{"id":"70352f8b-c92d-4569-8b57-75375ce25819","arxiv_id":"2508.11793","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In self-aligning adhesive particles, flocking makes a cluster's persistence length grow with mass, switching aggregation from a t^1/2 diffusion-limited regime to a t^2 collective ballistic regime.","lead":"This paper shows that sticky, self-propelled particles that steer to move in the same direction clump together much faster once they form moving flocks. It explains why cell clusters in tissue sometimes show time-squared growth instead of the slower random-walk growth.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The analytic z=2 derivation uses a ballistic kernel that vanishes identically at γ=0; z=2 rests on unstated orientational disorder, so the theory as written does not derive the claimed exponent.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing gap: the ballistic kernel in Supplement Eq. S24 vanishes when γ=0, so the written theory cannot produce z=2 without an additional orientation-disorder term. I agree with that assessment. The concern is real and central, because the paper's headline mechanism is the analytic connection between flocking (γ=0) and z=2. However, the paper has independent support that prevents the concern from being fatal: direct simulations show z ≈ 2 in the collective regime (Fig. 1a, Fig. 4a), single-cluster MSD data show V_c becoming mass-independent and τ_p ∼ M (Fig. 3b, 3d), and the lp/lc criterion is consistent with a crossover to ballistic aggregation. The missing premise is eminently repairable by writing the kernel with velocity directions, and doing so yields z=2. Thus the appropriate verdict remains CONDITIONAL, as the reader already concluded; no verdict change is needed. The concrete test proposed above would either confirm that the orientation-averaged kernel reproduces the simulated exponent or, if it does not, would force a revision of the analytic explanation. I did not identify any additional concern of comparable weight: the absence of error bars weakens quantitative comparisons but does not undermine the central mechanism, and the claim of explaining experimental exponents is qualitative rather than central to the scaling derivation. The paper is honest about its modeling assumptions and about the need for future experimental parameter determination.","tokens_in":12055,"tokens_out":3304,"duration_ms":38617,"concrete_test":"Replace Eq. S24 with the orientation-resolved kernel K(M,M′) = B ⟨|n(M)−n(M′)|⟩ (R(M)+R(M′))^(d−1), with V_c = v0 and the angular average computed for isotropically distributed cluster orientations (e.g., ⟨|n−n′|⟩ = 4/π in 2D), then re-derive the scaling exponent from Eq. S14. If this yields ⟨M⟩ ∼ t^2 for d=df=2, the concern is resolved. As a complementary numerical check, directly measure the merger-rate kernel from the J=10, Pe=1 simulations by binning cluster pairs by mass and recording collision rates; if K(M,M) is nonzero and scales as (R(M)+R(M)) rather than vanishing, the orientation-disorder term is present and the corrected kernel is the right description.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that collective ballistic aggregation with mass-independent cluster speed (γ=0) yields z = 2/(1−2γ) = 2. This relies on the ballistic kernel K(M,M′) = B |V_c(M) − V_c(M′)| (R(M)+R(M′))^(d−1), written in the Supplement as K ∝ |M^γ − M′^γ| (M^(1/df)+M′^(1/df))^(d−1) (Eq. S24). At exactly the value γ=0 that defines the CBA regime, this kernel is identically zero: clusters of equal speed never collide, the integral I2 in Eq. S27 vanishes, and Eq. S26 gives d⟨M⟩/dt = 0 rather than ⟨M⟩ ∼ t^2. The z=2 prediction is therefore not derivable from the equations as presented. The result can be rescued by adding the physical premise that cluster velocities have different directions: with V_c mass-independent but orientations independent, the mean relative speed is ⟨|n−n′|⟩ V_c, which is a nonzero constant, and the effective kernel becomes K ∼ (R+R′)^(d−1), homogeneous of degree (d−1)/df. For d=df=2 this gives K ∼ M^(1/2) and, through the scaling argument, z=2. But this orientation-disorder term is not present in Eq. S24, nor is it stated in the main text where 'mass-independent cluster speed (γ=0)' is presented as the sufficient condition for z=2. The simulations may well be correct, but the analytic mechanism as written has a gap at the precise point used to derive the headline exponent.