{"id":"cc0f0691-6f98-4663-972b-8e33d015b59e","arxiv_id":"2507.02058","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Cosmological zoom-in simulations find that 10^5 solar mass Pop III.1 black hole seeds reach ~10^7 solar masses by z=8, with AGN feedback, especially radiation, regulating growth and launching fast outflows.","lead":"New computer simulations show that black hole seeds born from the first stars can grow to ten million solar masses by cosmic dawn, even while AGN feedback is active. The result may explain why the James Webb Space Telescope sees black holes that are over-massive compared to their small host galaxies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bondi-Hoyle accretion with no sub-grid angular momentum is the load-bearing assumption: if rotational support at the sink kernel suppresses inflow, the ~1e7 Msun final masses at z=8 are not robust.","rationale":"I agree with the reader's weakest assumption: the Bondi-Hoyle prescription dominates the growth history and is therefore the most load-bearing element of the central claim. The mass scale reached by z=8 in all models is set by this accretion rate; feedback only modulates it, and the lack of an angular-momentum barrier is a well-known reason Bondi rates in galaxy simulations overestimate true nuclear accretion. The paper's self-reported statement that the Bondi radius is resolved is questionable for the early 1e5 Msun seed, whose Bondi radius can be at or below the cell scale, and the absence of a convergence or sink-radius test makes the quantitative claim '~1e7 Msun by z=8' unsettled. The proposed check would directly probe whether unresolved rotational support suppresses the growth. Since the reader's CONDITIONAL verdict already flags this assumption, no change in verdict is needed.","tokens_in":30447,"tokens_out":7092,"duration_ms":87963,"concrete_test":"From the ThermKinRad run, at the z ~ 8.5 super-Eddington peak, compute the mass-weighted specific angular momentum j of the 2109 cloud particles relative to the BH and compare it with j_Bondi = sqrt(G M_BH r_Bondi). If j_cloud/j_Bondi is large, unresolved angular momentum is important. Then rerun ThermKinRad (a) with maximum refinement increased by one level (Δx ~ 0.8 pc physical at z=8) and (b) in a separate run with an accretion limiter that reduces the Bondi rate by the ratio j_Bondi/j_cloud or a standard sub-grid disk prescription. If the BH mass at z=8 drops by more than ~0.5 dex in either run, the central growth claim is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2 (SMBHs Accretion) computes the Bondi-Hoyle rate (Eqs. 4-9) from gas properties averaged over cloud particles inside r_cloud = 4Δx, spanning roughly a few to ~40 pc depending on redshift, with no sub-grid angular momentum, viscous torque, or unresolved disk model. In the feedback-regulated runs the accretion rate is neither artificially boosted nor Eddington-capped, so the quoted growth is a direct prediction of the Bondi formula. For a 10^5 Msun seed, the Bondi radius at c_s ~ 10-30 km/s is only ~0.5-4 pc, comparable to or below the maximum-refinement cell size, making the paper's statement that 'the spatial resolution ... is sufficient to capture the Bondi radius' at best marginal and only for the most massive seeds. High-z gas is turbulent and often rotationally supported; the specific angular momentum at the sink-kernel radius can be much larger than the angular momentum at the Bondi radius, in which case gas would circularize into a disk instead of accreting at the Bondi rate. Since all four self-regulated models and their final BH masses scale directly with this rate, a factor-of-several overestimate of the accretion rate would pull the final masses below the claimed ~1e7 Msun by z=8. The paper offers no convergence test, no sink-radius sensitivity study, and no angular-momentum-aware accretion limiter, leaving the central growth claim resting on an untested sub-grid assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"NA","tokens_in":30775,"tokens_out":7012,"duration_ms":82870,"significance":"If the central growth claim is robust, this is a valuable contribution: it provides a physically motivated pathway from Pop III.1 seeds to the overmassive BHs seen by JWST at z>8, and it isolates the role of radiative AGN feedback in driving galaxy-scale outflows. Strengths include the systematic