arxiv: 2604.21143 · v1 · submitted 2026-04-22 · 🧮 math.PR · math.AP

Quantitative homogenization for the critical long-range random conductance model

Ahmed Bou-Rabee , Paul Dario This is my paper

Pith reviewed 2026-05-09 23:00 UTC · model grok-4.3

classification 🧮 math.PR math.AP

keywords long-range random conductancecritical exponentquantitative homogenizationquenched invariance principleanomalous diffusionrandom walkBrownian motion

0 comments

The pith

At the critical long-range jump exponent the random conductance model homogenizes to the Laplacian at rate 1/sqrt(|ln ε|) with diffusivity given by the mean conductance alone.

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

The paper studies the long-range random conductance model on the integer lattice where the jump rate between distant sites decays exactly as the critical power |x-y| to the minus (d+2). It proves that the associated discrete elliptic equation converges quantitatively to the ordinary Laplacian, with an error that decays as one over the square root of the absolute logarithm of the mesh size. From this it follows that the random walk converges almost surely to Brownian motion, but only after rescaling time by the slightly superdiffusive factor sqrt(t log t). The limiting diffusion constant equals the average conductance and receives no correction from any corrector field at leading order.

Core claim

For i.i.d. uniformly elliptic conductances a(x,y), the long-range operator with kernel a(x,y)|x-y|^{-(d+2)} homogenizes to a multiple of the discrete Laplacian at the quantitative rate 1/sqrt(|ln ε|). Consequently the quenched invariance principle holds under the scaling sqrt(t log t), and the effective diffusivity is exactly the mean of a with no leading-order corrector contribution.

What carries the argument

The critical long-range random conductance operator with |x-y|^{-(d+2)} decay together with the quantitative homogenization error estimate that controls the difference from the Laplacian without a leading corrector term.

If this is right

The random walk satisfies a quenched invariance principle under the scaling sqrt(t log t).
The effective diffusivity equals the spatial average of the conductances with no corrector correction.
Homogenization error is controlled at the explicit rate 1/sqrt(|ln ε|).
The model sits precisely between stable Lévy processes (faster decay) and standard Brownian motion (slower decay).

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The logarithmic divergence appears to suppress the usual corrector fluctuation that dominates error estimates in short-range homogenization.
The same scaling argument may extend to other lattice models whose jump kernels have exactly logarithmic second-moment divergence.
Direct Monte-Carlo simulation of the walk on grids of size 10^4 or larger could check whether the position variance grows as t log t.

Load-bearing premise

The conductances are independent and identically distributed with uniform positive lower and upper bounds, and the decay exponent is set exactly at d+2 so that the second moment of the jump kernel diverges only logarithmically.

What would settle it

Numerical solution of the elliptic equation on large finite boxes would falsify the claim if the observed homogenization error fails to decay proportionally to 1/sqrt(|ln ε|) or if the effective diffusivity deviates from the plain average conductance by an amount that does not vanish as the box size grows.

Figures

Figures reproduced from arXiv: 2604.21143 by Ahmed Bou-Rabee, Paul Dario.

**Figure 1.** Figure 1: A sample path of the critical long-range random walk on Z 2 with jump kernel |z| −4 and random conductances, shown at three successive scales: 500 steps (left), 5,000 steps (center), and 100,000 steps (right). By Theorem 1.2, the walk converges to Brownian motion under √ tlog t scaling. 1 arXiv:2604.21143v1 [math.PR] 22 Apr 2026 [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗

**Figure 2.** Figure 2: Path counting (Lemma 2.3) for z = (3, 2) in Z 2 . The reference path γz→0 (blue) has |z|1 = 5 edges. Every path γv→u with v − u = z is a translate of γz→0. Two translates (green dashed, purple dotted) both use the target edge e (red), each because a different reference edge of γz→0 lands on e. Since each reference edge determines at most one translate, at most |z|1 paths use e. (2) For every η > 0 and ever… view at source ↗

