On the gradient dynamics associated with wetting models

Henri Elad Altman; Jean-Dominique Deuschel; Tal Orenshtein

REVIEW 2 major objections 4 minor 17 references

On the gradient dynamics associated with wetting models

T0 review · 2 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 Rescaled wetting dynamics are tight, and a new tilted Bessel path measure converges to reflecting Brownian motion.

desk verdict A useful, honest paper: it proves the tightness conjecture for shrinking-strip wetting dynamics and gives a genuinely new continuous wetting measure converging to reflecting Brownian motion, but the proof of the tightness theorem has a repairable gap and the a=0 SDE step is under-justified. read the letter →

arxiv 1908.08850 v2 pith:ILOZCIA4 submitted 2019-08-23 math.PR

Jean-Dominique Deuschel , Henri Elad Altman , Tal Orenshtein This is my paper · ORCID

classification math.PR MSC 60K3560H1560J55

keywords wettingmodelsshrinkingstripgradientdynamicsscalinglimitsreflectingBrownianmotionmeanderBesselSPDEslocaltimes

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 proves two approximation results for critical wetting models, interfaces pulled toward a wall by a thin strip of pinning. It shows that the rescaled reversible gradient dynamics of the discrete shrinking-strip model are tight in $C([0,T],H^{-1}(0,1))$, settling a conjecture from earlier work. It also constructs a continuous wetting measure $P^{1,\eta}_a$, given as a 3-dimensional Bessel law tilted by an explicit local-time factor, and proves that as $\eta\to0$ it converges weakly to the law of reflecting Brownian motion on $[0,1]$. A sympathetic reader would care because these results make a concrete bridge between discrete interface models and a continuum SPDE universality class, and they give a computable pathwise approximation of reflecting Brownian motion.

What carries the argument

The load-bearing device is an explicit Radon–Nikodym density with a local-time tilt: against the 3-dimensional Bessel law, $P^{1,\eta}_a$ weights paths by $\frac{X_1\wedge\eta}{X_1}\frac{a}{a\wedge\eta}\exp(L^\eta_1/(2\eta))$. Itô–Tanaka rewrites this density as the Girsanov factor of a martingale, identifying it with the SDE whose drift is $1_{\{X\le\eta\}}/X$; comparison of squared Bessel processes then lets the proof pass $\eta\to0$. On the discrete side, the machinery is the Lyons–Zheng decomposition, which expresses time-reversal-stable martingale increments with quadratic variation fixed by the gradient form; the whole tightness proof reduces to bounding $N^2\sum_{k=1}^N \langle h,x_k^N\rangle^2$ uniformly in $N$.

What would settle it

Compute, for a sine mode $e_1$ and growing $N$, the quantity $N^2\sum_{k=1}^N \langle e_1,x_k^N\rangle^2$. If this grows without bound, the uniform quadratic-variation estimate underpinning Theorem 2.1 fails and the tightness claim has no support. A second check: simulate $X_t^\eta = a + \int_0^t 1_{\{X_s\le \eta\}}X_s^{-1}\,ds + B_t$ for decreasing $\eta$ and compare the empirical law at time 1 to the Bessel-1 density; Theorem 4.1 asserts the distance goes to zero.

Watch

Extended reading notes

Core claim

The paper's central claim is that the continuous wetting measure $P^{1,\eta}_a$, defined on path space by the explicit density $\frac{X_1\wedge\eta}{X_1}\frac{a}{a\wedge\eta}\exp\bigl(\frac{1}{2\eta}L^\eta_1\bigr)$ with respect to the law of a 3-dimensional Bessel process started at $a$, is the law of the unique strong solution of the truncated-drift SDE $X_t = a + \int_0^t \frac{1_{\{X_s\le \eta\}}}{X_s}\,ds + B_t$, and that as $\eta\to0$ these laws converge weakly to $P^1_a$, the law of a reflecting Brownian motion on $[0,1]$. It also claims that the rescaled reversible dynamics of the discrete shrinking-strip wetting model form a tight family in $C([0,T],H^{-1}(0,1))$, so the sequence has subsequential limiting dynamics. In both cases the stated result is a genuine approximation theorem: the continuous family $P^{1,\eta}_a$ interpolates monotonically between the Bessel-3 law and the Bessel-1 law, and the discrete tightness was previously an open conjecture.