variation of feedback physics in a consistent code framework, self-consistent modeling of the Pop III.1 progenitor and its preheating effect, a freely moving BH with resolved dynamical friction, and direct comparison to current JWST samples. The growth predictions are not circular: the feedback efficiencies and seed mass are not tuned to the JWST data. However, the conclusions rest on a single halo and on a Bondi-Hoyle accretion prescription with no angular-momentum-aware subgrid model, and the paper lacks resolution or convergence tests. These gaps make the quantitative claims (e.g., ~10^7 Msun by z=8, outflow velocities and extents) less secure than the presentation suggests.","major_comments":[{"comment":"The Bondi-Hoyle accretion prescription is the load-bearing element for the central growth claim, but it contains no subgrid angular-momentum limiter and the claim that the resolution is sufficient to capture the Bondi radius is only marginal. For a 10^5 Msun seed, the Bondi radius is ~4 pc for c_s = 10 km/s and ~0.5 pc for c_s = 30 km/s, while the physical cell size at z~8 is about 1.6 pc and r_cloud = 4 Delta x is only a few pc. If gas at the sink-kernel scale is rotationally supported, the true accretion rate can be much lower than the Bondi rate, and a factor-of-several reduction would pull the final masses below the claimed ~10^7 Msun by z=8. The paper provides no convergence study, no sink-radius sensitivity test, and no comparison with an angular-momentum-aware accretion model. I ask the authors to include at least one resolution test and one test with a rotational-support limiter (e.g., the prescriptions of Rosas-Guevara et al. 2015 or Tremmel et al. 2017) to establish that the central growth result is not an artifact of the unresolved accretion model.","section":"Sec. 2, Eqs. (4)-(9)"},{"comment":"The outflow claim is attributed to 'AGN-driven radiation pressure' (Sec. 3.4, Fig. 5 caption), but Sec. 2's description of AGN radiation only states that ionizing photon energy is released (Eq. 11, based on Bieri et al. 2017). No radiation-pressure or photon-momentum deposition term is described in the methods. If the code does include radiation pressure, this should be stated explicitly; if it does not, the attribution should be revised to photoionization heating and the resulting pressure gradients, which is the mechanism actually supported by the equations presented. This matters because the fast, kpc-scale outflows in ThermKinRad are a headline result.","section":"Sec. 3.4 and Fig. 5"},{"comment":"All conclusions are drawn from a single zoom-in halo, and the paper acknowledges (Sec. 3.1) that differences among the four feedback-regulated models are 'comparable to those arising from stochastic effects.' Without at least one resolution test and ideally a second halo or an explicit discussion of halo-to-halo variance, the broad claims 'across all our models' and the quantitative agreement with JWST samples in Figs. 3 and 9 are not fully supported. This is not a request for a full statistical sample, but the lack of any convergence test leaves the central claim vulnerable to numerical resolution effects.","section":"Sec. 2 (Initial conditions) and Sec. 3.1"}],"minor_comments":[{"comment":"The abstract states that seeds reach ~10^7 Msun by z=8 across all models, but the Edd_lim model reaches ~5x10^8 Msun by z~9, and the NoAGN and Therm runs are stopped at z~11, so their z=8 values are not shown. Please qualify the statement to reflect the actual model sample and final redshift.","section":"Abstract and Sec. 3.1"},{"comment":"The model names are used inconsistently: ThermHEKin/ThermHEKinRad vs ThermHKin/ThermHKinRad, and ThermKinRT vs ThermKinRad. Please adopt a single naming convention in the text, tables, figures, and captions.","section":"Throughout"},{"comment":"The quoted cell width of ~14 pc and cloud radius of ~40 pc should specify whether they are comoving or physical; this is directly relevant to the Bondi-radius resolution argument. At z=8, the physical cell size is ~1.6 pc, which is important for evaluating the claim in Sec. 2.","section":"Sec. 2"},{"comment":"The caption contains sentence fragments and a typo ('ThemKinRad'); it also lists symbols without defining them in the caption. Please make the caption self-contained and grammatically complete.","section":"Fig. 1 caption"},{"comment":"There