{"id":"ac12d97f-1367-4807-a1f2-531418ea2e3d","arxiv_id":"2411.18775","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":5,"one_line_summary":"A particle in a heterogeneous Brownian bath converges, as the bath grows, to fractional Brownian motion and related Gaussian processes with random diffusion coefficient and random Hurst exponent.","lead":"This paper proves that a test particle pushed by a huge number of Brownian particles with randomly distributed masses moves, in the large-N limit, like a general Gaussian process whose memory is set by the mass distribution, including fractional Brownian motion. It supplies a mechanical origin for superstatistical and random-Hurst models of anomalous diffusion that were previously used only as fitting tools.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The limit theorem appears mathematically sound, but the physical-interpretation claim that conservation of momentum is met is false for system (6), which would undermine the advertised 'dynamical foundation' if not corrected.","rationale":"The mathematical core of the paper, Theorem 2.1 and Theorem 3.1, is amply supported by the detailed proofs in Section 4: the three-step approximation, the conditional Gaussian structure, the fourth-moment Borel-Cantelli argument, and the tightness estimates are all internally consistent given the stated assumptions. I checked the exponent bookkeeping in Lemmas 4.2-4.6 and the parameter ranges in the examples; they match Assumption 2.6 and the covariance constant. Thus no mathematical error was found that would invalidate the limit theorem. The load-bearing concern is instead about the physical-interpretation claim that anchors the paper's stated goal. The abstract and Section 2 assert that conservation of momentum and energy are met, but system (6) is a thermostatted Langevin system: the surround particles have independent damping and noise, so the total momentum of test-particle plus surround is not conserved. The coupling terms alone do not form an action-reaction pair, as beta_k U and alpha_k V do not generally balance. This is not an error in the proof, since the proof never uses momentum conservation, but it is a real overstatement of the physical foundation: the model is an open system in a thermal bath, not a closed conservative system whose large-N limit produces fBm. The reader flagged this issue in the rationale but chose the exponent balance as the weakest assumption; I regard the conservation overclaim as the more concrete and easily verified defect. Since the math survives and the fix is a revision of the physical claims rather than a change of theorems, the reader's conditional verdict remains appropriate, so I do not recommend altering the verdict. A simple momentum-balance calculation would settle the matter and strengthen the paper by forcing an accurate description of the model as a non-conservative, bath-driven system.","tokens_in":28568,"tokens_out":33739,"duration_ms":304627,"concrete_test":"Compute the total momentum P_t = M V^N_t + sum_k m_{k,N} U^{k,N}_t from system (6) and evaluate P_{t+h} - P_t exactly. The alpha V terms cancel, but the terms sum_k beta_{k,N} U^{k,N}_t dt and sum_k m_{k,N} sqrt(2sigma) dW^k_t remain, so E[(P_{t+h} - P_t)^2] > 0 for every h > 0; this shows momentum is not conserved. This one-line computation settles the conservation claim directly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract and Section 2 state that the considered particle system satisfies conservation of momentum and energy. For system (6), computing the total momentum P_t = M V^N_t + sum_k m_{k,N} U^{k,N}_t yields dP_t = sum_k (beta_{k,N} - gamma m_{k,N}^2) U^{k,N}_t dt + sum_k m_{k,N} sqrt(2sigma) dW^k_t, because the alpha V terms cancel between the equations for V^N and U^{k,N}, while the thermostat damping -gamma m U and the independent noise remain. Hence P_t is not conserved; it is driven by the thermal forcing of the bath. This is not fatal for the limit theorem, whose proof uses only the linear SDE structure, but it invalidates the claim that the system is a conservative mechanical foundation for fBm. The paper should either restrict the word 'conservation' to the coupling terms and the fluctuation-dissipation relation, or present a genuinely momentum-conserving Hamiltonian or Newtonian system with a scaling limit. As written, the advertised physical foundation overstates what system (6) actually delivers.