Interface fluctuations in non equilibrium stationary states: the SOS approximation

Anna De Masi; Immacolata Merola; Stefano Olla

REVIEW 3 major objections 4 minor 12 references

Interface fluctuations in non equilibrium stationary states: the SOS approximation

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A driven 2D interface in the SOS approximation fluctuates on the N^{1/4} scale, and its scaling limit is a stationary Ornstein-Uhlenbeck process.

desk verdict Real SOS result with a fixable but real proof slip in the finite-dimensional convergence step; deserves a major-revision referee, not a desk reject. read the letter →

arxiv 1908.02920 v1 pith:KU3UIATT submitted 2019-08-08 math.PR math-phmath.MP

Anna De Masi , Immacolata Merola , Stefano Olla This is my paper · ORCID

classification math.PRmath-phmath.MP MSC 60K3560F1782B2482C22

keywords nonequilibriumstationarystatesSOSmodelinterfacefluctuationsOrnstein-UhlenbeckprocessN^{1/4}scalingeigenfunctionMarkovchainGinzburg-Landaumodels

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

This paper studies the interface in a two-dimensional solid-on-solid (SOS) approximation of a boundary-driven Ginzburg-Landau/Ising model whose non-equilibrium stationary state was constructed in a companion paper [4]. The central claim is that a stationary current makes the interface far more rigid than in thermal equilibrium: the rescaled interface position $\tilde S_N(t)=s_{[tN^{1/2}]}/N^{1/4}$ converges in law to a stationary Ornstein-Uhlenbeck process, so fluctuations are of order $N^{1/4}$ rather than the $\sqrt N$ order of the equilibrium 2D Ising interface. The proof works by showing that the principal eigenfunction of a compact transfer kernel converges to a Gaussian profile $e^{-r^2/(2\sigma)}$, and that the associated eigenvalue satisfies $\lambda_N^{\sqrt N}\to e^{-\sigma/2}$. The argument matters because it gives a rigorous instance where the non-local, non-equilibrium nature of the stationary state still allows an explicit and universal scaling limit.

What carries the argument

The load-bearing object is the positive compact transfer operator $T_N$ with kernel $T_N(s,s')=e^{-(s^2+s'^2)/(2N)}\pi(s-s')$. By the Krein-Rutman theorem it has a strictly positive eigenfunction $h_N$ with eigenvalue $\lambda_N$; these define the reversible Markov chain $p_N(s,s')=h_N(s')T_N(s,s')/(\lambda_N h_N(s))$, whose law under the Gibbs state is exactly the SOS statistical weight. The argument then runs through three steps: Gaussian bounds on $h_N(s)\le C N^{-1/8} e^{-cs^2/\sqrt N}$, compactness of the rescaled functions $\tilde h_N(r)=N^{1/8}h_N([rN^{1/4}])$ in $L^2(\mathbb R)$, and identification of every limit point via the Feynman-Kac fixed-point equation (4.6), whose solution $u(r)=e^{-r^2/(2\sigma)}$ with $\lambda=e^{-\sigma/2}$ is the Gaussian profile. This fixed-point equation is the central identity that connects the discrete spectral problem to the Ornstein-Uhlenbeck limit.

What would settle it

Solve equation (4.6) numerically on a grid with positive non-Gaussian initial data and see whether iteration converges to a fixed point other than $e^{-r^2/(2\sigma)}$; exhibiting any such positive fixed point would falsify the uniqueness assertion on which Proposition 4.3 and Theorem 2.1 rest.

Watch

Extended reading notes

Core claim

The paper's discovery is Theorem 2.1: starting from the SOS Hamiltonian $H_N(s)=N^{-1}\sum s_x^2+\sum |s_x-s_{x-1}|$ with boundary distribution given by the principal eigenfunction of $T_N(s,s')=e^{-(s^2+s'^2)/(2N)}\pi(s-s')$, the process $\tilde S_N(t)=s_{[tN^{1/2}]}/N^{1/4}$ converges in law on $C([0,1])$ to the stationary Ornstein-Uhlenbeck process with variance $\sigma/2$, where $\sigma^2$ is the variance of the symmetric increment distribution $\pi$. The same theorem gives $\lambda_N^{\sqrt N}\to e^{-\sigma/2}$. In physical terms, the interface is pinned by the slowly varying magnetic field inherited from the non-equilibrium stationary state: the quadratic potential term suppresses the $\sqrt N$ equilibrium wandering down to $N^{1/4}$, and the limiting process is universal in the sense that only the variance of the increments survives in the limit.