Load-bearing premise

The proof of the discrete tightness result depends on a sharp collective estimate for the tent-shaped coordinate functions—the sum of squared overlaps with any fixed height profile must be of order $N^{-2}$—and the text's displayed Cauchy–Schwarz bound does not by itself give that estimate.

Editorial extensions

If this is right

The rescaled discrete dynamics have subsequential weak limits in $C([0,T],H^{-1}(0,1))$, so any future proof of the Bessel-SPDE limit only needs identification of the limit, not precompactness.
The continuous family $P^{1,\eta}_a$ gives an explicit monotone interpolation $P^1_a \preceq P^{1,\eta}_a \preceq P^{1,\eta'}_a \preceq P^3_a$ for $\eta \le \eta'$, so reflecting Brownian motion is the $\eta\to0$ endpoint of a continuous family of Bessel-type laws.
The mollified continuous wetting measures have a well-defined reversible dynamics described by an SPDE with reflection at 0 and an attractive term near level $\eta$; sending $\eta\to0$ formally recovers the conjectured Bessel-1 dynamics.
The static convergence of $P^{1,\eta}_a$ works on any finite time horizon, giving a finite-time continuous approximation of reflecting Brownian motion that is not available as an absolutely continuous tilt on the infinite horizon.

Reading between the lines

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

A direct route to the open conjectures is to prove that the rescaled discrete integration-by-parts boundary term converges to the Bessel-SPDE boundary term at generator level; the paper identifies this as the missing step.
The monotone interpolation $P^1_a \preceq P^{1,\eta}_a \preceq P^3_a$ suggests a broader universality: any pinning shape with the same effective critical parameter should produce the same $\eta\to0$ law, making the specific choice of $\phi_a$ irrelevant for the scaling limit.
Because the continuous wetting measure is an explicit tilt of a 3-dimensional Bessel law, it can serve as an importance-sampling proposal for reflecting Brownian motion, and the comparison argument gives a checkable way to measure how the bias disappears as $\eta\to0$.

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

Dear colleague — two things stand out. Theorem 2.1 answers the tightness conjecture from [DO19] for reversible gradient dynamics of shrinking-strip wetting models, and Theorem 4.1 introduces a genuinely new continuous wetting measure — a Brownian meander tilted by local time near zero — and proves it converges to reflecting Brownian motion. The explicit density, the BES(3)-to-reflecting-BM interpolation, and the fact that the conjectures are clearly labeled as conjectures are all real positives. The paper is a serious step toward the conjectured Bessel SPDE limit.

It deserves a serious referee. The Girsanov/Itô–Tanaka computation behind Theorem 4.1 is clean, and the convergence argument is essentially sound. The conjectures in Sections 3 and 4.3 are honest about what is open, and the self-citation to [EAZ19] is not a problem because the relevant statements are explicitly conjectural.

There are two soft spots, both repairable. In the proof of Theorem 2.1, the quadratic variation bound after Eq. (2.2) does not follow from Cauchy–Schwarz plus ‖x_k^N‖² ≤ 2N^{-2}. That route gives Σ |⟨h,x_k^N⟩|² ≤ 2N^{-1}‖h‖², so the factor 2N² leaves a divergent N-dependence. What is needed is the Bessel-sequence bound Σ |⟨h,x_k^N⟩|² ≤ C N^{-2}‖h‖², which is true for these hat functions because their supports overlap at most twice, but it is not stated. Without that estimate the BDG bound and the H^{-1} tightness conclusion do not go through as written. This is a genuine gap, but local and easily fixed.

