{"id":"998ea9d8-ab1b-451a-ab3b-f96ba3213771","arxiv_id":"2501.10039","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A facilitated-advection lattice model predicts that the abruptness of yielding in glasses is controlled by a dynamic correlation length that diverges as quenched disorder vanishes.","lead":"This paper proposes a lattice model in which the microscopic relaxation rate of a glass under oscillatory shear is coupled between neighboring sites, and shows that adding disorder converts the abrupt yielding transition into a gradual, rounded one. The model predicts a dynamic correlation length that grows as disorder shrinks and diverges at zero disorder, which would explain why some samples yield abruptly while others flow gradually.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Fig. 6c averages ξ* only over gradually yielding runs, so the claimed divergence of ξ with vanishing disorder may be an artifact of conditioning; abrupt runs with ξ of one lattice spacing are excluded.","rationale":"The paper is a serious attempt to unify abrupt and gradual yielding via a disorder-dependent correlation length, and the mean-field cusp/catastrophe analysis is internally consistent. The numerical model qualitatively reproduces the observed coexistence and the sharp-to-gradual crossover with disorder. The load-bearing step, however, is the inference from Fig. 6 that ξ* grows and would diverge as d→0, and from Fig. 4 that the abrupt/gradual crossover is controlled by N vs ξ*. That inference depends on comparing N*(d)—an ensemble threshold over all runs—with ξ*(d)—an average over the gradual subset only. The caption explicitly excludes abrupt-yielding simulations from the averaged dataset; at small d these are the majority of runs, and Section VII states they have ξ of order one lattice spacing. This selection can create the appearance of a diverging correlation length even if the true disorder-averaged correlation length is bounded. The statement that with vanishing disorder ξ diverges and yielding becomes discontinuous is therefore not directly demonstrated by the data; it is an extrapolation, and the very phenomenon (abruptness) is what triggers the exclusion of the runs where ξ is small. This is a circularity risk in the central claim. A clean re-analysis with an unbiased estimator (all realizations, or a susceptibility peak) would settle whether the divergence survives. If it does not, the claimed universality of the N-vs-ξ criterion would collapse to a statement about selected realizations only. I therefore keep the reader's CONDITIONAL verdict: the model is worth further work, but the central quantitative claim is not yet established. I partially agree with the reader: the FA coupling in Eq. 4 is indeed ad hoc and not derived from microscopic dynamics, but the more immediate threat to the central claim is the selection bias in the main supporting figure. The coupling-form concern would require a different model or a derivation; the selection-bias concern can be tested on the existing data and is thus the more decisive check.","tokens_in":27478,"tokens_out":5812,"duration_ms":63166,"concrete_test":"Recompute Fig. 6c using all independent realizations, not just gradually yielding runs, defining ξ* as the maximum of the disorder-averaged correlation function over strain; if the resulting ξ*(d) no longer grows as d decreases (or no longer tracks N*(d)), the claimed divergence is an artifact of selection and the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that abrupt yielding is a finite-size effect controlled by a diverging correlation length ξ—rests on two pillars: (i) N*(d) from Weibull fits tracks ξ*(d), and (ii) ξ* diverges as d→0. Both are compromised by the selection rule in Fig. 6c: ξ* is computed only from realizations that exhibited gradual yielding, and the caption states that runs with abrupt yielding were excluded from the averaged dataset. At small d, a large fraction of finite-N runs are abrupt; Section VII reports that in those runs c(r) decays on a single lattice spacing, so ξ ≈ 1. Excluding them necessarily inflates the average and can manufacture the apparent growth/divergence of ξ* with decreasing d. The comparison with N* is also not apples-to-apples: N*(d) is extracted from Ψ(d), the fraction of gradual runs over all independent realizations, while ξ*(d) is the mean over the gradual subset only. Moreover, abruptly yielding runs have ξ of order one, so the statement that abruptness occurs when ξ exceeds N is an inference about a hidden ξ, not a directly observed one. Finally, the claim that in the thermodynamic limit the model predicts gradual yielding for all finite d is an extrapolation from N ≤ 512 and d ≥ 0.02, with no scaling collapse, error bars on ξ*, or fitted divergence exponent; the Maxwell-rule yield strain is matched for one parameter set, and the underlying potential is admitted to