• Aucun résultat trouvé

We provide a modified proof of the absence of a wetting transition in the square-potential case, which does not require the aforementioned lower bound

N/A
N/A
Protected

Academic year: 2022

Partager "We provide a modified proof of the absence of a wetting transition in the square-potential case, which does not require the aforementioned lower bound"

Copied!
3
0
0

Texte intégral

(1)

NOTE ON BOLTHAUSEN-DEUSCHEL-ZEITOUNI’S PAPER ON THE

ABSENCE OF A WETTING TRANSITION FOR A PINNED HARMONIC CRYSTAL IN DIMENSIONS THREE AND LARGER

LOREN COQUILLE AND PIOTR MIŁO ´S

ABSTRACT. The article [1] provides a proof of the absence of a wetting transition for the discrete Gaussian free field conditioned to stay positive, and undergoing a weak delta-pinning at height 0.

The proof is generalized to the case of a square pinning-potential replacing the delta-pinning, but it relies on a lower bound on the probability for the field to stay above the support of the potential, the proof of which appears to be incorrect. We provide a modified proof of the absence of a wetting transition in the square-potential case, which does not require the aforementioned lower bound. This shows that the results of [1] hold true.

1. DEFINITIONS AND NOTATIONS

We keep the notations of [1] except for the field which we callφinstead ofX. LetAbe a finite subset ofZd, letφ=(φx)x∈Zd ∈RZ

d and the Hamiltonian defined as HA(φ)= 1

8d

X

x,y∈A∪∂A:|x−y|=1

x−φy)2 (1)

where∂Ais the outer boundary ofA. The following probability measure onRAdefines the discrete Gaussian free field onA(with zero boundary condition) :

PA(dφ)= 1

ZAe−HA(φ)Aδ0(dφAc) (2) wheredφA=Q

x∈Axandδ0is the Dirac mass at 0. The partition functionZAis the normalization ZA = R

RAexp(−H(φA))dφA. We will also need the following definition of a setAbeing ∆-sparse (morally meaning that it has only one pinned point per cell of side-length∆), which we reproduce from [1, page 1215] :

Definition 1. LetN ∈ Z, ∆> 0,ΛN = {−bNc/2, . . . ,bNc/2}d and letlN = {zi}|li=N1| denote a finite collection of pointszi ∈ ΛN such that for eachy ∈ΛN∩∆Zdthere is exacly onez ∈lN such that

|z−y|<∆/10. LetAl

N = ΛN\lN.

2. LOWER BOUND ON THE PROBABILITY OF THE HARD WALL CONDITION

The proof of [1, Theorem 6] relies on [1, Proposition 3]. Unfortunately, the proof provided in the paper, when applied witht>0 provides a lower bound which is a little bit weaker than what is clamed, namely

Proposition 2. Correction of [1, Proposition 3] :

Assume d ≥ 3and let t ≥ 0. Then there exist three constants c1,c2,c3 > 0depending on t, and c4>0independent of t, such that, for all∆integer large enough

lim inf

N→∞ inf

l

N

1

(2N+1)d logPA

l N

(Xi≥t,i∈Al

N)≥ −dlog∆

d +c1log log∆

d −c2ec4t

log

d(log∆)c3 (3) The statement of [1, Proposition 3] only contains the first two terms. The dependence in t vanishes between equations (2.4) and (2.5) in [1]. Note that fort = 0 the third term is irrelevant and the bound coincides with the one stated in the paper.

1

(2)

NOTE ON THE ABSENCE OF A WETTING TRANSITION FOR A PINNED GFF INd3 2

3. PROOF OF THE ABSENCE OF A WETTING TRANSITION IN THE SQUARE-POTENTIAL CASE

Let us introduce the following notations ξˆN = X

x∈ΛN

1[x|≤a], ξ˜N = X

x∈ΛN

1[φx∈[0,a]],

+A ={φx ≥0,∀x∈A}, Ω+N ={φx ≥0,∀x∈ΛN} A={x∈ΛNx ∈[0,a]}

and the following probability meaure with square-potential pinning : P˜N,a,b(dφ)= 1

N,a,bexp







−H(φ)+ X

x∈ΛN

b1[φx∈[0,a]]







ΛNδ0(dφΛc

N) in contrast with the measure used in [1] :

N,a,b(dφ)= 1 ZˆN,a,bexp







−H(φ)+ X

x∈ΛN

b1[φx∈[−a,a]]







ΛNδ0(dφΛc

N).

Observe that

N,a,b( ˜ξN < Nd|Ω+N)= PˆN,a,b( ˆξN < Nd|Ω+N) Theorem 3. (Absence of wetting transition, [1, Theorem 6])

Assume d ≥3and let a,b>0be arbitrary. Then there existsb,a, ηb,a>0such that

N,a,b( ˜ξN > b,aNd|Ω+N)≥1−exp(−ηb,aNd). (4) provided N is large enough.

