{"id":"4c024d40-6d03-4430-9871-062ea4419154","arxiv_id":"2507.00815","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Star cluster simulations show that spatial clumpiness fades by about 2.5 free-fall times, kinematic clumpiness lingers, and binary stars speed up mass segregation.","lead":"This paper measures how fast young star clusters lose their clumpy structure by analyzing three computer simulations of star-forming clouds. It finds that spatial clumpiness disappears in about 2.5 cloud free-fall times, while motion clumpiness lingers and binary stars accelerate mass segregation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The universal 'spatial substructure erased at ~2.5 tff' claim is not fully established: M2 stops at 2.5 tff and M3 has not lost substructure by its final snapshot, so longer runs or a qualified statement are needed.","rationale":"Good-faith reading: the paper uses state-of-the-art Torch simulations with coupled MHD, N-body dynamics, primordial binaries, and stellar feedback; the substructure statistics (Q, Moran's I) are standard and applied carefully, including binary center-of-mass reduction and outlier clipping. The kinematic claim (substructure persists to the end) is directly supported by the simulated time span. The mass-segregation binary comparison is a secondary claim and, as the reader notes, the controlled nature of the comparison is not demonstrated. However, the most load-bearing element for the abstract's headline is the '2.5 tff' number, and the currently presented evidence for it is truncated: only M1 runs well past the claimed crossing, M2 ends exactly at the crossing, and M3 has not crossed by its final output. The internal tension between the text ('for all clusters') and the figure caption ('lost in both M1 and M2') makes this a concrete, non-manufactured weakness. The fix is straightforward—longer runs or a qualified claim—so the conditional verdict remains appropriate; no change to the reader's verdict is needed. I disagree with the reader's choice of weakest assumption because the substructure timescale, not the binary comparison, is the paper's central quantitative result.","tokens_in":13922,"tokens_out":5372,"duration_ms":53067,"concrete_test":"Extend simulations M2 and M3 to at least 3.5-4.0 tff, or until the blue Q curve for the final main cluster has remained above 0.8 for several consecutive snapshots, then measure the first time Q crosses 0.8 for each simulation. If M3's crossing occurs after 2.5 tff or does not occur, revise the claim to a lower-mass or limited-density statement; if it crosses near 2.5 tff, the universal claim is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim is that spatial substructure is erased on a timescale of approximately 2.5 times the initial free-fall time (abstract; Section 3.1). The supporting evidence in Figure 4 is incomplete. Simulation M1 runs to ~4 tff and clearly crosses the Q = 0.8 threshold near 2.5 tff, but M2 ends at ~2.5 tff, so the crossing is seen only at the last output, with no post-crossing baseline; M3 also ends at ~2.5 tff and, per the figure caption, has not lost its spatial substructure (\"Spatial substructure has been lost in both M1 and M2...\"). The text nevertheless states that the crossing \"roughly coincides with 2.5 free-fall times for all clusters.\" If M3's Q remains below 0.8 at 2.5 tff, the erasure timescale depends on cloud mass or surface density, and the abstract's universal statement is not supported by the data. The claimed timescale may be an artifact of the limited runtime rather than a robust physical result.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses three magnetohydrodynamic plus direct N-body simulations of star-forming giant molecular clouds (M1, M2, M3, from Cournoyer-Cloutier et al. 2024) to follow the assembly of the most massive cluster in each cloud. The authors measure spatial substructure with the Q statistic, kinematic substructure with Moran's I, and mass segregation with a modified Lambda_MSR statistic. They report that spatial substructure is erased at roughly 2.5 initial free-fall times, kinematic substructure persists longer, and primordial binaries enhance and accelerate dynamical mass segregation. The last claim rests on a comparison with the binary-free simulations of Polak et al. (2025).","tokens_in":14233,"tokens_out":6156,"duration_ms":68398,"significance":"If the 2.5 tff spatial-substructure timescale is robust, it gives observers a direct chronological diagnostic: significant spatial substructure in a young cluster would imply that the cluster is still in the first few free-fall times of assembly. The paper also contributes to the debate on whether early dynamical mass segregation can occur, particularly with a realistic primordial binary population. Strengths of the analysis are its use of standard, externally benchmarked statistics, the transparent treatment of binaries for the substructure metrics, and the direct link to Gaia-era observational tests. However, the key quantitative claim is only partially supported by the runs as presented, and the binary comparison is not shown to be a controlled