The map-making and C_l pipeline are standard and look correctly assembled, and the robustness checks over Nside and frequency are sensible. I also want to credit the authors for openly discussing Poisson noise in Sec. 4.2 and even noting that their spectral shapes match the Poisson-noise predictions of Cusin et al. (2019). That is honest.\n\nThe soft spot is load-bearing. The paper never separates the angular power spectrum into a clustered LSS term and a Poisson shot-noise term. For a weighted catalog of discrete sources, the shot-noise contribution is constant in C_l, so log10[l(l+1)C_l/(2π)] rises with slope ~2 in log10 l. That is exactly the 'characteristic slope' the abstract and Sec. 5 advertise. The authors acknowledge in Sec. 4.2 that the 1-bin approximation retains the typical shape and matches Poisson predictions, but they do not compute the shot-noise floor or subtract it. Since the three source types share host galaxies, their weights are strongly correlated, so the factor-of-2 alignment after subtracting nearby sources could be a common shot-noise floor rather than evidence that the residual anisotropy traces LSS. The upper-limit claim also extrapolates beyond the 250 Mpc/h volume, and the fiducial r2 model was chosen for maximum anisotropy. Those are conditioning issues, not fatal flaws, but they matter.\n\nMinor issues: no slope fit with uncertainties, no code or data released, and the interpolation assumptions for the lightcone construction are reasonable but untested against a higher-time-resolution run.\n\nWho is this for? People working on SGWB anisotropy predictions for ET, CE, and LISA. The mock catalogs themselves could be useful. The paper deserves a serious referee, but the referee should ask for a shot-noise decomposition and a more cautious statement about LSS tracing. I would not cite this for the slope-2 claim without checking whether it survives shot-noise subtraction.","headline":"Competent mock-catalog study whose headline slope and cross-type alignment are probably dominated by Poisson shot noise, so the LSS-tracing interpretation is not yet established.","tokens_in":18183,"tokens_out":1205,"would_cite":true,"duration_ms":16223,"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":"The paper claims that the anisotropy of the gravitational-wave background from local compact-binary mergers is set almost entirely by the radial distribution of the nearest sources, and that after subtracting those sources the residual…","keywords":["gravitational waves","gravitational wave sources","cosmic anisotropy","gravitational wave astronomy","stochastic gravitational wave background","angular power spectrum","compact binary mergers","large-scale structure"],"falsifier":"Recompute the residual spectra using a larger simulation box (roughly twice the current $500\\,\\mathrm{Mpc}/h$ side length) with a proper lightcone output: if the seven-catalogue scatter or the offset between source types after subtracting sub-50-Mpc sources grows beyond a factor of about 2 at any multipole up to 1000, the universality claim is falsified. Observationally, if a future ground-based detector network measures the SGWB angular power spectrum in the 20–100 Hz band and the residual anisotropy after subtracting resolved nearby mergers disagrees with these mock spectra by more than a factor of a few over $\\ell = 1$–1000, the prediction is falsified.","tokens_in":17087,"feed_emoji":"🌌","tokens_out":14443,"duration_ms":124279,"temperature":0.7,"pith_summary":"Using seven all-sky mock catalogues of local merging black-hole and neutron-star binaries, this paper asks where the stochastic gravitational-wave background (SGWB) is brightest and how that unevenness depends on the source population. It finds that the angular power spectra of the SGWB overdensity for binary black holes, neutron-star–black-hole binaries, and binary neutron stars all rise approximately linearly with $\\log_{10}\\ell$ at small scales, with a characteristic slope near 2. The large differences between the seven catalogues come almost entirely from the radial (distance) distribution of sources closer than about 50 Mpc/h, which dominate because the signal falls as inverse-square distance. Removing those nearby sources brings the spectra of all three source types into agreement within a factor of about 2 across $\\ell = 1$–1000. If these mock skies represent the real Universe, the residual anisotropy is a map of the cosmic web, and the