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a test particle coupled to N underdamped Langevin bath particles with random, heterogeneous masses and with coupling constants that scale as \\(\\alpha_{k,N}\\simeq C_\\alpha N^{-a}\\), \\(\\beta_{k,N}\\simeq C_\\beta N^{-b}\\). The main result (Theorem 2.1) states that, as \\(N\\to\\infty\\), the test-particle position converges to a centered Gaussian process with stationary increments and covariance \\(D_0(v(t)+v(s)-v(|t-s|))\\), where \\(v\\) is determined by the limiting mass distribution; for a stable-subordinator mass distribution this gives fractional Brownian motion with \\(H\\in(1/2,1)\\). Section 3 adds a random coupling coefficient \\(A\\) and a random Hurst-type parameter \\(H\\), yielding conditionally Gaussian superstatistical models, and Remark 3.2 derives associated Kolmogorov--Fokker--Planck equations. The proof is self-contained and proceeds through three explicit approximation steps, using mixing convergence and detailed \\(L^2\\) error estimates.","tokens_in":28845,"tokens_out":16369,"duration_ms":229869,"significance":"The mathematical core is a genuine limit theorem: the proof is self-contained, all lemmas carry explicit error rates, the mass distribution is an input rather than a fitted target, and no part of the argument is circular. This gives a rigorous route from a particle system to fractional Brownian motion and to a broad family of related Gaussian and conditionally Gaussian processes, which is of real value for the anomalous-diffusion literature. However, two load-bearing presentation points need attention: the advertised conservation of momentum is false for system (6), and the covariance constant in the main theorem statement is inconsistent with the proof and with the examples. These issues are correctable but currently undermine the physical interpretation and the exact statement of the limit.","major_comments":[{"comment":"The limiting covariance prefactor in Theorem 2.1 and Lemma 4.4 is inconsistent with the proof. From equations (34) and (35), one obtains \\(E[\\operatorname{Cov}(\\tilde Z^N_t,\\tilde Z^N_s\\mid M)]\\to (\\sigma/\\gamma^2)(C_\\beta^2C_\\delta/C_\\alpha^2)(v(t)+v(s)-v(|t-s|))\\), because the factor \\(1/2\\) in (34) combines with the factor \\(2\\sigma/\\gamma^2\\) in (35) and because \\(2(a-b)-\\delta=1\\) kills the \\(N\\) dependence. The theorem and Lemma 4.4 instead display \\(\\bigl(2\\sigma C_\\beta^2C_\\delta/(\\gamma^2C_\\alpha^2)\\bigr)^{1/2}\\) as the prefactor. Remark 2.1's constant \\(D=\\sigma C_\\beta^2C_\\delta/(\\gamma^2C_\\alpha^2)\\) agrees with the derivation from (34)--(35), while the theorem statement and the formulas in Example 3.1(i) do not. This must be reconciled throughout the paper, including the displayed special cases.","section":"Theorem 2.1, Lemma 4.4, Example 3.1"},{"comment":"The paper claims that conservation of momentum is met by system (6). This is not the case. For \\(P_t=M V^N_t+\\sum_{k=1}^N m_{k,N}U^{k,N}_t\\), direct differentiation gives \\(dP_t=\\sum_{k=1}^N(\\beta_{k,N}-\\gamma m_{k,N}^2)U^{k,N}_t\\,dt+\\sum_{k=1}^N m_{k,N}\\sqrt{2\\sigma}\\,dW^k_t\\): the \\(\\alpha\\) terms cancel, while the \\(\\beta\\) coupling and the independent thermal noise remain. Hence total momentum is not conserved. The limit theorem does not use momentum conservation, so the mathematical result survives, but the advertised 'dynamical foundation' overstates what system (6) delivers. Please either restrict the conservation statement to the coupling forces and the fluctuation--dissipation relation, or provide a genuinely momentum-conserving Hamiltonian or Newtonian system with the same scaling limit.","section":"Abstract, Section 2"},{"comment":"The derivation of the generalized Kolmogorov--Fokker--Planck equation differentiates \\(\\mathbb E[e^{-ADv_H(s)p^2/2}]\\) with respect to \\(s\\) and then applies the chain rule and Fubini's theorem. For the general \\(v_h\\) arising from Bernstein functions, \\(v_h\\) is only continuous and nondecreasing, not necessarily differentiable. The argument therefore requires additional regularity assumptions on \\(v_h\\) and dominated-convergence conditions for the interchange of expectation and differentiation. Please state these hypotheses explicitly, or present the result as a mild/integral evolution equation rather than a differential one.","section":"Remark 3.2"}],"minor_comments":[{"comment":"For \\(\\nu(dy)=e^{-y}y^{-1}dy\\) the Laplace exponent is \\(\\Phi(\\lambda)=\\log(1+\\lambda)\\), not \\(\\log(1-\\lambda)\\); the displayed formula is undefined for \\(\\lambda>1\\) and is not a Bernstein function.","section":"Example 2.2"},{"comment":"The symbol \\(H\\) is used both for the random vector and for its state space \\(\\mathcal H\\); this is confusing in Example 3.1 and should be resolved by a distinct notation for the state space.","section":"Section 3"},{"comment":"The exact exponent balance \\(2(a-b)-\\delta=1\\) and the inequality \\(b>d\\) are imposed as scaling assumptions, and the proof shows that they are precisely what makes the error estimates vanish. It would be helpful to state explicitly whether they are also necessary and to comment on what happens if they fail by a small amount, since the advertised physical foundation currently rests on this tuning.","section":"Assumption 2.6"},{"comment":"The text contains the typo 'Schwarz space'; it should be 'Schwartz space'.","section":"Remark 3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper contains a substantial and mostly rigorous limit theorem, but the covariance prefactor in the main theorem and in the superstatistical examples is internally inconsistent with the proof, and the momentum-conservation claim is false for the stated system. Both issues are fixable within the manuscript's scope, but they affect the central statements and the advertised physical message, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline: this is a rigorous limit theorem, the first of its kind, turning an underdamped Langevin system with heterogeneous masses into fBm and, with extra randomization, superstatistical fBm with random Hurst parameter. The proof is clean and self-contained; explicit error estimates, mixing convergence, and the examples all check out. The stable-subordinator case yields H in (1/2,1); mixtures give time-dependent anomalous exponents; the random-Cα and random-H extension is a natural consequence of Theorem 2.1. The paper deserves a serious referee.\n\nThe soft spot is the advertised physical interpretation. The abstract and Section 2 claim conservation of momentum and energy. For system (6), total momentum P_t = M V_t^N + Σ m_k U_t^{k,N} obeys dP_t = Σ(β_k − γ m_k²) U^k dt + Σ m_k sqrt(2σ) dW^k; the α-coupling cancels, but the thermostat damping and noise do not. So the system does not conserve momentum. This does not affect the limit theorem, whose proof uses only the linear SDE structure, but it does undercut the 'dynamical foundation' language. The authors should either say the coupling terms are constructed to conserve momentum in the interaction (action–reaction), or present the system as a Langevin thermostat around a conservative coupling. As written, the claim is too strong.\n\nMinor issues: Assumption 2.6 imposes the exact balance 2(a−b)−δ=1 and b>d. That's a tuning condition, typical for this kind of scaling limit, but the paper would be clearer if it said so explicitly rather than presenting it as if the mechanics forced it. In Remark 3.2, the KFP derivation needs a bit more regularity on v_H and the Laplace transform to justify differentiating under the expectation and dividing by it. Also, in Example 2.2(iii), the gamma subordinator's Laplace exponent is Φ(λ)=log(1+λ); the paper writes log(1−λ), which is a typo (the second listed pair is fine).\n\nBottom line: the reviewer's conditional verdict is right. The mathematics holds; the interpretation needs reworking. I'd send it to review.