{"id":"125ed7a3-bf9e-4112-a034-90f64bcb878e","arxiv_id":"2412.08280","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Collision-driven black hole seeds formed in nuclear star clusters can grow to a billion solar masses and reproduce the observed high-mass black hole mass function in a galaxy formation model.","lead":"This paper models how supermassive black holes could begin as dense star clusters in galaxy centers, where stars collide and collapse into a black hole seed. The simulations grow those seeds to billions of solar masses and roughly match the observed population of the most massive black holes, suggesting the collision channel can contribute to real black holes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The predicted SMBH population rests entirely on a uniform rescaling of NSC radii (epsilon_r) that is never validated against the observed radius-mass relation; with epsilon_r=1 no seeds form, so the BHMF match is conditional on this ad hoc premise.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the entire seed population disappears when epsilon_r=1, and the adopted radius rescaling is motivated but not validated inside the model. I considered alternative concerns, including possible mass non-conservation in Eq. 18 and the absence of an explicit Eddington cap on the Bondi-Hoyle-Lyttleton accretion, but these are less central to the stated strongest claim because the BHMF match at high masses is primarily governed by when and how many seeds form, which is set by the radius rescaling. The proposed test directly probes whether the seed-forming NSCs are consistent with the observed radius distribution; if the result changes dramatically under a plausible scatter in radii, then the high-mass BHMF agreement is an artifact of the fixed epsilon_r assumption. The paper is transparent about the exploratory nature of the model, and the conditional verdict remains appropriate.","tokens_in":1018,"tokens_out":667,"duration_ms":131994,"concrete_test":"Rerun model A4 (or a representative seed-forming model) with epsilon_r sampled from the observed scatter in the NSC radius-mass relation (Eq. 2, e.g., a log-normal with sigma ~0.2 dex) instead of a fixed value, keeping all other parameters fixed. If the seed masses in Table 2 and the BHMF above 1e8 Msun change by more than an order of magnitude, the central result is controlled by the unvalidated epsilon_r prescription rather than by the physics of collision-based seeding.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that collision-based seeding in NSCs produces a BHMF comparable to Vika et al. (2009) above 1e8 Msun is controlled by the seed-production rate, which is nonzero only for epsilon_r < 1 (Section 2.2.1). The model applies a constant factor epsilon_r to the observed NSC radius relation (Eq. 2), and for the seed-forming models (epsilon_r=0.1-0.5) this implies present-day NSCs that are 2-10 times smaller than the observed mean relation at all masses and at all cosmic times. Banerjee & Kroupa (2017) motivate early-time expansion of young massive clusters, not a permanent uniform compression of the entire NSC population; a scatter-based or time-evolving radius treatment could yield a very different seed population. Since Table 2 shows that heavy seeds (up to 1.7e5 Msun) form only in epsilon_r=0.1 models, the high-mass end of the BHMF is a direct product of this unvalidated assumption. The paper itself notes the importance of the mass-radius correlation but does not check the resulting radii against a direct NSC census.