{"id":"00f6b95a-4c43-48fa-a3be-dc58a2d07a6c","arxiv_id":"2505.23768","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Supersonic turbulence, with Mach numbers 1.8 to 4.2 scaling with halo mass, is common in 15 simulated minihalos and fragments their central gas into Jeans-unstable clumps.","lead":"Simulations of 15 primordial minihalos show that gas falling into dark matter halos at redshift 17-20 becomes supersonically turbulent, fragmenting the star-forming cloud into multiple clumps. This suggests turbulence is common in the first star-forming environments and may lower the masses of the first stars.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The simulated accretion flows are seeded from unresolved TNG50-1 minihalos: splitting one-particle halos cannot create the physical infall that is claimed to drive turbulence and fragmentation.","rationale":"The reader's weakest assumption concerned the 10 ckpc boundary and truncation of the accretion streams. My stress-test identifies a related but more fundamental problem: even with perfect boundary conditions, the initial conditions do not contain the physical information needed to produce the claimed result. TNG50-1 at z=20 resolves the large-scale environment, but the minihalos themselves are at or below the mass resolution: a 4.9e5 Msun halo corresponds to one dark-matter particle. Particle splitting is a valid technique for refining an already-resolved fluid element, but it cannot generate the small-scale density and velocity perturbations that are absent in the parent simulation. The turbulence and fragmentation observed in the re-simulation could therefore be controlled by the splitting algorithm and numerical relaxation instead of by gravitational infall from the cosmic web. This is a correctness risk, not a matter of disagreement with prior work, and it is directly testable. I do not think it invalidates the paper outright; a proper high-resolution zoom of the same halos would settle the issue, and the authors may also add resolution-convergence and random-seed tests. Because the reader already recommended conditional acceptance pending tests, my analysis reinforces that verdict without moving it, hence UNCHANGED.","tokens_in":11296,"tokens_out":9739,"duration_ms":112403,"concrete_test":"Re-simulate at least one low-mass halo (e.g., halo A) and one high-mass halo (e.g., halo O) using genuine high-resolution cosmological zoom-in initial conditions generated from the TNG50-1 z=127 initial conditions with the same Lagrangian patch (e.g., via MUSIC), including full cosmological expansion and external tides, and compare the Mach-number distribution, the driving scale of the kinetic-energy spectrum (Section 3.3), and the clump masses in Table 2 with the split-TNG runs. If the proper zoom does not produce supersonic turbulence with similar Mach numbers and clump masses, the reported signal is an artifact of splitting unresolved TNG particles. A cheaper partial check is to rerun one halo with a different particle-splitting random seed; if clump masses change by more than a factor of two, the fragmentation is numerically seeded.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that infall-driven supersonic turbulence is common in minihalos and fragments the first star-forming clouds—requires the simulated accretion flows to be physical. Sections 2.2–2.3 do not establish this. TNG50-1 has gas and dark-matter mass resolutions of about 8.4e4 Msun and 4.5e5 Msun, respectively, yet the selected halos in Table 1 have virial masses as low as 4.9e5 Msun with dark-matter masses of 4.28e5–5.49e5 Msun for halos A and B: these are one-particle halos in the parent simulation. The super-Lagrangian splitting increases resolution by a factor of ~1e5, but splitting an unresolved particle into many particles cannot introduce new density or velocity structure; it only re-samples a smooth kernel. The anisotropic, filamentary accretion that is claimed to drive turbulence may therefore be seeded by the splitting geometry, random sub-particle offsets, and subsequent numerical relaxation rather than by true cosmic-web accretion. No resolution-convergence test is presented, and the paper does not state whether gravitational softening is reduced after splitting; if TNG softening is retained, the dark-matter potential is unresolved at the virial radii of 0.1–0.3 kpc. This directly undermines the characteristic Mach numbers 1.8–4.2 and the 2.6–66.5 Msun Jeans-unstable clumps that form the basis of the paper's headline conclusions.