\n\nBest,","headline":"Rigorous limit theorem deriving fBm and superstatistical fBm with random Hurst parameter from a heterogeneous-mass Langevin system; the conservation-of-momentum claim is overstated and should be fixed, but the mathematics holds up.","tokens_in":29318,"tokens_out":5932,"would_cite":true,"duration_ms":50612,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60F17","60G15","60G22","60J60"],"pacs":[],"model":"deepseek-v4-flash","headline":"A test particle kicked by many unequal-mass Brownian particles converges, as N grows, to a Gaussian process whose covariance is set by the mass distribution; for a power-law mass law the limit is fractional Brownian motion with Hurst…","keywords":["fractional Brownian motion","superstatistical fractional Brownian motion","generalized grey Brownian motion","randomly scaled Gaussian processes","random Hurst parameter","time-dependent random Hurst parameter","anomalous diffusion","ensemble heterogeneity"],"falsifier":"Simulate system (6) with couplings $\\alpha_{k,N}\\simeq C_\\alpha N^{-a}$, $\\beta_{k,N}\\simeq C_\\beta N^{-b}$ and masses drawn from the $(2H-1)$-stable subordinator construction; the theorem predicts convergence to the stated fBm covariance when $2(a-b)-\\delta=1$ and $b>d$, so observing that same covariance when the balance is broken would refute the claim, while divergence or a different covariance under the broken balance would support it.","tokens_in":28377,"feed_emoji":"🎲","tokens_out":11949,"duration_ms":99072,"temperature":0.7,"pith_summary":"The paper tries to show that fractional Brownian motion and its random-coefficient, random-Hurst neighbors are limiting behaviors of a concrete mechanical system, not just fitting models. A test particle of mass $M$ is coupled to $N$ underdamped Langevin particles whose masses are random and whose couplings decay as prescribed powers of $N$. The authors prove that as $N\\to\\infty$ the test-particle position converges to a centered Gaussian process with covariance $D(v(t)+v(s)-v(|t-s|))$, where $v$ is read off from the limiting mass distribution. Choosing the surround masses according to a stable subordinator's L\\'evy measure yields fBm with $H\\in(1/2,1)$, while mixtures of power laws yield a sum of independent fBms whose effective anomalous exponent changes with time. Randomizing the test-particle coupling and the mass law through a random environment parameter $H$ then produces superstatistical fBm and models with time-dependent random Hurst parameter, together with their generalized Fokker--Planck equations.","feed_headline":"Unequal particle masses make a test particle diffuse like fBm","feed_subtitle":"This theorem gives a mechanical origin for random-Hurst and random-coefficient anomalous diffusion.","key_machinery":"The load-bearing object is the covariance identity $\\operatorname{Cov}(Z_t,Z_s)=D(v(t)+v(s)-v(|t-s|))$ with $v(t)=\\int_0^t \\dot v(\\tau)d\\tau$, where $\\dot v$ is the pointwise limit of $e_N(t)/m^*_N$ and $e_N(t)=\\int(1-e^{-\\gamma y t})y^{-2}\\mu_N(dy)$. This $e_N$ is an integrated Laplace transform of the mass law, so the family of limits is organized by Bernstein functions when the masses come from subordinator L\\'evy measures. The proof mechanism is conditional Gaussianity: given the random masses, the surround velocities are Ornstein--Uhlenbeck processes, and the paper proves $\\sigma(M)$-mixing convergence, with the exact exponent balance $2(a-b)-\\delta=1$ making the error terms from the cross-interaction and coupling vanish as $N\\to\\infty$.","core_discovery":"The paper's central claim is Theorem 2.1: under scaling assumptions, the position process $X^N$ solving the Langevin system (6) converges in finite-dimensional distributions (and, with an extra regularity condition, in $C[0,t_0]$) to a centered Gaussian process $Z$ with covariance $\\operatorname{Cov}(Z_t,Z_s)=\\left(2\\sigma C_\\beta^2 C_\\delta/(\\gamma^2 C_\\alpha^2)\\right)^{1/2}(v(t)+v(s)-v(|t-s|))$, where $v(t)=\\int_0^t \\dot v(\\tau)d\\tau$ and $\\dot v$ is the monotone limit of $e_N(t)/m^*_N$, with $e_N(t)=\\int_{(0,\\infty)}(1-e^{-\\gamma y t})y^{-2}\\mu_N(dy)$. The variance function $v$ is the fingerprint of the