Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

A consistency-stability approach to scaling limits of zero-range processes

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves explicit, time-uniform convergence rates for the hydrodynamic limit of the symmetric zero-range process in dimensions 1 and 2, in a Monge-Kantorovich distance and in relative entropy.

desk verdict A clean abstract framework and an unproven central theorem: Section 5.7's random-walk estimate assumes a uniform increment bound that Theorem 3.1 never states. read the letter →

arxiv 2412.16714 v1 pith:LBE5UUIA submitted 2024-12-21 math.PR math-phmath.APmath.MP

Daniel Marahrens , Angeliki Menegaki , Clément Mouhot This is my paper · ORCID
classification math.PRmath-phmath.APmath.MP MSC 60K3582C2260F0535K55
keywords zero-rangeprocesshydrodynamiclimitquantitativeerrorestimatesMonge-KantorovichdistancerelativeentropyparabolicscalinglocalGibbsmeasureconsistency-stabilitymethod
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves a quantitative hydrodynamic limit for the symmetric zero-range process — a lattice gas in which a particle jumps from a site at a rate depending only on the occupation of that site — in dimensions $1$ and $2$ under parabolic scaling. The claim is that the law of the process stays close, with an explicit algebraic or logarithmic rate, to the local Gibbs measure (product state matching the macroscopic density profile) built from the solution of the nonlinear diffusion equation. In dimension $1$ the rate is $N^{-1/6+\varepsilon_0}$ and in dimension $2$ it is $(\ln N)^{-1/8+\varepsilon_0}$, uniformly in time. The distance is a Monge-Kantorovich (dual Lipschitz) distance whose macroscopic counterpart is the classical $L^1$ stability estimate for conservation laws, and the same rate is obtained for relative entropy. The method is a consistency-stability scheme that avoids block estimates and is presented as a general template for symmetric attractive particle systems.

What carries the argument

The load-bearing object is the variation-of-constants formula for the difference between the density $F_t^N$ of the particle system and the density $G_t^N$ of the local Gibbs measure: $$F_t^N - G_t^N = $e^{{tL_N^*}}$(F_0^N - G_0^N) + \int_0^t $e^{{(t-\tau)L_N^*}}$\bigl(L_N^* G_\tau^N - L_\infty^* G_\tau^N\bigr)\,d\tau.$$ The first term is bounded by the microscopic stability assumption, which says the basic coupling contracts the $\ell^1$ distance between configurations; the second term is the Monge-Kantorovich consistency error, comparing the microscopic generator with the macroscopic evolution. That error is reduced through an intermediate box scale $\ell$: local averages are projected onto constant-mass hyperplanes, a spectral-gap inequality on the box controls the non-equilibrium part, a quantitative local limit theorem (equivalence of ensembles) controls the projected part, and recurrence of the difference random walk gives the final decay in dimensions $1$ and $2$.

What would settle it

One concrete check: take the admissible jump rate $g(k)=\lfloor(k+1)/2\rfloor$, which satisfies the theorem's stated growth condition but has flat steps where $g(k+1)=g(k)$, and compute the probability bound (5.16) for the coupled difference process; at flat sites the holding rate vanishes, so the claimed exponential bound on the number of jumps cannot hold, and a positive probability of too few jumps would show the stated rates need a stronger hypothesis.

Watch

Extended reading notes

Core claim

On the torus in dimensions $1$ and $2$, for a centred symmetric zero-range process with a non-decreasing Lipschitz jump rate satisfying a coarse growth condition, and for initial profiles $f_0 \in C^3(\mathbb{T}^d)$ bounded below, the paper's Theorem 3.1 asserts that under parabolic scaling the law $\mu_t^N$ and the local Gibbs measure $\vartheta^N_{f_t}$ built from the solution $f_t$ of $\partial_t f = A:\nabla^2\sigma(f)$ satisfy $$\sup_{t\in[0,T]}\|\mu_t^N - \vartheta^N_{f_t}\|_{\mathrm{Lip}^*} \leq \|\mu_0^N - \vartheta^N_{f_0}\|_{\mathrm{Lip}^*} + C $N^{{-1/6+\varepsilon_0}}$$$ in $d=1$, with $C(\ln N)^{-1/8+\varepsilon_0}$ in $d=2$, and the same bound holds for the rescaled relative entropy $H^N(\mu_t^N|\vartheta^N_{f_t})$. The estimate is uniform in time. The proof verifies four abstract assumptions for this model: microscopic stability, macroscopic stability, Monge-Kantorovich consistency, and entropic consistency. The consistency error is controlled by a site-by-site computation, an intermediate mesoscopic scale, a local spectral-gap inequality, a quantitative local limit theorem for the equivalence of ensembles, and, in dimensions $1$ and $2$, recurrence estimates for the difference of two configurations coupled by the basic coupling.