is a typo 'sftar formation' in the last paragraph; also, 'ThermHEKinRad' appears where 'ThermHKinRad' or 'ThermHEKinRad' should be used consistently.","section":"Sec. 3.5"},{"comment":"The abstract reports stellar masses of ~10^8.5 Msun, while Sec. 4 states a host galaxy stellar mass of ~5x10^8 Msun; please reconcile these values or clarify which mass (e.g., within 0.15 Rvir vs total) is being quoted.","section":"Abstract and Sec. 4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of MNRAS and addresses a timely topic. The main concern is that the central growth result depends on a single halo and on a Bondi-Hoyle accretion model without angular-momentum regulation, and the current text does not provide the resolution or sensitivity tests needed to assess the robustness of the claim. The Pop III.1 progenitor modeling is cited to 'Sanati et al. submitted' and 'in prep'; the editor may wish to verify that these references are available and contain the necessary details. No concerns about citation fairness beyond the heavy reliance on the authors' own submitted work."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe paper you asked about is worth a serious look. What's genuinely new is not the claim that heavy seeds can grow fast — that's been argued before — but the systematic decomposition of AGN feedback modes in a radiative-MHD zoom-in, with self-consistent preheating from a Pop III.1 progenitor. The result that radiative feedback, rather than kinetic winds, is what launches the kpc-scale, >2500 km/s outflows and modestly suppresses star formation is a useful and credible addition. The comparison to JWST overmassive BHs is presented as broad agreement, not as a fitted model, which is the right posture.\n\nThe main soft spot is the accretion prescription. The Bondi-Hoyle rate is computed from gas averaged over a ~40 pc cloud kernel, while the Bondi radius for a 1e5 M_sun seed is sub-pc to a few pc. The paper says the resolution is sufficient to capture the Bondi radius; that is at best marginal and only for the most massive seeds. With no sub-grid angular momentum treatment and no sink-radius or resolution test, the final masses of ~1e7 M_sun could be overestimated by a factor of a few if rotation suppresses inflow. That is a real limitation, but I would not call it fatal: the feedback-regulated models reach 1e7 even with strong feedback, and even a factor of a few reduction puts them at a few 1e6 M_sun, still overmassive relative to their hosts and still consistent with the observed population at roughly the right level.\n\nTwo smaller issues. The claim that seeds reach ~1e7 M_sun 'across all models' is undercut because the NoAGN and Therm runs stop at z~11; they are reference runs, but the wording needs qualification. And the single-halo realization limits the population-level statements that peek through in the discussion. No convergence tests, no multiple halos, no public data. Those are all things a referee should ask for.\n\nIf the choice is desk rejection or a serious referee, I'd send it to review. The simulation effort is substantial, the feedback comparison is systematic, and the qualitative conclusions are probably right. The referee should push on the accretion prescription and ask for at least one resolution test and the data behind the key mass-growth curves.\n\nBest,\n[Your name]","headline":"A solid simulation study with a genuinely useful feedback decomposition; the masses rest on an untested Bondi-Hoyle assumption and a single halo, but the qualitative conclusions should survive a serious referee.","tokens_in":31361,"tokens_out":3936,"would_cite":true,"duration_ms":45068,"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":"In cosmological simulations, 10^5 solar-mass seeds of Population III.1 origin grow to roughly 10^7 solar masses by z=8 under every AGN feedback model tested, with radiative feedback driving outflows beyond 2500 km/s.","keywords":["supermassive black hole growth","Population III.1 stars","AGN feedback","cosmological zoom-in simulations","radiative transfer","high-redshift galaxies","super-Eddington accretion","overmassive black holes"],"falsifier":"Repeat the most complete feedback simulation with an accretion model that accounts for angular momentum transport at scales below about 40 pc, for example by limiting accretion to the rate at which a resolved nuclear