surround-mass distribution: when $\\mu_N$ is built from the L\\'evy measure of the $(2H-1)$-stable subordinator, $v(t)\\propto t^{2H}$, so the limit is fBm with Hurst parameter $H\\in(1/2,1)$. Mixtures of power laws give a sum of independent fBms and a time-dependent anomalous exponent; tempered or exponential mass laws interpolate between ballistic and superdiffusive or classical regimes; a deterministic mass choice recovers a Wiener process. The paper then randomizes the coupling constant $A$ and the mass distribution through a random $H$, obtaining conditionally Gaussian limits $\\sqrt{A}\\,G^{(H)}$ that include a randomly scaled (superstatistical) fBm with random Hurst parameter.","pith_inferences":["A testable extension the authors leave implicit: the measured variance function $v(t)$ of a tracer should map back to the effective crowd-mass distribution through $e_N(t)/m^*_N$, so single-particle-tracking data on $v(t)$ constrain the environmental heterogeneity rather than leaving Hurst parameter free.","The tuning condition $2(a-b)-\\delta=1$ suggests anomalous scaling may be confined to an intermediate window in particle number and coupling strength; finite-$N$ simulations of system (6) away from that window could show crossovers that the infinite-$N$ theorem does not describe.","The Bernstein-function organization of the examples indicates the construction generalizes: any mass law whose $e_N(t)/m^*_N$ converges to a nondecreasing Bernstein-like function should generate a valid Gaussian limit, potentially yielding new anomalous-diffusion processes by choosing $\\Phi$ at will.","In the random-$H$ construction, each value $h$ can be read as a local environment patch; spatial segmentation of trajectories should then reproduce the conditional covariance $v_h$ within patches, a prediction accessible to experiments with spatially varying crowding."],"forward_implications":["Superstatistical fBm, previously proposed as a phenomenological model, is obtained as the $N\\to\\infty$ limit of a momentum- and energy-conserving Langevin system, giving it a dynamical foundation.","The same theorem produces fBm-like Gaussian limits whose anomalous exponent changes with time, such as sums of independent fBms with different Hurst parameters.","With random coupling $A$ and random environment variable $H$, the limit $\\sqrt{A}\\,G^{(H)}$ covers both random diffusion coefficient and random Hurst parameter in one conditionally Gaussian process.","The associated Kolmogorov--Fokker--Planck equations are generalized evolution equations with pseudo-differential operators, so users of these anomalous-diffusion models also have their generators.","Because the limiting covariance is determined by the mass distribution, the listed choices of mass laws give concrete predictions: fBm, mixtures of fBms, ballistic-to-superdiffusive and ballistic-to-classical crossovers, and ordinary diffusion."],"supporting_citations":[{"why":"Supplies the stable/mixing convergence machinery used to pass from the conditionally Gaussian limits to the unconditional process limit.","marker":"[22]"},{"why":"Provides the Bernstein-function and subordinator examples whose variance functions yield fBm, tempered-stable crossovers, and classical diffusion.","marker":"[48]"},{"why":"The preliminary centre-of-mass superposition of Ornstein--Uhlenbeck processes that this paper makes rigorous through an explicit particle system.","marker":"[14]"},{"why":"Gives the fluctuation-dissipation and conservation-law relations encoded in the Langevin dynamics of the surround particles.","marker":"[18]"},{"why":"Defines the class of stochastic processes, including generalized grey Brownian motion, that the random-scaling step embeds into.","marker":"[41]"},{"why":"One of the superstatistical fBm model sources whose physical basis the random-coefficient limit establishes.","marker":"[38]"},{"why":"Provides statistical analysis and experimental motivation for a population of diffusion coefficients in superstatistical fBm.","marker":"[33]"},{"why":"An earlier superstatistical model with random diffusion coefficient that the present dynamics now justifies at the