Skip to main content

Full text of "Superdiffusivity of asymmetric exclusion process in dimensions one and two"

See other formats


Superdiffusivity of asymmetric exclusion process in 

dimensions one and two 



C. Landimt J. QuastelJ M. Salmhofer} and H.-T. Yau§ 

IMPA, University of Toronto, Max-Planck-Institut, Leipzig, and Courant Institute 

February 1, 2008 



Abstract 

We prove that the diffusion coefficient for the asymmetric exclusion process diverges 
at least as fast as t^^^ in dimension d = 1 and (logt)^/^ in d = 2. The method applies 
to nearest and non-nearest neighbor asymmetric exclusion processes. 



1 Introduction 



Asymmetric exclusion is a Markov process on {0, 1}^'' consisting of interacting continuous 
time random walks with asymmetric jump rates. There is at most one particle allowed per 
site. A particle at a site x waits for an exponential time and then jumps to y provided the 
site is not occupied. Otherwise the jump is suppressed and the process starts again. The 
jump is attempted at rate pijj — x). In this article the jump law p(-) is assumed to have a 
nonzero mean, so that there is transport of the system. 

Consider the system in equilibrium with a Bernoulli product measure of density p as the 
invariant measure. Define the time dependent correlation function in equilibrium by 

If we choose p = 1/2, there is no net global drift, i.e., J^x-^'^i^^^) ~ ^- Otherwise one needs 
to subtract a net drift, which complicates but does not change the results or methods. Our 
main question is the behaviour for large t of the diffusion coefficient. 



X 



In dimensions d > 3, the diffusion coefficient was proved to be bounded |^ for general 
asymmetric simple exclusion processes. Based on mode coupling theory, Beijeren, Kutner 
and Spohn B conjectured that D{t) ~ (logt)^/^ in dimension d = 2 and D(t) ~ t^^^ in 



*Partially supported by CNRS, UMR 6085, landim@inipa.br 
^Partially supported by NSERC, quastel@math.toronto.edu 
■I- mns @ mis . mpg . de 

^Partially supported by NSF grant DMS-0072098, yau@cims.nyu.edu 



1 



d = 1. Similar predictions we made in for the Kardar-Parisi-Zhang equation in ci = 1, 
which, when differentiated and appropriately discretized, yields the asymmetric exclusion 
process. 

This problem has received much attention recently in the context of integrable systems. 
The main quantity analyzed there is fluctuation of the current across the origin in d = 1 
for the totally asymmetric simple exclusion process (only nearest neighbor jumps to the 
right), starting from the special initial configuration with all sites to the left of the origin 
occupied and all sites to the right of the origin empty. Johansson ^ observed that in this 
special situation the current across the origin can be mapped into a last passage percolation 
problem. By analyzing this problem asymptotically, Johansson proved that the variance 
of the current is of order t^/^. In the case of discrete time, Baik and Rains analyze an 
extended version of the last passage percolation problem and obtain fluctuations of order t", 
where a = l/3ora = l/2 depending on the parameters of the model. Both the approaches 
of and are related to the earlier results of Baik-Deift- Johansson [|l| on the distribution 
of the length of the longest increasing subsequence in random permutations. 

In |]10| (see also Prahofer and Spohn succeeded in mapping the current of the 



totally asymmetric simple exclusion process into a last passage percolation problem for a 
general class of initial data, including the equilibrium case considered in this article. For the 
discrete time case, the extended problem is closely related to the work [Q], but the boundary 
conditions are different. For continuous time, besides the boundary condition issue, one 
needs to extend the result of from the geometric to the exponential distribution. 

To relate these results to our problem, the variance of the current across the origin is 
proportional to 

J2\^\S{x,t) (1.1) 

X 

Therefore, Johansson's result on the variance of the current can be interpreted as the spread- 
ing of S{x, t) being of order t^/^. If we combine the work of [10| and 0, neglect various issues 



discussed above, and extrapolate to the second moment, we obtain a growth of the second 
moment as f^^^, consistent with the conjectured D(t) ~ t^^^. 

The results based on integrable systems give not just the variance of the current across 
the origin, but also its limiting distribution. The main restrictions appear to be the rigid 
requirements on the fine details of the dynamics and on the initial data, the restriction to 
one space dimension, and the special quantities which can be analysed. In fact, even for the 
totally asymmetric simple exclusion process in = 1 there was previously no proof that D{t) 
diverges as t ^ oo. And for general asymmetric exclusion processes all of these problems 
were completely open. In this article, we present a method to study the diffusion coefficient 
of general asymmetric exclusion processes. Using this we obtain without too much work 
lower bounds D{t) > (logt)^/^ in dimension d = 2 and D(t) > t^/^ m. d = 1. We have 
restricted the proof to the case p = 1/2, but a similar proof works for all densities away from 
zero or one. 