Proof. Let us compute the probability of the complement event and provide bounds on the numer- ator and the denominator corresponding to the conditional probability :

N,a,b( ˜ξN < Nd|Ω+N)= P˜N,a,b( ˜ξN < Nd∩Ω+N)

N,a,b(Ω+N) (5)

3.1. Lower bound on the denominator.

N,a,b(Ω+N)FKG≥ ZN

N,a,b X

A⊂ΛN

(eb−1)|A|PN(A ⊃A)

| {z }

(∗)

PN(Ω+Ac|A ⊃ A)

| {z }

≥PN(+Ac|φ≡0onA)

=PAc(+Ac)

PN(Ω+A|A ⊃A)

| {z }

=1

(6)

WritingA={x1, . . . ,x|A|},

(∗)= PN(φ∈[0,a]on A)=

|A|

Y

i=1

PNxi ∈[0,a]|φxi+1, . . . , φx|A| ∈[0,a]) (7)

|A|

Y

i=1

PNxi ∈[0,a]|φxi+1, . . . , φx|A| =0) (8)

=

|A|

Y

i=1

P{xi+1,...,x|A|}cxi ∈[0,a])≥[c(1/2∧a)]|A| (9)

for somec=c(d)>0, since the variance of the free field is bounded ind ≥3.

Hence,

N,a,b(Ω+N)≥ ZN

N,a,b X

A⊂ΛN

exp(J0|A|)PAc(Ω+Ac) (10) withJ0 =log(eb−1)+logc+log(1/2∧a).

(3)

NOTE ON THE ABSENCE OF A WETTING TRANSITION FOR A PINNED GFF INd3 3

3.2. Upper bound on the numerator.

N,a,b( ˜ξN < Nd∩Ω+N)= ZN

N,a,b X

A:|A|<Nd

(eb−1)|A|P| N(A ⊃{z }A)

(1/2∧a)|A|

PN(Ω+N∩ {φ >a on Ac}|A ⊃A)

| {z }

≤1

(11)

≤ ZN

N,a,b]{A:|A|< Nd}exp(JNd) (12) withJ =log(eb−1)+log(1/2∧a), where]Xdenotes the cardinality of the setX.

3.3. Upper bound on(5).

N,a,b( ˜ξN < Nd|Ω+N)≤ exp(JNd)]{A:|A|< Nd} P

A⊂ΛNexp(J0|A|)PAc(Ω+Ac) (13) And now we proceed similarily as for the proof withδ-pinning potential:

1

Nd log ˜PN,a,b( ˜ξN < Nd|Ω+N)= 1 Nd log

exp(JNd)]{A:|A|< Nd}

(14)

− 1

Nd log X

A⊂ΛN

exp(J0|A|)PAc(Ω+Ac) (15) The right hand side of (14) can be bounded by(J+1−log) asNtends to infinity (by a rough approximation and the Stirling formula), which in turn can be made as close to 0 as we want by choosing =(J) sufficiently small. See [1].

To bound (15) we use [1, Proposition 3] witht=0 which matches to our Proposition 2 : (15)≤ − 1

Nd log X

A⊂ΛN:A is−sparse

exp(J0|A|)PAc(Ω+Ac) (16)

≤ − 1 Nd

N

d

[(dlog∆ +c0)+J0−dlog∆ +c1log log∆]

!

(17)

=−J0+c0+c1log log∆

d <0 for∆ = ∆(J0) large enough. (18) where ∆-sparseness corresponds to Definition 1 : a set A ⊂ ΛN is ∆-sparse if it equals Al

N, for some setlN.

REFERENCES

[1] E. Bolthausen, J. D. Deuschel, and O. Zeitouni. Absence of a wetting transition for a pinned harmonic crystal in dimensions three and larger.J. Math. Phys., 41(3):1211–1223, 2000. Probabilistic techniques in equilibrium and nonequilibrium statistical physics.

L. COQUILLE, INSTITUTFOURIER, UMR 5582DUCNRS, UNIVERSITE DE´ GRENOBLEALPES, 100RUE DES

MATHEMATIQUES´ , 38610 GIERES` , FRANCE

E-mail address:loren.coquille@univ-grenoble-alpes.fr P. MIŁOS´, MIMUW, BANACHA2, 02-097 WARSZAWA, POLAND

E-mail address:pmilos@mimuw.edu.pl

Références

Documents relatifs

Their linearity properties were recently studied by Drakakis, Requena, and McGuire, who conjectured that the absolute values of the Fourier coefficients of an Exponential Welch

We present a proof of the Generalized Riemann hypothesis (GRH) based on asymptotic expansions and operations on series.. The advan- tage of our method is that it only uses

On effectue ensuite une opération appelée DifférenceCarrée : pour chaque côté du carré, on calcule la différence positive ou écart entre les deux nombres a et

As the deterministic bounds, the Bayes- ian Cram´er-Rao bound and the Bayesian Bhattacharyya bound are small error bounds contrary to the Bobrovsky-Zaka¨ı bound and the

Ilias, Immersions minimales, premiere valeur propre du Lapla- cien et volume conforme, Mathematische Annalen, 275(1986), 257-267; Une ine- galit´e du type ”Reilly” pour

Similarly to the case where P is the Laplace operator (see [9]), we apply a scaling argument on this estimate to derive a linear Strichartz estimate on T L 2 on the time interval

The existence of a positive lower bound of the compressi- bility is proved in the case of general quantum lattice systems at high temperature.. The compressibility

For every p-pass Shellsort algorithm and every increment sequence, every subset of n!/2 n input permutations of n keys contains an input permutation that uses ⍀ ( pn 1⫹(1⫺⑀)/p