{"id":"173bc458-732c-42a7-aaa1-00fdba3117a3","arxiv_id":"2501.19197","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Velocity autocorrelation in a uniformly heated granular gas shows waiting-time dependence (aging), with slower decay at longer waiting times.","lead":"This paper simulates a 3D granular gas kept away from equilibrium by a white-noise thermostat and measures how velocity correlations decay over time. It reports that the decay depends on how long the system has been running, a signature of aging, which matters for understanding memory in dissipative systems.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Aging evidence confounded: C(τw,τ) is plotted vs absolute τ, so any lag-dependent stationary correlation will appear to slow with τw; no fixed-lag collapse test is presented.","rationale":"The reader's verdict was CONDITIONAL, with the weakest assumption being the uniform-kick thermostat vs Gaussian white noise. I read the paper in good faith and believe the thermostat discrepancy is real but secondary. The primary issue is that the observable used to infer aging is not analyzed in a way that can discriminate aging from ordinary stationary lag dependence. The paper explicitly claims a non-equilibrium steady state (Section IV: 'This process continues until a state is reached where the losses from dissipation are counterbalanced by the gains from noise'; Fig. 2 shows temperatures stabilizing). In such a state, two-time statistics are invariant under time translation. Therefore the fact that C depends on τw is, by itself, empty: even a perfectly stationary, non-aging process would produce C(τw, τ) that depends on τw when plotted as a function of absolute τ, because the lag τ − τw differs. The demonstration must be that C(τw, τw + Δ), with Δ fixed, changes with τw. No such demonstration appears. The paper's own description of the decay as exponential at early times and slower later makes the visual artifact quantitative: for larger τw, the curve begins in the slower tail, so the apparent slope in τ is smaller. Thus Figs. 3–4 and the abstract's 'compelling evidence' are logically insufficient. This is not a claim that the simulation data are wrong; it is a claim that the reported analysis does not establish the headline result. The remedy is straightforward and fully computational: re-plot the existing trajectories versus lag, or as a function of τw at fixed lag. If the curves collapse, the paper's conclusion should be revised. If they do not collapse after the temperature plateau, then the aging interpretation has real support, though the thermostat details (Eq. 14, uniform kicks, unspecified dt) and the contradictory sentence about a 'decreasing coefficient of restitution' would still need correction. Given this, I do not change the reader's conditional verdict; I would, however, make the re-analysis a mandatory condition rather than a stylistic clarification, because the central claim depends on it. I therefore mark agreement_with_reader as disagree: the reader's identified weakest assumption (uniform vs Gaussian thermostat) is not the most load-bearing concern.","tokens_in":7021,"tokens_out":10883,"duration_ms":115611,"concrete_test":"Using the saved per-particle velocity histories from the 50 independent runs, recompute the normalized autocorrelation at fixed lags: C_n(τw, Δ) = C(τw, τw + Δ)/C(τw, τw) for Δ = 1, 10, 50, 100 and τw = 10, 50, 100, 200, 500, 1000. Plot C_n versus τw for each Δ, and overlay C_n versus Δ for all τw after the temperature plateau of Fig. 2. If all curves collapse onto one master curve (flat in τw), the central aging claim is an artifact of plotting against absolute τ. If they systematically separate in τw at fixed Δ after the plateau, genuine aging remains. Repeat the runs with Gaussian white-noise kicks (standard normal with variance A dt) instead of the uniform ξ in Eq. (14) to confirm the result is thermostat-independent.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing flaw is statistical and precedes the thermostat question. Section IV asserts that the system settles into a non-equilibrium steady state. A steady-state process is time-translation invariant, so any two-time autocorrelation must satisfy C(τw, τ) = C(0, τ − τw). Under this condition, plotting the normalized C(τw, τ)/C(τw, τw) against the absolute time τ necessarily produces different-looking curves for different τw: the horizontal offset τw shifts each curve into a different region of the lag-dependent decay, and if that decay is non-exponential (as the paper itself reports: exponential at early times, slower than exponential later), the later-τw curves appear to decay more slowly. That is exactly the visual pattern in Figs. 3–4 and exactly what the abstract calls 'compelling evidence of aging.' It is not. To establish aging one must show that C(τw, τw + Δ) depends on τw at fixed lag Δ, or equivalently that the curves do not collapse when plotted versus Δ = τ − τw. No such test appears anywhere. If the system is not in fact stationary at the waiting times used, then the observed τw dependence is transient relaxation toward the steady state, not steady-state aging. Either way, the central claim is unsupported as presented.