{"id":"e1a8df4c-dfac-4bca-a63f-f479785bf24e","arxiv_id":"1908.05779","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"LISA detection rates for merging giant black holes built from the Illustris simulation are about 0.5 to 1 per year, roughly an order of magnitude below many semi-analytic predictions.","lead":"Using the Illustris galaxy simulation, this paper predicts that LISA will detect roughly 0.5 to 1 merging giant black hole binaries per year, lower than most earlier estimates. Because the simulation omits black holes below 100,000 solar masses, the authors treat this as a lower limit for planning LISA science.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed 'lower limit' status is unsafe: aligned maximal spins (a=0.8) are optimistic for the unequal-mass binaries that contribute substantially to the rate, so a conservative spin choice could push the detection rate below the quoted 0.5-1 yr^-1 range.","rationale":"I read the paper as a transparent, model-dependent rate forecast: it uses the Illustris simulation and two published sub-grid binary evolution prescriptions, with a new extraction that retains lower-mass mergers, and it compares two LISA configurations. The internal pipeline is careful, and the paper's caveats are extensive, including the ad hoc seeding and the missing sub-1e5 Msun population. The reader's conditional verdict is appropriate. My stress-test focuses on a different, more internal point than the reader's seeding concern. The abstract makes a strong claim: the rates should be treated as a lower limit. For that to be true, the detectability calculation must be conservative or at least not systematically optimistic. The choice of a=0.8 aligned spins is systematically optimistic, as acknowledged in Section 3.2.1, and it preferentially boosts the very unequal-mass binaries that appear in the detected population (Figure 8). The paper never tests the sensitivity of the headline rates to this choice, and the Monte Carlo does not sample over spin orientations. This is a correctness risk specific to the central claim, not a disagreement with consensus. If the concrete test shows the rate drops below 0.5 yr^-1 with zero spins, the paper's headline would need to be revised to a range that is not advertised as a lower limit. Because the concern is testable with a modest post-processing rerun and does not undermine the broader modeling contribution, I keep the reader's CONDITIONAL verdict unchanged rather than rejecting the paper.","tokens_in":27989,"tokens_out":8296,"duration_ms":87348,"concrete_test":"Recompute the integral (Equation 27) and Monte Carlo detection rates from the existing catalogs with the identical pipeline but set the dimensionless spins to a1=a2=0 (and also a=-0.8) instead of 0.8 in the PhenomD SNR calculation, keeping all cuts, masses, redshifts, and the LISA PL sensitivity fixed. If the resulting 'All' detection rates for DA17 and K17 fall below 0.5 yr^-1, the quoted range and the 'lower limit' wording in the abstract are not robust to the spin assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline rate '~0.5-1 yr^-1' and its characterization as a lower limit rest on a detectability calculation that fixes both black hole spins to a=0.8, aligned with the orbital angular momentum, for every binary (Section 3.2.1). The authors explicitly call this 'the optimistic case' for unequal-mass systems, noting the signal peak can change by an order of magnitude between spin-down and spin-up configurations. The detectable population, however, includes a substantial tail of mass ratios down to q~1e-2 (Figure 8), precisely where spin matters most. Since the same observation time and SNR threshold (rho=8) are used, an overestimated SNR translates directly into an overestimated detection volume and rate. The missing low-mass population (m<1e5 Msun) motivates the 'lower limit' label, but the paper does not quantify whether that upward correction outweighs the downward correction from replacing optimistic spins with zero, isotropic, or antialigned spins. Thus the central claim that the true rate is near or above 0.5-1 yr^-1 is not established; the model as run gives an optimistic-detectability rate, not a conservative lower bound.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses the Illustris-1 cosmological simulation to build a catalog of massive black hole (MBH) mergers, applying a new post-processing extraction that retains binaries down to the seed mass of 1.42e5 M_sun. It evolves the merger population to coalescence under three prescriptions (no delay, ND; DA17; K17) and computes LISA detection rates using PhenomD waveforms, the LISA proposal sensitivity curve, a Galactic background model, and an SNR threshold of rho = 8, with both an integral rate calculation and Monte Carlo catalog generation. The main reported results are detection rates of about 0.75 yr^-1 (ND), 0.70 yr^-1 (DA17), and 0.44 yr^-1 (K17), summarized as ~0.5-1 yr^-1, with the claim that this should be treated as a lower limit because masses below 1e5 M_sun are missing from the sample.","tokens_in":28257,"tokens_out":5489,"duration_ms":56192,"significance":"If the results