Load-bearing premise

The load-bearing premise is that the fixed-point equation (4.6) has a unique positive solution up to normalization, namely $u(r)=e^{-r^2/(2\sigma)}$; the paper asserts this uniqueness but does not prove or cite it, so if another positive solution existed, different subsequences of $\tilde h_N$ could converge to different limits and Theorem 2.1 would not follow.

Editorial extensions

If this is right

In the SOS approximation of the non-equilibrium stationary state, interface fluctuations are of order $N^{1/4}$, not the $\sqrt N$ order of the equilibrium 2D Ising interface.
The scaling limit is a stationary Ornstein-Uhlenbeck process with variance $\sigma/2$, depending on the single-step distribution only through its variance.
The eigenvalue asymptotics $\lambda_N^{\sqrt N}\to e^{-\sigma/2}$ give a quantitative relation between the spectral gap of the transfer operator and the amplitude of stationary fluctuations.
The result supports the general picture that a stationary current created by boundary reservoirs strongly suppresses interface wandering, at least when the interface is a single graph.
Because the limit is Gaussian and Markovian, the SOS interface in the non-equilibrium stationary state admits an explicit effective description as an Ornstein-Uhlenbeck process with variance set by the step distribution.

Reading between the lines

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

We infer that the $N^{1/4}$ scaling should persist in the full Ginzburg-Landau/Ising model of [4] as long as the interface is a single graph, since the SOS computation shows the pinning is caused by the quadratic potential from the slowly varying field rather than by special features of the step distribution.
The unproved uniqueness in Proposition 4.3 suggests a concrete analytic target: proving that the Feynman-Kac operator $u\mapsto \mathbb E_r(e^{-\frac12\int_0^1 B_s^2 ds}u(B_1))$ is a contraction in a suitable cone would close the gap and also give quantitative speed of convergence for $\tilde h_N$.
We infer that the equivalence between the SOS Gibbs state and the reversible Markov chain of (2.6) opens a simulation-friendly route: generating the chain and measuring $N^{-1/4}$ fluctuations tests the theory directly and would also probe finite-$N$ corrections to the $e^{-\sigma/2}$ law.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The paper proves that in the SOS approximation of a boundary-driven 2D interface, stationary fluctuations scale as N^{1/4} and converge to a stationary Ornstein-Uhlenbeck process with variance σ/2, with λ_N^{√N} → e^{-σ/2}. This is a meaningful first rigorous, albeit SOS-level, demonstration that a steady current stiffens an interface. The authors do not overclaim the physical scope.

What's new and good: the eigenvalue asymptotic is original, and the proof is mostly self-contained. The key estimates—Theorem 3.2's exponential decay of the eigenfunction and Lemma 4.1—are given in real detail. Tightness (Prop 5.2) is fine. The uniqueness gap flagged in Prop 4.3 is real but minor: (4.6) is a Krein-Rutman equation for a compact, strictly positive Feynman-Kac semigroup, so uniqueness up to normalization is standard and can be cited rather than proved.

The soft spot is bigger than that. The stress-test note is correct: the prefactor in (4.12) and in Proposition 5.1 is wrong for t<1. Iterating the eigenfunction relation over n=[t√N] steps gives λ_N^{-n}, and since λ_N^{√N}→e^{-σ/2}, the correct prefactor is e^{σ t/2}, not e^{σ/2} independent of t. Proposition 5.1's e^{kσ/2} should be e^{σ τ_k/2} (or a product e^{σ t_i/2}). The printed integrand also drops the shift r; the OU limit should involve (σB_s + r), not σB_s alone. As written, the displayed proof of Theorem 2.1 does not establish the OU limit for general t—the stationarity check for t=1, φ=1 gives ∫ψ(r)g(r)g(σr)dr instead of ∫ψ(r)g^2(r)dr.

This is a load-bearing error, but it is clearly fixable: replace the prefactor, shift the Brownian motion, and the rest of the proof structure holds. The paper deserves a serious referee. I would send it out, asking the authors to fix the finite-dimensional computation and add the Krein-Rutman citation for Prop 4.3.

Who is this for: probabilists and statistical physicists working on NESS interfaces. It is a useful contribution once corrected; I would cite the corrected version, not the current one.