experiment.","major_comments":[{"comment":"The central claim that spatial substructure is erased at approximately 2.5 tff is not supported for all three simulations. In Figure 4, M2 and M3 extend only to about 2.5 tff, so the threshold crossing for M2 is seen only at the final output with no post-crossing baseline, and M3 has not lost its substructure by its last snapshot according to the caption, which states that substructure has been lost in M1 and M2 only. This is in tension with the Section 4 statement that the behavior of the three simulations is similar. The abstract and the text of Section 3.1 should either be qualified to the simulations that actually show the crossing, or the simulations should be extended beyond 2.5 tff.","section":"Section 3.1 / Figure 4 / Abstract"},{"comment":"The conclusion that primordial binaries enhance and accelerate mass segregation depends on the comparison to Polak et al. (2025) being a controlled experiment. The manuscript states that those simulations were run in the same framework without the primordial binary prescription, but it never shows that the initial cloud masses, surface densities, resolutions, feedback implementations, and analysis choices are identical to those used here. If the initial conditions differ, the earlier and stronger mass segregation could be driven by those differences rather than by binaries. Please either provide a side-by-side comparison of the relevant simulation settings or soften the causal claim to a hypothesis.","section":"Section 3.3"},{"comment":"The modified Lambda_MSR calculation never defines what is meant by 'massive stars'. The text says that 25 stars are randomly sampled from the total number of massive stars, but no mass threshold or selection rule is given. Without this definition, the mass-segregation curves in Figure 6 are not reproducible and the reader cannot assess how the statistic was computed. Please specify the selection criterion (for example, stars above a particular mass or the top percentile by mass).","section":"Section 2.5"}],"minor_comments":[{"comment":"There is a typo: 'After is has formed' should read 'After it has formed'.","section":"Section 2.1"},{"comment":"The legend entries in the captions appear to be incomplete or garbled; the authors should state clearly which line is the current most-massive cluster and which line traces the eventual most-massive cluster in the last snapshot.","section":"Figure 4 and Figure 5 captions"},{"comment":"The keyword list contains a typographical artifact ('Star clusters (1567) –' with a double hyphen); this should be cleaned up.","section":"Abstract / Keywords"},{"comment":"The limitation statement that M2 and M3 have not finished forming stars and that cluster assembly is likely incomplete is appropriate, but it directly contradicts the abstract's universal 2.5 tff claim; this should be reconciled in revision.","section":"Section 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest about many of its limitations, but the abstract overstates the universality of the 2.5 tff erasure timescale given that M3 has not yet lost substructure and M2 stops at the crossing. The Polak et al. comparison is the weakest link in the mass-segregation argument; if the authors can demonstrate equivalence of initial conditions, the conclusion would be much stronger. The missing definition of 'massive' in the Lambda_MSR calculation is a straightforward but necessary fix for reproducibility."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a straightforward, transparent application of two standard substructure statistics (Q and Moran's I) to Torch simulations that include primordial binaries. The analysis is careful in the right places: binaries are collapsed to their center of mass before computing the statistics, velocity outlier removal follows Arnold & Wright, and the three projections are checked for orientation effects. That gives me confidence in the evolutionary curves they report. The genuinely new results are the quantitative erasure timescales and the claim that binaries accelerate early mass segregation.\n\nCredit where due: the Q and I evolution plots are clearly presented, the distinction between each snapshot's biggest cluster and the stars in the final cluster is a nice way to separate assembly from observational selection, and the binary reduction test in the mass segregation section is a sensible robustness check. The comparison to Polak et al. is physically motivated: binaries have larger interaction cross sections, so it is plausible they speed up segregation.\n\nThe soft spot is exactly what the stress test flags. The abstract's universal approximately 2.5 tff is not supported by Figure 4. M1 crosses Q=0.8 near 2.5 tff and keeps going. M2 ends almost exactly at 2.5 tff, so there is no post-crossing baseline to show the crossing was not a transient spike. M3, by the authors' own caption, has not lost its spatial substructure by the final snapshot at 2.5 tff. The text says the crossing roughly coincides with 2.5 free fall times for all clusters, which is contradicted by M3. This needs either longer runs, especially for M2 and M3, or a qualified statement that the erasure timescale is only constrained in M1 and may depend on cloud mass or surface density. As is, the abstract overstates the result.