are robust, the paper provides an important hydrodynamic-simulation-based benchmark for LISA MBH rate predictions, roughly an order of magnitude below many semi-analytic predictions and closer to EAGLE-based estimates. The strengths are the clearly documented advanced extraction, use of two published and independent binary-evolution prescriptions, transparent Monte Carlo construction, standard waveform and sensitivity choices, and explicit comparison with prior LISA rate calculations. The central result is internally consistent. However, the 'lower limit' interpretation is not currently supported because the detectability calculation is optimistic with respect to black hole spin, and because several dominant systematics are not quantified. The significance is therefore moderate pending revision of the central claim.","major_comments":[{"comment":"The headline rate '~0.5-1 yr^-1' and its characterization as a lower limit are not supported by the detectability calculation as presented. The calculation fixes a1 = a2 = 0.8, aligned with the orbital angular momentum, for every binary, and the text explicitly describes this as 'the optimistic case' for unequal-mass systems, where the signal peak can change by an order of magnitude between spin-down and spin-up configurations. The detected population includes mass ratios down to q ~ 1e-2 (Figure 8), exactly the regime where spin matters most; since the SNR in Eq. (23) scales detection volume, a conservative spin choice (zero, isotropic, or anti-aligned) could push the DA17 and K17 rates below the quoted 0.5 yr^-1. The authors should either recompute rates under a conservative spin model or replace the 'lower limit' language with a statement that the rates are model-dependent estimates under an optimistic detectability assumption.","section":"Section 3.2.1 and Abstract"},{"comment":"The stated upward correction from missing low-mass systems is asserted but not quantified against the downward correction from optimistic spins and other modeling choices. The seeding prescription (1.42e5 M_sun seed in halos above 7.1e10 M_sun) is labeled ad hoc in Section 5, and it directly sets the completeness limit invoked by the 'lower limit' claim. Since the paper does not estimate the magnitude of the missing m < 1e5 M_sun contribution, nor combine it with the spin sensitivity, the conclusion that the true LISA rate is close to or above ~0.5-1 yr^-1 is not established. A systematic-error table or a bracketed rate range covering both upward and downward corrections would make the central claim defensible.","section":"Sections 2, 3.3.1, and 5"},{"comment":"The quoted numerical errors (<0.01 in the integral calculation, ~1% in the Monte Carlo) reflect only redshift binning and Poisson sampling, not the dominant systematics: one Illustris volume, one seeding prescription, and two specific delay models. Given that the between-model spread alone is a factor of about 1.6 (0.44 vs 0.70 yr^-1), presenting the result with two-decimal precision understates the uncertainty. The manuscript should state this limitation more explicitly, even if a full Bayesian treatment is beyond the scope.","section":"Equation (27) and Tables 1, A1"}],"minor_comments":[{"comment":"The text contains the typo 'redshits' where 'redshifts' is intended.","section":"Section 5"},{"comment":"The text contains the typo 'surpression' where 'suppression' is intended.","section":"Section 6"},{"comment":"The caption contains 'Similary' and 'negligble'; these should be 'Similarly' and 'negligible', and 'inpiral-only' should be 'inspiral-only'.","section":"Table A1 caption"},{"comment":"The phrase 'can be see' should be 'can be seen'.","section":"Figure A2 caption"},{"comment":"The Robson & Cornish (2017) reference is incomplete; an arXiv identifier or journal citation should be supplied.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The spin assumption is the deciding issue; I would support publication after the authors either add conservative spin tests or reframe the central claim. The self-citation of K17 and BOWIE/gwsnrcalc is not a concern here because these are published models with independent grounding."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou should know two things about this one. First, the genuinely new piece is the post-processing: by replacing the old 10^6 Msun mass cut with a check on whether the merging black holes existed before their host's fly-by interaction, they keep 8,265 extra mergers and double the LISA detection rate. That matters because it is exactly the near-seed-mass binaries that LISA is most sensitive to, and prior Illustris-based work threw them out. Second, the paper's own headline '0.5–1 per year, treated as a lower limit' is the weakest part of the paper. Do not repeat that claim without a caveat.