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies aggregation of adhesive active Brownian disks with self-alignment, varying the alignment strength J and the Péclet number Pe. It identifies three kinetic regimes: diffusion-limited cluster aggregation (DLCA, z between 1/2 and 1), non-collective ballistic aggregation (z ≈ 1), and a collective ballistic aggregation (CBA) regime with z ≈ 2. Single-cluster simulations are used to extract mass-dependent cluster speed, persistence time, and persistence length, and a criterion comparing persistence length with intercluster distance is proposed to locate the crossover to ballistic aggregation. A generalized Smoluchowski coagulation theory with diffusion- or velocity-based kernels is then used to express the aggregation exponent z in terms of measured single-cluster exponents, and the results are summarized in a Pe–J regime diagram.","tokens_in":12416,"tokens_out":6761,"duration_ms":75035,"significance":"If the CBA mechanism is correct, the paper provides a simple and appealing explanation for anomalously fast aggregation exponents observed in adhesive active matter: flocking makes the cluster persistence length grow linearly with mass while the intercluster distance grows only as the square root of mass, so large clusters move in a collective ballistic regime with M(t) ~ t^2. The paper's strengths are its systematic simulations, the clean single-cluster observable lp(M), the dynamical-scaling collapse of P(M,t), and the unified regime diagram. The main theoretical derivation of the headline exponent z=2, however, has a gap at the precise point γ=0, because the ballistic kernel written in the Supplement vanishes identically for mass-independent cluster speed. The result is likely repairable by adding an explicit orientational-disorder premise, but as written the central claim is not derived from the stated equations.","major_comments":[{"comment":"The ballistic kernel is written as K(M,M′) = B |M^γ − M′^γ| (M^{1/df} + M′^{1/df})^{d−1}. For the CBA regime the paper uses γ=0, corresponding to mass-independent cluster speed. At γ=0 the factor |M^γ − M′^γ| is identically zero, so I2 in Eq. (S27) vanishes, C2=0, and Eq. (S26) gives d⟨M⟩/dt = 0 rather than ⟨M⟩ ~ t^2. Consequently z = 2/(1−2γ) evaluated at γ=0 is not derivable from the kernel as stated. The intended physical picture must be that clusters have equal scalar speeds but randomly oriented velocities, so the mean relative speed is a nonzero, mass-independent constant. With that premise the kernel becomes K ~ (R+R′)^{d−1}, which for d=df=2 is homogeneous of degree 1/2 and indeed gives z=2. The authors should state this orientational-disorder premise explicitly and revise Eq. (S24) accordingly, since the current text presents γ=0 as a sufficient condition.","section":"Supplement, Eqs. (S24)–(S31); main text after Eq. (S31)"},{"comment":"The agreement between measured and predicted aggregation exponents is a consistency check between two observables of the same simulation, not an independent prediction. The exponents γ, ν, and α are extracted from single-cluster simulations of the same model and then inserted into the Smoluchowski scaling formulas to obtain z. This is acceptable as a scaling analysis, but the text should be more precise: it says 'Smoluchowski theory predicts z=2', whereas in fact the theory uses the measured mass-dependence of cluster motion as input. Please clarify which ingredients are measured and which are derived, so that readers can judge the explanatory content of the framework.","section":"Main text, 'To explain z...' paragraph and Fig. 3"}],"minor_comments":[{"comment":"The word 'yelding' should be 'yielding'.","section":"Main text, after 'We obtain z = 2/(1−2γ)'"},{"comment":"The captions of Figs. S1 and S2 appear inconsistent with the surrounding text: the text says Fig. S1 is for low Pe and Fig. S2 for high Pe, but the Fig. S1 caption states 'high Péclet number regime (Pe=10^3)'. Please correct the labeling.","section":"Supplement, Figs. S1 and S2"},{"comment":"The measured exponents γ, ν, and α are central inputs to the theory, yet the panels in Fig. 3 show no error bars or fit ranges. Adding uncertainty estimates or at least stating how the power-law fits were performed would strengthen the quantitative claims.","section":"Fig. 3"},{"comment":"The statement that masses M < 225.73 are in the ballistic regime and larger masses follow DLCA is based on the 2lp/lc criterion, but the M(t) data in Fig. 4a do not show a visible crossover in the displayed time window. A quantitative comparison of the predicted crossover mass with the simulation time scale would make this argument more convincing.","section":"High persistence, J=0 paragraph"},{"comment":"There is no data- or code-availability statement; providing simulation details or a reproducibility statement would be helpful, especially because several claims (robustness to packing fraction, robustness to alignment rule) are mentioned without a figure.