{"id":"12311e6b-0f69-40a6-bc2b-aa3a4f6c4f87","arxiv_id":"2411.12390","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Resetting an unstable resource-redistribution model frequently enough makes its mean and variance converge, turning a non-ergodic process into a stationary one.","lead":"This paper studies what happens when a popular model of resource redistribution, the reallocating geometric Brownian motion, is repeatedly interrupted by stochastic resets that send the system back to a starting point. The authors find that frequent enough resetting can stabilize an otherwise unstable and unequal wealth distribution, and they identify the exact resetting rate where this stabilization begins.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The infinite self-averaging claim rests on cross-moment equations (Eq. 14–15) whose resetting terms are wrong for both common and independent resetting; the factor-of-two error invalidates the RN(t) calculation.","rationale":"The reader's weakest assumption correctly flags the mean-field closure and the use of the stationary mean in the Fokker-Planck equation, but my stress-test identifies a sharper, more load-bearing problem: the cross-moment equations used to compute the relative variance are algebraically wrong. The central claim has two parts: the moment thresholds in Table 1 and the infinite self-averaging time. The moment thresholds can likely survive under the correct independent-resetting model, because the stationary second-moment condition r>2(mu-tau)+sigma^2 follows from the exact single-particle equation once self-averaging is assumed. However, the self-averaging claim is not proven by the paper: the RN(t) calculation in Section 3.1 is based on Eq. 15, which double-counts the resetting rate in the v equation and uses an incorrect reset source in the q equation. This is not merely a missing justification; it is a concrete inconsistency that can be exhibited in the tau=0 limit, where the paper claims to recover known srGBM results but instead obtains a different stationary second moment. The simulation evidence in Figures 1, 3, and 5 supports the moment stabilization, and the paper provides reproducible code, so I do not recommend rejection. But the analytical derivation of the self-averaging claim must be corrected and the resetting mechanism specified before the central classification is fully established. Hence CONDITIONAL remains the appropriate verdict, with the condition being a corrected derivation of Eqs. 14–16 and a clear statement of common versus independent resetting.","tokens_in":16985,"tokens_out":22272,"duration_ms":211348,"concrete_test":"Re-derive Eq. 15 from Eq. 8 for the case of independent Poisson resets and compare with Monte Carlo simulation of the N-particle RGBM at the claimed threshold r_c=2(mu-tau)+sigma^2. As an analytical checkpoint, set tau=0: the exact single-particle second moment of GBM with resetting is v_st = r x0^2/(r-2mu-sigma^2), while Eq. 15 yields v_st = r x0^2/[2(r-mu)-sigma^2]. If the corrected v,q ODEs match the simulated RN(t) and Eq. 15 does not, then Section 3.1 and the infinite self-averaging claim must be revised; if RN(t) still vanishes only above r_c, Table 1 survives but needs a different derivation.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The headline 'infinite self-averaging time' is established through the relative variance RN(t) in Eqs. 13–16, which requires the cross moments v(t)=E[xi^2] and q(t)=E[xi xj]. The equations for v and q (Eq. 15, from Eq. 14) are internally inconsistent with Eq. 8 and with any definite resetting mechanism. For f=xi^2, Eq. 8 gives a reset contribution r(E[x0^2]-E[xi^2]), i.e. a term -r v, so v' should contain 2mu+sigma^2-r (plus mean-field terms). Eq. 15 instead contains 2mu+sigma^2-2r, a factor of two in the resetting rate. For f=xi xj with independent per-agent resets, the jump contribution is 2r x0 E[x]-2r q (either particle can reset); Eq. 15 uses r x0^2 instead of 2r x0 E[x] while keeping the -2r q coefficient. With common resets, the q equation should have -r q and r x0^2, not -2r q. Thus Eq. 15 corresponds to neither mechanism. The tau=0 limit is a direct witness: the standard srGBM second moment is v_st = r x0^2/(r-2mu-sigma^2), whereas Eq. 15 gives v_st = r x0^2/[2(r-mu)-sigma^2]; for mu=0.021, sigma^2=0.01, r=0.1 these differ by more than a factor of 3, and Eq. 15 can even make RN negative. Because RN(t)->0 is the only analytical support for 'effectively infinite self-averaging time,' this part of the central claim is currently unsupported, even though the stationary moment threshold in Table 1 may survive under the correct independent-resetting equations.