Referee Report

3 major / 4 minor

Summary. The paper studies the SOS approximation of an interface in a two-dimensional non-equilibrium stationary state produced by boundary reservoirs. The model is a one-dimensional interface s_x with Hamiltonian (1.1), containing a quadratic pinning potential and nearest-neighbor gradients. The authors represent the stationary measure as the marginal of a Doob h-transform of a random walk, using the positive eigenfunction of the kernel T_N defined in (2.2). Their main result, Theorem 2.1, states that the diffusively rescaled interface S̃_N(t)=s_{[tN^{1/2}]}/N^{1/4} converges in law to the stationary Ornstein-Uhlenbeck process with variance σ/2, and that λ_N^{√N}→e^{-σ/2}. The proof is organized into three parts: spectral estimates on h_N and λ_N (Section 3), convergence and identification of the rescaled eigenfunction (Section 4), and tightness plus finite-dimensional convergence (Section 5).

Significance. If the proof is completed, this is a significant contribution: it is one of the few rigorous results on interface fluctuations in a non-equilibrium stationary state, and it exhibits a concrete mechanism—a Doob transform of a random walk with a quadratic killing potential—that produces the N^{1/4} scaling and an Ornstein-Uhlenbeck limit. The dependence only on the variance σ^2 of the increment distribution is a clean universality statement, and the result is explicitly falsifiable within the SOS approximation. The paper is self-contained after the model is introduced, and the key estimates (Theorems 3.1 and 3.2, Lemma 4.1) are given in detail. However, the identification of the limit and the finite-dimensional convergence contain errors that must be fixed before the main theorem is established.

major comments (3)

[Eq. (4.12), Section 4] Iterating the eigenfunction relation (2.3) over [t√N] steps yields a prefactor λ_N^{-[t√N]}, and the claimed λ_N^{√N}→e^{-σ/2} then gives e^{σt/2} in the limit, not the t-independent e^{σ/2} printed in (4.12). This is not a bookkeeping typo: for t<1 and φ=1, the fixed-point identity (4.6) with g(r)=e^{-r^2/(2σ)} gives E_r(e^{-1/2∫_0^t B_s^2 ds}g(B_t))=e^{-σt/2}g(r), so the printed prefactor produces e^{σ(1-t)/2}∫ψ(r)g^2(r)dr instead of the stationary one-time marginal ∫ψ(r)g^2(r)dr. In addition, the Brownian motion in (4.12) must be the variance-σ^2 process starting at r (i.e. B_s=r+σW_s), and the arguments σB_t in φ and g should be B_t; as written the expression is inconsistent with (4.9)–(4.11).
[Proposition 5.1, Section 5] The displayed finite-dimensional formula uses a single factor λ_N^{-k√N} and concludes e^{kσ/2}. But the elapsed time from 0 to τ_k is τ_k, and the product of the per-block eigenfunction factors is λ_N^{-([τ_1√N]+...+[τ_k√N])}, whose limit is e^{στ_k/2}; this equals e^{kσ/2} only when all τ_i=1, which is not the general case. Consequently the displayed proof does not establish convergence of the finite-dimensional distributions to the claimed OU process. The formula also omits the successive Feynman-Kac factors over the intervals (τ_{i-1},τ_i] and the correct Brownian shifts r_{i-1}+σW_s; these must be written precisely before the limit can be evaluated.
[Proposition 4.3, Section 4] The proposition asserts uniqueness of the positive solution of the fixed-point equation (4.6), but no proof or reference is supplied. This uniqueness is load-bearing: without it, different subsequences of h̃_N could converge to different positive eigenfunctions, and Corollary 4.4 and the identification λ=e^{-σ/2} would not follow. The assertion is plausibly true and can be justified by applying the Krein-Rutman theorem to the compact, strictly positive Feynman-Kac operator on the right-hand side of (4.6), but the proof should be included in the manuscript.

minor comments (4)