be unexplained (Section VI).","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a lattice model with facilitated advection coupling to describe the dynamics of yielding in glassy materials under oscillatory shear. In mean-field, the model reduces to a Van der Waals-like equation of state whose non-monotonic regime maps onto a cusp catastrophe, predicting abrupt yielding with hysteresis. Adding quenched disorder in the coupling constants produces gradual, rounded yielding with coexistence of solid-like and fluid-like domains. The authors report a correlation length ξ of local relaxation rates that peaks at yielding and grows as disorder decreases, and they argue that abrupt yielding occurs when ξ exceeds the system size, making sharp yielding a finite-size effect in this model. The paper also proposes a Maxwell-rule interpretation of the yield strain in the thermodynamic limit, based on an effective potential V(ρ) obtained by integrating the mean-field equation of state.","tokens_in":27832,"tokens_out":2550,"duration_ms":29356,"significance":"If the central scenario were fully established, the paper would offer a unified mesoscopic mechanism for both abrupt and gradual yielding, connecting a dynamically emerging correlation length to the abruptness of the transition. The mean-field algebra is internally consistent, and the simulations qualitatively reproduce rounded transitions, bimodal rate distributions, and growing domain sizes with decreasing disorder. However, the decisive quantitative evidence—the claimed divergence of ξ as d→0 and the lengthscale competition with N—is currently compromised by a conditioning artifact in the sampling of ξ*, as detailed below. With more careful statistics, the qualitative scenario remains plausible and worth testing, but the paper in its present form does not support the central claim as stated.","major_comments":[{"comment":"The central claim that abrupt yielding is controlled by the competition between N and ξ* rests on ξ*(d) growing as d decreases. However, Fig. 6c's caption states that full datapoints are averaged only over the 10 independent simulations that exhibited gradual yielding, and that simulations with abrupt yielding were excluded from the averaged dataset. Section VII itself reports that abruptly yielding runs have ξ as small as a single lattice spacing. Because at small d most runs are abrupt (Fig. 4b shows Ψ decreasing strongly with d), the conditional average over the gradual subset can inflate the mean ξ* and manufacture the apparent growth with decreasing d. The comparison with N*(d) is also not apples-to-apples: N*(d) is extracted from Ψ(d), the fraction of gradual runs over all realizations (Fig. 4c), while ξ*(d) is averaged only over the gradual subset. Thus the conclusion that ξ* tracks N* and that abruptness corresponds to ξ exceeding the system size is not supported by the data as presented.","section":"Section VII, Fig. 6c"},{"comment":"The conclusion that ξ* diverges as d→0 (stated in the abstract and reiterated in Section VIII) is an extrapolation from simulations with N=512 and d between 0.02 and 0.2. No scaling fit, no error bars on ξ*, no fitted divergence exponent, and no finite-size scaling collapse are provided. The data in Fig. 6c cover only a limited range of d, and the highest-disorder points are open (partially excluded) symbols. The authors should either perform a scaling analysis with system-size dependence or explicitly soften the claim to a suggestion, rather than asserting a divergence.","section":"Sections VII and VIII"},{"comment":"The Maxwell-rule yield strain γth^(M), which is used as the thermodynamic-limit yield point and as the reference in Figs. 3 and SM6, is obtained by globally minimizing the potential V(ρ) of Eq. (12). The authors themselves state that the physical meaning of V is 'yet to be unveiled' and that its global minimization does not have straightforward physical relevance. Since V is introduced by formal analogy and its global minimum is not derived from microscopic dynamics, the first-order-transition interpretation and the identification of γth^(M) with the simulated yield point are not theoretically grounded. A derivation of the global-minimization criterion, or at least a stability argument showing why nucleation selects the Maxwell set, is needed before this claim can be accepted.","section":"Section VI, Eq. (12)"}],"minor_comments":[{"comment":"The open symbols are described only as 'disorders for which one or more simulations exhibited abrupt yielding,' but the fraction of excluded runs is not given; adding that fraction (or showing Ψ(d) alongside) would make the conditioning transparent.","section":"Fig. 6c"},{"comment":"The caption states that dashed lines represent simulations where yielding was abrupt due to finite size effects, but the text uses dashed lines also for small-disorder mean-field-like behavior; please