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports event-driven molecular dynamics simulations of 500,000 inelastic hard spheres in a cubic box with periodic boundary conditions and a white-noise thermostat. It defines a two-time velocity autocorrelation function C(τw, τ) and presents normalized curves for restitution coefficients r = 0.80, 0.85, 0.90, and 0.95 and waiting times τw = 10, 50, 100, and 200. The authors observe that the decay of the normalized autocorrelation becomes slower as τw increases and interpret this as evidence of aging. They also compare the decay with predictions for a constant restitution coefficient and conclude that the system retains stronger velocity memory at larger waiting times, especially for strong dissipation.","tokens_in":7438,"tokens_out":6187,"duration_ms":57533,"significance":"If the aging claim were established, the paper would provide useful numerical evidence for two-time velocity correlations in uniformly heated granular gases and would extend existing work on the homogeneous cooling state. The strengths of the manuscript are the large system size (N = 500,000), the average over 50 independent initial conditions, and the systematic variation of the restitution coefficient. However, the quantitative evidence for the central claim is currently missing: no fixed-lag collapse test is performed, no statistical uncertainties are shown, and the thermostat implementation is not clearly connected to the stated Gaussian white-noise model. The claim is therefore plausible but not demonstrated as presented; with additional analysis, the manuscript could become publishable.","major_comments":[{"comment":"The central aging claim is not supported by the plotted data as presented. The normalized autocorrelation C̄(τw, τ)/C̄(τw, τw) is shown against the absolute time τ, so curves with different τw are horizontally shifted relative to one another. For any time-translation-invariant process, C(τw, τ) = C(0, τ − τw), and plotting against τ will produce different-looking curves whenever the decay is not purely exponential; the paper itself reports exponential early decay followed by slower-than-exponential decay. The observed widening separation is therefore exactly what a stationary process would produce, and it cannot be used as evidence of aging. To establish aging, the authors must show that at fixed lag Δ = τ − τw the correlation depends on τw, or equivalently that the curves do not collapse when plotted versus Δ. They should also verify that the system has reached a steady state before the earliest waiting time used; otherwise the τw dependence is transient relaxation toward the steady state rather than steady-state aging.","section":"Section IV, Figs. 3–4"},{"comment":"The thermostat implementation is not equivalent to the stated Gaussian white-noise force of Eqs. (8)–(10) unless additional conditions are met. Equation (14) adds √A√dt ξ with ξ uniformly distributed on [−1/2, 1/2], whose variance is 1/12 rather than 1, and the time step dt is never specified. A fixed-timestep update is also at tension with the event-driven algorithm, which normally advances collision-to-collision. The authors should report dt, state how the uniform noise is mapped to the Gaussian noise (e.g., by matching second moments or by using a Gaussian random number generator), and justify that the resulting steady state is the same as that of the stated stochastic equations.","section":"Section IV, Eq. (14)"},{"comment":"The text contains a direct contradiction about the restitution coefficient. It states that 'the system is not evolving with a constant coefficient of restitution; instead, the coefficient of restitution decreases as the system evolves due to the cooling of the granular gas,' but the simulations are described as one run for each fixed r = 0.80, 0.85, 0.90, and 0.95. If r is in fact time-dependent, the protocol must be specified; if r is fixed, the sentence should be removed. This ambiguity directly affects the comparison with constant-r results in Section V and must be resolved.","section":"Section IV, paragraph after Eq. (15)"},{"comment":"No statistical uncertainties are reported despite the claim of averages over 50 independent initial conditions. On the semilog and linear-log scales