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper implements an in-situ Nuclear Star Cluster (NSC) formation model in the semi-analytic galaxy formation code Galacticus, and couples it to a collision-based supermassive black hole (SMBH) seeding recipe. NSC gas is fed from the spheroid at a rate A_res times the spheroid star formation rate, NSC sizes follow the observed r_NSC-M_dyn relation with an ad hoc rescaling factor epsilon_r, and a BH seed forms when the NSC stellar mass exceeds a critical collision mass Mcrit; the seed mass is set to 0.5 times the NSC stellar mass and subsequently grows by Bondi/disk accretion. The authors run a grid over A_res, epsilon_r, and a minimum NSC mass threshold, and compare the predicted NSC scaling relations, NSC mass function, BHMF, and NSC-SMBH coexistence fractions with local observations. The headline claim is that compact NSCs (epsilon_r = 0.1-0.5) form BH seeds of roughly 5x10^2 to 1.7x10^5 solar masses that grow to SMBHs up to about 10^9 solar masses, and that the predicted z=0 BHMF is comparable to the Vika et al. (2009) observational BHMF above 10^8 solar masses.","tokens_in":33403,"tokens_out":10560,"duration_ms":110276,"significance":"If the central claim holds, the paper provides a useful proof-of-concept that collision-based seeding in dense NSCs can produce a cosmologically relevant SMBH population within a semi-analytic framework. The work is transparent: the parameter grid is clearly specified, the implementation is modular, and Appendix A presents convergence tests for both the NSC and BH mass functions. The significance is limited, however, by three structural features: every SMBH-forming model requires an ad hoc uniform compression of NSC radii, the 'observed' NSC mass function used for validation is itself derived by assuming a constant M_NSC/M_gal scaling rather than being a direct census, and the heaviest seeds in this implementation form only at low redshift (z ~ 0.2-0.35), so the channel as modeled does not directly address high-redshift SMBH formation. These caveats are partially acknowledged in the manuscript, but they materially weaken the strength of the headline BHMF comparison.","major_comments":[{"comment":"The entire seed-forming parameter space depends on the rescaling r_NSC -> epsilon_r r_NSC. With epsilon_r = 1, none of the models forms any BH seed (Table 2 lists no seed for A1, B1, C1, D1, and their epsilon_r = 1 counterparts), and the heaviest seeds, up to ~1.7x10^5 M_sun, occur only for epsilon_r = 0.1. The headline BHMF comparison is therefore conditional on the assumption that a substantial part of the NSC population has radii 2-10 times smaller than the mean relation of Eq. (2) at all masses and all times. The cited motivation, Banerjee and Kroupa (2017), concerns the early expansion of young massive clusters, not a permanent uniform compression of the whole NSC population. The authors should validate this assumption, for example by comparing the model's predicted z=0 NSC radius distribution with the observed scatter in the r_NSC-M_dyn relation, or by implementing a physically motivated, time-dependent NSC radius evolution and showing how the BHMF responds.","section":"Sec. 2.2.1, Eq. (2), Table 2"},{"comment":"The central claim of a 'comparable' BHMF above 10^8 M_sun is not stated consistently. Section 3.4 says the BHMF from Vika et al. (2009) is comparable at masses above 10^8 M_sun, while Section 4 states that at higher masses the model tends to underpredict the observed population. The quantitative summary in Table B.1 reports average deviations with ±1 dex scatter over the full mass range, but no goodness-of-fit statistic restricted to the claimed 10^8-10^9 M_sun range is provided. Given that the low-mass BHMF is overpredicted by roughly two orders of magnitude at 10^6-10^7 M_sun, the paper should either report a quantitative agreement measure for the high-mass range (for example, chi-square or a Kolmogorov-Smirnov test over 10^8-10^9 M_sun) or consistently describe the high-mass behavior as an underprediction.","section":"Sec. 3.4, Sec. 4, Table B.1"},{"comment":"The 'observed' NSC mass function used for validation is not a direct NSC census; it is derived from the GAMA galaxy stellar mass function by assuming a constant scaling M_NSC = 10^-3 M_galaxy (Georgiev et al. 2016). This proxy presupposes a linear scaling between NSC mass and galaxy mass, which is precisely the kind of relation the model is being tested against. The agreement in shape therefore cannot validate the absolute normalization or the low-mass behavior of the model. The manuscript already acknowledges the lack of a direct NSC sample; I recommend supplementing the comparison with direct NSC samples, for