disk can lose angular momentum; if the black hole then ends z=8 below roughly $10^{7}$ solar masses, the central claim is refuted. An observational counterpart would be long-cadence monitoring of z~8 overmassive black holes: if sustained super-Eddington accretion, rather than short episodic bursts, is required to explain their masses, the simulated feedback cycle is disfavoured.","tokens_in":30248,"feed_emoji":"🕳️","tokens_out":5119,"duration_ms":51149,"temperature":0.7,"pith_summary":"This paper asks whether active galactic nucleus (AGN) feedback prevents the first heavy black hole seeds from growing into supermassive black holes by the time the Universe is about 600 million years old. Using high-resolution zoom-in cosmological simulations, the authors start with a $10^{5}$ solar-mass seed left behind by a Population III.1 star and switch on AGN feedback in progressive stages: none, Eddington-limited thermal, thermal plus kinetic winds, and finally radiation. In every feedback model the seed reaches a mass near $10^{7}$ solar masses by redshift 8, so feedback regulates but does not stop the growth. The simulation with radiation feedback also launches fast, mass-loaded outflows extending to about 50 kpc, and the predicted overmassive black holes match moderate-luminosity quasars observed by JWST. The result matters because it offers a concrete path from heavy seeds to the supermassive black holes already seen in the early Universe.","feed_headline":"Black hole seeds reach 10^7 solar masses by z=8 despite AGN feedback","feed_subtitle":"Simulations show feedback regulates, not prevents, early supermassive growth, matching JWST's overmassive quasars.","key_machinery":"The argument is carried by a suite of zoom-in cosmological simulations that follow a single dark matter minihalo from z=100 to z=8 with roughly 14 pc maximum cell size, resolving the Bondi radius around the black hole. The seed is a $10^{5}$ solar-mass sink particle whose progenitor is a Population III.1 star; the star's ionizing feedback is modelled self-consistently before collapse, capturing the preheating that sets the initial accretion environment. Black hole growth uses the Bondi-Hoyle-Lyttleton rate computed from cloud particles within roughly 40 pc, and AGN feedback is injected as thermal energy, kinetic winds, and radiation through a radiative transfer scheme with three photon groups. The decisive comparison is across model variants that build up feedback components one at a time, isolating which mechanism regulates accretion and which launches outflows.","core_discovery":"The central claim is that heavy seeds of $10^{5}$ solar masses, formed from the collapse of Population III.1 stars after their ionizing radiation preheats the host minihalo, grow efficiently to roughly $10^{7}$ solar masses by z=8 regardless of how AGN feedback is modelled. The growth is not a single sustained burst: it proceeds through cycles in which super-Eddington accretion episodes, reaching up to about 86 times the Eddington rate, trigger AGN feedback that temporarily suppresses accretion before the next inflow. Across the feedback-regulated models the black hole-to-stellar mass ratio stays between 0.01 and 1, consistent with the overmassive black holes that JWST finds in low-mass galaxies at z>4. The authors conclude that AGN feedback, especially the radiative component, shapes the host galaxy by driving high-velocity winds and slightly suppressing star formation, while allowing the seed to reach supermassive scales within the reionisation era.","pith_inferences":["If the Bondi prescription overestimates the gas supply because unresolved angular momentum keeps gas in a disc around the black hole, the same seeds would grow more slowly; a direct test would be to repeat one run with accretion limited by resolved angular momentum transport and check whether 10^7 solar masses by z=8 survives.","The preheating result extends beyond Pop III.1 seeds: any formation pathway in which the host halo is heated before the black hole forms, such as externally irradiated direct-collapse halos, should show a similar short delay followed by rapid Eddington-limited growth.","The predicted outflow velocities and mass-loading factors give a concrete target for future observations: the fastest outflows should appear in galaxies where the black hole is undergoing a post-merger super-Eddington