arxiv: 2604.21341 · v1 · submitted 2026-04-23 · ❄️ cond-mat.stat-mech · math.PR· physics.comp-ph

Recognition: unknown

Fractals of Simple Random Walks in Two Dimensions: A Monte Carlo Study

Jiang Zhou , Ziru Deng , Pengcheng Hou

Authors on Pith no claims yet

Pith reviewed 2026-05-08 13:53 UTC · model grok-4.3

classification ❄️ cond-mat.stat-mech math.PRphysics.comp-ph

keywords simple random walksfractal geometrychemical distanceMonte Carlo simulationSLEGaussian free fieldlogarithmic fractalshull dimension

0 comments

The pith

Simple random walk clusters in two dimensions have spanning chemical distance scaling as L (ln L) to the 1/4 power.

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

This Monte Carlo study examines the geometry of clusters generated by long simple random walks on a two-dimensional lattice. It confirms a logarithmic scaling for the cluster mass and a hull fractal dimension of exactly 4/3 consistent with SLE theory. The key finding is that the chemical distance needed to traverse the cluster grows proportionally to L multiplied by the fourth root of the logarithm of L, matching the expected upper limit for percolation clusters defined by level sets of the Gaussian free field.

Core claim

Clusters of discrete-time simple random walks with L^2 steps on periodic L x L lattices have mass satisfying M/L^2 approximately equal to (ln L)^{-1} times [pi/2 + b (ln L)^{-2}] with b approximately -(pi/2)^{-2}. The hull fractal dimension is 4/3. The spanning chemical distance scales as S ~ L (ln L)^{1/4}, which is the theoretical upper bound for level-set percolation clusters on the two-dimensional Gaussian free field. This indicates the clusters are marginal logarithmic fractals with conformally invariant frontiers.

What carries the argument

The spanning chemical distance S of the random walk cluster, analyzed through finite-size scaling on periodic lattices to establish the asymptotic form S ~ L (ln L)^{1/4}.

If this is right

The external frontier of the cluster is conformally invariant and belongs to the Brownian frontier universality class via SLE_{8/3}.
The cluster contains highly efficient asymptotically linear connective paths.
The observed scaling places the random walk cluster exactly at the theoretical upper bound for chemical distances in level-set percolation on the two-dimensional Gaussian free field.
The mass scaling M/L^2 ~ (ln L)^{-1} [pi/2 + b (ln L)^{-2}] confirms the marginal logarithmic fractal character with the specific correction coefficient.

Where Pith is reading between the lines

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

This numerical match suggests random walk clusters may function as discrete approximations to level sets of the Gaussian free field.
The precise agreement with the SLE_{8/3} prediction for the hull indicates that conformal invariance emerges in the scaling limit of these discrete clusters.
Further analytic work could seek to prove the chemical distance bound directly from the random walk definition using SLE techniques.

Load-bearing premise

The scaling forms extracted from finite-size Monte Carlo data on periodic L by L lattices continue to hold in the infinite-volume limit without additional correction terms that would alter the reported exponents.

What would settle it

A simulation on substantially larger lattices showing that the power of ln L in the chemical distance scaling deviates from exactly 1/4 or that the ratio S/L fails to grow as (ln L)^{1/4}.

Figures

Figures reproduced from arXiv: 2604.21341 by Jiang Zhou, Pengcheng Hou, Ziru Deng.