example the volume-limited samples of Sanchez-Janssen et al. (2019) or the sample of Pechetti et al. (2020), with selection effects accounted for, or at minimum clearly labeling the derived mass function as a conditional proxy rather than an observational benchmark.","section":"Sec. 3.2"},{"comment":"The paper attributes the low-mass BHMF overprediction to an excess of galaxies predicted by Galacticus relative to the Baldry et al. (2012) sample, but Appendix B reports a galaxy stellar mass function overprediction of only 0.5-0.6 dex, i.e., a factor of 3-4, while the BHMF is overpredicted by about 10^2 at 10^6-10^7 M_sun. A factor-of-100 discrepancy cannot be explained by the stated galaxy normalization offset alone. The authors should quantify how much of the BHMF excess follows from the galaxy mass function offset, from the epsilon_r-driven seeding recipe, and from the accretion model, for example by rerunning model A4 with weights that match the observed galaxy mass function or by decomposing the BHMF excess into occupation fraction and seed mass contributions.","section":"Sec. 3.4, Appendix B"}],"minor_comments":[{"comment":"In the text 'Phi_bullet ~ 5 [Mpc^-1 dex^-1]' the unit should read Mpc^-3 dex^-1.","section":"Sec. 3.4"},{"comment":"The sentence 'the most massive NSC is is NGC 4461' contains a duplicated verb, and 'There is a clearly deviation' should be 'There is a clear deviation'.","section":"Sec. 3.2"},{"comment":"The non-seeding condition is written as 'Mthreshold >= M_NSC'; since the seeding condition is M_NSC >= Mthreshold, the non-seeding condition should be M_NSC < Mthreshold, i.e., 'Mthreshold > M_NSC'.","section":"Fig. 6 caption"},{"comment":"The caption lists three values of epsilon_r (0.5, 0.2, 0.1) for the two models D3 and D4 shown, and the phrase 'Mthreshold = 10^3 M_sun 10^3 M_sun' repeats the mass; the caption should be cleaned up.","section":"Fig. 5 caption"},{"comment":"The G4 row is missing its seed-mass range; Section 3.4 states that G4 forms SMBHs up to ~10^9 M_sun, so the blank entry appears to be an omission.","section":"Table 2"},{"comment":"The chi-square values mentioned in the text and in Figure 1 are not quoted in the text, so the reader cannot assess the fit quality or the degrees of freedom used.","section":"Sec. 3.1"},{"comment":"The fate of the initial 10 M_sun BH seed is not described once the new collision-formed seed replaces it; the text should clarify whether it merges with the new seed or remains as a separate object.","section":"Sec. 2.2.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a reasonable proof-of-concept for collision-based NSC seeding in a SAM, but the headline BHMF comparison rests on an unvalidated, uniform radius rescaling (epsilon_r) that is never checked against the observed NSC radius-mass relation. The authors should either validate this assumption with direct observational comparisons or reframe the paper as an explicitly conditional exploration of compact-NSC models. The indirect NSC mass function benchmark and the inconsistency between Section 3.4 and Section 4 regarding the high-mass BHMF also need to be addressed before the central claim can be considered robust. No concerns about citation ethics; the use of prior same-group papers is transparent."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this is the first population synthesis of the Escala (2021) / Vergara et al. (2023) collision-seeding recipe inside a SAM. That is genuinely new. Prior work was single-cluster or analytic; here you get z=0 BHMFs, seed mass ranges, occupation fractions, and scaling relations from Galacticus. The paper is honest and internally consistent, reports convergence tests, and does not hide that the low-mass BHMF overshoots Vika et al. by factors of 10–100. The high-mass overlap above 1e8 Msun is a meaningful result if the seeding premise holds.