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper analyzes the Reallocating Geometric Brownian Motion (RGBM) model subject to stochastic resetting, focusing on the negative reallocation regime (τ<0) where the standard model is non-stationary and non-ergodic. The authors derive moment equations, identify three long-time regimes for the first and second moments as functions of the resetting rate r, and compute a relative variance R_N(t) to argue that beyond a critical resetting rate the self-averaging time becomes effectively infinite. They complement the analysis with numerical simulations and apply the results to wealth mobility and inequality measures.","tokens_in":17279,"tokens_out":18580,"duration_ms":170699,"significance":"If the central claims were fully substantiated, the paper would be a useful contribution to the stochastic resetting literature: it generalizes srGBM results to a mean-field reallocation model, provides a simple regime diagram (Table 1), and connects the theory to economic mobility measures. The numerical simulations, the explicit GitHub code, and the recovery of the τ=0 srGBM limit are strengths. However, the analytical derivation of the self-averaging time rests on cross-moment equations that contain resetting terms inconsistent with the model's dynamics, and the ensemble mean is given by two incompatible expressions; these issues must be resolved before the regime classification and the 'infinite self-averaging time' claim can be accepted.","major_comments":[{"comment":"The resetting contributions in the cross-moment equations are inconsistent with Eq. (8). For f = x_i^2, Eq. (8) yields a reset term r(x_0^2 - v), so dv/dt should contain -r v, not -2r v as in Eq. (15). For f = x_i x_j with independent per-agent resetting, the jump contribution is 2r x_0 E[x] - 2r q (and r(x_0^2 - q) for common resetting), whereas Eq. (15) uses r x_0^2 - 2r q, matching neither mechanism. As a result, the relative variance R_N(t) in Eq. (16) and the associated critical resetting rate for t_c → ∞ are not reliably derived. A direct witness is the τ = 0 limit, where Eq. (15) gives v_st = r x_0^2/[2(r-μ)-σ^2] instead of the standard srGBM result r x_0^2/(r-2μ-σ^2), a discrepancy of more than a factor of 3 for the parameters used in the paper (μ=0.021, σ^2=0.01, r=0.1). This leaves the analytical support for the 'effectively infinite self-averaging time' claim in Sec. 3.1 and Table 1 currently unsupported.","section":"Sec. 3.1, Eqs. (14)-(16)"},{"comment":"The paper gives two different expressions for the ensemble mean that are not equivalent for τ≠0. Eq. (10) follows from Eq. (8) under the self-averaging closure E[⟨x⟩_N]=E[x] and yields dE[x]/dt = (μ-r)E[x] + r x_0, with no τ in the transient rate. Eq. (21), derived from the Fokker-Planck equation (19) with ⟨x⟩_N replaced by its stationary value, has transient rate r - μ + τ. The exact finite-N moment equation for m(t)=E[x_i] from Eq. (4) gives the Eq. (10) form because the τ terms cancel when E[⟨x⟩_N]=E[x]. Consequently, the first row of Table 1 (convergence for r > μ-τ) appears incorrect; the mean converges for r > μ. This changes the boundary between the first and second regimes, although the stationary value (Eq. (23)) is unaffected.","section":"Sec. 3 and Sec. 4.1, Eqs. (10) and (21)"},{"comment":"The Fokker-Planck equation (19) substitutes the stationary mean-field value ⟨x⟩_N = r/(r-μ) x_0 (Eq. (20)) into the time-dependent drift. This is only justified in the long-time stationary limit, not for the transient dynamics. Using Eq. (19) to derive the time-dependent moments (21)-(22) and to compare with simulations (Fig. 3) is therefore not valid in the transient; this is the source of the spurious τ dependence in the mean that leads to the incorrect threshold in Table 1.","section":"Sec. 4.1, Eq. (19)"},{"comment":"The derivation of the critical resetting rate mixes exact finite-N moment equations with the self-averaging assumption (Eq. (11)) in a way that is not transparent. Equations (14)-(15) for v(t) and q(t) close exactly for finite N without invoking Eq. (11), whereas the denominator E[x_i(t)] in Eq. (16) is taken from Eq. (10), which does assume E[⟨x⟩_N]=E[x]. The paper should separate these ingredients and state which results are exact and which are self-consistent approximations; as written, the logic is difficult to verify and the 'infinite self-averaging time' conclusion relies on a closure that is itself under test.","section":"Sec. 3.1"}],"minor_comments":[{"comment":"There is a stray equals sign in the displayed relation; it should read