**Figure 1.** Figure 1: FIG. 1. Typical sRW trajectory of view at source ↗

**Figure 2.** Figure 2: FIG. 2. Rescaled cluster mass view at source ↗

**Figure 4.** Figure 4: FIG. 4. Plot of ln view at source ↗

**Figure 5.** Figure 5: FIG. 5. Finite-size scaling of the chemical distance view at source ↗

read the original abstract

We present a Monte Carlo study of the fractal geometry of clusters formed by discrete-time simple random walks (sRW) of $L^2$ steps on a periodic square $L\times L$ lattice. We verify with high precision that the asymptotic behavior of the cluster mass follows $M/L^2 \simeq (\ln L)^{-1} [\frac{\pi}{2}+b (\ln L)^{-2}]$, with $b\approx -(\pi/2)^{-2}$, demonstrating marginal ``logarithmic fractals". We further determine the fractal dimension of the hull to be $d_{\rm hull}=1.333\,29(14)=4/3$, in excellent agreement with the prediction of Schramm-Loewner evolution ($\rm SLE_{8/3}$) for the Brownian frontier universality class. More importantly, we analyze the chemical distance $S$ spanning the cluster and obtain strong evidence that it asymptotically scales as $S\sim L(\ln L)^{1/4}$, lying exactly on the theoretical upper bound for the chemical distance for level-set percolation clusters on the two-dimensional Gaussian free field. Our numerical results show that the sRW cluster exhibits a conformally invariant external frontier and contains highly efficient asymptotically linear connective paths.

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

Monte Carlo confirmation of SLE hull dimension plus new numerical scaling for chemical distance in random walk clusters, with the usual caveats on log corrections.

read the letter

The main takeaway here is that this Monte Carlo paper gives a high-precision check on the hull dimension of simple random walk clusters in two dimensions and presents what looks like the first numerical support for the chemical distance scaling S ~ L (ln L)^{1/4} that sits on the GFF upper bound. They do a good job with the hull dimension, getting 1.33329(14) which matches 4/3 to several digits. The mass scaling with the logarithmic correction also comes out close to the expected b value. That's solid verification work. The chemical distance part is the fresh angle. If their extrapolation from finite lattices holds, it adds concrete evidence that these random walk clusters behave like the level sets in the Gaussian free field for connectivity purposes. The claim that the external frontier is conformally invariant and has efficient paths is consistent with the SLE picture. Where it could be softer is in the handling of the logarithmic corrections for the chemical distance. Those terms approach their asymptotics slowly, and without seeing the exact fitting procedure, error analysis, or tests for alternative correction terms, it's hard to be fully confident that the 1/4 exponent is isolated cleanly. The abstract mentions strong evidence but leaves the details to the full text. If the paper includes checks like varying the fit range or adding subleading terms, that would strengthen it a lot. This kind of work is aimed at people in 2D statistical mechanics and conformal field theory who want numerical backing for theoretical predictions. It doesn't introduce new methods but refines our picture of random walk geometry. I think it deserves peer review. A referee could verify the numerics and suggest any needed robustness tests on the scaling fits.

Referee Report

3 major / 2 minor

Summary. The manuscript presents a Monte Carlo study of clusters formed by simple random walks of L² steps on periodic L×L lattices in two dimensions. It reports high-precision numerical verification that the cluster mass satisfies M/L² ≃ (ln L)^{-1} [π/2 + b (ln L)^{-2}] with b ≈ -(π/2)^{-2}, that the hull fractal dimension is d_hull = 1.33329(14) = 4/3 in agreement with SLE_{8/3}, and that the spanning chemical distance scales asymptotically as S ∼ L (ln L)^{1/4}, matching the theoretical upper bound for level-set percolation clusters on the 2D Gaussian free field. The work concludes that the clusters possess a conformally invariant external frontier and contain highly efficient asymptotically linear connective paths.

Significance. If the asymptotic scalings hold, the results supply high-precision numerical confirmation of marginal logarithmic fractals for 2D random-walk clusters and an exact match to the GFF level-set upper bound on chemical distance. The reported agreement with the SLE_{8/3} prediction for the hull dimension and the demonstration of efficient connective paths would strengthen the link between random-walk geometry and conformally invariant curves. The Monte Carlo approach with large L is a positive feature, though the central claims depend on finite-size extrapolations whose robustness must be demonstrated.

major comments (3)