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the z=2 derivation gap: the ballistic kernel in Eq. (S24) vanishes at γ=0, so the headline exponent is not derived from the equations as written. The fix is local and conceptually clear (add orientational disorder to the relative-velocity term), and the simulation results appear internally consistent, so I do not see grounds for rejection. I would ask the authors to revise the derivation and to temper the 'predicts' language, since the input exponents come from the same model."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Give this a serious referee, but make sure the referee asks for a fix to the analytic derivation. The core simulation result is strong: above the flocking transition, cluster persistence length grows about linearly with mass while the mean intercluster distance grows as M^(1/2), so 2lp/lc crosses one and the aggregation switches from DLCA to a fast collective ballistic regime with z ≈ 2. That mechanism is new, and the single-cluster analysis of Vc(M), τp(M), D(M) is a clean step beyond Mones et al. and Beatrici et al. The Pe–J regime diagram is useful. I also appreciate the dynamical scaling collapse of the cluster-size distributions, which supports the Smoluchowski framework.\n\nThe soft spot is real, and it is exactly where the stress-test note points. The main text writes the ballistic kernel as |Vc(M) - Vc(M')|, which for vector velocities does not vanish when speeds are mass-independent—directions matter. But the supplement replaces that with the scalar |M^γ - M'^γ|, and at γ=0 that factor is identically zero. The integral I2 in Eq. S27 then vanishes, so the derivation gives d⟨M⟩/dt = 0, not ⟨M⟩ ∼ t^2. The authors get z=2 by plugging γ=0 into the formula z=2/(1−2γ), but that formula was derived assuming the kernel integral is nonzero. The fix is short: add the physical premise that cluster velocities have independent directions, which gives a nonzero mass-independent relative speed and an effective kernel ∼(R+R')^(d−1), homogeneous of degree 1/2 in d=2, reproducing z=2. As written, the analytic support for the headline exponent has a gap.\n\nOther issues are minor. The reported exponents have no error bars, though the power laws are visually clear. The comparison with experimental exponents is qualitative; the paper overstates a bit when it says 'explain the broad range.' And the exponents γ, ν, α are measured from the same model that is then used in Smoluchowski theory, so it is a consistency check rather than a parameter-free prediction. None of that bothers me much.\n\nBottom line: the simulations are likely correct and the mechanism is plausible. The derivation needs one paragraph's worth of repair. I would not cite the current version, but a corrected version would be citable. Send it to peer review.","headline":"Nice simulation study with a plausible persistence-length mechanism for z≈2 in adhesive active matter, but the analytic derivation of the headline exponent has a real gap at γ=0.","tokens_in":12974,"tokens_out":6375,"would_cite":false,"duration_ms":63461,"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":"Strong self-alignment in adhesive active matter triggers collective ballistic aggregation, in which the average cluster mass grows as M(t) ~ t², faster than diffusion- or persistence-limited aggregation.","keywords":["adhesive active matter","aggregation kinetics","collective motion","flocking transition","Smoluchowski coagulation","self-alignment","cluster persistence length","cell aggregation"],"falsifier":"Measure the coalescence rate of two equal-mass clusters moving at the same speed in random directions in the strong-alignment regime; if that rate tends to zero as the speed difference vanishes, the t² law lacks the required source of relative motion and the collective-ballistic explanation, as written, fails.","tokens_in":11831,"feed_emoji":"🧫","tokens_out":15527,"duration_ms":149183,"temperature":0.7,"pith_summary":"The paper is trying to establish why sticky, self-propelled particles—motile cells in tissue formation being the motivating case—can aggregate much faster than classical coagulation theory predicts, with the mean cluster mass growing as t² rather than $t^{{1/2}}$ or t. The answer it proposes is a flocking transition: strong