**Figure 3.** Figure 3: The divergence-free cycle Cu+v,u for u = (0, 0) and v = (3, 2) in Z 2 . One unit of flow travels from u to u + v along the long-range edge (red, +1) and returns via the nearest-neighbour path γu+v→u (blue), which decreases coordinates lexicographically. The net flow at every vertex is zero. Fix a total order ≺ on εZ d (for instance, the lexicographic order), and define Iε := (u, v) ∈ U ε × (εZ d \ {0}) :… view at source ↗

**Figure 4.** Figure 4: The single-scale averaging argument (Lemma 4.2). The function h (red bump) is supported in U ε . Each application of Tk averages h at points shifted by ≈ ε3 k in the first coordinate. After N = ⌈diam(U)/(ε3 k )⌉ iterations, the sampled points lie outside U, so T N k h = 0 on U ε . The resulting prefactor N2 · (ε3 k ) 2 ε d = C diam(U) 2 ε d is independent of the scale k: this is special to the critical exp… view at source ↗

read the original abstract

We consider the long-range random conductance model on $\mathbb{Z}^d$ at the critical exponent: the jump rate between sites $x$ and $y$ decays as $\mathbf{a}(x,y) |x-y|^{-(d+2)}$, where $\mathbf{a}(x,y)$ are i.i.d. uniformly elliptic conductances. Below the critical exponent $(d+2)$ the walk converges to a stable process; above it, to Brownian motion with diffusive $\sqrt{t}$ scaling. At criticality the second moment of the jump kernel diverges logarithmically. We establish quantitative homogenization of the associated elliptic equation to the Laplacian at the rate $1/\sqrt{|\ln\varepsilon|}$. As a consequence, we deduce quenched convergence of the random walk to Brownian motion under the anomalous $\sqrt{t \log t}$ scaling. Unlike in standard homogenization, the effective diffusivity is determined by the mean conductance alone, with no corrector contribution at leading order.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper gives a quantitative homogenization rate of 1/sqrt|ln ε| at the critical long-range conductance model, with diffusivity set by the mean alone and no leading corrector.

read the letter

The central result here is a homogenization rate of 1 over square root of absolute log epsilon for the elliptic equation in the long-range random conductance model exactly at the critical exponent d+2. From that they extract quenched convergence of the walk to Brownian motion under the sqrt(t log t) scaling, and they note that the effective diffusivity is just the average conductance with nothing extra from a corrector at leading order. That combination looks new relative to the subcritical stable regime and the supercritical diffusive one. The setup uses standard i.i.d. uniformly elliptic multipliers on the |x-y|^{-(d+2)} kernel, so ergodicity gives the mean directly and the log divergence of the second moment dictates the anomalous scaling once the error is controlled. They handle the distinction between regimes cleanly and focus on the boundary case where the divergence is only logarithmic. The assumptions are minimal and the claim avoids fitting or circularity. The main soft spot is that the rate itself is slow, which is expected from the log term but means any application would need to live with fairly coarse error bounds; without the full estimates it is hard to see exactly how they close the variance or Green function controls at that precision. Still, nothing in the argument structure contradicts itself or relies on hidden assumptions. This is for people working on homogenization of non-local operators or random walks in random media with long-range jumps. A reader who already knows the sub- and super-critical results will see the gap being filled. It is worth sending to a serious referee because the statement is precise, the regime is natural, and the conclusion is falsifiable once the proof is checked.

Referee Report

2 major / 2 minor

Summary. The manuscript studies the long-range random conductance model on Z^d at the critical exponent, with i.i.d. uniformly elliptic conductances a(x,y) multiplying the kernel |x-y|^{-(d+2)}. It proves quantitative homogenization of the associated elliptic equation to the Laplacian at the explicit rate 1/sqrt(|ln ε|). As a consequence, it obtains quenched convergence of the random walk to Brownian motion under the anomalous scaling sqrt(t log t). The effective diffusivity equals the mean conductance with no leading-order corrector contribution, in contrast to standard homogenization.