1.1 The model and main results 

Denote the particle configuration by = {rjx}x&i'i where rj^ is equal to 1 if site x is occupied 
and is equal to otherwise. Denote by rj^'^ the configuration obtained from r] by exchanging 



2 



the occupation variables at x and y : 

{Vz if ^ 7^ X, y, 
T]^ ifz = yand 
riy \l z — X. 

We assume that the jump law p(-) is local, p{z) = for \z\ > L for some L < oo, and that 
there is transport of mass, zp{z) ^ 0. The generator of the asymmetric simple exclusion 
process is given by 

miv) = J2 p(y- - vy)[f{v'''') - fm ■ (1-2) 

For each p in [0, 1], denote by the Bernoulli product measure on (0, 1}^'' with density p 
and by < ■, ■ >p the inner product in L^(i/p). The probability measures i/p are invariant for 
the process. 

For two cylinder functions /, g and a density p, denote by {f;g)p the covariance of / and 
g with respect to i/p : 

{f;g)p = {fg)p- if) pig) p- 

Let Pp denote the law of the asymmetric exclusion process starting from the equilibrium 
measure Up. Expectation with respect to Pp is denoted by Ep. Let 

Sp{x,t)^Ep[{ri,{t)-ri,{0)}riom 

denote the time dependent correlation functions in equilibrium with density p. Denote by x 
the compressibility given by 

X = x{p) = '^{vx;vo)p- 

X 

In our setting, x(p) = p(l — p). 

The bulk diffusion coefficient is defined by 

Di,j{p,t) = J2^i^j^p(^^'^) -xMivjtn, (1.3) 

where v in W'' is the velocity defined by 

vt ^ -V] xEp\{r),{t) - Vx{0)}r)o{0)] . (1.4) 

xezd 

To simplify the notation we now specialize to the special case of the totally asymmetric 
simple exclusion process to the right in d = 1 and, in d = 2, jumps only to the nearest 
neighbor to the right in the Xi coordinate, and jumps to both nearest neighbors in the X2 
coordinate with symmetric jump rule. More precisely, we take 

miv) = J2'^x{l-Vx+i)[f{r]'''^')-fm, d^l; (1.5) 
xez 

{CfM = J2 Vx[l - 77.+eJ[/(77^'^+^^) - fm + lifiv'^''-''') - mi d = 2. (1.6) 

xez'^ 



3 



where we have combined the symmetric jump in the X2-axis into the last term. We emphasize 
that the resuh and method in this paper apply to all asymmetric exclusion processes; the 
special choice is made only to simplify the notation. The velocity of the totally asymmetric 
simple exclusion process is explicitly computed as v = 2(1 — 2p) in d = 1 and v = 2(1 — 2p)ei 
in d = 2. We further assume for simplicity that the density p = 1/2 so that the velocity is 



zero. 



Denote the instantaneous currents (i.e., the difference between the rate at which a particle 
jumps from x to x + ei and the rate at which a particle jumps from x + ei to x) by w^.x+er 



w. 



x,x+ei 



Vx[^ — Vx+ei]^ Wx^x+e2 — 2 [^3;'+e2 ~ Vx 



We have the conservation law 



;i-7) 



. 



Let Wi{ri) denote the renormalized current in the i-th direction: 

d 



Note the subtraction of the linear term in this definition. We have 

wi{r]) = {r]o - p){r]e, - p) + p[Ve^ - Vo], ^2(77) = ^[r^ea - Vo] 



A function / on {0, l}^'* will be called local if it only depends on the variables at finitely 
many sites. For local functions / and g we define the semi-inner product 



{{9, h)), 



^ <T^g; h>p= ^ <T^h; g>i_ 



1.9) 



x& 



Since the density p is fixed to be 1/2 in this article, we will henceforth leave out the subscript. 
All but a finite number of terms in this sum vanish because Vp is a product measure and (?, 
h are local. From this inner product, we define the seminorm: 



((/,/))• 



1.10) 



Note that gradient terms g = r^h — h and all degree one functions vanish in this norm. 
Therefore we shall identify the currents w with their degree two parts: For the rest of the 
article we set 



W^l(^) = (^0 -P)(^?ei -P), W2(r/)=0 



1.11) 



Fix a unit vector I G M'^. A simple calculation using Ito's formula ^ allows one to rewrite 
the diffusivity as 



1 

X 



/ i-w{r]{s))d6 
Jo 



1.12) 



4 



This is a variant of the Green-Kubo formula ||^. In = 1 of course D{t) is a scalar. In our 
special case in d = 2, since W2 = 0, D{t) is a matrix with all entries zero except 







Recall that e~^^ f{t)dt ~ as A ^ means, in some weak sense, that f{t) ~ t"^^ 
as t 00. Throughout the following A will always be a positive real number. We can now 
state the main result. 