Load-bearing premise

The argument's weakest premise is that every allowed jump changes the jump rate by at least a fixed positive amount; the stated hypotheses only guarantee a coarser growth condition, and the key probability estimate (5.16) that controls the consistency rates is asserted from that stronger premise without proof.

Editorial extensions

If this is right

  • The empirical density field converges quantitatively to the solution of the nonlinear diffusion equation: the paper's Remark 1 transfers the Lip* bound into an explicit probability estimate for the empirical measure against test functions.
  • The same rate is obtained in relative entropy without any logarithmic Sobolev inequality, so the entropic route does not depend on that strong functional inequality.
  • Because the abstract theorem only needs the four consistency-stability assumptions, the method applies beyond the zero-range process to other symmetric attractive models; the paper notes that simple exclusion fits the same assumptions.
  • The paper's Remark 4 notes that in dimension 1, for test functions of the form $N^{-1}\sum_x \eta_x\phi_x$ with $\phi\in C^1$, the jump-distance method gives a better consistency rate, and higher-order local limit expansions are expected to bring the general rate close to the optimal $N^{-1/2}$.
  • The uniformity in time means the quantitative closeness does not degrade as the macroscopic profile relaxes to its equilibrium.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Not claimed by the paper: the argument's random-walk step could be repaired under the theorem's stated hypotheses, because the coarse growth condition on $g$ does not imply the uniform per-step lower bound used in Section 5.7; closing this gap is a concrete open problem.
  • Not claimed by the paper: the same consistency-stability layout may handle degenerate limit equations such as the porous medium equation, where quantitative relative entropy methods are currently unavailable; the paper states this direction is under investigation.
  • Not claimed by the paper: if the jump-distance contraction of Section 5.7 were established in dimension $2$, the logarithmic rate would likely become algebraic, since the random-walk recurrence input would no longer be the bottleneck.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes an abstract consistency-stability framework for quantitative hydrodynamic limits. Theorem 2.1 bounds the dual-Lipschitz distance (and relative entropy) between the microscopic law and a local Gibbs measure in terms of the corresponding initial discrepancy and a consistency error, assuming microscopic stability, macroscopic stability, and consistency conditions (H1)-(H4). The framework is then applied to the symmetric zero-range process on the d-dimensional torus with d=1,2 under parabolic scaling. Theorem 3.1 claims explicit rates N^{-1/6+epsilon} in d=1 and (ln N)^{-1/8+epsilon} in d=2 for both the Monge-Kantorovich distance and the relative entropy, uniformly in time. The proof of the application verifies the abstract assumptions for the ZRP; the main technical work is the Monge-Kantorovich consistency estimate in Section 5, whose core is a random-walk recurrence estimate in Section 5.7.

Significance. If fully established, the result would provide a quantitative hydrodynamic limit for a genuinely nonlinear zero-range process without block estimates, with rates in a weak distance and in relative entropy, uniform in time. The abstract framework of Section 2 is clean and the microscopic stability and entropic-consistency parts of the ZRP verification are careful and potentially reusable. The paper also identifies an interesting direction for degenerate limit equations. However, as detailed in the major comments, the central consistency estimate relies on an unproved and, under the stated hypotheses, unjustified random-walk bound. The main theorem is therefore not established as stated, and the paper is not suitable for publication in its current form.