Significance. If the central estimates hold, the result supplies a sharp quantitative bridge between subcritical stable Lévy processes and supercritical diffusive behavior, precisely capturing the logarithmic divergence of the second moment of the jump kernel. The explicit rate and the absence of a leading corrector are technically notable and may inform related models of anomalous diffusion in random media. The quenched invariance principle under the corrected scaling is a direct and falsifiable consequence once the homogenization error is controlled.

major comments (2)

[§3] §3 (or the section containing the main homogenization theorem): the proof that the corrector contribution vanishes at leading order while still obtaining the 1/sqrt(|ln ε|) rate must be checked for circularity; the logarithmic divergence is controlled via the i.i.d. assumption, but it is unclear whether the variance estimates remain uniform when the second-moment integral is replaced by its truncated version.
[Theorem 1.2] Theorem 1.2 (quenched convergence statement): the passage from the elliptic homogenization rate to the random-walk invariance principle under sqrt(t log t) scaling requires a quantitative control on the Green function or on the martingale approximation; the manuscript should exhibit the precise error term that converts the 1/sqrt(|ln ε|) bound into the claimed almost-sure convergence.

minor comments (2)

[Abstract] The abstract and introduction should explicitly state whether the rate 1/sqrt(|ln ε|) is expected to be optimal (e.g., via a matching lower bound or heuristic) or only an upper bound.
[§2] Notation for the rescaled conductances and the truncated kernels should be introduced once and used consistently; several places appear to reuse the symbol a_ε without redefinition.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The comments highlight important points on the logical structure of the estimates and the quantitative passage to the invariance principle. We address each major comment below and will revise the manuscript to incorporate clarifications and explicit error terms where needed.

read point-by-point responses

Referee: [§3] §3 (or the section containing the main homogenization theorem): the proof that the corrector contribution vanishes at leading order while still obtaining the 1/sqrt(|ln ε|) rate must be checked for circularity; the logarithmic divergence is controlled via the i.i.d. assumption, but it is unclear whether the variance estimates remain uniform when the second-moment integral is replaced by its truncated version.

Authors: The argument is not circular. The variance estimates for the corrector are first obtained for the truncated kernel using i.i.d. concentration inequalities (Bernstein-type bounds on the sum of independent terms), which hold uniformly in the truncation level because the truncation threshold is chosen after the logarithmic divergence is bounded by the mean conductance alone. The contribution of the tail is then controlled separately by the same 1/sqrt(|ln ε|) rate via a deterministic comparison to the averaged kernel. We will add a short paragraph at the beginning of §3 that explicitly states this order of estimates and the uniformity of the variance bound. revision: partial
Referee: [Theorem 1.2] Theorem 1.2 (quenched convergence statement): the passage from the elliptic homogenization rate to the random-walk invariance principle under sqrt(t log t) scaling requires a quantitative control on the Green function or on the martingale approximation; the manuscript should exhibit the precise error term that converts the 1/sqrt(|ln ε|) bound into the claimed almost-sure convergence.

Authors: We agree that the quantitative link deserves an explicit statement. In the proof of Theorem 1.2 the homogenization error enters the martingale approximation through the Green function representation of the position; the resulting error is of order 1/sqrt(log t) uniformly on compact time intervals after the sqrt(t log t) rescaling. Almost-sure convergence then follows from a Borel–Cantelli argument on dyadic times. We will insert a new lemma (or subsection) that records this error propagation and the precise constant dependence on the homogenization rate. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation proceeds from i.i.d. uniform ellipticity and the critical exponent (d+2) to control the homogenization error at rate 1/√|ln ε| via quantitative estimates that do not presuppose the target diffusivity or scaling. The claim that effective diffusivity equals the mean conductance (no leading corrector) is obtained as a direct consequence of the logarithmic divergence of the second moment together with the error bound, rather than by fitting or redefinition. The anomalous √(t log t) quenched convergence follows from the homogenization result by standard arguments. No load-bearing step reduces to a self-citation, ansatz smuggled via prior work, or renaming of a known pattern; the argument is self-contained against the stated assumptions and ergodicity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The result rests on i.i.d. uniform ellipticity of conductances and the mathematical fact of logarithmic second-moment divergence at exponent d+2; these are domain-standard assumptions with no new free parameters or invented entities.

axioms (2)

domain assumption Conductances a(x,y) are i.i.d. and uniformly elliptic, i.e., bounded between fixed positive constants λ and Λ independent of x,y.
Explicitly stated in the abstract as the model setup.
standard math At the critical exponent d+2 the second moment of the jump kernel diverges logarithmically.
This is a direct integral/sum computation for the kernel |x|^{-(d+2)} and is invoked to explain why standard diffusive scaling fails.