episode.","Because the simulations reach 10^7 solar masses by z=8, the same model could be run forward to test whether these black holes end up as the central engines of local dwarf galaxies or grow into the seeds of more massive systems."],"forward_implications":["Seeds of 10^5 solar masses reach roughly 10^7 solar masses by z=8 even with strong AGN feedback, so no artificially boosted accretion is needed to explain moderate-luminosity high-redshift quasars.","AGN feedback regulates black hole growth through high-accretion, strong-feedback, low-accretion cycles rather than shutting it off, keeping accretion near the Eddington limit.","Radiative feedback, not thermal or kinetic feedback alone, is required to launch the fastest outflows, which exceed 2500 km/s and extend to about 50 kpc.","The simulated black-hole-to-stellar-mass ratios, from 0.01 to 1, reproduce the overmassive black holes seen by JWST and deviate from local scaling relations in the same direction.","Super-Eddington episodes are short-lived and often merger-triggered, leaving average Eddington ratios of about 0.2 to 0.3 over the full growth history."],"supporting_citations":[{"why":"Supplies the adaptive mesh refinement code used to run the cosmological simulations.","marker":"Teyssier 2002"},{"why":"Supplies the black hole sink particle formation and accretion implementation adapted for the seeds.","marker":"Dubois et al. 2010"},{"why":"Supplies the radiative transfer scheme used to propagate stellar and AGN ionizing radiation.","marker":"Rosdahl et al. 2013"},{"why":"Supplies the dual-mode quasar/radio AGN feedback framework that the paper extends.","marker":"Dubois et al. 2012"},{"why":"Provides the method for releasing AGN radiation into the radiative transfer photon bins.","marker":"Bieri et al. 2017"},{"why":"Provides the mechanical supernova feedback model and motivates the resolved escape of ionizing radiation.","marker":"Kimm & Cen 2014"},{"why":"Provides the cosmological framework for Pop III.1 heavy seed formation in isolated minihalos.","marker":"Banik et al. 2019"},{"why":"Supplies the JADES observational sample of high-redshift infant black holes used for comparison.","marker":"Maiolino et al. 2023"},{"why":"Supplies observed quasar progenitor masses at z>8 used for comparison with the simulated black hole masses.","marker":"Larson et al. 2023"}],"fun_headline_variants":["Black hole seeds reach 10^7 solar masses by z=8 despite feedback","Super-Eddington bursts drive 2500 km/s winds, yet seeds grow to 10^7","AGN feedback regulates but never halts early black hole growth","JWST overmassive black holes explained by feedback-resistant seed growth"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole result rests on the assumption that gas measured about 40 parsecs from the black hole is actually available to fall in at the computed accretion rate, with no unresolved angular momentum or small-scale feedback withholding it.","fun_headline_variants_meta":{"raw":{"variants":["Black hole seeds reach 10^7 solar masses by z=8 despite feedback","Super-Eddington bursts drive 2500 km/s winds, yet seeds grow to 10^7","AGN feedback regulates but never halts early black hole growth","JWST overmassive black holes explained by feedback-resistant seed growth"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001351,"raw_usage":{"total_tokens":5554,"prompt_tokens":1084,"completion_tokens":4470,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":700,"completion_tokens_details":{"reasoning_tokens":4385}},"tokens_in":700,"tokens_out":4470,"duration_ms":30159,"temperature":1.0,"reasoning_tokens":4385,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:39:18.273370+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the most complete feedback simulation with an accretion model that accounts for angular momentum transport at scales below about 40 pc, for example by limiting accretion to the rate at which a resolved nuclear disk can lose angular momentum; if the black hole then ends z=8 below roughly $10^{7}$ solar masses, the central claim is refuted. An observational counterpart would be long-cadence monitoring of z~8 overmassive black holes: if sustained super-Eddington accretion, rather than short episodic bursts, is required to explain their masses, the simulated feedback cycle is disfavoured.","supporting_citations":[],"review_version":1}