Theorem 1 There exists C > Q so that for sufficiently small A > 0, 

/•oo 

d=l: / e"^* tD{t)dt > CX-^~-^, (1.13) 
Jo 



00 



d = 2: / e-^' tDii{t)dt>CX-^\logX\ ' . (1.14) 



The conjectured behavior for t large is D(t) ~ t^^^ in c? = 1 and -Dii(t) ~ (logt)^/^ in 
d = 2. This theorem says that in a certain average, asymptotic sense D(t) > t^^^ in 0? = 1 
and Dnit) > (logty/^ in d = 2. 

From the definition, we can rewrite the diffusion coefficient as 

t 2 i"^ r 

tDn{t) = - + - / ((e"^wi,wi)) duds 
2 xJo Jo 

in d = 2, with an analogous formula in ci = 1 (just drop the subscripts). Thus 

I e'^UDn(t)dt = — + - dt / e"^* {(e''^wi,wi)) duds 
Jo 2A^ xJo io io 

dt e-^(*-") ( /" rfs ) } ((e-^"e"^u;i , wi)) 



= ^ + ^''^"'((^'1' (A-/:)-V)). (1.15) 

Therefore, Theorem |l] follows from the following estimate on the resolvent. 

Lemma 1.1 There exists C > such that for sufficiently small A > 0, 

d=l: {{w,{X- C)-^w)) >CX-^/^; (1.16) 

d = 2: {{wi,{X- Cy^wi)) >C\logX\^^\ (1.17) 

2 Duality and resolvent hierarchy 

Let C* denote the adjoint of C with respect to the inner product of L'^{i^) and S = {C + C*)/2 
and A= {C — C*)/2 be the symmetric and antisymmetric parts of C so that 

C = S + A. (2.1) 



5 



Denote by C = C{p) the space of z/p-mean zero local functions. For a finite subset A of Z*^, 
denote by ^\ the t/p-mean zero local function defined by 

U-U^., ^^-^==- (2-2) 

Note that the collection {^a} where A ranges over finite subsets of Z*^, forms an orthonormal 
basis of L^(z/p). Denote by Ain the space of local functions of degree n, i.e., the space 
generated by monomials of degree n : 

Mn ^ {feC; J2fAU: /Ae^}- (2.3) 

|A|=n 

Note that in the definition all but a finite number of coefficients f\ vanish because / is 
assumed to be local. Denote by 

Cn = Ui<j<nMj (2.4) 

the space of cylinder functions of degree less than or equal to n. All i/p-mean zero local 
functions / can be decomposed uniquely as a finite linear combination of cylinder functions 
of finite degree : C — U„>iA^„. Any / e LP'ivp) can be written by degree, 

/ = (/l,/2,/3, ■ ■ ■) 

with e Mn- 

For / e Mn represented as / = EA,|A|=n /aCa we have 

oo 

11/11' = E E (2.5) 

n=l AcZd,|A|=na;eZd 

Note that this effectively reduces the degree by one. A simple computation shows that in 
this basis the symmetric part S ol C maps Ain into itself and is given by 



1 

^fiv) ^ "2^X1 E [fnu{x+ej} - fnu{x}] [^au{x+ej} - ^au{x}] ■ (2.6) 

j=i xgz<* n \n\=n-i 

Cln{x,x+ej}=<l> 

The asymmetric part A can be decomposed into three pieces ^ = + A+ — A*^ where 
^0 : Mn Mn into itself, A+ : Mn Mn+i and A^ : Mn Mn-i is the adjoint of 

■^ofiv) = — ^— ^ E E [/ou{x+ei} - fnu{x}] [^nu{x+ei} + ^nu{x}] , 

x&d- _njf2|=n-l 
nn{a;,a;+ei}=0 

^+/('7) = --\/p(l ~ P) E [/nu{x+ei} - /hum] Cnu{x,x+ei}- 

xezd n |a|=n-i 

nn{i,a;+ei|=0 

In our special case p = 1/2, we have = and thus A = A+ — A'^. We can also identify 
/ = J2a \A\=n /a^a with a symmetric function of n variables, 

/(x„...,x„) = |^^--> if^^v^^.fo^^v^i' i,j-h...,n; ^2.7) 
otherwise. 