unify the notation and clarify the criterion for labeling a simulation as abrupt.","section":"Section V, Fig. 3a caption"},{"comment":"The statement 'in the thermodynamic limit, N→∞, our model predicts gradual yielding with coexistence for all finite disorders' is made without a scaling analysis or a control parameter for the thermodynamic limit; it should be phrased as a conjecture supported by the Weibull trends, not a proven result.","section":"End of Section V"},{"comment":"There are several typographical errors: 'univoquely' (Introduction), 'gradudal' (Section V), and 'the reminder of this paper' (Introduction). These should be corrected.","section":"Introduction and Section V"},{"comment":"The definition of ξ via the integral of c(r) and the KWW fit is clear, but no details are given on the fit range, the quality of the fits, or how the integral behaves when c(r) does not decay to zero within the simulation box; a brief discussion of these practical issues would strengthen the measurement.","section":"Section VII, Eq. (7) and KWW fit"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely question, and the qualitative scenario—disorder rounding a first-order dynamic transition and generating a lengthscale—is attractive. However, the key quantitative evidence for the central claim (divergence of ξ and the N vs ξ competition) is compromised by the conditional averaging in Fig. 6c. This is not merely a presentation issue; the inference changes qualitatively if abruptly yielding runs are included. The authors should be asked to recompute ξ* over all realizations, or to explicitly model the biased average, and to provide a scaling analysis for the claimed divergence. The Maxwell-rule step also needs a physical justification beyond formal analogy. Given that these issues affect the paper's main message, major revision is appropriate; the manuscript is not ready for acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Stefano, quick take on Aime and Truzzolillo, arXiv:2501.10039. The paper has a genuinely nice mean-field core: a lattice model with facilitated advection that maps yielding onto a cusp catastrophe, with a Maxwell-rule strain selected by an effective potential. The disorder extension is new in this context, and the qualitative picture—disorder rounds the transition, domains of slow and fast dynamics coexist, and the domain size grows as disorder decreases—is plausible and supported by the simulations at face value. The writing is clear and the authors are honest about what is speculative.\n\nThe soft spot is the central quantitative claim. The paper argues that ξ, the correlation length of dynamic heterogeneities, diverges as disorder d→0, and that abrupt yielding is a finite-size effect when ξ exceeds the system size N. But Fig. 6c averages ξ* only over runs that exhibited gradual yielding; abrupt runs, where ξ is about one lattice spacing, are excluded. At small d, abrupt runs are common, so excluding them inflates the average and can manufacture the apparent divergence. The comparison with N*(d) is also not apples-to-apples: N* comes from the fraction of gradual runs over all realizations, while ξ* is the mean over the gradual subset only. So the central claim is not established by the data as presented.\n\nThere are also smaller issues: the divergence is extrapolated from d between 0.02 and 0.2 with no scaling fit or error bars, and the Maxwell-rule match is for one representative parameter set with an admittedly unexplained potential. These are fixable in revision—include all runs, report error bars, fit a divergence, and be explicit that ξ is defined on the subset of gradual realizations.\n\nBottom line: the model is a useful playground and the mean-field catastrophe picture is likely to be cited, but the 'diverging lengthscale controls brittleness' story needs better evidence. Worth sending to a serious referee—the idea is significant enough that a desk rejection would be wrong—but the referee should push hard on Fig. 6c and the extrapolation. For a reading group, it's a good case study in selection bias.","headline":"Nice mean-field catastrophe picture, but the central claim of a diverging correlation length is undercut by selection bias in the averaging.","tokens_in":28313,"tokens_out":2560,"would_cite":false,"duration_ms":25034,"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":"A lattice model shows that the abruptness of yielding in glassy materials is set by a diverging correlation length.","keywords":["yielding transition","glassy materials","dynamic heterogeneity","correlation length","cusp catastrophe","lattice model","disorder","oscillatory shear"],"falsifier":"Measure the connected correlation length of local relaxation rates (for example by imaging tracer motion or stroboscopic DLS in a colloidal glass under oscillatory shear) while systematically reducing sample disorder by deeper