[Chemical-distance analysis (results section)] The headline claim that S ∼ L (ln L)^{1/4} exactly (abstract and results section on chemical distance) rests on fits to finite-L data on periodic tori. The manuscript must demonstrate that the extracted exponent remains stable when subleading corrections (additive constant, (ln L)^β with β ≠ 1/4, or 1/ln L terms) are included in the fit, because logarithmic corrections converge slowly and any unaccounted term can shift the apparent leading exponent by an amount comparable to the reported statistical precision.
[Mass-scaling results] The mass-scaling fit M/L² ≃ (ln L)^{-1} [π/2 + b (ln L)^{-2}] with b ≈ -(π/2)^{-2} is presented as high-precision verification, but the text supplies no explicit table or figure showing the effective exponent or the scaled quantity versus 1/ln L together with the goodness-of-fit when alternative correction terms are tested.
[Abstract and numerical-methods section] The abstract states “strong evidence” for the chemical-distance scaling without quoting error bars on the effective exponent, the range of L used, or the χ² of the fit; these quantitative details are required to assess whether the data truly isolate the (ln L)^{1/4} form in the infinite-volume limit.

minor comments (2)

Clarify the precise definition of the chemical distance S (minimal path length along the cluster or something else) and whether it is measured on the lattice or in the continuum limit.
The hull-dimension value is given to five decimal places; the text should state the fitting window in L and the extrapolation method used to obtain the quoted uncertainty.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the careful and constructive report. The comments on the robustness of the logarithmic fits and the presentation of quantitative details are well taken. We address each major comment below and have prepared revisions that strengthen the analysis without altering the core conclusions.

read point-by-point responses

Referee: [Chemical-distance analysis (results section)] The headline claim that S ∼ L (ln L)^{1/4} exactly (abstract and results section on chemical distance) rests on fits to finite-L data on periodic tori. The manuscript must demonstrate that the extracted exponent remains stable when subleading corrections (additive constant, (ln L)^β with β ≠ 1/4, or 1/ln L terms) are included in the fit, because logarithmic corrections converge slowly and any unaccounted term can shift the apparent leading exponent by an amount comparable to the reported statistical precision.

Authors: We agree that explicit checks for stability under subleading corrections are necessary given the slow convergence of logarithmic terms. We have now performed additional least-squares fits to forms that include an additive constant, a free power β in (ln L)^β, and a 1/ln L correction term. In all cases the leading exponent remains consistent with 1/4 within one standard deviation, and the χ² per degree of freedom remains acceptable (1.05–1.25). A new table summarizing the fit parameters and goodness-of-fit for each model will be added to the results section, together with a brief discussion of the stability. revision: yes
Referee: [Mass-scaling results] The mass-scaling fit M/L² ≃ (ln L)^{-1} [π/2 + b (ln L)^{-2}] with b ≈ -(π/2)^{-2} is presented as high-precision verification, but the text supplies no explicit table or figure showing the effective exponent or the scaled quantity versus 1/ln L together with the goodness-of-fit when alternative correction terms are tested.

Authors: The referee is correct that diagnostic plots and alternative-fit comparisons were omitted. We will add a figure showing the scaled quantity (M/L²) ln L plotted against 1/ln L, which should approach π/2 with the expected slope if the (ln L)^{-2} correction dominates. We have also computed the effective exponent and tested alternative corrections (e.g., (ln L)^{-3} or constant terms); the extracted b remains close to −(π/2)^{-2} with comparable χ². These diagnostics and the associated table of fit qualities will be included in the revised manuscript. revision: yes
Referee: [Abstract and numerical-methods section] The abstract states “strong evidence” for the chemical-distance scaling without quoting error bars on the effective exponent, the range of L used, or the χ² of the fit; these quantitative details are required to assess whether the data truly isolate the (ln L)^{1/4} form in the infinite-volume limit.

Authors: We acknowledge that the abstract would benefit from more quantitative information. Because of length constraints we will revise the abstract to read “strong numerical evidence that S scales as L (ln L)^{0.25(2)} for L up to 2048 (χ²/dof ≈ 1.2)”. The full range of L, error bars, and χ² values are already reported in the numerical-methods and results sections; we will add a cross-reference in the abstract to make this explicit. This constitutes a partial revision that supplies the requested details while respecting abstract formatting limits. revision: partial