\n\nThe mass segregation comparison has a secondary problem: Section 3.3 describes Polak et al. as using the same framework without a primordial binary prescription, but never shows that the initial conditions are otherwise identical. If resolution, cloud mass, or feedback treatment differ, the different segregation timing could come from those differences. That is fixable with one paragraph or table, but right now it is an uncontrolled comparison.\n\nThe audience is the embedded-cluster formation community and anyone interpreting Gaia-based substructure measurements. For that audience, this paper is a useful reference, but the headline timescale should be treated as provisional. It deserves a serious referee. I would send it, and I would want the referee to require a qualified statement about the 2.5 tff claim and details on the Polak comparison before publication.","headline":"Worth a serious referee, but the 2.5 tff spatial erasure timescale is only strictly established for M1; M3 has not lost substructure by its final snapshot.","tokens_in":14727,"tokens_out":4080,"would_cite":true,"duration_ms":44363,"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":"Star clusters lose their spatial clumpiness within about 2.5 free-fall times of their natal cloud, while kinematic substructure lingers, and primordial binaries accelerate mass segregation.","keywords":["star cluster formation","substructure evolution","mass segregation","primordial binaries","Q statistic","Moran's I statistic","hierarchical assembly","free-fall timescale"],"falsifier":"Run the same three initial clouds through the same code with primordial binary formation switched off, holding gas masses, densities, resolutions, feedback implementations, and cluster-selection criteria fixed; if the Lambda_MSR evolution curves match the binary-included runs, the claim that binaries accelerate mass segregation is disproved.","tokens_in":13751,"feed_emoji":"🌌","tokens_out":5803,"duration_ms":60333,"temperature":0.7,"pith_summary":"Star clusters form hierarchically: stars are born in clumps and sub-clusters that later merge, and this process should leave observable traces in young clusters. This paper uses radiation-magnetohydrodynamical simulations with star-by-star N-body dynamics, primordial binaries, and stellar feedback to measure how those traces decay. The central finding is that spatial clumpiness, measured by the Q statistic, disappears in roughly 2.5 initial free-fall times of the cloud, while kinematic substructure, measured by Moran's I, lingers throughout the simulated assembly. The paper also argues that a population of primordial binaries makes dynamical mass segregation stronger and sets in earlier than in simulations without binaries. If correct, the results give observers a clock: seeing spatial substructure means the cluster is still in its first few free-fall times, and seeing kinematic substructure without spatial substructure means assembly is ongoing but nearly complete.","feed_headline":"Star cluster substructure fades in 2.5 free-fall times","feed_subtitle":"Kinematic order lags spatial order, and binary stars make mass segregation arrive earlier and stronger.","key_machinery":"The analysis is carried by three dimensionless statistics. The Q statistic, the ratio of the normalized minimum spanning tree edge length to the mean projected separation, classifies a stellar distribution as substructured (Q < 0.8) or smooth and centrally concentrated (Q > 0.8). Moran's I, a spatial autocorrelation measure computed on the stellar velocities with inverse-distance weights, detects kinematic substructure, with a value near zero or -1/(N-1) indicating a fully mixed population. The modified mass segregation ratio Lambda_MSR compares the mean minimum spanning tree length of randomly sampled 25-star subgroups of the massive stars against random samples of the full population, avoiding the N-dependence of the original ratio. To keep these statistics honest, the authors replace each binary system with a single particle at its center of mass, preventing the anti-correlated orbital motions of companions from masquerading as substructure.","core_discovery":"The paper establishes a clean time-ordering of substructure erasure during cluster assembly. Across three clouds spanning a factor of sixteen in gas mass, the Q parameter of the most massive cluster rises above the smoothness threshold of 0.8 at about 2.5 times the cloud's initial free-fall time, meaning the stars have lost their filament-born clumpiness. Moran's I for the same clusters declines steadily but never reaches the zero value expected for a fully mixed population, asymptoting near I ~ 0.1 by the end of the simulations. Using a modified Lambda_MSR statistic, the authors find mass segregation grows before and around the time of cluster collapse, and it is consistently stronger and earlier than in a comparison simulation set without primordial binaries; they attribute this to binaries acting as larger gravitational targets that eject low-mass stars