\n\nThe calculation itself is careful. They take 17,535 mergers from Illustris-1, evolve them to coalescence with two published, independent delay prescriptions (DA17 and K17), check consistency between an integral rate calculation and a 10,000-realization Monte Carlo, and use standard SNR integration against the LISA proposal curve. The rates agree with the only other hydro-based forecast in their table (Salcido et al. on EAGLE) and are an order of magnitude below the older SAM predictions, which is a useful quantitative statement for mission planning. Credit where due: the paper is explicit about its own limitations, flagging the Illustris seeding as ad hoc, the missing dwarf-galaxy population, and the known optimistic spin choice.\n\nThe soft spots are real but not fatal. The 'lower limit' claim is the big one. They set every binary to a=0.8 aligned spins, which they admit in Section 3.2.1 is the optimistic case for unequal-mass systems, and the detected population includes mass ratios down to about 1e-2. The missing low-mass population pushes the rate up; the spin choice pushes it down. They never quantify which dominates, so 'lower limit' is not established. Treat 0.5–1/yr as a central estimate with order-of-magnitude uncertainty. Related point: the statistical error bars are small (<0.01), but the dominant systematics—seeding, delay prescriptions, spins—are not propagated into the quoted rates. Minor: the processed catalogs and code were not released, so independent verification is limited. The K17 and gwsnrcalc self-citations point to published work, so I do not hold those against them.\n\nThe stress-tester's concern lands but should be kept in proportion: it is an interpretive flaw in the framing, not a methodological failure. The rate calculation is sound; the headline overreaches.\n\nNet: this deserves a serious referee. I would cite it as a reference point for LISA rate forecasts, and it is a good reading-group piece for discussing which way the systematics cut.","headline":"Worth a serious referee: the advanced extraction doubles the Illustris-based LISA rate forecast and the calculation is careful, but the 'lower limit' label does not survive the paper's own optimistic spin choice.","tokens_in":28792,"tokens_out":5873,"would_cite":true,"duration_ms":50860,"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":"This paper predicts that LISA will detect roughly 0.5 to 1 massive black hole merger per year, an order of magnitude below many earlier estimates, because the simulated seed population lacks lighter black holes.","keywords":["LISA","massive black hole binaries","gravitational wave detection rates","cosmological simulation","black hole seeding","binary evolution timescales","Monte Carlo catalogs","black hole spectroscopy"],"falsifier":"Compare the predicted rate against one from any cosmological simulation with seeds of $10^3$--$10^4$ solar masses run through identical delay prescriptions; if that rate exceeds several per year, or if LISA itself detects more than about one massive black hole merger per year in its first years of operation, the lower-limit claim would be incomplete.","tokens_in":27813,"feed_emoji":"🛰️","tokens_out":9066,"duration_ms":77594,"temperature":0.7,"pith_summary":"This paper asks how often the space-based gravitational-wave detector LISA will see two massive black holes spiral together and merge, and what the detected population will look like. Using merger histories from a large hydrodynamical cosmological simulation, it evolves the simulated binaries to coalescence with two different physical prescriptions for how the pair sheds energy. The central result is a detection rate of roughly $0.5$ to $1$ per year for binaries with total mass above $10^5$ solar masses, which the authors explicitly treat as a lower limit because the simulation cannot resolve lighter black holes. If correct, LISA's first years would yield about one massive black hole merger per year, an order of magnitude below several earlier predictions.","feed_headline":"LISA may catch about one black hole merger per year","feed_subtitle":"Simulation-based rates are an order of magnitude below earlier estimates because light seeding is missing.","key_machinery":"The load-bearing objects are the merger catalog and the coalescence-time prescriptions. The catalog comes from identifying every simulated pair of black hole particles that merge at kiloparsec scales, then filtering out spurious mergers caused by halo-finder mistakes by tracking whether a host galaxy has lost its central black hole; this advanced extraction retains 17,535 mergers down to the seed mass. Two sub-grid prescriptions then assign each binary a delay from formation to coalescence: DA17, a three-stage analytic model with dynamical friction, stellar hardening, and gravitational-wave emission, and K17, a numerical integration with dynamical friction, a loss-cone hardening rate, and a circumbinary gas disk. These delays, together with Monte Carlo Poisson sampling of coalescence times and waveform signal-to-noise ratios built from a phenomenological waveform model, convert the simulation's merger history into mock LISA detection catalogs.","core_discovery":"The paper claims that the no-delay baseline, in which every simulated galactic merger immediately produces a gravitational-wave source, already gives a LISA detection rate of about $0.98$ per year; adding two physically motivated delay prescriptions lowers the all-signal detection rate to $0.70$ per year (DA17) and $0.44$ per year (K17). These rates are lower than most published estimates, and the authors argue the main reason is seeding: the simulation places seeds of $1.42\\times 10^5$ solar masses in massive halos, so the numerous lighter black holes that dominate LISA's accessible population are absent. A new extraction method that retains binaries down to the seed mass doubles the detection rate relative to older cuts at $10^6$ solar masses, showing that the low-mass tail is where most LISA events live. Because the sample omits masses below $10^5$ solar masses and ignores gas-driven migration and triple interactions, the quoted range is a lower limit.","pith_inferences":["If real seeds are lighter or form earlier, the missing low-mass binaries would push the rate substantially above $1$ per year, possibly toward the higher semi-analytic predictions.","The paper's own comparison to gas-rich and gas-poor constant delays suggests that adding gas-driven migration or triple-black-hole interactions would raise rates, so the quoted range is a floor under the paper's modeling choices.","A direct testable extension is to run the same pipeline on a simulation with $10^3$--$10^4$ solar-mass seeds; the low-mass peak of the detectable population is where the models separate most clearly."],"forward_implications":["If the rate is roughly $0.5$--$1$ per year, LISA's first years would yield only a handful of massive black hole mergers, making each event statistically precious.","Binaries with a component near $10^5$ solar masses dominate the detectable population, so LISA measurements would directly probe seed formation rather than only late-time growth.","The two delay prescriptions differ mainly through low-mass binaries, meaning the predicted rate and the detected mass distribution are jointly shaped by binary evolution physics and seeding physics.","Black hole spectroscopy should be possible for most detected events, giving a spectroscopy rate close to the overall detection rate."],"supporting_citations":[{"why":"Supplies the cosmological simulation output that provides the massive black hole merger history and host galaxy properties.","marker":"Vogelsberger et al. 2014a"},{"why":"Details the simulation's black hole seeding, accretion, and repositioning that define the initial population.","marker":"Sijacki et al. 2015"},{"why":"Establishes the procedure for extracting host-galaxy density profiles and velocity dispersions from merger events.","marker":"Blecha et al. 2016"},{"why":"Provides the K17 binary evolution prescription and the earlier merger catalog against which the new extraction is compared.","marker":"Kelley et al. 2017a"},{"why":"Supplies the DA17 delay prescription with dynamical friction, hardening, and gravitational-wave timescales.","marker":"Dosopoulou & Antonini 2017"},{"why":"Defines the proposed LISA sensitivity curve and mission parameters used for detectability.","marker":"Amaro-Seoane et al. 2017"},{"why":"Provides the waveform and signal-to-noise generation code and the sensitivity-curve construction used here.","marker":"Katz & Larson 2019"},{"why":"Supplies, with Husa et al. 2016, the PhenomD waveform family used for inspiral-merger-ringdown signals.","marker":"Khan et al. 2016"},{"why":"Supplies the black hole spectroscopy detectability thresholds and a high-rate prediction to compare against.","marker":"Berti et al. 2016"},{"why":"Provides a hydrodynamic-simulation rate prediction and the variance estimate used for the integral rates.","marker":"Salcido et al. 2016"}],"fun_headline_variants":["LISA might see under one merger per year","Missing light seeds halve LISA's merger count","LISA detection likely 0.5-1 per year, lower bound","Low-mass black holes missing from LISA rate estimates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes the simulation's seeding prescription—one black hole of $1.42\\times10^5$ solar masses in every halo above $7.1\\times10^{10}$ solar masses—produces a representative massive black hole population; if real seeds are smaller or form earlier, the missing low-mass binaries would raise the rate.","fun_headline_variants_meta":{"raw":{"variants":["LISA might see under one merger per year","Missing light seeds halve LISA's merger count","LISA detection likely 0.5-1 per year, lower bound","Low-mass black holes missing from LISA rate estimates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1370,"prompt_tokens":1003,"completion_tokens":367,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":619,"completion_tokens_details":{"reasoning_tokens":300}},"tokens_in":619,"tokens_out":367,"duration_ms":4497,"temperature":1.0,"reasoning_tokens":300,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:05:10.832511+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the predicted rate against one from any cosmological simulation with seeds of $10^3$--$10^4$ solar masses run through identical delay prescriptions; if that rate exceeds several per year, or if LISA itself detects more than about one massive black hole merger per year in its first years of operation, the lower-limit claim would be incomplete.","supporting_citations":[],"review_version":1}