major comments (3)
  1. [Section 5.7, Eq. (5.16)] The proof of the 'too few jumps' estimate (5.16) explicitly assumes 0 < g_*1 <= |g(k+1)-g(k)| <= g_*2 < infinity for every k. This assumption is not part of Theorem 3.1, whose hypotheses are monotonicity, global Lipschitz regularity, g(0)=0, g(n)>0, and the coarse growth condition g(n')-g(n) >= c_g for n' >= n+n0. The coarse condition permits flat stretches of length n0-1 on which adjacent increments vanish, so a uniform positive per-step lower bound is not a consequence of the assumptions. Since (5.16) controls the J_slow term in the decomposition of J_N^t and is used to derive (5.20)-(5.21), Proposition 5.1 and hence Theorem 3.1 are not proved for general Lipschitz test functions under the stated hypotheses.
  2. [Section 5.7, random-walk argument] The random-walk estimate treats x_-(t) and x_+(t) as a symmetric random walk in a fixed random environment, with jump rates (g(eta_{x_-}) - g(eta_{x_-}-1))_+ and (g(zeta_{x_+}+1) - g(zeta_{x_+}))_+. The manuscript's own phrase 'provided we consider eta as given' acknowledges that the environment is frozen in this reasoning, but no argument is provided to transfer the fixed-environment waiting-time and return-probability estimates to the actual dynamic environment, where the occupancies eta and zeta evolve and the jump rates change in time. Under the theorem's hypotheses, a site's g-increment may vanish on long stretches, so the lower bound on the number of jumps of the coupled difference is not justified for the actual process.
  3. [Section 5.7, Eqs. (5.20)-(5.21)] The return-probability estimates quoted from [LL10, Prop 4.2.4] are applied to the difference process x_+(t)-x_-(t), but it is not verified that this process satisfies the hypotheses of the cited result (in particular, symmetry or aperiodicity are not established from the assumptions on p). Moreover, the control of the unbounded factor g(eta_{y1}) is left to the reader ('We leave this easy calculation to the reader'), yet it is part of the quantitative estimate that determines the final rate in Theorem 3.1. These omissions are substantial because the resulting bounds (5.20)-(5.21) directly feed into the consistency error in Proposition 5.1.
minor comments (5)
  1. [Section 3, Theorem 3.1] The statement contains the duplicated phrase 'jump rate jump rate g'; it should read 'the jump rate g'.
  2. [Introduction, Section 1.1] There are typographical errors: 'threrein' should be 'therein', 'mesure' should be 'measure', and 'opened' should be 'open'.
  3. [References] The reference [DMOW] is missing publication details (year and journal or arXiv number); please complete it.
  4. [Appendix A, Theorem A.1] The notation E_{vartheta^{ell,m}} and E_{vartheta^ell_m} is confusing, because the difference E_{vartheta^{ell,m}}[g(eta_{y0})] - E_{vartheta^ell_m}[g(eta_{y0})] appears to compare a conditional measure with itself unless the two symbols are intended to denote different measures; please distinguish them clearly.
  5. [Section 5.5, Eq. (5.8)] The flattening error is ultimately bounded after choosing M = 2theta' ln(N ell^{-d-1}), but this optimization is not shown in the text; please state explicitly how the two terms of (5.8) are balanced.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the consistency-stability decomposition is verified for the ZRP by direct model computations and external standard results, with no fitted parameter or self-citation chain driving the claim.

full rationale

The derivation is self-contained in the relevant sense. Theorem 2.1 is an abstract Duhamel/decomposition argument whose conclusion follows once (H1)-(H4) hold. These hypotheses are then checked directly for the ZRP: (H1) is proven from the standard coupling in Section 4.1 via (4.2); (H2) uses the external standard reference [KL99] for the equilibrium structure plus standard parabolic regularity; (H3) is proven in Proposition 5.1 by a site-by-site computation, an intermediate scale, a Poincar\'e inequality ([LSV96]) and the local limit theorem ([Pet75], [KL99]); and (H4) is proven by direct integration against the PDE in Section 6.2. No step is defined in terms of the target conclusion, and no parameter is fitted to data: the rates N^{-1/6} and (ln N)^{-1/8} are obtained by optimizing the displayed error terms in Step 6, Section 5.9. The only self-reference, [MMM24], is an announcement of forthcoming work, not load-bearing evidence. I flag a separate, non-circular correctness concern: Method 2 in Section 5.7 proves (5.16) under the extra assumption '0 < g_*1 <= |g(k)-g(k+1)| <= g_*2 < infinity', which is not implied by Theorem 3.1's coarse growth condition g(n')-g(n) >= c_g for n' >= n+n0, and the random-walk environment is only argued for fixed eta while eta evolves dynamically. This is an unproven estimate and therefore a proof gap, but it is not circularity: the claimed conclusion is not assumed as an input or manufactured by a self-citation chain.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The central result inherits its validity from several external results; the only genuinely questionable extra ingredient is the uniform lower bound on g increments (axiom 3).