quoted spectra serve as a theoretical upper limit on the anisotropy from compact-binary coalescences.","feed_headline":"Remove the nearest mergers and the wave sky reveals cosmic structure","feed_subtitle":"Seven simulated skies agree within a factor of two once sources inside 50 Mpc are subtracted.","key_machinery":"The load-bearing mechanism is the construction of the all-sky mock lightcone catalogues. An observer is placed at one of seven positions in an eightfold-replicated $(500\\,\\mathrm{Mpc}/h)^3$ N-body simulation box; a galaxy enters the catalogue when it crosses the observer's past lightcone between two stored snapshots. The crossing time is found by solving two linear-interpolation equations — galaxy position and comoving distance both grow linearly with cosmic time between snapshots — and the galaxy's sky position and merger rate are linearly interpolated. Each galaxy contributes to the SGWB with weight $R_k/[(1+z_k)d_{c,k}^2]$, so nearby galaxies dominate the skymap. The overdensity $\\delta_{\\mathrm{GW}} = (\\Omega_{\\mathrm{GW}} - \\langle\\Omega_{\\mathrm{GW}}\\rangle)/\\langle\\Omega_{\\mathrm{GW}}\\rangle$ is pixelised on an equal-area sphere and expanded into spherical harmonics to give $C_\\ell$.","core_discovery":"On the paper's own terms, the central finding is a quantitative statement about the angular power spectrum $C_\\ell$ of the SGWB overdensity. Expressed as $\\log_{10}[\\ell(\\ell+1)C_\\ell/(2\\pi)]$, the spectra for BBHs, NSBHs and BNSs all show an approximately linear increase with $\\log_{10}\\ell$ at high multipoles ($\\ell \\gtrsim 30$–300), with slope $\\beta \\approx 2$. Across the seven mock catalogues the spectra vary by up to one to two orders of magnitude at $\\ell \\gtrsim 80$, and the paper attributes this scatter to fluctuations in the radial distribution of the nearest sources (within 50 Mpc/h), which are weighted by $d_c^{-2}$. After subtracting those nearby sources, the spectra of the seven catalogues shrink to within a factor of about 4 at low $\\ell$ and about 2 at $\\ell = 20$–1000, and the three source types become aligned within a factor of about 2 across $\\ell = 1$–1000. The paper concludes that the SGWB anisotropy is primarily influenced by the distribution of GW-source host galaxies and can probe large-scale structure, and that including more distant sources would only dilute the anisotropy, making these spectra an upper limit.","pith_inferences":["A natural extension, not explored in the paper, would be to apply the same lightcone construction to a larger simulation box; if the factor-of-2 agreement persists with a box larger than (500 Mpc/h)^3, the residual spectrum could be treated as a converged cosmic-web prediction rather than a catalogue-dependent estimate.","Because the radial distribution of the nearest sources is the dominant source of variance, real-world subtraction of individually detected nearby mergers from LIGO/Virgo/KAGRA data may shrink the theoretical uncertainty in the SGWB more than improving binary population synthesis models would.","The factor-of-2 alignment across source types suggests the residual SGWB is essentially a biased tracer of the galaxy density field; cross-correlating the mock residual map with a galaxy catalogue would be a direct test of this interpretation.","The $\\beta \\approx 2$ slope is degenerate with shot noise, so the claim that the residual spectrum probes large-scale structure could be sharpened by measuring a higher-order statistic (e.g., the bispectrum) of the same skymaps to separate Poisson clustering from genuine large-scale clustering."],"forward_implications":["If the residual anisotropy after subtracting sources within 50 Mpc/h traces the host-galaxy distribution, then future GW detectors that resolve and subtract nearby CBCs can use the leftover SGWB as a complementary tracer of large-scale structure, alongside galaxy surveys.","The spectra presented in the paper set a theoretical upper limit on the anisotropic SGWB from stellar-mass compact-binary mergers, so any cosmological SGWB component (for example from cosmic strings) that predicts larger anisotropy at these multipoles could in principle be distinguished.","The agreement of the three source types after subtraction implies that population-synthesis uncertainties in the relative merger rates matter less for the residual anisotropy than the spatial distribution of the host galaxies does.","The characteristic slope $\\beta \\approx 