E[f(x)] = lim_{N→∞} ⟨f(x)⟩_N.","section":"Eq. (11)"},{"comment":"In the second line of Eq. (14), the second sum should be over k≠j E[x_k x_j] rather than over k≠j E[x_k x_i].","section":"Eq. (14)"},{"comment":"The condition for t_c = ∞ should be stated precisely, e.g., R_N(t) < 1 for all t, and an explicit formula or numerical value for the critical rate r_c should be provided rather than only plots.","section":"Sec. 3.1"},{"comment":"The notation ⟨x⟩_N is described as both 'ensemble or population average' and later as the empirical average; please clarify the distinction between ⟨x⟩_N and E[x] throughout.","section":"Sec. 2"},{"comment":"The x-axis label in the left panel appears truncated ('0 10'); please check the axis range and labels.","section":"Fig. 5"},{"comment":"The text states that the critical point for variance convergence occurs at r_c = 2(μ-τ)+σ^2, but this is not derived from the corrected moment equations; please ensure consistency with the corrected R_N(t) analysis.","section":"Sec. 4.2"}],"recommendation":"major_revision","confidential_remarks":"The paper's core idea is publishable if the moment calculations are corrected. The current version has load-bearing errors in the cross-moment equations (Eq. 15), in the ensemble mean expressions (Eq. 10 vs Eq. 21), and in the use of the stationary mean field in the transient Fokker-Planck equation. These issues affect the central claims about the regime boundaries and the infinite self-averaging time, so I cannot recommend acceptance in the present form. The authors should re-derive the moment equations with the correct resetting terms and re-evaluate Table 1 and the R_N(t) analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe short version: this is a useful paper with a load-bearing algebraic error in its most advertised claim. The combination of stochastic resetting with reallocating geometric Brownian motion is new, and the three-regime moment classification with the tau-shifted threshold r_c = 2(mu - tau) + sigma^2 is plausible and supported by simulations. But the paper's route to \"infinite self-averaging time\" runs through cross-moment equations that contain a factor-of-two mistake in the resetting terms. As written, that headline claim does not follow from the model.\n\nCredit where due: the resetting + RGBM combination is a natural extension that was missing from the literature; the simulations are provided via a public GitHub repository; and the authors are candid about the limitation that resetting eliminates the multiplicative growth of mean wealth, which is a real empirical constraint. The exposition is readable and the figures are informative.\n\nThe soft spots. Eq. 10 and Eq. 21 give two different transient expressions for the ensemble mean, with decay rates differing by tau; they cannot both be right. The model never specifies whether resetting is common or independent, and the moment equations in Eq. 14-15 are compatible with neither. For v = E[x_i^2], Eq. 8 gives a reset contribution -r v, but Eq. 15 uses -2r v. For q = E[x_i x_j], independent resets give +2r x0 E[x] - 2r q, and common resets give +r x0^2 - r q; Eq. 15 instead uses r x0^2 with -2r q. The tau=0 limit is the cleanest witness: Eq. 15 does not reproduce the known reset-GBM second moment from [67]. Since RN(t) -> 0 is the only analytical support for the infinite self-averaging time, that part of the central claim is unsupported. The moment thresholds in Table 1 may survive a corrected derivation, but the paper does not currently establish the self-averaging result. The mean-field ansatz E[f(x)] = <f(x)>_N is also invoked to close the equations and then used to derive the rate at which it holds; that is a self-consistency argument, not a proof.\n\nWho should read it: modelers of wealth and resource dynamics who want the tau-shifted moment thresholds and a concrete numerical demonstration of resetting-induced stabilization. I would send it to a serious referee, but with the explicit expectation of major revision: fix the moment equations, reconcile the two mean formulas, and redo the self-averaging analysis. The raw material is there; the advertised conclusion needs an honest redo before it can be cited.","headline":"A useful extension with a correct-looking moment classification, but the advertised infinite-self-averaging-time result rests on cross-moment equations with a factor-of-two resetting error.","tokens_in":17926,"tokens_out":6193,"would_cite":false,"duration_ms":56135,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H10","60J60","82C31"],"pacs":["05.40.