Circularity Check

0 steps flagged

Purely numerical Monte Carlo study with no load-bearing derivation or self-referential reduction

full rationale

The paper performs Monte Carlo simulations of random walks on finite L×L tori and extracts scaling forms (M/L², d_hull, S) by fitting the generated data. No analytical derivation chain exists that reduces a claimed result to its own inputs by construction. The reported b ≈ −(π/2)^{-2} is a numerical fit compared to an external theoretical expression; the S ∼ L (ln L)^{1/4} claim is likewise a fit to simulation output benchmarked against an independent GFF upper bound. No self-citation is load-bearing, no ansatz is smuggled, and no fitted parameter is relabeled as a prediction. The study is self-contained against external benchmarks and therefore exhibits no circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper is a numerical Monte Carlo study; no explicit free parameters, axioms, or invented entities are introduced in the abstract. The reported b coefficient is presented as approximately matching a theoretical value rather than being freely fitted.

pith-pipeline@v0.9.0 · 5525 in / 1131 out tokens · 57003 ms · 2026-05-08T13:53:07.587618+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

42 extracted references · 4 canonical work pages

[1]

Spitzer,Principles of Random Walk, 2nd ed., Gradu- ate Texts in Mathematics, Vol

F. Spitzer,Principles of Random Walk, 2nd ed., Gradu- ate Texts in Mathematics, Vol. 34 (Springer, New York, 1976)

1976
[2]

G. F. Lawler and V. Limic,Random Walk: A Modern In- troduction, Cambridge Studies in Advanced Mathemat- ics, Vol. 123 (Cambridge University Press, Cambridge, 2010)

2010
[3]

B. D. Hughes,Random Walks and Random Environ- ments, Oxford Science Publications, Vol. 1 (Oxford Uni- versity Press, Oxford, 1995)

1995
[4]

R´ ev´ esz,Random Walk in Random and Non-Random Environments, 3rd ed

P. R´ ev´ esz,Random Walk in Random and Non-Random Environments, 3rd ed. (World Scientific, Singapore, 2013)

2013
[5]

E. W. Montroll and G. H. Weiss, Journal of Mathematical Physics6, 167 (1965)

1965
[6]

Havlin and D

S. Havlin and D. ben Avraham, Advances in Physics36, 695 (1987)

1987
[7]

ben Avraham and S

D. ben Avraham and S. Havlin,Diffusion and Reactions in Fractals and Disordered Systems(Cambridge Univer- sity Press, Cambridge, 2000)

2000
[8]

de Gennes,Scaling Concepts in Polymer Physics (Cornell University Press, Ithaca, 1979)

P.-G. de Gennes,Scaling Concepts in Polymer Physics (Cornell University Press, Ithaca, 1979)

1979
[9]

B. B. Mandelbrot,The Fractal Geometry of Nature(W. H. Freeman, New York, 1982)

1982
[10]

Y. Deng, W. Zhang, T. M. Garoni, A. D. Sokal, and A. Sportiello, Physical Review E81, 020102 (2010)

2010
[11]

Erd˝ os and S

P. Erd˝ os and S. J. Taylor, Acta Mathematica Academiae Scientiarum Hungaricae11, 137 (1963)

1963
[12]

N. C. Jain and W. E. Pruitt, Zeitschrift f¨ ur Wahrschein- lichkeitstheorie und Verwandte Gebiete16, 279 (1970)

1970
[13]

Schramm, Israel Journal of Mathematics118, 221 (2000)

O. Schramm, Israel Journal of Mathematics118, 221 (2000)

2000
[14]

Lawler,Conformally Invariant Processes in the Plane, Mathematical Surveys and Monographs, Vol

G. Lawler,Conformally Invariant Processes in the Plane, Mathematical Surveys and Monographs, Vol. 114 (Amer- ican Mathematical Society, Providence, Rhode Island, 2008)

2008
[15]

E. A. Soares and A. A. Moreira, Physical Review E112, 034105 (2025)

2025
[16]

G. F. Lawler, O. Schramm, and W. Werner, Mathemat- ical Research Letters8, 401 (2001)

2001
[17]

Werner, inLectures on Probability Theory and Statis- tics, Lecture Notes in Mathematics, Vol

W. Werner, inLectures on Probability Theory and Statis- tics, Lecture Notes in Mathematics, Vol. 1840 (Springer, Berlin, Heidelberg, 2004) pp. 107–195

2004
[18]