pith-pipeline@v0.9.0 · 5456 in / 1581 out tokens · 58133 ms · 2026-05-09T23:00:00.968742+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

34 extracted references · 3 canonical work pages

[1]

Andres, J.-D

S. Andres, J.-D. Deuschel, and M. Slowik,Invariance principle for the random conductance model in a degenerate ergodic environment, Ann. Probab.43(2015), 1866–1891

2015
[2]

Superdiffusive central limit theorem for a Brownian particle in a critically-correlated incompressible random drift, September 2024

S. Armstrong, A. Bou-Rabee, and T. Kuusi,Superdiffusive central limit theorem for a Brownian particle in a critically-correlated incompressible random drift, preprint, arXiv:2404.01115, 2024

work page arXiv 2024
[3]

Armstrong and T

S. Armstrong and T. Kuusi,Elliptic homogenization from qualitative to quantitative, preprint, arXiv:2210.06488, 2022

work page arXiv 2022
[4]

Armstrong, T

S. Armstrong, T. Kuusi, and J.-C. Mourrat,Quantitative Stochastic Homogenization and Large-Scale Regularity, Grundlehren der mathematischen Wissenschaften, vol. 352, Springer, 2019

2019
[5]

Armstrong and C

S. Armstrong and C. Smart,Quantitative stochastic homogenization of convex integral functionals, Ann. Sci. ´Ec. Norm. Sup´ er.49(2016), 423–481. 28 AHMED BOU-RABEE AND PAUL DARIO

2016
[6]

R. F. Bass and D. A. Levin,Transition probabilities for symmetric jump processes, Trans. Amer. Math. Soc.354 (2002), 2933–2953

2002
[7]

Billingsley,Convergence of Probability Measures, 2nd ed., Wiley Series in Probability and Statistics, John Wiley & Sons, 1999

P. Billingsley,Convergence of Probability Measures, 2nd ed., Wiley Series in Probability and Statistics, John Wiley & Sons, 1999

1999
[8]

Biskup,Recent progress on the random conductance model, Probab

M. Biskup,Recent progress on the random conductance model, Probab. Surv.8(2011), 294–373

2011
[9]

Biskup, X

M. Biskup, X. Chen, T. Kumagai, and J. Wang,Quenched invariance principle for a class of random conductance models with long-range jumps, Probab. Theory Related Fields180(2021), 847–889

2021
[10]

Bouchaud and A

J.-P. Bouchaud and A. Georges,Anomalous diffusion in disordered media: statistical mechanisms, models and physical applications, Phys. Rep.195(1990), 127–293

1990
[11]

Cannizzaro, Q

G. Cannizzaro, Q. Moulard, and F. Toninelli,Superdiffusive central limit theorem for the stochastic Burgers equation at the critical dimension, preprint, arXiv:2501.00344, 2025

work page arXiv 2025
[12]

Chen, Z.-Q

X. Chen, Z.-Q. Chen, T. Kumagai, and J. Wang,Quenched invariance principle for long range random walks in balanced random environments, Ann. Inst. Henri Poincar´ e Probab. Stat.57(2021), 2243–2267

2021
[13]

Z.-Q. Chen, X. Chen, T. Kumagai, and J. Wang,Homogenization of symmetric stable-like processes in stationary ergodic media, SIAM J. Math. Anal.53(2021), 2957–3001

2021
[14]

Chen, Z.-Q

X. Chen, Z.-Q. Chen, T. Kumagai, and J. Wang,Quantitative stochastic homogenization for random conductance models with stable-like jumps, Probab. Theory Related Fields191(2025), 627–669

2025
[15]

Chen and T

Z.-Q. Chen and T. Kumagai,Heat kernel estimates for jump processes of mixed types on metric measure spaces, Probab. Theory Related Fields140(2008), 277–317

2008
[16]

Z.-Q. Chen, T. Kumagai, and J. Wang,Heat kernel estimates for general symmetric pure jump Dirichlet forms, Ann. Sc. Norm. Super. Pisa Cl. Sci. (5)23(2022), 1091–1140