6 



With this notation we have 



n+l 



i=l jj^i kytj 
• • • ; 2^n+l) f^'^li • • • ; -^j ~l~ , . . . , Xj, . . . Xfi-\-l) f^'^li • • • 1 -^ii • • • 1 •^ji ■ • • 1 2'n+l) 

where 5(0) = 1 and zero otherwise, and Xj indicates the absence of Xj in the vector. Also, 



... , x„) = ^ ^ ^ JJ ( 1 - + crej - Xfc) ) 

1=1 o-=± j=l k^i 

X [/(xi, . . . Xi + acj, . . . , x„) - /(xi, . . . , 



(2.9) 

Note that S is the discrete Laplacian with Neumann boundary condition on 

£i = {x„ := (xi, . . . , Xn) : Xi 7^ Xj, for i ^ j}. (2.10) 

The current w G C2 and in our notation W{o,ei} = 1/4 and = for A 7^ {0, ei}, |A| =2. 
The resolvent equation (A — C)u = w becomes the hierarchy 



A\us + (A - 5)^2 = w, 
A\_Uk+i + (A — S)uk — A+Uk-i = 0, A; > 3. 



(2.11) 
(2.12) 



The hierarchy starts at degree 2 instead of 1 because the ((■,■)) inner product effectively 
reduces the degree by one. Any term A+ui is trivial in the sense of (|2.5| ) and hence the 
degree one term plays no role and we can set mi = 0. In the same way, we disregard the 
degree one part of the current, and take wa = for all finite subsets A C Z*^ except {0, ei}, 
and W{o,ei} = 1/4. 

Consider the truncated equation up to the degree n, 



A^us + (A - S)u2 =w, 
A\uk-\-i + (A — S)uk — A+Uk-i =0, n — 1 > /c > 3 

(A - S)Un - A+Un-l =0. 

We can solve the final equation of ( p.l3| ) by 

Mn = (A - Sy^A+Un-l- 

Substituting this into the equation of degree n — 1, we have 

r 1 ~^ 

Un-1= {X-S)+AliX-S)-^A+ Un-2- 

Solving iteratively we arrive at 



(2.13) 



U2 



(A-5) +^;|(A-5) + --- + ^;((A-5) + ^;(A-5)^U+) A- 



w. 



7 



Let TTn '■ C ^ Cn denote the projection onto degree < n and = 7r„£7r„. The truncated 
equation represents the solution of (A — Cn)u = w and hence {{u2,w)) = {{w, (A — 
For example 

w))={{w,iX-Cs)-'w)). (2.14) 
w))={{w,{\-C,)-'w)). 

Since ^+(A — C)^^A+ is nonnegative, we have the monotonicity inequality 

{{w, (A - Csr'w)) <{{w, (A - C,)-'w)) <■■■< {{w, (A - Cy'w)) 

<■■■< {{w, (A - C,)-'w)) < {{w, (A - C2)-'w)). (2.15) 

To check that {{w, {X — C)^^w)) is in fact the limit of these upper and lower bounds we use 
the variational formula. For any matrix M, let Ms denote the symmetric part (M + M*)/2. 
The identity {[Af-^] = M*(M,)-^M always holds, and thus we have 

{{w, (A - C)-'w)) = sup {2{{w, /)) - (((A - £)/, (A - S)-\X - £)/))} . (2.16) 

/ 

Note that 



{w, 



\-S + Al{\-S)-^A- 



{{w, X-S + AUX-S + Al{X-S)-^A 



A, 



(((A - C)f, (A - S)-\X - £)/)) = ((/, (A - S)f)) + {{Af, (A - SyUf)). 



Hence 



(K(A-£)-M) 



sup inf {2{{w - A*g, /)) - ((/, (A - S)f)) + {{g, (A 

/ 9 



S)g))}. (2.17) 



Let a„ denote the supremum restricted to / G C„, and a" denote the infimum restricted 



and a" | {{w, (A — C) ^w)). By straightforward 
— Cn)^^w)) < a", giving the desired result. 



to G C„ so that a„ ] {{w, (A — C)~^w)) 
computation one checks that a„ < {{w, (A 

In what follows we will present a general approach to the equations (|2.13| ) which, from 
( p.l5| ) and ( |2.14| ) give a nontrivial lower bound on the diffusion coefficient without too much 
work. Because it gives a sequence of upper and lower bounds, the method has the potential 
to give the full conjectured scaling of the diffusion coefficient. 