level of a limit theorem.","marker":"[24]"},{"why":"The multifractional Brownian motion with stochastically varying exponent that the random-Hurst example of Section 3 is calibrated to reproduce.","marker":"[2]"},{"why":"Supplies the tightness criterion used to upgrade finite-dimensional convergence to convergence in the space of continuous paths.","marker":"[26]"}],"fun_headline_variants":["Mass distribution dictates diffusion law in many-body limit","Random Hurst exponent emerges from particle mass spread","Mass bath noise yields fBm with random Hurst parameter","Unequal masses lead to fractional Brownian motion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof requires the exact exponent balance $2(a-b)-\\delta=1$ and the inequality $b>d$; these are tuning conditions imposed on the coupling and mass scalings, and if the scalings deviate, the error terms do not vanish and the claimed limit need not hold.","fun_headline_variants_meta":{"raw":{"variants":["Mass distribution dictates diffusion law in many-body limit","Random Hurst exponent emerges from particle mass spread","Mass bath noise yields fBm with random Hurst parameter","Unequal masses lead to fractional Brownian motion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000618,"raw_usage":{"total_tokens":3006,"prompt_tokens":1223,"completion_tokens":1783,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":839,"completion_tokens_details":{"reasoning_tokens":1721}},"tokens_in":839,"tokens_out":1783,"duration_ms":10219,"temperature":1.0,"reasoning_tokens":1721,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:54:19.859669+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate system (6) with couplings $\\alpha_{k,N}\\simeq C_\\alpha N^{-a}$, $\\beta_{k,N}\\simeq C_\\beta N^{-b}$ and masses drawn from the $(2H-1)$-stable subordinator construction; the theorem predicts convergence to the stated fBm covariance when $2(a-b)-\\delta=1$ and $b>d$, so observing that same covariance when the balance is broken would refute the claim, while divergence or a different covariance under the broken balance would support it.","supporting_citations":[{"cited_title":"H¨ ausler and H","cited_arxiv_id":null,"evidence_quote":"Supplies the stable/mixing convergence machinery used to pass from the conditionally Gaussian limits to the unconditional process limit."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Bernstein-function and subordinator examples whose variance functions yield fBm, tempered-stable crossovers, and classical diffusion."},{"cited_title":"D’Ovidio, S","cited_arxiv_id":null,"evidence_quote":"The preliminary centre-of-mass superposition of Ornstein--Uhlenbeck processes that this paper makes rigorous through an explicit particle system."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the fluctuation-dissipation and conservation-law relations encoded in the Langevin dynamics of the surround particles."},{"cited_title":"Mura and G","cited_arxiv_id":null,"evidence_quote":"Defines the class of stochastic processes, including generalized grey Brownian motion, that the random-scaling step embeds into."},{"cited_title":"Molina-Garc ´ ıa, T","cited_arxiv_id":null,"evidence_quote":"One of the superstatistical fBm model sources whose physical basis the random-coefficient limit establishes."},{"cited_title":"Ma´ cka la and M","cited_arxiv_id":null,"evidence_quote":"Provides statistical analysis and experimental motivation for a population of diffusion coefficients in superstatistical fBm."},{"cited_title":"Itto and C","cited_arxiv_id":null,"evidence_quote":"An earlier superstatistical model with random diffusion coefficient that the present dynamics now justifies at the level of a limit theorem."},{"cited_title":"Balcerek, S","cited_arxiv_id":null,"evidence_quote":"The multifractional Brownian motion with stochastically varying exponent that the random-Hurst example of Section 3 is calibrated to reproduce."},{"cited_title":"Kallenberg","cited_arxiv_id":null,"evidence_quote":"Supplies the tightness criterion used to upgrade finite-dimensional convergence to convergence in the space of continuous paths."}],"review_version":1}