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents GIZMO meshless-finite-mass zoom-in simulations of 15 minihalos with virial masses ~4.9e5–7.7e6 Msun at z ~ 17–20, initialised by extracting ~10 ckpc comoving spheres from the TNG50-1 simulation and applying a particle-splitting technique that raises the nominal mass resolution to 0.19 Msun for gas and 80.9 Msun for dark matter. The simulations include non-equilibrium primordial chemistry and cooling via GRACKLE. The authors report that gas infall drives predominantly supersonic turbulence with characteristic Mach numbers 1.8–4.2, increasing with halo mass, that the kinetic energy spectra are broadly Kolmogorov-like, and that the turbulent clouds fragment into multiple Jeans-unstable clumps with masses 2.6–66.5 Msun, which they interpret as the seeds of Pop III star formation.","tokens_in":11614,"tokens_out":5892,"duration_ms":54531,"significance":"If the accretion flows are physical, the paper would make a valuable contribution by extending a single-halo study to a 15-halo sample and by proposing infall-driven turbulence as a generic mechanism that fragments primordial star-forming clouds and lowers the expected Pop III mass scale. The study uses a well-tested code and a detailed primordial cooling network, and it is honest about the fact that the runs are stopped before actual star formation. However, because the parent halos are at or below the mass resolution of TNG50-1 and because no resolution or boundary convergence tests are provided, the quantitative conclusions are not yet established. As it stands, the paper is a suggestive numerical exploration rather than a demonstrated generic result.","major_comments":[{"comment":"The TNG50-1 dark matter mass resolution is ~4.5e5 Msun, while halo A has a total virial mass of 4.90e5 Msun and a dark matter mass of 4.28e5 Msun, and halo B has a dark matter mass of 5.49e5 Msun. Halos A and B are therefore represented by at most one dark matter particle in the parent simulation. Splitting a single particle into many sub-particles cannot generate the anisotropic, filamentary accretion flows that the paper claims drive the turbulence; it only samples a smooth kernel with random offsets, and any small-scale structure that appears afterwards may be a numerical relaxation artifact. The authors should show, e.g., by re-simulating with a parent simulation that resolves the halos or by testing different splitting seeds and comparing against a proper zoom-in with full cosmological boundary conditions, that the infall pattern and the resulting Mach numbers are physical. This is load-bearing for the central claim because the headline numbers (M 1.8–4.2, clump masses 2.6–66.5 Msun) are produced by these initial conditions.","section":"Section 2.3 and Table 1"},{"comment":"The paper states only that \"we extract a comoving spherical volume of radius ~10 ckpc from IllustrisTNG\" and map it directly onto GIZMO as zoom-in initial conditions. It does not state how the outer boundary of this volume is treated, whether the tidal field from surrounding large-scale structure is retained, or whether cosmological expansion continues after extraction. Since the accretion streams that are claimed to drive the turbulence enter through this boundary, truncation could either suppress or artificially focus the infall. Please specify the boundary conditions and provide a test (e.g., varying the extraction radius or including the surrounding tidal field) showing that the accretion pattern and turbulence statistics are unchanged.","section":"Section 2.2"},{"comment":"No resolution convergence study is presented. The particle-splitting algorithm introduces random directions for the child-particle offsets, and the manuscript does not state the gravitational softening lengths used after splitting or whether the original TNG softening is retained. For a paper whose main results are quantitative (the Mach number–mass relation, the clump mass function, and the kinetic energy spectrum slope), the authors should demonstrate convergence with respect to particle mass, splitting seed, and softening. Without such tests, the numerical robustness of the reported values is unverified.","section":"Section 2.3"},{"comment":"The simulations stop when the timestep collapses, and the text acknowledges that \"to follow the collapse further, higher resolution or the introduction of sink particles is necessary.