2022
[17]

X. Chen, T. Kumagai, and J. Wang,Random conductance models with stable-like jumps: heat kernel estimates and Harnack inequalities, J. Funct. Anal.279(2020), 108656

2020
[18]

X. Chen, T. Kumagai, and J. Wang,Random conductance models with stable-like jumps: quenched invariance principle, Ann. Appl. Probab.31(2021), 1180–1231

2021
[19]

S. N. Ethier and T. G. Kurtz,Markov Processes: Characterization and Convergence, Wiley Series in Probability and Mathematical Statistics, John Wiley & Sons, 1986

1986
[20]

Flegel, M

F. Flegel, M. Heida, and M. Slowik,Homogenization theory for the random conductance model with degenerate ergodic weights and unbounded-range jumps, Ann. Inst. Henri Poincar´ e Probab. Stat.55(2019), 1226–1257

2019
[21]

Gilbarg and N

D. Gilbarg and N. S. Trudinger,Elliptic Partial Differential Equations of Second Order, Classics in Mathematics, Springer, 2001

2001
[22]

Gloria, S

A. Gloria, S. Neukamm, and F. Otto,Quantification of ergodicity in stochastic homogenization: optimal bounds via spectral gap on Glauber dynamics, Invent. Math.199(2015), 455–515

2015
[23]

Gloria and F

A. Gloria and F. Otto,An optimal variance estimate in stochastic homogenization of discrete elliptic equations, Ann. Probab.39(2011), 779–856

2011
[24]

Gloria and F

A. Gloria and F. Otto,An optimal error estimate in stochastic homogenization of discrete elliptic equations, Ann. Appl. Probab.22(2012), 1–28

2012
[25]

Kassmann, A

M. Kassmann, A. Piatnitski, and E. Zhizhina,Homogenization of L´ evy-type operators with oscillating coefficients, SIAM J. Math. Anal.51(2019), 3641–3665

2019
[26]

S. M. Kozlov,Averaging of random operators, Mat. Sb. (N.S.)109(151)(1979), no. 2, 188–202

1979
[27]

Murugan and L

M. Murugan and L. Saloff-Coste,Anomalous threshold behavior of long range random walks, Electron. J. Probab. 20(2015), no. 74, 1–21

2015
[28]

Murugan and L

M. Murugan and L. Saloff-Coste,Davies’ method for anomalous diffusions, Proc. Amer. Math. Soc.145(2017), 1793–1804

2017
[29]

Murugan and L

M. Murugan and L. Saloff-Coste,Heat kernel estimates for anomalous heavy-tailed random walks, Ann. Inst. Henri Poincar´ e Probab. Stat.55(2019), 697–719

2019
[30]

G. C. Papanicolaou and S. R. S. Varadhan,Boundary value problems with rapidly oscillating random coefficients, in: Random Fields, Vol. I, II (Esztergom, 1979), Colloq. Math. Soc. J´ anos Bolyai, vol. 27, North-Holland, Amsterdam, 1981, pp. 835–873

1979
[31]

Piatnitski and E

A. Piatnitski and E. Zhizhina,Periodic homogenization of nonlocal operators with a convolution-type kernel, SIAM J. Math. Anal.49(2017), 64–81

2017
[32]

R. W. Schwab,Periodic homogenization for nonlinear integro-differential equations, SIAM J. Math. Anal.42 (2010), 2652–2680

2010
[33]

Sz´ asz and T

D. Sz´ asz and T. Varj´ u,Limit laws and recurrence for the planar Lorentz process with infinite horizon, J. Stat. Phys.129(2007), 59–80

2007
[34]

V. V. Yurinski˘ ı,On a Dirichlet problem with random coefficients, in: Stochastic Differential Systems (Proc. IFIP-WG 7/1 Working Conf., Vilnius, 1978), Lecture Notes in Control and Information Sci., vol. 25, Springer, Berlin–New York, 1980, pp. 344–353. Department of Mathematics, University of Pennsylvania, Philadelphia, PA 19104 Email address:ahmedmb@sa...

1978