Second, SDE (4.5) is used for a = 0 with a singular drift 1_{X≤η}/X, and the paper does not prove pathwise uniqueness or strong existence there. This is probably minor — the convergence P^{1,η}_0 → P^1_0 can likely be obtained from the a > 0 case via the Imhof relation or a direct Girsanov argument — but the proof as written leans on an unstated assumption.

Bottom line: the main results look true, the paper is well situated in the literature, and the gaps are genuine but local. I would send it to peer review and ask for a revised proof of the frame estimate and a justification of the a = 0 SDE step. For people working on wetting models, gradient interface dynamics, or Bessel SPDEs, this is worth reading.

Referee Report

2 major / 4 minor

Summary. The paper studies scaling limits of gradient (Langevin) dynamics associated with critical wetting models. In the discrete setting, for the strip-pinned wetting model of [DO19], the authors prove that the diffusively rescaled reversible processes (Y^N) are tight in C([0,T],H^{-1}(0,1)) (Theorem 2.1), using the Lyons–Zheng decomposition and an estimate of the martingale quadratic variation. In the continuum setting, they introduce the tilted Bessel measure P^{1,η}_a(dX) = (X_1∧η)/X_1 · a/(a∧η) exp(L^η_1/(2η)) P^3_a(dX), identify it as the law of the SDE dX_t = 1_{X_t≤η}/X_t dt + dB_t, and prove weak convergence to the law P^1_a of a one-dimensional Bessel process started from a as η→0 (Theorem 4.1). They also derive an integration-by-parts formula for the discrete measures, conjecture that the limit of (Y^N) is the reversible gradient dynamics of reflected Brownian motion, and formulate a Bessel-SPDE conjecture for a further continuous approximation.

Significance. If the proofs are completed, the paper makes two useful contributions. The continuum result of Theorem 4.1 is particularly clean: the density is an explicit exponential martingale (4.6), the family P^{1,η}_a monotonically interpolates between P^3_a and P^1_a, and the proof works on any finite time interval while correctly identifying the failure of the infinite-horizon analogue. The paper is also honest about its limitations: Conjectures 3.4 and 4.5 are clearly labeled as open, as are the convergence of the discrete integration-by-parts formula and the strong Feller property. Even if the conjectures remain unresolved, the explicit static approximation of reflected Brownian motion by a local-time-tilted Bessel/meander law is a valuable and potentially reusable result.

major comments (2)

[Section 2, proof of Theorem 2.1, after Eq. (2.2)] The claimed uniform bound on the sharp bracket is not obtained by the argument as written. Cauchy–Schwarz together with ‖x_k^N‖² ≤ 2N^{-2} gives Σ_{k=1}^N |⟨h,x_k^N⟩|² ≤ 2N^{-1}‖h‖², and hence the prefactor 2N² in (2.2) yields a bound of order N, not a constant. The argument needs the frame-type estimate Σ_{k=1}^N |⟨h,x_k^N⟩|² ≤ C N^{-2}‖h‖², which is true (it follows from x_k^N ≤ N^{-1/2}1_{I_k}, with I_k=[(k-1)/N,(k+1)/N], and the uniform bound on the overlap of the intervals I_k), but it is neither stated nor proved. Since this estimate is exactly what removes the N-dependence from the BDG bound and thus supports the H^{-1}(0,1) tightness conclusion, it should be added.
[Section 4.2, proof of Theorem 4.1, passage from Eq. (4.7) to the limit] The convergence of the SDE for Z^η is not fully justified. The text has only pointwise a.s. convergence Z^η→Z and the a.s. bound Z≥Z^0>0 for a.e. t. To let η→0 in (4.7) one must prove convergence of the stochastic integrals, e.g. by E∫_0^1(√Z^η_s−√Z_s)² ds→0, using the domination 0≤(√Z^η_s−√Z_s)²≤Z^η_s≤Z^∞_s and the integrability of Z^∞, and one must justify the vanishing of the occupation drift 2∫_0^t 1_{Z^η_s≤η²}ds by dominated convergence. These are repairable but should be written out. In addition, the sentence 'the only possible subsequential weak limit... is given by the law of X' is used before the continuity of the pointwise limit X has been established; the argument should be reordered so that the SDE identification also yields continuity of X.