\n\nThe soft spot is exactly where the reader and stress test point: the whole seed population requires epsilon_r < 1. With epsilon_r = 1, no model forms any seed. The model rescales the observed r_NSC–M_dyn relation uniformly by a factor 2–10 at all masses and all cosmic times. The motivation from Banerjee & Kroupa (2017) is that young massive clusters expand over time; that does not mean the present-day NSC population is permanently more compact by a constant factor. The paper acknowledges the mass–radius correlation is crucial but never checks the resulting radii against a direct NSC census. So the quantitative BHMF, and especially the heavy seeds that only appear at epsilon_r = 0.1, are conditional on an assumption that could easily be wrong. The stress test is on target.\n\nThe NSC mass function comparison also uses a derived baseline (M_NSC = 1e-3 M_gal from Baldry et al. 2012), not a direct observational sample. The paper is upfront that the low-mass end is a lower limit, but it weakens the claim of shape agreement. No code or parameter files are released, which makes the Galacticus implementation hard to reproduce.\n\nCredit where due: the equations are clear, the parameter exploration is systematic, and the occupation fractions and seed mass ranges will be useful for anyone working on seeding channels. The paper is careful to label the comparison as preliminary. It is not circular—the Vika et al. BHMF was not used to fit the model.\n\nFor a referee: this deserves serious review, not a desk reject. But my recommendation would be major revision: either validate or replace the uniform epsilon_r assumption with a scatter-based or time-evolving radius treatment and compare against observed NSC radii, or explicitly reframe the paper as a proof-of-concept conditioned on unobserved compact NSCs. Releasing the Galacticus configuration would strengthen it a lot. As it stands, I would not cite the quantitative BHMF as a prediction, but I would use the qualitative map of where collision seeding can contribute.","headline":"A transparent SAM exploration of NSC collision seeding that produces plausible high-mass BHMFs, but the entire seed population rests on an unvalidated radius rescaling that should be reframed as proof-of-concept.","tokens_in":33903,"tokens_out":2489,"would_cite":false,"duration_ms":28865,"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":"Collisions between stars in compact nuclear star clusters can seed black holes that grow to ~10^9 solar masses, and the predicted local black-hole mass function matches observations above 10^8 solar masses.","keywords":["galaxies: evolution","galaxies: formation","galaxies: nuclei","Galaxy: center","quasars: supermassive black holes","nuclear star clusters","black hole seeding","stellar collisions"],"falsifier":"Measure the radii of young nuclear star clusters at high redshift (or infer their initial radii from the present-day expansion of young massive clusters) and check whether compact clusters with $r_{\\rm NSC}\\lesssim0.1$ pc exist; if no NSC population is 2–10 times more compact than the local $r_{\\rm NSC}$–$M_{\\rm dyn}$ relation at the epochs when seeds must form, the critical-mass condition $M_{\\rm NSC}\\ge M_{\\rm crit}$ is essentially never met and the predicted black-hole population collapses.","tokens_in":32922,"feed_emoji":"🕳️","tokens_out":10200,"duration_ms":90823,"temperature":0.7,"pith_summary":"This paper argues that supermassive black holes can be seeded by runaway stellar collisions inside nuclear star clusters that form in situ from gas funneled to galaxy centers. Implementing this channel in the semi-analytic model Galacticus, the authors show that only initially compact clusters — those a factor 2 to 10 smaller than the present-day radius relation — pass the critical-mass threshold and produce seeds of roughly $5\\times10^2$ to $1.7\\times10^5\\,M_\\odot$. Those seeds then accrete and grow to black holes as massive as $\\sim10^9\\,M_\\odot$ by $z=0$, producing a black-hole mass function that tracks the observed one above $10^8\\,M_\\odot$ and overpredicts it below. The nuclear star cluster population formed in the same runs has a mass-function shape comparable to observations and overlaps the observed scaling relations, though quantitative agreement is still limited. If this channel is real, it offers a single origin for the coexistence of nuclear star