used, the differences between curves at successive waiting times can easily be within sampling error. Standard errors, confidence bands, or at least a statement of the run-to-run spread are needed to assess whether the observed τw dependence is significant.","section":"Section IV, Figs. 3–4"},{"comment":"Equation (12) as written, C(τw, τ) = C(τw, τw) exp[(1 + ε)/(2d)(τ − τw)], grows with τ − τw because 1 + ε > 0, and therefore cannot represent the 'decay behavior' described in the text; the sign or normalization is likely wrong. Equation (13) has an unusual nested-exponential structure and is not used explicitly in the figures. These equations should be corrected and either derived or compared quantitatively with the simulation data.","section":"Section III, Eqs. (12)–(13)"}],"minor_comments":[{"comment":"The definition C(τw, τ) = (1/N) Σ_i v_i(τw)·⟨v_i(τ)⟩ is notationally incorrect: the ensemble average should act on the product v_i(τw)·v_i(τ), not on v_i(τ) alone, whose average vanishes by momentum conservation. Also, the bar symbol C̄ used in the figures is never defined.","section":"Section III, Eq. (11)"},{"comment":"The text says results for τw = 0 are plotted, but the figure legends show only τw = 10, 50, 100, and 200; please clarify what is actually shown.","section":"Section IV, Figs. 3–4"},{"comment":"The variable τ is introduced as a cumulative collision count in Eq. (6), but in Eq. (14) and in the figures it is treated as a continuous time; specify the conversion between real time and τ used in the simulations.","section":"Section II, Eq. (6) and simulation figures"},{"comment":"The phrase 'dependence on both waiting time tw and correlation time t in an independent manner' is vague; the precise statement should be that C(τw, τ) is not a function of τ − τw alone.","section":"Abstract and Section III"},{"comment":"There are numerous typographical errors (e.g., 'g as' in the abstract) and inconsistent notation between t and τ; the manuscript needs careful proofreading.","section":"Entire manuscript"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal, but the central claim requires a fixed-lag analysis and error bars before it can be accepted. The internal contradiction about the restitution coefficient and the thermostat ambiguity are fixable in a revision, so I do not see grounds for rejection. I found no evidence of citation problems or duplicate publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the central claim—aging in the velocity autocorrelation—is not supported by the evidence as plotted. The paper plots the normalized VACF against absolute time τ for several waiting times τw. In a steady state, C(τw, τ) must be a function of τ − τw alone. Plotting against τ therefore shifts each curve horizontally; if the lag-dependent decay is non-exponential (which the paper itself reports), larger τw will appear to decay more slowly even with zero aging. The paper never shows a fixed-lag comparison or a collapse test, so the 'compelling evidence' in the abstract is an artifact of the plotting convention. This is the load-bearing flaw.\n\nCredit where due: the simulation is large-scale (N = 500,000, 50 independent runs), event-driven, and the temperature evolution to a steady state is checked against Haff's law. That is solid groundwork. The raw VACF data may well contain real aging; the paper just does not demonstrate it.\n\nSoft spots, in order: (1) Missing fixed-lag analysis—the decisive test. (2) An internal contradiction: Section IV says results are for a constant coefficient of restitution and then says the coefficient decreases as the system evolves. The simulation presumably uses fixed r per run; this needs to be resolved. (3) The thermostat is described as Gaussian white noise, but Eq. (14) draws ξ from a uniform distribution and the time step dt is never given. The uniform distribution has different kurtosis; this may be fine in the small-dt limit, but it needs justification. (4) No error bars on any correlation curve, so we cannot tell whether the waiting-time differences exceed statistical noise.\n\nThe citation pattern is appropriate—the Ben-Naim–Krapivsky HCS results are the right comparison—though Eqs. (12)–(13) look misprinted.