self-alignment makes each cluster move coherently, and the distance a cluster travels while keeping its direction grows linearly with cluster mass, while the typical gap between clusters grows only as the square root of mass. As soon as the persistence length overtakes the intercluster gap, collisions become ballistic and the system enters a collective ballistic aggregation regime that is asymptotically self-sustaining. The same persistence-length criterion, supported by a generalized Smoluchowski coagulation theory, organizes all measured exponents (z from 1/2 to 2) into a regime diagram in the plane of self-alignment strength and single-particle persistence.","feed_headline":"Clusters that flock grow at a t² rate","feed_subtitle":"A persistence-length comparison tells when sticky particles switch from slow diffusion to fast collective merging.","key_machinery":"The load-bearing comparison is between two lengths: the cluster persistence length lp = Vcτp, the distance a cluster travels before its direction decorrelates, and the mean intercluster distance lc = (M/ρ)^{1/2} set by the global density. Self-alignment changes the mass scalings of single-cluster dynamics—strong alignment drives the polar order parameter toward 1, saturates Vc at v0 (γ = 0), and makes τp grow linearly with M (ν = 1)—so lp ∝ M while lc ∝ $M^{{1/2}}$, and the criterion 2lp/lc > 1 marks entry into collective ballistic aggregation, with lp/lc ∝ $M^{{1/2}}$ making that regime self-sustaining. The companion machinery is a generalized Smoluchowski coagulation theory whose kernels encode these scalings, giving z = 1/(1−α) for diffusion-limited aggregation and z = 2/(1−2γ) for ballistic aggregation in two dimensions.","core_discovery":"The paper's central claim is that the anomalous aggregation exponent z ≈ 2 (mean cluster mass growing as M(t) ~ t²) is produced by a flocking transition inside clusters, not by the persistence of individual particles. Single-cluster simulations in the strong-alignment limit show that the cluster speed becomes independent of mass, Vc = v0 (exponent γ = 0), while the persistence time grows linearly with mass, τp ∝ M (exponent ν = 1). The persistence length lp = Vcτp therefore scales as M, while the average intercluster distance lc = (M/ρ)^{1/2} scales as $M^{{1/2}}$, so the ratio lp/lc grows as $M^{{1/2}}$; once the criterion 2lp/lc > 1 is met, clusters collide ballistically and keep doing so as they grow, making the t² regime asymptotic rather than transient. A generalized Smoluchowski coagulation equation with diffusion kernel ∝ (D(M)+D(M′))(R+R′)^{d−2} and ballistic kernel ∝ |Vc(M)−Vc(M′)|(R+R′)^{d−1} yields the exponent formulas z = 1/(1−α) and z = 2/(1−2γ) in d = 2, reproducing the simulations and uniting diffusion-limited (z ≈ 1/2–1), non-collective ballistic (z ≈ 1), and collective ballistic (z ≈ 2) aggregation in a single picture.","pith_inferences":["The paper leaves implicit that the ballistic kernel written in the Supplement, |Vc(M) − Vc(M′)|, vanishes when the cluster speed is mass-independent (γ = 0); obtaining z = 2 from the written equations requires the added premise that equal-speed clusters still approach each other with nonzero relative velocity because their directions differ.","With that premise made explicit, the Smoluchowski formalism extends naturally to an angle-averaged kernel proportional to Vc, which would give an explicit d-dimensional prediction for z(γ, d) instead of the collinear-kernel formula z = d/(1 − dγ).","A testable extension suggested by the mechanism: reducing adhesion so that clusters can fragment should cut off the linear growth of lp with M and lower the measured exponent—the paper lists fragmentation as future work.","The persistence-length criterion maps onto an experimental protocol: measuring Vc and τp for cell aggregates of increasing size should show lp(M) turning linear at the same alignment strength at which the aggregation exponent jumps toward 2."],"forward_implications":["Cell-sorting and tissue experiments that report aggregation exponents between 1 and 2 can be interpreted as sitting in the crossover between diffusion-limited and collective ballistic aggregation, without invoking new physics.","The criterion 2lp/lc > 1 is directly measurable: tracking cluster centroids to obtain Vc(M) and τp(M) predicts the aggregation regime of a given adhesive active system.","Because lp/lc keeps growing as M^{1/2} in the collective regime, the t² law is self-stabilizing: larger clusters are even further into the ballistic regime, so no parameter tuning is needed to sustain fast