2$ at high $\\ell$ is the signature expected from Poisson shot noise, so the small-scale part of the spectrum should be interpreted as a noise floor rather than as a clustered signal."],"supporting_citations":[{"why":"It supplies the Phinney formula that relates the gravitational-wave energy density to the source population, and it is the basis of Eq. (2).","marker":"Phinney 2001"},{"why":"It provides the GABE-COSMIC mean merger rates and the original event catalogue, which this paper extends onto the celestial sphere.","marker":"Li et al. (2024)"},{"why":"It introduces the mock-lightcone approach of summing galaxy contributions to the anisotropic SGWB and gives the comparison spectra for BBHs, NSBHs and BNSs.","marker":"Jenkins et al. (2018)"},{"why":"It supplies the angular power spectrum formalism and earlier all-sky spectra for stellar-mass compact binaries that the present results are compared against.","marker":"Jenkins et al. (2019a)"},{"why":"It gives an independent theoretical prediction of the anisotropic SGWB from compact binaries, used here to assess the amplitude of the spectra.","marker":"Cusin et al. (2018)"},{"why":"It predicts that the high-multipole part of the angular power spectrum is dominated by Poisson noise from the azimuthal source distribution, which the paper uses to interpret the slope-2 shape.","marker":"Cusin et al. (2019)"},{"why":"It supplies the cosmological N-body simulation volume from which the replicated box and past-lightcone catalogues are built.","marker":"Springel et al. (2005)"},{"why":"It supports the straight-line interpolation of galaxy positions between snapshots, a key assumption in the lightcone construction.","marker":"Kitzbichler & White (2007)"},{"why":"It provides the HEALPix equal-area pixelisation and spherical harmonic analysis used to compute the angular power spectra.","marker":"Górski et al. (2005)"}],"fun_headline_variants":["Wave background spectra align to a slope of two once nearby sources are cut","Subtracting nearest binaries shrinks wave background variations to factor two","After removing close binaries, all mock wave backgrounds line up","The wave sky reveals cosmic structure once 50 Mpc sources are removed","All wave backgrounds share a universal slope after nearby source removal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument stands on the assumption that the mock catalogues reproduce the real three-dimensional distribution of merging compact binaries in the local universe: galaxy positions and merger rates are linearly interpolated between simulation snapshots, and the adopted binary population synthesis rates stand in for the true local population — if nearby galaxies are misplaced or those rates are wrong, the claimed dominance of the sub-50-Mpc radial distribution and the factor-of-2 agreement after subtraction would not transfer to the real sky.","fun_headline_variants_meta":{"raw":{"variants":["Wave background spectra align to a slope of two once nearby sources are cut","Subtracting nearest binaries shrinks wave background variations to factor two","After removing close binaries, all mock wave backgrounds line up","The wave sky reveals cosmic structure once 50 Mpc sources are removed","All wave backgrounds share a universal slope after nearby source removal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001304,"raw_usage":{"total_tokens":5416,"prompt_tokens":1142,"completion_tokens":4274,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":758,"completion_tokens_details":{"reasoning_tokens":4186}},"tokens_in":758,"tokens_out":4274,"duration_ms":33252,"temperature":1.0,"reasoning_tokens":4186,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T16:32:15.642826+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the residual spectra using a larger simulation box (roughly twice the current $500\\,\\mathrm{Mpc}/h$ side length) with a proper lightcone output: if the seven-catalogue scatter or the offset between source types after subtracting sub-50-Mpc sources grows beyond a factor of about 2 at any multipole up to 1000, the universality claim is falsified. Observationally, if a future ground-based detector network measures the SGWB angular power spectrum in the 20–100 Hz band and the residual anisotropy after subtracting resolved nearby mergers disagrees with these mock spectra by more than a factor of a few over $\\ell = 1$–1000, the prediction is falsified.","supporting_citations":[],"review_version":1}