\n\nWho this is for: someone working on driven granular gases might be interested in a potential waiting-time dependence of VACF, but as presented the evidence is not trustworthy. I would not cite it. Recommendation: reject this version; invite resubmission after a fixed-lag analysis, with error bars and thermostat parameters clearly stated. Not worth a full referee cycle in its current form.","headline":"The aging claim rests on plotting against absolute time; without a fixed-lag analysis, the central conclusion is an artifact.","tokens_in":7792,"tokens_out":3407,"would_cite":false,"duration_ms":34818,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Velocity correlations in a uniformly heated granular gas age with waiting time.","keywords":["aging","granular gas","velocity autocorrelation","molecular dynamics","event-driven simulation","white-noise thermostat","nonequilibrium steady state","inelastic hard spheres"],"falsifier":"Measure the steady-state velocity distribution and the two-time autocorrelation using a true Gaussian white-noise thermostat and repeat the run with several significantly smaller time steps. If the normalized $C(\\tau_w,\\tau)$ curves collapse onto a single function of $\\tau-\\tau_w$, or if the decay rate changes systematically with $\\mathrm{d}t$, the aging signature is not intrinsic to the heated granular gas. A collapse onto a master curve in $\\tau/\\tau_w$ would instead confirm genuine aging.","tokens_in":6862,"feed_emoji":"⏳","tokens_out":8588,"duration_ms":74966,"temperature":0.7,"pith_summary":"This paper uses event-driven molecular dynamics to simulate a three-dimensional gas of inelastic hard spheres kept out of equilibrium by a white-noise thermostat. The authors compute the two-time velocity autocorrelation function $C(\\tau_w,\\tau)$ and find that its normalized decay becomes slower as the waiting time $\\tau_w$ grows, so the correlation depends on $\\tau_w$ and $\\tau$ separately rather than only on the elapsed time $\\tau-\\tau_w$. They take this explicit waiting-time dependence as evidence that the system exhibits aging. The result matters because it shows that a driven dissipative gas retains a growing memory of its earlier velocities, a signature usually associated with glasses and other non-equilibrium systems, and it offers a numerical benchmark for theories of granular hydrodynamics and aging.","feed_headline":"Aging detected in velocity correlations of heated granular gas","feed_subtitle":"Slower decay with longer waiting time shows a 3D dissipative gas remembers its past velocities.","key_machinery":"The central object is the two-time velocity autocorrelation function $C(\\tau_w,\\tau)=\\frac{1}{N}\\sum_{i}\\mathbf{v}_i(\\tau_w)\\cdot\\langle\\mathbf{v}_i(\\tau)\\rangle$, evaluated after the system reaches its nonequilibrium steady state. The simulation uses event-driven molecular dynamics for $N=500{,}000$ identical inelastic hard spheres with periodic boundaries, and a white-noise thermostat implemented by adding a random velocity kick $\\sqrt{A}\\,\\sqrt{\\mathrm{d}t}\\,\\xi$ to each particle at every time step, followed by a shift to the center-of-mass frame. The diagnostic is the set of decay curves of $C(\\tau_w,\\tau)$ at fixed $\\tau_w$: an equilibrium system would collapse them to a function of $\\tau-\\tau_w$, while here they separate by $\\tau_w$, which the paper reads as aging. Haff's law for the cooling stage sets the temperature scale and frames the comparison with the freely cooling gas.","core_discovery":"The central discovery is that in a three-dimensional uniformly heated granular gas, the normalized velocity autocorrelation function decays exponentially at short times and then crosses over to a slower-than-exponential decay, with the decay rate decreasing as the waiting time $\\tau_w$ increases. This two-time structure means the system's velocity memory cannot be described by a function of the time difference alone; the explicit dependence on $\\tau_w$ is the paper's evidence of aging. The effect is most pronounced for the most inelastic case studied, $r=0.80$, where the decay is slowest, while weaker dissipation ($r=0.95$) shows faster decorrelation. The authors compare with the known behavior of freely cooling gases and with systems governed by a constant restitution coefficient, and report that the heated system's decay is consistently faster than in the constant-$r$ case while still displaying aging.","pith_inferences":["One could test whether the aging data obey a scaling collapse, for example $C(\\tau_w,\\tau)/C(\\tau_w,\\tau_w)$ as a function of $\\tau/\\tau_w$; the paper does not report such a collapse, but if it works it would connect this system to the aging phenomenology of glasses and spin glasses.","A direct measurement of the spatial velocity correlation length as a function of $\\tau_w$ would test the paper's attribution of slow decay to growing velocity-field correlations; if the correlation length stays constant while the decay slows, the proposed mechanism would need revision.","Because the thermostat uses uniform rather than Gaussian kicks, repeating the simulation with a proper Gaussian white-noise integrator would separate heating-protocol effects from physics; the unspecified time step $\\mathrm{d}t$ leaves