free parameters (2)
  • intermediate coarse-graining scale ell = ell = N^{1/3} (d=1); ell = (ln N)^{1/8} (d=2)
    Introduced in Step 2 to form ell-cubes; chosen at the end to minimize the sum of error terms. It is an optimization parameter, not an empirical fit, but the final rate depends on it.
  • cutoff M(N) in flattening error = M = 2 theta' ln(N ell^{-d-1})
    Used in equations (5.7)-(5.8) to split large and small local densities. No scientific content; a proof device.
assumptions (6)
  • standard math The one-site measure n_lambda and sigma satisfy smoothness and coercivity: sigma is C^infinity, sigma'(R+) subset [Lambda, Lambda^{-1}].
    Quoted from [KL99, Chapter 2, Section 3] in Section 4.3; needed for macroscopic well-posedness and for the consistency estimates.
  • domain assumption For f0 in C^3(T^d), the PDE partial_t f = A : nabla^2 sigma(f) has a global smooth solution converging exponentially fast to its mean in C^infinity.
    Asserted in Section 4.3 via De Giorgi-Nash, Schauder and spectral gap; used to make R1(t) integrable and the error estimates uniform in time.
  • ad hoc to paper There exist 0 < g_*1 <= g_*2 < infinity such that g_*1 <= |g(k+1)-g(k)| <= g_*2 for all k.
    Used in Section 5.7 to lower-bound the jump rate of the difference process. It is stated as following from the hypotheses, but the hypotheses only give an averaged growth condition; this uniform per-step lower bound is an extra assumption.
  • standard math Poincare inequality holds on ell-cubes with constant proportional to ell^2 and independent of particle number.
    Cited to [LSV96] in Section 5.7 Step 4; needed to bound the fluctuating part of the test function.
  • standard math The local limit theorem / quantitative equivalence of ensembles (Theorem A.1) holds.
    Proved in Appendix A using [Pet75, KL99]; used in Step 5 to project onto constant mass.
  • standard math For symmetric random walks in Z^d, d=1,2, the probability of not returning to the origin after n steps is O(n^{-1/2}) (d=1) and O(1/ln n) (d=2).
    Cited to [LL10, Prop 4.2.4] in Section 5.7; used in equations (5.20)-(5.21).

how reviews work

0 comments
Cite this review

Pith. "Pith review of A consistency-stability approach to scaling limits of zero-range processes." pith.science (2026). https://pith.science/paper/LBE5UUIA

@misc{pith2026241216714,
  author       = {Pith},
  title        = {Pith review of: A consistency-stability approach to scaling limits of zero-range processes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LBE5UUIA}},
  note         = {Machine review of arXiv:2412.16714}
}
read the original abstract

We propose a simple quantitative method for studying the hydrodynamic limit of interacting particle systems on lattices. It is applied to the diffusive scaling of the symmetric Zero-Range Process (in dimensions one and two). The rate of convergence is estimated in a Monge-Kantorovich distance asymptotic to the L^1 stability estimate of Kruzkhov, as well as in relative entropy; and it is uniform in time. The method avoids the use of the so-called ``block estimates''. It is based on a modulated Monge-Kantorovich distance estimate and microscopic stability properties.

Figures

Figures reproduced from arXiv: 2412.16714 by the authors.

Figure 1
Figure 1. The functional setting. The downward right arrow simply embeds the atomic elements δf ∈ P(M+(T d )) for some f ∈ M+(T d ) into M+(T d ) by the mapping δf 7→ f [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Hydrodynamic Limit of the Symmetric Zero-Range Process with Slow Boundary

    math.PR 2025-08 conditional novelty 6.0 of 10

    For theta >= 1, the empirical density of the symmetric zero-range process with slow boundary reservoirs converges in probability to the unique weak solution of the nonlinear heat equation with Robin (theta = 1) or Neu...