annealing: if the peak correlation length does not grow, or if abrupt yielding does not track the condition ξ* ≈ sample size, the central claim is wrong. A direct numerical test is to replace the α_ij/(Γ_i Γ_j) coupling with a linear (Γ_i − Γ_j) coupling and check whether the bimodal coexistence and the ξ divergence survive.","tokens_in":27238,"feed_emoji":"🔬","tokens_out":6318,"duration_ms":60260,"temperature":0.7,"pith_summary":"The paper tries to explain why some glassy materials yield abruptly while others yield gradually, and to give the sharp-to-gradual switch a concrete mechanism. It builds a lattice model of local relaxation rates under oscillatory shear, adding a dynamic-facilitation coupling that makes nearby regions relax in a coordinated way. In the mean-field version, the model reproduces the observed solid-to-fluid dynamic transition as a cusp-catastrophe equation of state, with discontinuous yielding and hysteresis. Once disorder in the coupling is switched on, the transition becomes continuous and rounded: fluid-like domains nucleate at weakly coupled sites and coexist with solid-like domains over a finite strain window. The paper's central claim is that this coexistence carries an emergent correlation length ξ which grows as disorder decreases and diverges at zero disorder, making abrupt yielding the finite-size shadow of a diverging dynamic heterogeneity length.","feed_headline":"Yielding turns abrupt when dynamic correlations outgrow the sample","feed_subtitle":"A lattice model shows brittle yielding in glasses is a finite-size effect of a diverging heterogeneity length.","key_machinery":"The load-bearing object is the facilitated-advection (FA) coupling in Eq. 4: the shear-induced relaxation of a lattice site is slowed by a term α_ij/(Γ_i Γ_j) that penalizes rate differences between neighbours, borrowing the idea of dynamic facilitation from quiescent glasses. In mean field this reduces to a Van der Waals-like equation of state, which the authors recast as a cusp catastrophe manifold $ρ^{3}$ + aρ + b = 0 with an effective potential V(ρ) = $ρ^{4}$/4 + $aρ^{2}$/2 + bρ. The potential's two minima give the solid and fluid branches, the fold set gives the spinodal thresholds γ_th^±, and the Maxwell set (equal-depth minima) gives the yield strain γ_th^(M) selected in large disordered systems. The emergent correlation length is extracted from the connected spatial correlation of log Γ_i, together with finite-size statistics linking disorder to domain size.","core_discovery":"The discovery is that the sharp-versus-gradual dilemma of yielding can be stated as a competition between two lengths. In the model, local relaxation rates Γ_i satisfy a coupled equation whose disorder-free mean field is a Van der Waals-like equation of state on a cusp catastrophe manifold: for sufficiently glassy parameters, increasing strain amplitude brings the system to a fold where Γ jumps discontinuously. Introducing quenched disorder in the coupling constants rounds the transition: the lattice splits into solid-like and fluid-like domains whose coexistence reproduces the two-mode correlation functions seen in experiments. The correlation length ξ of the local rates peaks at yielding and increases as the disorder variance tends to zero, so a finite sample is abrupt exactly when ξ exceeds the system size. The authors conclude that, in the thermodynamic limit, the model predicts gradual yielding with coexistence for any finite disorder, while the abruptness seen in small systems is a finite-size effect controlled by the same emergent length scale.","pith_inferences":["If the central claim is right, then 'brittle' yielding in laboratory samples is not a distinct intrinsic phase but a consequence of the dynamic heterogeneity length exceeding the sample size; the same material should become ductile in a larger sample.","The effective potential V(ρ) was introduced by formal analogy, but its Maxwell-rule behavior suggests it may capture a true out-of-equilibrium free energy; testing whether its barrier height predicts the nucleation rate of fluid domains would extend the model to time-dependent and stress-controlled yielding.","The model is not limited to oscillatory shear: replacing the Γ_sh ∝ ωγ_0^n ansatz with a steady-shear counterpart should produce an analogous correlation length in creep or shear-rate sweeps, an extension the paper leaves implicit.","Systems with spatially correlated disorder, such as shear bands or gradients in annealing, would break the assumption of uncorrelated coupling constants; the Weibull analysis suggests the low-coupling tail, not the variance alone, controls nucleation, so tuning that tail could sharpen or suppress brittleness without changing the average."],"forward_implications":["In the thermodynamic limit N→∞, the model predicts gradual, rounded yielding with solid/fluid coexistence for every finite