minor comments (4)

[Proof of Theorem 2.1] The phrase 'reﬂecting Brownian bridge started from 0' appears to be a typo; elsewhere in the paper P^1_0 denotes the law of reflecting Brownian motion (one-dimensional Bessel process) started from 0, not a bridge.
[Section 4.2, Eq. (4.5)] The assertion that pathwise uniqueness follows from monotonicity of x ↦ 1_{x≤η}/x is not a standard criterion as stated, because the coefficient is singular at 0 and discontinuous at η; please replace it by a reference or proof (e.g., via the Bessel SDE and Yamada–Watanabe).
[Proposition 4.4] The Cauchy–Schwarz argument as written yields only L¹-boundedness of exp(X_1/η−M^ǫ_1), which does not by itself imply uniform integrability; the conclusion follows if one notes that both factors are bounded in L^p for every p<∞, but this should be stated explicitly.
[Remark 4.3] The claim that 1/(2η) is the smallest coefficient α for which exp(αL^η_∞) is not integrable is not proved; if kept, it should be accompanied by a reference or a short derivation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the dynamical tightness and continuous wetting convergence are proved against independent static benchmarks and explicit SDE arguments; self-citations are not load-bearing in a circular sense.

full rationale

The derivation chain is not circular. Theorem 2.1 proves tightness of the rescaled reversible dynamics using the Lyons-Zheng decomposition and the explicit quadratic variation computation (2.2); the only author-overlapping input, [DO19, Thm 1.5], supplies the static one-time marginal convergence of Pf_{phi_{a_N},N} to P^1_0, which is a different, independently established scaling limit and is used as an outside benchmark rather than as the dynamical conclusion being proved. The continuous approximation in Theorem 4.1 is self-contained: P^{1,eta}_a is defined by an explicit Radon-Nikodym density, the paper derives the equivalent SDE (4.5) by a Girsanov and Ito-Tanaka computation, and convergence to P^1_a is obtained from comparison, tightness of the squared processes, and strong uniqueness of the limiting SDE; no fitted parameter is renamed a prediction and no equation is defined in terms of the target limit. Conjectures 3.4 and 4.5 invoke [EAZ19] and related work but are explicitly labeled conjectures and are not used in the proofs of the theorems. The Cauchy-Schwarz estimate in the proof of Theorem 2.1 appears too weak to give the stated uniform bound, but that is a repairable correctness gap in an estimate, not an equivalence-by-construction or a self-citation loading; hence it does not raise the circularity score.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper's proven results are analytic: they rely on standard stochastic calculus (Girsanov, Itô-Tanaka, comparison, Kolmogorov tightness), on the static scaling limit of [DO19], and on Dirichlet form constructions. No free parameters are fitted; the small parameters η, ε, a_N are model parameters that are sent to 0. No new physical entities are introduced.

assumptions (5)

domain assumption Static scaling limit of the shrinking-strip wetting measure to reflecting Brownian motion (Theorem 1.5 of [DO19]), used for tightness of the one-time marginals in Theorem 2.1.
The proof of Theorem 2.1 needs the marginals (Y^N_t) to be tight in H^{-1}; this follows from [DO19] under Condition (A) and a_N = o(N^{-1/2}). If this external result failed, the tightness claim would not follow.
standard math The gradient SDE (1.2) with reflection has a unique strong solution and the associated reversible Markov process exists with the stated Dirichlet form.
The paper constructs the dynamics via the Dirichlet form, relying on standard theory (e.g., [FOT10], [Zam02]); no proof of well-posedness is given in the paper.
domain assumption The SDE X_t = a + ∫_0^t 1_{X_s≤η}/X_s ds + B_t has a unique strong solution for all a ≥ 0, including a = 0 where the drift is singular.
Invoked in Section 4.2 to identify P^{1,η}_a as the law of X^η and to run the comparison argument. The cited Yamada-Watanabe theorem is applied although the drift coefficient is undefined at 0, so this premise requires an additional approximation or a dedicated reference.
standard math The hat functions x_k^N form a frame for H_N with upper frame bound O(N^{-2}), so that Σ_k |⟨h, x_k^N⟩|^2 ≤ C N^{-2} ||h||^2 uniformly in N.
Used in the proof of Theorem 2.1 to get the uniform quadratic variation bound; the paper's stated Cauchy-Schwarz argument is insufficient, so this frame property is an implicit background fact.
standard math The comparison theorem and Itô-Tanaka formula apply to the truncated Bessel drift and to the convex function log((x∧η)/x).
The identification of the limit process in Theorem 4.1 uses the comparison principle from [RY13, Chapter IX] and the Itô-Tanaka formula from [RY13, Chapter VI].