\" The conclusion that the identified Jeans-unstable clumps will \"collapse imminently to form stars\" is therefore an extrapolation beyond what is simulated. Please either soften the claim to \"candidate prestellar cores\" or justify the extrapolation by showing that the clumps remain Jeans-unstable over several dynamical times and that the collapse is not a numerical artifact at the resolution limit.","section":"Section 3.5"}],"minor_comments":[{"comment":"The caption contains a duplicated word and a typo: \"Gas accretion accretion onto minhalos\" should be \"Gas accretion onto minihalos.\"","section":"Figure 4 caption"},{"comment":"The text says \"The read line shows the best fit profile,\" which should be \"red line,\" and the fit parameters should be reported in the text or caption.","section":"Figure 6"},{"comment":"The text defines the characteristic Mach number as the peak of the mass fraction versus Mach number distribution, but Figure 6 appears to plot a mean Mach number; please clarify which quantity is used and how it is computed.","section":"Section 3.4"},{"comment":"The Jeans mass is said to follow \"Equation 2 in Chen et al. (2024)\", but no definition is given in this paper; include the expression for M_J so the table is self-contained.","section":"Table 2"},{"comment":"There are two papers involving \"Chen, K.-J.\" with 2024 and 2025, and the text cites both \"Tung & Chen (2024)\" and \"Tang & Chen (2024)\" in different places; please verify the names and avoid ambiguity (also note that the reference list contains duplicate Hirano et al. 2014 entries).","section":"References"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper gives a first systematic look at turbulence in 15 minihalos, using TNG50-1 initial conditions re-simulated with GIZMO after a ~1e5 particle split. The new result is the sample and the correlation: characteristic Mach numbers run 1.8–4.2 and increase with halo mass, and the central clouds fragment into multiple clumps in the 2.6–66.5 Msun range. That is a genuine extension of the single-halo study by Chen et al. (2025), and the power-spectrum analysis showing a Kolmogorov-like inertial range is a nice piece of evidence.\n\nWhat the paper does well: it is clearly written, uses standard primordial gas chemistry (GRACKLE), a standard clump finder, and the authors are upfront about the relationship to the companion work. The driving scale of a few virial radii found from the spectra is a plausible and interesting claim.\n\nThe soft spots are serious and they sit at the base of the argument. The smallest halos in Table 1 are unresolved in TNG50-1: halo A has a dark matter mass of 4.28e5 Msun against a resolution of ~4.5e5 Msun, i.e. less than one particle. Splitting a single particle into ~1e5 pieces does not create the anisotropic filamentary inflow that the turbulence interpretation requires; it re-samples a smooth kernel. The paper gives no resolution convergence test, no statement about how the 10 ckpc extraction boundary is handled, and no mention of whether cosmological expansion remains active after the initial conditions are mapped. Without those, the Mach number–mass trend, especially at the low-mass end, could be partly a numerical artifact. This is not a small worry in a footnote; it is the basis for the headline claim.\n\nA second, minor issue: the abstract says the clumps are 'soon collapsing to form the first stars,' but the simulations stop when the timestep collapses. The clump masses are upper limits, as the authors note in the methods, but the abstract and conclusion do not always carry the caveat.\n\nThere is no reason to suspect the authors are hiding anything; the dependence on companion papers is transparent. The need is for verification, not a rewrite.\n\nWho should read this? Anyone working on Population III star formation or minihalo zoom-ins. If the initial-condition problem can be fixed with a convergence study or a re-run starting from a better-resolved parent simulation, this would be a solid contribution. Right now, I would not cite the central claim in my own work.\n\nRecommendation: worth peer review, but only with referees who will push hard on the unresolved-halo issue and the boundary treatment. That is the difference between an incremental but useful paper and a misleading one.","headline":"A plausible but unverified extension of the single-halo turbulence claim to 15 minihalos; the unresolved TNG initial conditions and missing convergence tests leave the central result shaky, yet the paper deserves peer review with demand for verification.","tokens_in":12136,"tokens_out":4204,"would_cite":false,"duration_ms":41128,"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":"Supersonic turbulence, driven by infalling gas, fragments the first star-forming clouds into collapsing clumps.","keywords":["Cosmology","Population III stars","Turbulence","Star