disorder.","Abrupt (brittle) yielding in a finite sample is predicted whenever the peak correlation length ξ* exceeds the system size N.","The yield strain of a large, weakly disordered system is set by the Maxwell rule, γ_th^(M) = γ_c (4Kr−3)^(−1/n), rather than by the mean-field spinodal thresholds.","The model predicts a genuine critical point at Kr = 1 separating discontinuous from continuous yielding, with second-order behavior at the glass transition.","Because one disorder parameter d controls the whole sharp-to-gradual crossover, the model offers a testable route to design samples with prescribed yielding abruptness."],"supporting_citations":[{"why":"Sets the experimental phenomenology: stroboscopic rheo-DLS showing a two-mode dynamic transition at yielding in several soft glasses, the data the model is built to reproduce.","marker":"[46]"},{"why":"Provides the Van der Waals analogy, the law of corresponding states, the Maxwell rule, and the free-energy minimization logic used to define the yield strain.","marker":"[128]"},{"why":"Supplies the cusp catastrophe manifold classification (bifurcation sets, Maxwell sets) that organizes the mean-field equation of state.","marker":"[142]"},{"why":"Establishes the connection between brittleness, soft spots, and system-spanning plastic events in amorphous solids that the disorder argument extends to the dynamic transition.","marker":"[68]"},{"why":"Provides numerical evidence that structural disorder suppresses spatial correlations and ductilizes yielding, the baseline the model's disorder parameter is compared with.","marker":"[103]"},{"why":"Documents rounded and continuous first-order transitions in spin systems with coupling disorder, the analogy invoked for coexistence at finite disorder.","marker":"[145]"}],"fun_headline_variants":["Yielding's abruptness traced to a diverging correlation length","Disorder smooths yielding; finite samples snap when length grows","Glassy yielding: a finite-size tale of a hidden length scale","Sharp yielding is a finite-size effect, model shows","The abrupt yielding of glasses is a sample-size illusion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire mechanism rests on the specific facilitated-advection coupling term proportional to α_ij/(Γ_i Γ_j) in Eq. 4, which is introduced by analogy with dynamic facilitation rather than derived from any underlying microscopic dynamics, so a different coupling form could change or erase the cusp, the bimodal coexistence, and the divergence of ξ.","fun_headline_variants_meta":{"raw":{"variants":["Yielding's abruptness traced to a diverging correlation length","Disorder smooths yielding; finite samples snap when length grows","Glassy yielding: a finite-size tale of a hidden length scale","Sharp yielding is a finite-size effect, model shows","The abrupt yielding of glasses is a sample-size illusion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000212,"raw_usage":{"total_tokens":1413,"prompt_tokens":936,"completion_tokens":477,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":552,"completion_tokens_details":{"reasoning_tokens":394}},"tokens_in":552,"tokens_out":477,"duration_ms":4820,"temperature":1.0,"reasoning_tokens":394,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T19:24:05.612056+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the connected correlation length of local relaxation rates (for example by imaging tracer motion or stroboscopic DLS in a colloidal glass under oscillatory shear) while systematically reducing sample disorder by deeper annealing: if the peak correlation length does not grow, or if abrupt yielding does not track the condition ξ* ≈ sample size, the central claim is wrong. A direct numerical test is to replace the α_ij/(Γ_i Γ_j) coupling with a linear (Γ_i − Γ_j) coupling and check whether the bimodal coexistence and the ξ divergence survive.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Van der Waals analogy, the law of corresponding states, the Maxwell rule, and the free-energy minimization logic used to define the yield strain."},{"cited_title":"Okni ´nski, Catastrophe Theory, Comprehensive Chemical Kinetics No","cited_arxiv_id":null,"evidence_quote":"Supplies the cusp catastrophe manifold classification (bifurcation sets, Maxwell sets) that organizes the mean-field equation of state."},{"cited_title":"Richard, M","cited_arxiv_id":null,"evidence_quote":"Provides numerical evidence that structural disorder suppresses spatial correlations and ductilizes yielding, the baseline the model's disorder parameter is compared with."},{"cited_title":"Emergent scales and spatial correlations at the yielding transition of glassy materials","cited_arxiv_id":"2501.10039","evidence_quote":"Documents rounded and continuous first-order transitions in spin systems with coupling disorder, the analogy invoked for coexistence at finite disorder."}],"review_version":1}