[Throughout] The manuscript text contains numerous OCR/rendering artifacts (e.g. 'N^{1/slash.left4}', '/summation.disp', '/parallel.alt1', '/parenleft.alt2'); these should be cleaned up before publication.
[Eqs. (4.9)–(4.12), Section 4] The notation for the Brownian motion is inconsistent: in (4.9)–(4.11) the limit process is r+σB_s with B_s standard, while (4.12) writes σB_s without the shift. Please standardize, for instance by declaring that B_s in (4.12) denotes the Brownian motion with variance σ^2 starting at r, or by writing r+σW_s explicitly.
[Proposition 5.2, Section 5] In the proof of the fourth-moment bound, the local limit estimate (5.3) is quoted for all s; the uniformity of the constant over the relevant range of s should be stated explicitly, since the bound is then used under summation over s.
[Section 1] In the heuristics leading to the SOS Hamiltonian (1.1), the reduction of ∑_{i=1}^{|s_x|} i/N to s_x^2/N should be described as an approximation up to lower-order boundary terms, since the exact identity is ∑_{i=1}^{m} i/N = m(m+1)/(2N).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the OU limit is derived from the stated SOS model, not fitted or imported.

full rationale

The paper's derivation is self-contained once the SOS Hamiltonian (1.1) is adopted. The variance parameter σ is the variance of the input jump distribution π, and the N^{1/4} scaling and the Ornstein-Uhlenbeck limit are consequences of eigenfunction estimates, Krein-Rutman theory, and explicit martingale computations (Sections 3–5), not of fitting a parameter to the target result. The only self-citation with any load-bearing role is [4], which supplies the NESS structure motivating the SOS Hamiltonian; this is a model input rather than an ingredient in the proof of Theorem 2.1. Moreover, [4] is an external mathematical result about a broader class of models, so citing it does not make the present derivation circular. The unproved uniqueness assertion in Proposition 4.3 and the possible prefactor issue in (4.12)/Proposition 5.1 flagged by the skeptic are mathematical correctness concerns, not instances of a prediction reducing to its own input by construction.

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

No free parameters or invented entities. The model's variance sigma^2 is an input from the increment distribution pi, not fitted; the proof uses auxiliary parameters (e.g., alpha=N^{-1/2} in Theorem 3.1) but these are internal to the estimates and do not affect the statement of the result. The derivation is self-contained after the model is posited; all dependence on prior work is through standard theorems plus the cited NESS structure theorem [4].

assumptions (6)

standard math Krein-Rutman theorem guarantees a positive maximal eigenfunction h_N and eigenvalue lambda_N for the compact, positive operator T_N
Used in Theorem 3.1 to define the eigenfunction that enters the boundary condition and the Markov chain (2.3)-(2.6).
standard math Local central limit theorem for aperiodic random walks with finite fourth moments (Lawler-Limic, Theorem 2.1.1)
Used in Theorem 3.2 (3.8) and Proposition 5.2 (5.3) to control transition probabilities.
domain assumption The NESS of the Ginzburg-Landau model is a Gibbs state with slowly varying chemical potential, as proved in [4] by the same authors
This is the bridge from the physical NESS problem to the SOS Hamiltonian (1.1); if [4] failed, the physical interpretation would change, though the mathematical theorem for the SOS model would stand.
standard math Donsker invariance principle for the random walk sum
Used in (4.9) to pass to Brownian motion in Proposition 4.3.
standard math Ito's formula and exponential martingale for Brownian motion
Used in Proposition 4.3 to identify the unique fixed point u(r) and lambda.
standard math Kolmogorov-Riesz compactness criterion and Billingsley's tightness criterion
Used in Proposition 4.2 and Proposition 5.2 to obtain compactness and tightness.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Interface fluctuations in non equilibrium stationary states: the SOS approximation." pith.science (2026). https://pith.science/paper/KU3UIATT

@misc{pith2026190802920,
  author       = {Pith},
  title        = {Pith review of: Interface fluctuations in non equilibrium stationary states: the SOS approximation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KU3UIATT}},
  note         = {Machine review of arXiv:1908.02920}
}

abstract

We study the $2d$ stationary fluctuations of the interface in the SOS approximation of the non equilibrium stationary state found in \cite{DOP}. We prove that the interface fluctuations are of order $N^{1/4}$, $N$ the size of the system. We also prove that the scaling limit is a stationary Ornstein-Uhlenbeck process.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 11 canonical work pages

[1]

Bertini, A

L. Bertini, A. De Sole, D. Gabrielli, G. Jona-Lasinio, C. Landim (2015 ) Macroscopic Fluc- tiation Theory, Rev. Mod. Phys. 87, 593

work page 2015
[2]

Bernardin, S

C. Bernardin, S. Olla (2005) Fourier law for a microscopic model of heat conduction Journ. Stat. Phys., 121, 271-289

work page 2005
[3]