formation","Metal-poor stars","Supersonic turbulence","minihalos","primordial gas"],"falsifier":"Re-run the same 15 minihalos with the same mass resolution while keeping them embedded in the full cosmological environment, including external tidal fields and continued cosmic expansion, and compare the mass-weighted Mach numbers and the number of bound Jeans-unstable clumps; if supersonic turbulence at Mach $\\sim2$--$4$ and the clumpy fragmentation do not appear, the claim that this is a common feature of minihalos fails.","tokens_in":11102,"feed_emoji":"🌌","tokens_out":12334,"duration_ms":98323,"temperature":0.7,"pith_summary":"Using high-resolution zoom-in simulations of 15 minihalos at redshift $z\\sim17$--$20$, this paper argues that supersonic turbulence is a common and natural product of minihalo assembly. Infalling gas does not settle into a smooth spherical cloud; anisotropic accretion streams collide and shear, generating characteristic Mach numbers between about 1.8 and 4.2 that grow with halo mass. The turbulence fragments the central primordial gas into clumpy filaments, and the densest bound clumps, with masses from 2.6 to 66.5 $M_\\odot$, exceed their Jeans masses and are collapsing to form the first stars. If correct, the birth environment of Population III stars is a turbulent, multi-core cloud rather than a single collapsing disk, with direct consequences for the masses and initial mass function of the first stars.","feed_headline":"Supersonic turbulence fragments the first star-forming clouds","feed_subtitle":"Fifteen minihalos at z≈17–20 show Mach 2–4 infall that breaks each central cloud into collapsing clumps.","key_machinery":"The driving mechanism is infall-driven turbulence: gas streams falling into a minihalo's dark matter potential well arrive anisotropically, collide, and convert gravitational binding energy into a supersonic velocity field. The numerical enabler is a particle-splitting refinement that multiplies the mass resolution of the parent cosmological initial conditions by a factor of $\\sim10^5$, reaching gas particles of $\\sim0.2\\,M_\\odot$ and dark matter particles of $\\sim80\\,M_\\odot$, so that accretion streams and the resulting eddies are actually resolved. A clump finder that traces isodensity contours then isolates the dense, Jeans-unstable cores whose masses set the predicted star-formation sites.","core_discovery":"The central discovery is that the first star-forming gas clouds are shaped by supersonic turbulence that arises from gravitational infall, not by quiescent cooling flows. Across 15 minihalos with virial masses from $4.9\\times10^5$ to $7.7\\times10^6\\,M_\\odot$ at $z\\sim17$--$20$, the characteristic Mach number of the halo gas is 1.8--4.2 and increases with halo mass; the kinetic energy spectra follow a Kolmogorov $k^{-5/3}$ cascade with energy injected at scales of roughly three to five virial radii. The turbulence breaks the central cloud into multiple dense clumps, and the most massive gravitationally bound clump in each halo has a mass of 2.6--66.5 $M_\\odot$, always above its Jeans mass. The authors present this as evidence that earlier idealized simulations underestimated the role of turbulence because they lacked resolution on the 100--1000 pc scales where accretion flows develop.","pith_inferences":["The paper stops when the first bound clump begins collapsing, so the mapping from clump mass to final stellar mass is not tested; follow-up runs with sink particles could reveal whether each clump yields one star or several, and whether later accretion pushes masses upward.","If infall-driven turbulence is as generic as claimed, the first-star initial mass function may be set by the turbulent fragmentation scale rather than by the thermal Jeans mass alone, which would change predicted supernova yields and the chemical fingerprints preserved in extremely metal-poor stars.","A testable extension is to embed the same extraction spheres in the full cosmological volume, with external tides and continuing expansion, and check whether the Mach numbers and clump statistics survive the change of boundary conditions.","The same refinement strategy could be applied to more massive, atomic-cooling halos to see whether infall-driven turbulence persists in environments thought to produce direct-collapse black holes."],"forward_implications":["Supersonic turbulence (Mach $\\sim2$--$4$) is the rule, not the exception, in minihalos at $z\\sim17$--$20$, with more massive halos driving stronger turbulence.","The central star-forming cloud fragments into multiple clumps instead of one