aggregation.","Systems in collective ballistic aggregation develop a power-law cluster-mass distribution with dynamical scaling (exponent λ ≈ 1.1), whereas diffusion-limited aggregation retains a characteristic cluster size, giving experiments a statistical fingerprint that distinguishes the regimes.","High single-particle persistence without alignment produces a long-lived transient with z ≈ 1 before the crossover to diffusion-limited behavior, so early-time exponents in biological assays need not reflect the asymptotic regime."],"supporting_citations":[{"why":"Prior simulation study that reported z ≈ 2 in adhesive active particles with self-alignment; its unexplained finding is the phenomenon this paper explains.","marker":"[19]"},{"why":"Earlier theory connecting internal alignment to the diffusivity–mass relation and aggregation exponent up to z = 1; the present work generalizes it to strong alignment.","marker":"[23]"},{"why":"Review supplying the Smoluchowski coagulation formalism, the diffusion- and ballistic-limited kernels, and the dynamical-scaling method used to derive the exponent formulas.","marker":"[13]"},{"why":"Cell-sorting experiments reporting aggregation exponents z > 1 that classical models cannot match; the target observations the framework unifies.","marker":"[12]"},{"why":"Original Smoluchowski coagulation equation, the basis of the generalized theory used to predict z in each regime.","marker":"[24]"},{"why":"Source of the adhesive active Brownian particle model of motile tissue cells on which the simulations are based.","marker":"[25]"}],"fun_headline_variants":["Flocking transition inside clusters yields t² aggregation","Persistence length beats intercluster distance, fast growth","Collective ballistic merging grows clusters at t² rate","Why adhesive active matter sometimes aggregates quadratically","Cluster alignment accelerates aggregation to t²"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The t² law rests on the unstated premise that clusters of equal mass and equal speed still meet at a nonzero rate because their motion directions differ; if collisions required a difference in speed, equal-speed clusters would never coalesce and the predicted fast regime would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Flocking transition inside clusters yields t² aggregation","Persistence length beats intercluster distance, fast growth","Collective ballistic merging grows clusters at t² rate","Why adhesive active matter sometimes aggregates quadratically","Cluster alignment accelerates aggregation to t²"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000319,"raw_usage":{"total_tokens":1789,"prompt_tokens":922,"completion_tokens":867,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":796}},"tokens_in":538,"tokens_out":867,"duration_ms":10348,"temperature":1.0,"reasoning_tokens":796,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:28:16.952331+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the coalescence rate of two equal-mass clusters moving at the same speed in random directions in the strong-alignment regime; if that rate tends to zero as the speed difference vanishes, the t² law lacks the required source of relative motion and the collective-ballistic explanation, as written, fails.","supporting_citations":[{"cited_title":"Mones, A","cited_arxiv_id":null,"evidence_quote":"Prior simulation study that reported z ≈ 2 in adhesive active particles with self-alignment; its unexplained finding is the phenomenon this paper explains."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier theory connecting internal alignment to the diffusivity–mass relation and aggregation exponent up to z = 1; the present work generalizes it to strong alignment."},{"cited_title":"Leyvraz, Physics Reports 383, 95 (2003)","cited_arxiv_id":null,"evidence_quote":"Review supplying the Smoluchowski coagulation formalism, the diffusion- and ballistic-limited kernels, and the dynamical-scaling method used to derive the exponent formulas."},{"cited_title":"Méhes, E","cited_arxiv_id":null,"evidence_quote":"Cell-sorting experiments reporting aggregation exponents z > 1 that classical models cannot match; the target observations the framework unifies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Original Smoluchowski coagulation equation, the basis of the generalized theory used to predict z in each regime."},{"cited_title":"Szabo, G","cited_arxiv_id":null,"evidence_quote":"Source of the adhesive active Brownian particle model of motile tissue cells on which the simulations are based."}],"review_version":1}