convergence in step size as an open check.","If the aging is genuine, effective transport coefficients extracted from one-time averages in driven granular gases may depend on preparation time, which could matter for industrial and astrophysical applications of granular flow."],"forward_implications":["If the aging claim holds, two-time velocity correlations in a heated granular gas cannot be reduced to a function of elapsed time alone; the system's memory grows with its age.","Stronger dissipation (lower restitution coefficient $r$) produces a longer-lived velocity memory, so aging is more pronounced in more inelastic gases.","The early-time exponential decay followed by a slower-than-exponential tail means no single relaxation time describes the dynamics.","Because the decay with a variable, cooling-induced effective restitution is faster than with a fixed restitution coefficient, the heating protocol itself enters the aging rate and must be controlled in comparisons."],"supporting_citations":[{"why":"Supplies the collision-frequency expression used in Haff's law for the cooling phase.","marker":"[21]"},{"why":"Provides the white-noise thermostat algorithm used to inject energy into the system.","marker":"[19]"},{"why":"Companion thermostat formulation for the heating mechanism used in the simulations.","marker":"[20]"},{"why":"Event-driven molecular dynamics method used to evolve the inelastic hard-sphere collisions.","marker":"[24]"},{"why":"Event-driven molecular dynamics implementation used for the hard-sphere system.","marker":"[25]"},{"why":"Gives Haff's law for the homogeneous cooling state that frames the temperature decay.","marker":"[7]"}],"fun_headline_variants":["Heated granular gas shows aging in velocity correlations","Velocity memory in 3D granular gas fades slower with age","Aging velocity correlations emerge in heated 3D granular gas","Longer waiting time slows velocity decay in heated granular gas","Aging in heated granular gas: velocity correlations decay slower with wait time"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The simulation heats the gas by adding uniform random velocity kicks at each discrete time step $\\mathrm{d}t$ and treats this as equivalent to the Gaussian white-noise thermostat in the equations of motion, without specifying $\\mathrm{d}t$ or showing that the steady state is the same; if this equivalence fails, the observed waiting-time dependence could be an artifact of the heating protocol rather than a genuine aging property of the granular gas.","fun_headline_variants_meta":{"raw":{"variants":["Heated granular gas shows aging in velocity correlations","Velocity memory in 3D granular gas fades slower with age","Aging velocity correlations emerge in heated 3D granular gas","Longer waiting time slows velocity decay in heated granular gas","Aging in heated granular gas: velocity correlations decay slower with wait time"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000709,"raw_usage":{"total_tokens":3203,"prompt_tokens":962,"completion_tokens":2241,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":578,"completion_tokens_details":{"reasoning_tokens":2156}},"tokens_in":578,"tokens_out":2241,"duration_ms":14687,"temperature":1.0,"reasoning_tokens":2156,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T20:56:29.347505+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the steady-state velocity distribution and the two-time autocorrelation using a true Gaussian white-noise thermostat and repeat the run with several significantly smaller time steps. If the normalized $C(\\tau_w,\\tau)$ curves collapse onto a single function of $\\tau-\\tau_w$, or if the decay rate changes systematically with $\\mathrm{d}t$, the aging signature is not intrinsic to the heated granular gas. A collapse onto a master curve in $\\tau/\\tau_w$ would instead confirm genuine aging.","supporting_citations":[{"cited_title":"Chapman and T","cited_arxiv_id":null,"evidence_quote":"Supplies the collision-frequency expression used in Haff's law for the cooling phase."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the white-noise thermostat algorithm used to inject energy into the system."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Companion thermostat formulation for the heating mechanism used in the simulations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Event-driven molecular dynamics method used to evolve the inelastic hard-sphere collisions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Event-driven molecular dynamics implementation used for the hard-sphere system."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives Haff's law for the homogeneous cooling state that frames the temperature decay."}],"review_version":1}