Billingsley Convergence of Probability measures Wiley, New York, 1968

P. Billingsley Convergence of Probability measures Wiley, New York, 1968

work page 1968
[4]

De Masi, S

A. De Masi, S. Olla, E. Presutti, (2019) A note on Fick’s law with phase transitions Journal Statistical Physics, 175, 203-21

work page 2019
[5]

Derrida, J

B. Derrida, J. L. Lebowitz, and E. R. Speer (2001) Free Energy Functional for Nonequi- librium Systems: An Exactly Solvable Case . Phys. Rev. Lett. 87, 150601

work page 2001
[6]

Derrida, J

B. Derrida, J. L. Lebowitz, and E. R. Speer(2002) Large Deviation of the Density Proﬁle in the Steady State of the Open Symmetric Simple Exclusion Pr ocess Journ. Stat. Phys. 107, Nos. 3/4, 599634

work page 2002
[7]

Gallavotti, (1972) The phase separation line in the two-dimensional Ising mode l, Com- mun

G. Gallavotti, (1972) The phase separation line in the two-dimensional Ising mode l, Com- mun. Math. Phys., 27 103-136

work page 1972
[8]

Kipnis, C

C. Kipnis, C. Marchioro, E. Presutti, (1982) Heat ﬂow in an exactly solvable model Journ. Stat. Phys. 27 6574

work page 1982

Show all 12 references

[9]

D. Ioﬀe, S. Shlosman, Y. Velenik, (2015) An invariance Principle to Ferrari-Spohn diﬀu- sions Commun. Math. Phys., 336, 905-932 13

2015
[10]

4, 385-394, doi.org/10.1016/j.exmath.2010.03.001

Harald Hanche-Olsen, Helge Holden, (2010), The KolmogorovRiesz compactness theorem , Expositiones Mathematicae, 28, n. 4, 385-394, doi.org/10.1016/j.exmath.2010.03.001

2010 doi
[11]

Krein, M.A

M.G. Krein, M.A. Rutman (1950) Linear operators leaving invariant a cone in a Banach space, Ann. Math.Soc. Transl. 26, 128

1950
[12]

G. F. Lawler, V. Limic, (2010) Random walk: a modern introduction Cambridge Studies in Advanced Mathematics 123. Anna De Masi, Dipartimento di Ingegneria e Scienze dell’Inf ormazione e Matem- atica,, Universit `a degli studi dell’Aquila, L’Aquila, 67100 Italy E-mail address : ...

2010

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Bertini, A

L. Bertini, A. De Sole, D. Gabrielli, G. Jona-Lasinio, C. Landim (2015 ) Macroscopic Fluc- tiation Theory, Rev. Mod. Phys. 87, 593

work page 2015

[2] [2]

Bernardin, S

C. Bernardin, S. Olla (2005) Fourier law for a microscopic model of heat conduction Journ. Stat. Phys., 121, 271-289

work page 2005

[3] [3]

Billingsley Convergence of Probability measures Wiley, New York, 1968

P. Billingsley Convergence of Probability measures Wiley, New York, 1968

work page 1968

[4] [4]

De Masi, S

A. De Masi, S. Olla, E. Presutti, (2019) A note on Fick’s law with phase transitions Journal Statistical Physics, 175, 203-21

work page 2019

[5] [5]

Derrida, J

B. Derrida, J. L. Lebowitz, and E. R. Speer (2001) Free Energy Functional for Nonequi- librium Systems: An Exactly Solvable Case . Phys. Rev. Lett. 87, 150601

work page 2001

[6] [6]

Derrida, J

B. Derrida, J. L. Lebowitz, and E. R. Speer(2002) Large Deviation of the Density Proﬁle in the Steady State of the Open Symmetric Simple Exclusion Pr ocess Journ. Stat. Phys. 107, Nos. 3/4, 599634

work page 2002

[7] [7]

Gallavotti, (1972) The phase separation line in the two-dimensional Ising mode l, Com- mun

G. Gallavotti, (1972) The phase separation line in the two-dimensional Ising mode l, Com- mun. Math. Phys., 27 103-136

work page 1972

[8] [8]

Kipnis, C

C. Kipnis, C. Marchioro, E. Presutti, (1982) Heat ﬂow in an exactly solvable model Journ. Stat. Phys. 27 6574

work page 1982