clusters and supermassive black holes in galaxies of roughly $10^{10}\\,M_\\odot$.","feed_headline":"Compact star clusters can seed billion-solar-mass black holes","feed_subtitle":"The model grows collision-born seeds to nearly 1e9 solar masses and matches observed black hole counts above 1e8.","key_machinery":"The load-bearing object is the critical mass $M_{\\rm crit}(r_{\\rm NSC})$ from Eq. (17), defined by setting the stellar collision timescale $t_{\\rm coll}$ equal to the cluster's mass-weighted age $t_H$; clusters above this mass are expected, via runaway collisions, to convert 10–50% of their mass into a central massive object. The model gathers NSC mass from the in-situ star formation reservoir and imposes the seed condition $M_{\\rm NSC}\\ge M_{\\rm crit}$ and $M_{\\rm NSC}\\ge M_{\\rm threshold}$, with seed mass $M_\\bullet = 0.5\\,M_{\\rm NSC}$; the radius rescaling $\\epsilon_r$ is what moves otherwise stable clusters across the threshold.","core_discovery":"The paper's central claim is that a collision-based seeding channel operating in nuclear star clusters can account for the observed population of supermassive black holes above $10^8\\,M_\\odot$. The formation criterion is the critical mass $M_{\\rm crit}(r_{\\rm NSC})$ of Eq. (17), the cluster mass at which the stellar collision timescale equals the mass-weighted age of the cluster; when $M_{\\rm NSC}^{\\rm stellar}\\geq M_{\\rm crit}$ and the cluster is well-sampled ($M_{\\rm NSC}^{\\rm stellar}\\geq M_{\\rm threshold}$), a seed of mass $M_\\bullet = \\epsilon_\\bullet M_{\\rm NSC}^{\\rm stellar}$ with $\\epsilon_\\bullet=0.5$ replaces the initial $10\\,M_\\odot$ black hole and begins accreting. Because the adopted present-day radius relation $r_{\\rm NSC}=r_0\\sqrt{M_{\\rm dyn}^{\\rm NSC}/10^6\\,M_\\odot}$ gives radii too large for collisions to matter, the model rescales radii by $\\epsilon_r$, with $\\epsilon_r=0.1$–$0.5$; these compact clusters form seeds from $5\\times10^2$ to $1.7\\times10^5\\,M_\\odot$, with the heaviest seeds appearing in the most compact systems and growing to a few times $10^9\\,M_\\odot$ by $z=0$. The resulting mass function agrees with the local SMBH mass function of Vika et al. (2009) above $10^8\\,M_\\odot$.","pith_inferences":["The paper does not follow this up, but its parameter grid implies a sharp observable dichotomy: if collision seeding is the main channel, SMBH occupation should be nearly complete in the most compact NSCs and nearly absent in the least compact ones, at fixed host stellar mass — a prediction that can be tested by resolved kinematical surveys of nearby galactic nuclei.","A testable extension is to compare the predicted seed-mass distribution with gravitational-wave merger rates: heavy seeds of $\\sim10^5\\,M_\\odot$ formed late would leave a distinctive imprint on the mass spectrum of binary black hole mergers detectable by future gravitational-wave observatories.","The model's reliance on $\\epsilon_r$ could be turned into a measurement: fitting the observed local black-hole mass function with $\\epsilon_r$ as a free parameter would constrain how much early expansion typical NSC progenitors undergo, connecting the seeding channel to cluster-formation theory.","If the overprediction below $10^8\\,M_\\odot$ is real rather than an artifact of the comparison sample, it suggests that many low-mass galaxies should contain black holes of $10^5$–$10^7\\,M_\\odot$ that current surveys might have missed; targeted X-ray or variability searches in dwarf galaxies would test this."],"forward_implications":["If the channel is correct, the low-mass end of the SMBH population is overproduced by the model: below $10^8\\,M_\\odot$ the predicted mass function exceeds the Vika et al. (2009) data by factors of 10–100, so other physics (feedback, accretion suppression, or additional seeds) must trim that population.","The heaviest seeds ($\\sim10^5\\,M_\\odot$) require the most compact clusters ($\\epsilon_r=0.1$) and form late in these halos, at $z\\sim0.35$; accordingly the channel contributes mainly to the local SMBH population rather than to the $z\\sim6$ quasar