Havlin, B

S. Havlin, B. Trus, G. H. Weiss, and D. Ben-Avraham, Journal of Physics A: Mathematical and General18, L247 (1985)

1985
[19]

Z. Zhou, J. Yang, Y. Deng, and R. M. Ziff, Physical Re- view E86, 061101 (2012). 7

2012
[20]

Schramm and S

O. Schramm and S. Sheffield, Acta Mathematica202, 21 (2009)

2009
[21]

Ding and L

J. Ding and L. Li, Communications in Mathematical Physics360, 523 (2018)

2018
[22]

Ding and M

J. Ding and M. Wirth, The Annals of Probability48, 10.1214/19-AOP1397 (2020)

work page doi:10.1214/19-aop1397 2020
[23]

J. Ding, M. Wirth, and H. Wu, Annales de l’Institut Henri Poincar´ e, Probabilit´ es et Statistiques58, 10.1214/21-AIHP1204 (2022)

work page doi:10.1214/21-aihp1204 2022
[24]

E. B. Dynkin, Journal of Functional Analysis50, 167 (1983)

1983
[25]

Sznitman,Topics in Occupation Times and Gaus- sian Free Fields, Zurich Lectures in Advanced Mathe- matics (European Mathematical Society, Z¨ urich, 2012)

A.-S. Sznitman,Topics in Occupation Times and Gaus- sian Free Fields, Zurich Lectures in Advanced Mathe- matics (European Mathematical Society, Z¨ urich, 2012)

2012
[26]

Le Jan,Markov Paths, Loops and Fields, Lecture Notes in Mathematics, Vol

Y. Le Jan,Markov Paths, Loops and Fields, Lecture Notes in Mathematics, Vol. 2026 (Springer, Berlin, 2011)

2026
[27]

Lupu, The Annals of Probability44, 2117 (2016)

T. Lupu, The Annals of Probability44, 2117 (2016)

2016
[28]

Cs´ aki and A

E. Cs´ aki and A. F¨ oldes, Entropy26, 1082 (2024)

2024
[29]

G. F. Lawler, O. Schramm, and W. Werner, Acta Math- ematica187, 237 (2001)

2001
[30]

G. F. Lawler, O. Schramm, and W. Werner, Acta Math- ematica187, 275 (2001)

2001
[31]

G. F. Lawler, O. Schramm, and W. Werner, Annales de l’Institut Henri Poincar´ e, Probabilit´ es et Statistiques38, 109 (2002)

2002
[32]

Friedman and Bogdan E

V. Beffara, The Annals of Probability36, 10.1214/07- AOP364 (2008)

work page doi:10.1214/07- 2008
[33]

Cardy, Annals of Physics318, 81 (2005), arXiv:cond- mat/0503313

J. Cardy, Annals of Physics318, 81 (2005), arXiv:cond- mat/0503313

work page arXiv 2005
[34]

Dembo, Y

A. Dembo, Y. Peres, J. Rosen, and O. Zeitouni, Annals of Mathematics160, 433 (2004)

2004
[35]

Grassberger, Journal of Physics A: Mathematical and General19, 2675 (1986)

P. Grassberger, Journal of Physics A: Mathematical and General19, 2675 (1986)

1986
[36]

Grossman and A

T. Grossman and A. Aharony, Journal of Physics A: Mathematical and General19, L745 (1986)

1986
[37]

R. F. Voss, Journal of Physics A: Mathematical and Gen- eral17, L373 (1984)

1984
[38]

Dvoretzky and P

A. Dvoretzky and P. Erd¨ os, inProceedings of the Sec- ond Berkeley Symposium on Mathematical Statistics and Probability, edited by J. Neyman (University of Califor- nia Press, 1951) pp. 353–368

1951
[39]

P. Hou, S. Fang, J. Wang, H. Hu, and Y. Deng, Physical Review E99, 042150 (2019)

2019
[40]

Cao and R

X. Cao and R. Santachiara, Physical Review Letters126, 120601 (2021)

2021
[41]

Kundu, S

A. Kundu, S. N. Majumdar, and G. Schehr, Physical Re- view Letters110, 220602 (2013)

2013
[42]

G. F. Lawler, Electronic Communications in Probability 1, 29 (1996)

1996