and redistribute energy. They conclude that primordial binaries enhance and accelerate dynamical mass segregation in young clusters.","pith_inferences":["If the 2.5 free-fall time scaling holds across environments, the Q statistic can be converted into an age estimator for embedded clusters that does not rely on stellar evolution models.","The residual kinematic substructure near I ~ 0.1 may serve as a fossil signature of hierarchical assembly that persists after spatial relaxation, potentially distinguishing formed-in-place clusters from merged ones in Gaia-era data.","A direct test would be to measure Moran's I in clusters spanning a range of dynamical ages; a plateau above zero in older embedded clusters would support the asymptotic behavior seen here.","The binary-acceleration mechanism suggests clusters with higher primordial binary fractions should mass-segregate faster, which is testable with resolved binary surveys in young clusters."],"forward_implications":["Observed spatial substructure in a young cluster indicates it is younger than roughly 2.5 free-fall times of its natal cloud.","Kinematic substructure is a longer-lived signature of hierarchical assembly and can reveal ongoing merging even when the spatial distribution looks smooth.","Dynamical mass segregation can appear before cluster collapse, so its presence in young clusters does not require primordial segregation.","The timescale for order is set by the cloud free-fall time and appears insensitive to cloud mass across the simulated range.","Primordial binaries are not just passive members but active agents that accelerate mass segregation during assembly."],"supporting_citations":[{"why":"Defines the Q statistic used throughout the paper to quantify spatial substructure.","marker":"A. Cartwright & A. P. Whitworth (2004)"},{"why":"Introduces Moran's I as a measure of kinematic substructure and shows its reliability under observational biases.","marker":"B. Arnold et al. (2022)"},{"why":"Provides the observed comparison sample of young clusters and the velocity-outlier removal procedure adopted here.","marker":"B. Arnold & N. J. Wright (2024)"},{"why":"Supplies the simulation suite, including initial conditions, binary sampling, and the star-by-star dynamical treatment.","marker":"C. Cournoyer-Cloutier et al. (2024)"},{"why":"Describes the cluster identification and merger-tree tracking method used to follow assembly histories.","marker":"C. Cournoyer-Cloutier et al. (2023)"},{"why":"Proposes the modified mass segregation ratio Lambda_MSR that removes the N-dependence of the original statistic.","marker":"L. Wei et al. (2025)"},{"why":"Provides the comparison simulations without primordial binaries that ground the mass segregation claim.","marker":"B. Polak et al. (2025)"},{"why":"Underlies the physical explanation that binaries have larger interaction cross-sections and thus accelerate dynamical encounters.","marker":"D. C. Heggie (1975)"},{"why":"Establishes the mass segregation ratio and the idea that collapse-induced dynamical segregation can happen in young clusters.","marker":"R. J. Allison et al. (2009)"}],"fun_headline_variants":["Spatial substructure fades in 2.5 free-fall times","Kinematic substructure persists after spatial erases","Primordial binaries enhance early mass segregation","Cluster assembly smooths spatial order, leaves kinematic"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's conclusion that binaries accelerate mass segregation rests on the assumption that the comparison simulations differ only in whether primordial binaries were included; if the clouds' masses, densities, resolutions, or feedback details also differ, those differences could explain the stronger and earlier segregation.","fun_headline_variants_meta":{"raw":{"variants":["Spatial substructure fades in 2.5 free-fall times","Kinematic substructure persists after spatial erases","Primordial binaries enhance early mass segregation","Cluster assembly smooths spatial order, leaves kinematic"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000608,"raw_usage":{"total_tokens":2805,"prompt_tokens":890,"completion_tokens":1915,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":506,"completion_tokens_details":{"reasoning_tokens":1851}},"tokens_in":506,"tokens_out":1915,"duration_ms":16611,"temperature":1.0,"reasoning_tokens":1851,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:06:16.092815+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same three initial clouds through the same code with primordial binary formation switched off, holding gas masses, densities, resolutions, feedback implementations, and cluster-selection criteria fixed; if the Lambda_MSR evolution curves match the binary-included runs, the claim that binaries accelerate mass segregation is disproved.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the observed comparison sample of young clusters and the velocity-outlier removal procedure adopted here."},{"cited_title":"R., Boyle, P","cited_arxiv_id":null,"evidence_quote":"Proposes the modified mass segregation ratio Lambda_MSR that removes the N-dependence of the original statistic."}],"review_version":1}