population.","The fraction of galaxies hosting both an NSC and an SMBH is set by $\\epsilon_r$: models with $\\epsilon_r=0.1$ give occupation fractions of 54–90%, while $\\epsilon_r=0.5$ gives a few percent, making NSC compactness a predictor of SMBH occupation.","The predicted scaling relations $M_{\\rm NSC}$–$M_{\\rm galaxy}$ and $M_{\\rm NSC}$–$\\sigma$ overlap the observed ones regardless of $A_{\\rm res}$, implying the in-situ channel alone can reproduce the shape of these relations even though the normalization depends on gas transfer efficiency."],"supporting_citations":[{"why":"Supplies the core framework: SMBHs form from failed NSCs whose average collision timescale is shorter than the system age.","marker":"Escala 2021"},{"why":"Provides the numerically calibrated critical-mass condition (Eq. 17) and the 10–50% conversion fraction of cluster mass into a central massive object.","marker":"Vergara et al. 2023"},{"why":"Motivates the radius-rescaling parameter $\\epsilon_r$ by showing young massive clusters with initial sizes below 0.3 pc expand by more than a factor of 10.","marker":"Banerjee & Kroupa 2017"},{"why":"Supplies the in-situ NSC formation model and the gas-transfer efficiency $A_{\\rm res}$ of order $10^{-2}$–$10^{-3}$.","marker":"Antonini et al. 2015"},{"why":"Provides the observed NSC radius relation ($r_0=3.3$ pc) and the observational sample of NSC properties used for comparison.","marker":"Neumayer et al. 2020"},{"why":"Gives the GAMA galaxy stellar mass function from which the observed NSC mass function is derived and to which the model is compared.","marker":"Baldry et al. 2012"},{"why":"Provides the observed local SMBH mass function that the model's black-hole mass function is compared against above $10^8\\,M_\\odot$.","marker":"Vika et al. 2009"},{"why":"Introduces Galacticus, the semi-analytic model into which the NSC and BH seeding recipes are implemented.","marker":"Benson 2012"}],"fun_headline_variants":["Collisions in compact star clusters spawn billion-solar-mass black holes","Nuclear star cluster collisions seed supermassive black holes","How dense star clusters give birth to giant black holes","Collision-born seeds explain supermassive black hole population","Compact clusters collide to make black hole seeds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Every predicted black hole assumes that nuclear star clusters were initially a factor 2 to 10 more compact than the present-day radius relation implies, because with no radius rescaling ($\\epsilon_r=1$) no model forms a single seed.","fun_headline_variants_meta":{"raw":{"variants":["Collisions in compact star clusters spawn billion-solar-mass black holes","Nuclear star cluster collisions seed supermassive black holes","How dense star clusters give birth to giant black holes","Collision-born seeds explain supermassive black hole population","Compact clusters collide to make black hole seeds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000591,"raw_usage":{"total_tokens":2936,"prompt_tokens":1272,"completion_tokens":1664,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":888,"completion_tokens_details":{"reasoning_tokens":1586}},"tokens_in":888,"tokens_out":1664,"duration_ms":13709,"temperature":1.0,"reasoning_tokens":1586,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T17:59:56.323700+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the radii of young nuclear star clusters at high redshift (or infer their initial radii from the present-day expansion of young massive clusters) and check whether compact clusters with $r_{\\rm NSC}\\lesssim0.1$ pc exist; if no NSC population is 2–10 times more compact than the local $r_{\\rm NSC}$–$M_{\\rm dyn}$ relation at the epochs when seeds must form, the critical-mass condition $M_{\\rm NSC}\\ge M_{\\rm crit}$ is essentially never met and the predicted black-hole population collapses.","supporting_citations":[{"cited_title":"P., Graham, A","cited_arxiv_id":null,"evidence_quote":"Provides the observed local SMBH mass function that the model's black-hole mass function is compared against above $10^8\\,M_\\odot$."}],"review_version":1}