-a","05.10.Gg"],"model":"deepseek-v4-flash","headline":"Stochastic resetting tames the mean-repulsive regime of the Reallocating Geometric Brownian Motion, making both first and second moments stationary above a critical rate.","keywords":["stochastic resetting","reallocating geometric Brownian motion","non-ergodicity","self-averaging","wealth inequality","economic mobility","Fokker-Planck equation","jump-diffusion process"],"falsifier":"Simulate $N$ agents with $\\tau<0$ and $r>2(\\mu-\\tau)+\\sigma^2$, compute the relative variance of the population mean $R_N(t)$ at long times, and compare the measured stationary mean and MSD with Eqs. (23)--(24); if $R_N(t)$ does not decay to zero or the moments do not match, the claimed infinite self-averaging time and stationary moments are false.","tokens_in":16703,"feed_emoji":"🔄","tokens_out":11345,"duration_ms":101893,"temperature":0.7,"pith_summary":"Stochastic resetting—periodically returning each agent's resource share to a common baseline—can stabilize the otherwise non-ergodic, mean-repulsive regime of the Reallocating Geometric Brownian Motion (RGBM), a standard model of resource redistribution in economics, ecology, and evolution. The paper establishes a threshold classification: for negative reallocation $\\tau<0$, the first moment converges when the resetting rate exceeds $\\mu-\\tau$, and the second moment converges when it exceeds $2(\\mu-\\tau)+\\sigma^2$. Above the larger threshold the full distribution becomes stationary and the population average tracks the ensemble average indefinitely, so the process behaves as if it were ergodic. This matters because it turns an unstable, concentrating dynamics into one whose mean and spread are predictable, and in wealth settings it shows a trade-off: resetting reduces inequality and raises mobility, but also lowers total growth.","feed_headline":"Stochastic resetting can make an unstable resource process stationary","feed_subtitle":"Above the critical resetting rate, both mean and variance converge and the process becomes self-averaging.","key_machinery":"The key object is the jump-diffusion representation of RGBM with resetting, where a Poisson process with intensity $r$ sends the process back to $x_0$ while the remaining time the dynamics follow RGBM (Eq. 4). Itô's formula applied with $f(x)=x$ and $f(x)=x_ix_j$ yields coupled moment equations for $v(t)=E[x_i^2]$ and $q(t)=E[x_ix_j]$, which close under the self-averaging ansatz $E[x]=\\langle x\\rangle_N$. The relative variance $R_N(t)=\\mathrm{var}(\\langle x\\rangle_N)/E[\\langle x\\rangle_N]^2$ then serves as the self-averaging criterion: the critical resetting rate is the one that makes $R_N(t)$ vanish in the long-time limit. Requiring the second-moment exponent to be negative gives $r_c=2(\\mu-\\tau)+\\sigma^2$, and the same condition makes the Fokker--Planck equation (19) well defined with a stationary mean field.","core_discovery":"The central claim is that in RGBM with resetting and negative reallocation, the long-time behavior is fully classified by two thresholds. For $r<\\mu-\\tau$, the process keeps the non-ergodic, diverging character of standard RGBM. For $\\mu-\\tau<r<2(\\mu-\\tau)+\\sigma^2$, the mean reaches a stationary value but the variance still diverges, so no stationary distribution exists. For $r>2(\\mu-\\tau)+\\sigma^2$, both the first and second moments converge to the closed-form stationary values in Eqs. (23)--(24), the Fokker--Planck density stabilizes, and the self-averaging time becomes effectively infinite, meaning the population average equals the ensemble average for all practical purposes. The paper backs this with numerical simulations and with mobility statistics: rank correlation and earnings elasticity fall as resetting increases, and the probability of top-1% states decreases and stabilizes.","pith_inferences":["The same threshold logic applied to the $n$th moment would predict convergence for $r>n(\\mu-\\tau)+n(n-1)\\sigma^2/2$; checking this numerically would reveal whether the stabilized distribution is fully captured by its first two moments or retains heavy tails.","The paper's trade-off between equality and growth suggests an optimal resetting rate for a social-welfare objective that weights both concentration and total wealth; the authors flag this balance but do not solve the optimization.","One could test the stabilization mechanism on other non-ergodic multiplicative processes, such as resetting to a distribution rather than a point, to see whether the threshold structure survives."],"forward_implications":["Above $r_c$, the stationary mean and mean-square displacement are given by the explicit formulas in Eqs. (23)--(24), so one