how reviews work

0 comments

Cite this review

Pith. "Pith review of On the gradient dynamics associated with wetting models." pith.science (2026). https://pith.science/paper/ILOZCIA4

@misc{pith2026190808850,
  author       = {Pith},
  title        = {Pith review of: On the gradient dynamics associated with wetting models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ILOZCIA4}},
  note         = {Machine review of arXiv:1908.08850}
}

read the original abstract

We consider several critical wetting models. In the discrete case, these probability laws are known to converge, after an appropriate rescaling, to the law of a reflecting Brownian motion, or of the modulus of a Brownian bridge, according to the boundary conditions. In the continuous case, a corresponding convergence result is proven in this paper, which allows to approximate the law of a reflecting Brownian motion by the law of Brownian meander tilted by its local time near the origin. On the other hand, these laws can be seen as the reversible probability measure of some Markov processes, namely, the dynamics which are encoded by integration by parts formulae. After proving the tightness of the associated reversible dynamics in the discrete case, based on heuristic considerations on the integration by parts formulae, we provide a conjecture on the limiting process, which we believe to satisfy a Bessel SPDE as introduced in a recent work by Elad Altman and Zambotti.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages

[1]

Sharp asymptotic behavior for wetting models in (1+1) -dimension

Francesco Caravenna, Giambattista Giacomin, and Lorenzo Zambotti. Sharp asymptotic behavior for wetting models in (1+1) -dimension. Electron. J. Probab. , 11:no. 14, 345--362, 2006

work page 2006
[2]

Deuschel, G

J.D. Deuschel, G. Giacomin, and L. Zambotti. Scaling limits of equilibrium wetting models in (1+1) -dimension. Probab. Theory Related Fields , 132(4):471--500, 2005

work page 2005
[3]

Deuschel and T

J.D. Deuschel and T. Orenshtein. Scaling limit of wetting models in 1+ 1 dimensions pinned to a shrinking strip. Stochastic Processes and their Applications , 2019

work page 2019
[4]

Bessel SPDEs with general Dirichlet boundary conditions

H. Elad Altman. Bessel SPDE s with general D irichlet boundary conditions. arXiv preprint arXiv:1908.02241 , 2019

work page Pith review arXiv 1908
[5]

Elad Altman and L

H. Elad Altman and L. Zambotti. Bessel SPDE s and renormalised local times. Probability Theory and Related Fields , Jun 2019

work page 2019
[6]

Ethier and T.G

S.N. Ethier and T.G. Kurtz. Markov processes: Characterization and convergence . Wiley Series in Probability and Mathematical Statistics: Probability and Mathematical Statistics. John Wiley & Sons, Inc., New York, 1986

work page 1986
[7]

Engelbert and G

H.J. Engelbert and G. Peskir. Stochastic differential equations for sticky B rownian motion. Stochastics An International Journal of Probability and Stochastic Processes , 86(6):993--1021, 2014

work page 2014
[8]

Fattler, M

T. Fattler, M. Grothaus, and R. Vo hall. Construction and analysis of a sticky reflected distorted B rownian motion. Ann. Inst. Henri Poincar\' e Probab. Stat. , 52(2):735--762, 2016

work page 2016

Show all 17 references

[9]