disk, so a single minihalo can host several simultaneous or sequential Pop III stars.","Because the bound clump masses ($2.6$--$66.5\\,M_\\odot$) sit below the $40$--$500\\,M_\\odot$ range of classic one-star-per-halo models, the predicted characteristic mass of the first stars shifts downward.","Each halo still contains roughly $10^2$--$10^3\\,M_\\odot$ of gas above the $10^5\\,\\mathrm{cm^{-3}}$ collapse threshold, so the total mass budget for the first stars is preserved even as it is split among several cores."],"supporting_citations":[{"why":"This is the single-halo study this paper extends; it first reported supersonic turbulence emerging in minihalo formation.","marker":"Chen et al. (2025)"},{"why":"This supplies the hydrodynamics method and the particle-splitting refinement used to reach the target mass resolution.","marker":"Hopkins (2015)"},{"why":"This provides the cosmological simulation suite from which the 15 minihalo initial conditions are drawn.","marker":"Nelson et al. (2018)"},{"why":"This supplies the non-equilibrium primordial chemistry and cooling network used to model the gas thermodynamics.","marker":"Smith et al. (2017)"},{"why":"This provides the analysis toolkit and clump-finder routine used to identify the dense star-forming clumps.","marker":"Turk et al. (2011)"},{"why":"This is an earlier simulation baseline that reported subsonic turbulence, which the paper argues suffered from insufficient large-scale resolution.","marker":"Abel et al. (1998)"},{"why":"This supplies the Jeans-mass formula used to show that the identified clumps exceed their Jeans masses.","marker":"Chen et al. (2024)"},{"why":"This predicts clumpy structures in turbulent primordial clouds, providing the morphological comparison for the clumps found here.","marker":"Tang & Chen (2024)"}],"fun_headline_variants":["First stars born from turbulent, clumpy gas clouds","Supersonic turbulence shapes the first star-forming halos","Simulations show turbulence fracturing primordial gas clouds","Mach 2-4 winds mark the first star-forming clouds","Infall-driven turbulence seeds clumpy first stars"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole picture rests on the assumption that the 10 comoving kiloparsec gas spheres cut from the parent cosmological run and re-evolved in isolation faithfully reproduce the real accretion streams; if the missing outside gravity and cosmic expansion change the inflow, the measured turbulence and fragmentation could be artifacts of the truncation.","fun_headline_variants_meta":{"raw":{"variants":["First stars born from turbulent, clumpy gas clouds","Supersonic turbulence shapes the first star-forming halos","Simulations show turbulence fracturing primordial gas clouds","Mach 2-4 winds mark the first star-forming clouds","Infall-driven turbulence seeds clumpy first stars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000297,"raw_usage":{"total_tokens":1778,"prompt_tokens":1057,"completion_tokens":721,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":673,"completion_tokens_details":{"reasoning_tokens":644}},"tokens_in":673,"tokens_out":721,"duration_ms":6411,"temperature":1.0,"reasoning_tokens":644,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:38:09.762512+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the same 15 minihalos with the same mass resolution while keeping them embedded in the full cosmological environment, including external tidal fields and continued cosmic expansion, and compare the mass-weighted Mach numbers and the number of bound Jeans-unstable clumps; if supersonic turbulence at Mach $\\sim2$--$4$ and the clumpy fragmentation do not appear, the claim that this is a common feature of minihalos fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This supplies the hydrodynamics method and the particle-splitting refinement used to reach the target mass resolution."},{"cited_title":"The Formation and Fragmentation of Primordial Molecular Clouds","cited_arxiv_id":"astro-ph/9810215","evidence_quote":"This is an earlier simulation baseline that reported subsonic turbulence, which the paper argues suffered from insufficient large-scale resolution."},{"cited_title":"J., et al","cited_arxiv_id":null,"evidence_quote":"This supplies the Jeans-mass formula used to show that the identified clumps exceed their Jeans masses."},{"cited_title":"Clumpy Structures within the Turbulent Primordial Cloud","cited_arxiv_id":"2303.00751","evidence_quote":"This predicts clumpy structures in turbulent primordial clouds, providing the morphological comparison for the clumps found here."}],"review_version":1}