can compute long-run resource levels and dispersion without simulation.","For resetting rates between $\\mu-\\tau$ and $r_c$, the mean is stationary but the variance diverges, so interventions that stabilize only the average still leave the distribution itself unstable.","In the stabilized regime, the equality between population average and ensemble average holds indefinitely, which makes the non-ergodic RGBM behave like an ergodic system for practical measurements.","In wealth-redistribution applications, increasing the resetting rate above $r_c$ lowers wealth concentration (fewer top-1% states) and increases mobility, at the cost of a smaller total pool."],"supporting_citations":[{"why":"Defines stochastic resetting and the stationary-state mechanism that the paper imports into RGBM.","marker":"[32]"},{"why":"Supplies the RGBM moment equations and non-ergodicity results that the resetting analysis extends.","marker":"[30]"},{"why":"Provides the no-reallocation srGBM moment regimes that the $\\tau=0$ limit of the new thresholds recovers.","marker":"[67]"},{"why":"Gives the Poissonian jump-diffusion representation that underlies the analytical moment derivation.","marker":"[69]"},{"why":"Defines the relative-variance self-averaging criterion used to locate the critical resetting rate.","marker":"[82]"},{"why":"Establishes the baseline mobility measures for RGBM against which the resetting effects are compared.","marker":"[66]"}],"fun_headline_variants":["Resetting tames unstable resource dynamics","Critical resetting rate stabilizes resource allocation","Stochastic resetting achieves stationary resource distribution","Above a critical rate, resetting yields a stable resource state","Optimal resetting reduces inequality in resource models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation leans on the assumption that the population average equals the ensemble average, $E[x]=\\langle x\\rangle_N$, including after replacing the time-dependent mean field by its stationary value in the Fokker--Planck equation; if this equality fails in the stabilized regime, the derived moments and thresholds do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Resetting tames unstable resource dynamics","Critical resetting rate stabilizes resource allocation","Stochastic resetting achieves stationary resource distribution","Above a critical rate, resetting yields a stable resource state","Optimal resetting reduces inequality in resource models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000285,"raw_usage":{"total_tokens":1669,"prompt_tokens":929,"completion_tokens":740,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":545,"completion_tokens_details":{"reasoning_tokens":670}},"tokens_in":545,"tokens_out":740,"duration_ms":8127,"temperature":1.0,"reasoning_tokens":670,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:36:08.946859+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate $N$ agents with $\\tau<0$ and $r>2(\\mu-\\tau)+\\sigma^2$, compute the relative variance of the population mean $R_N(t)$ at long times, and compare the measured stationary mean and MSD with Eqs. (23)--(24); if $R_N(t)$ does not decay to zero or the moments do not match, the claimed infinite self-averaging time and stationary moments are false.","supporting_citations":[{"cited_title":"Evans and Satya N","cited_arxiv_id":null,"evidence_quote":"Defines stochastic resetting and the stationary-state mechanism that the paper imports into RGBM."},{"cited_title":"Ergodicity breaking in wealth dynamics: The case of reallocating geometric brownian motion","cited_arxiv_id":null,"evidence_quote":"Supplies the RGBM moment equations and non-ergodicity results that the resetting analysis extends."},{"cited_title":"Geometric brownian motion under stochastic resetting: A stationary yet nonergodic process","cited_arxiv_id":null,"evidence_quote":"Provides the no-reallocation srGBM moment regimes that the $\\tau=0$ limit of the new thresholds recovers."},{"cited_title":"Stochastic representation of processes with resetting","cited_arxiv_id":null,"evidence_quote":"Gives the Poissonian jump-diffusion representation that underlies the analytical moment derivation."},{"cited_title":"Ergodicity economics","cited_arxiv_id":null,"evidence_quote":"Defines the relative-variance self-averaging criterion used to locate the critical resetting rate."},{"cited_title":"Measures of physical mixing evaluate the economic mobility of the typical individual","cited_arxiv_id":null,"evidence_quote":"Establishes the baseline mobility measures for RGBM against which the resetting effects are compared."}],"review_version":1}