Reference graph

Works this paper leans on

17 extracted references · 13 canonical work pages · cited by 1 Pith paper

  1. [1]

    Chang and H.-T

    C.-C. Chang and H.-T. Yau. Fluctuations of one dimensional ginzburg-landau models in nonequilibrium. Communications in Mathematical Physics , 145(2):209--234, 1992

    work page 1992

  2. [2]

    Dizdar, G

    D. Dizdar, G. Menz, F. Otto, and T. Wu. The quantitative hydrodynamic limit of the K awasaki dynamics. Preprint. arXiv:1807.09850

    arXiv

  3. [3]

    Funaki, C

    T. Funaki, C. Gu, and H. Wang. Quantitative homogenization and hydrodynamic limit of non-gradient exclusion process. arXiv:2404.12234, 2024

    arXiv 2024

  4. [4]

    Giunti, C

    A. Giunti, C. Gu, and J.-C. Mourrat. Quantitative homogenization of interacting particle systems. The Annals of Probability , 50(5):1885--1946, 9 2022

    work page 1946

  5. [5]

    Gon c alves, C

    P. Gon c alves, C. Landim, and C. Toninelli. Hydrodynamic limit for a particle system with degenerate rates . Annales de l'Institut Henri Poincaré, Probabilités et Statistiques , 45(4):887 -- 909, 2009

    work page 2009

  6. [6]

    Grunewald, F

    N. Grunewald, F. Otto, C. Villani, and M. Westdickenberg. A two-scale approach to logarithmic S obolev inequalities and the hydrodynamic limit. Ann. Inst. Henri Poincar\' e Probab. Stat. , 45(2):302--351, 2009

    work page 2009

  7. [7]

    M. Guo, G. Papanicolaou, and S. Varadhan. Nonlinear diffusion limit for a system with nearest neighbor interactions. Comm. Math. Phys. , 118(1):31--59, 1988

    work page 1988

  8. [8]

    Gess and Konarovskyi V

    B. Gess and Konarovskyi V. A quantitative central limit theorem for the simple symmetric exclusion process. arXiv:2408.01238, 2024

    arXiv 2024

Show all 17 references
  1. [9]

    Jara and O

    J. Jara and O. Menezes. Non-equilibrium fluctuations of interacting particle systems. arXiv:1810.09526, 2018

    2018 arXiv

  2. [10]

    Kipnis and C

    C. Kipnis and C. Landim. Scaling limits of interacting particle systems , volume 320 of Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences] . Springer-Verlag, Berlin, 1999

    1999

  3. [11]

    T. Liggett. Interacting Particle Systems . Springer Berlin Heidelberg, 1985

    1985

  4. [12]

    Lawler and Vlada Limic

    Gregory F. Lawler and Vlada Limic. Random Walk: A Modern Introduction . Cambridge Studies in Advanced Mathematics. Cambridge University Press, 2010

    2010

  5. [13]

    Landim, S

    C. Landim, S. Sethuraman, and S. Varadhan. Spectral gap for zero-range dynamics. Ann. Probab. , 24(4):1871--1902, 1996

    1902

  6. [14]

    Marahrens, A

    D. Marahrens, A. Menegaki, and C. Mouhot. Quantitative estimates in Y aus's relative entropy method and semigroup viewpoint. In progress , 2024

    2024

  7. [15]

    V. Petrov. Sums of independent random variables . Ergebnisse der Mathematik und ihrer Grenzgebiete [Results in Mathematics and Related Areas], Band 82. Springer-Verlag, New York-Heidelberg, 1975. Translated from the Russian by A. A. Brown

    1975

  8. [16]

    Rezakhanlou

    F. Rezakhanlou. Hydrodynamic limit for attractive particle systems on Z ^d . Comm. Math. Phys. , 140(3):417--448, 1991

    1991

  9. [17]

    H.-T. Yau. Relative entropy and hydrodynamics of G inzburg- L andau models. Lett. Math. Phys. , 22(1):63--80, 1991

    1991

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.