Fukushima, Y

M. Fukushima, Y. Oshima, and M. Takeda. Dirichlet forms and symmetric Markov processes , volume 19. Walter de Gruyter, 2010

2010
[10]

T. Funaki. Stochastic interface models. In \'E cole d'\'et\'e de Saint-Flour XXXIII-2003 , volume 1869 of Lecture Notes in Math. , pages 103--274. Springer, Berlin, 2005

2003
[11]

Grothaus and R

M. Grothaus and R. Vo hall. Strong F eller property of sticky reflected distorted B rownian motion. J. Theoret. Probab. , 31(2):827--852, 2018

2018
[12]

Weakly pinned random walk on the wall: pathwise descriptions of the phase transition

Yasuki Isozaki and Nobuo Yoshida. Weakly pinned random walk on the wall: pathwise descriptions of the phase transition. Stochastic Process. Appl. , 96(2):261--284, 2001

2001
[13]

Revuz and M

D. Revuz and M. Yor. Continuous martingales and B rownian motion , volume 293. Springer Science & Business Media, 2013

2013
[14]

Zambotti

L. Zambotti. Integration by parts formulae on convex sets of paths and applications to SPDE s with reflection. Probability Theory and Related Fields , 123(4):579--600, 2002

2002
[15]

Zambotti

L. Zambotti. Integration by parts on - B essel bridges, > 3 , and related S P D E s. The Annals of Probability , 31(1):323--348, 2003

2003
[16]

Zambotti

L. Zambotti. Fluctuations for a interface model with repulsion from a wall. Probability theory and related fields , 129(3):315--339, 2004

2004
[17]

Zambotti

L. Zambotti. Random O bstacle P roblems, \'E cole d' \'E t \'e de P robabilit \'e s de S aint- F lour XLV -2015 , volume 2181. Springer, 2017

2015

Pith tools

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

[1] [1]

Sharp asymptotic behavior for wetting models in (1+1) -dimension

Francesco Caravenna, Giambattista Giacomin, and Lorenzo Zambotti. Sharp asymptotic behavior for wetting models in (1+1) -dimension. Electron. J. Probab. , 11:no. 14, 345--362, 2006

work page 2006

[2] [2]

Deuschel, G

J.D. Deuschel, G. Giacomin, and L. Zambotti. Scaling limits of equilibrium wetting models in (1+1) -dimension. Probab. Theory Related Fields , 132(4):471--500, 2005

work page 2005

[3] [3]

Deuschel and T

J.D. Deuschel and T. Orenshtein. Scaling limit of wetting models in 1+ 1 dimensions pinned to a shrinking strip. Stochastic Processes and their Applications , 2019

work page 2019

[4] [4]

Bessel SPDEs with general Dirichlet boundary conditions

H. Elad Altman. Bessel SPDE s with general D irichlet boundary conditions. arXiv preprint arXiv:1908.02241 , 2019

work page Pith review arXiv 1908

[5] [5]

Elad Altman and L

H. Elad Altman and L. Zambotti. Bessel SPDE s and renormalised local times. Probability Theory and Related Fields , Jun 2019

work page 2019

[6] [6]

Ethier and T.G

S.N. Ethier and T.G. Kurtz. Markov processes: Characterization and convergence . Wiley Series in Probability and Mathematical Statistics: Probability and Mathematical Statistics. John Wiley & Sons, Inc., New York, 1986

work page 1986

[7] [7]

Engelbert and G

H.J. Engelbert and G. Peskir. Stochastic differential equations for sticky B rownian motion. Stochastics An International Journal of Probability and Stochastic Processes , 86(6):993--1021, 2014

work page 2014

[8] [8]

Fattler, M

T. Fattler, M. Grothaus, and R. Vo hall. Construction and analysis of a sticky reflected distorted B rownian motion. Ann. Inst. Henri Poincar\' e Probab. Stat. , 52(2):735--762, 2016

work page 2016