3 Degree 3 lower bounds 



From ( |2.15| ) and ( |2.14| ) of the previous section we have a lower bound at degree three. 



However, computations are complicated by the hard core exclusion. We now describe a 
method to remove the hard core restriction in the computation. We then perform the 
computations in Fourier space. The estimates which justify the removal of the hard core are 
presented in the next section. 

Recall (|2.1(]| ) the set Si consists of configurations of n particles, no two of which occupy 
the same site. In the previous section we consider functions on Si, and operators S and A-i- 
acting on them. By removal of the hard core, we mean replacing these by functions on Z'*" 
and operators acting on these. 



8 



Suppose that / is a function of particle configurations f{xi,... ,xn}^ i^iy ■ ■ ■ j^n) G ^i- We 
can extend / to all of Z""' by defining f{xi, . . . , x„) = f{xu...,x„} if (a^i, ... ,Xn) e £i and 
/(xi, . . . , Xn) = otherwise. Note that 



|^|=n ' XI,... ,Xn&,'^ 

For a function / : Z"'^ ^ M, we shall use the same symbol (/) to denote the expectation 

n! — ^ 

and write the inner product of two functions as (/,(?) = {fg). If / and g vanish on 
this coincides with the inner product introduced before. We also define, as before, ((/, g)) = 
^x€Zd{'^xf,g)- We use 7r„ for the projection onto C„. 

We now define Aj^F for symmetric functions not necessarily vanishing on Si by a formula 
analogous to (|2.8|) , except that we drop the product of delta functions: 

^ ra+l 

A+F{xi, ... , Xn+i) = 2 E E "^(^J ~ ~ ei)VlF{xi, ... , (3.1) 

i=i j^i 

Notice that (A^F) = if (F) < oo and hence the counting measure is invariant. The discrete 
Laplacian is given by 

n 

AF{xi, • • • , = E E E [-^(^1' + ^^a, ■■■ ,Xn) - F{xi, ... ,Xi,,... , Xn)] (3.2) 

i=l o-=± a=l,2 

and we define 

L = A + A. (3.3) 

Let L„ = TTnLiTn be the restriction of L to Cn. 

Throughout the rest of the paper we will use C(A) to denote a function of A > which 
has the property that for some C < oo, and for sufficiently small A, 

CW.{^l-^l' -J- ,3.4) 

In the next section we will prove the following lemma. 
Lemma 3.1 There exists a C(A) as in ( ^.41 ) such that 

^ ;{{w, (A - Ln)-'w)) < {{w, (A - Cn)-'w)) < C{\)nmw, (A - Ln)-'w)). (3.5) 



C(A)n2 

The special case n = 3 combined with ( |2.15| ) and ( p.l4| ) tells us that 



{{w,\X-A + Al{X-A)-^A+] ^w)) <C{X){{w,{X- C)-^w)). (3.6) 



9 



We define the Fourier transform of F : Z ^ M by 

for p„ G (]R'^/27rZ'^)"'. The Fourier transform of the discrete Laplacian acting on F is given 
by 

n d 

^(Pn) = - E E [^"'"'^ - 2 + e"*'*^"^] HpI. • • • ,Pn) = ^(Pn)^(Pn) 
j=l fc=l 

where a;(p„) = YTj=i^i'Pj)- In = 2 we will denote the d components of p by (r, s): 
= ('^i, • • • , fn) and s„ = (si, . . . , s„). In (i = 1, c<j(]9) = — [e*^ — 2 + e~*^] and in (i = 2, 
a;(p) = — [e*'' — 2 + e"*''] — [e*'' — 2 + e"*'']. Note that in both cases uj is real valued and 
nonnegative. If F is a symmetric function of two integer variables we have 

where a runs over permutations of degree three. By definition, we have 

((F,G)) = ^ j dp,dp2Fip,,p2)Gip,,p2)e'^^'+P-^' = J dpFip, -p)G{p, -p) 

In other words, when considering the inner product ((■,■)), we can consider the class of 
■ ■ ■ ,pn) defined only on the subspace J2jPj = mod lnU^ . 

Lemma 3.2 Suppose F G M.2- There exists a C < oo such that 
d=l : {{A+F,{\-A)-^A+F))<C f iu{u)[\ + uj{u)]-^^^ \F{u, -u)\^du; (3.7) 

d = 2 : {{A+F,{X-A)-^A+F))<C f u {e^ ■ u) \log{X + u;{u))\\F{u, -u)\^du. 

(3.8) 

Proof: Using the Schwarz inequality to bound the cross terms by the diagonal terms, we 
can bound A^F by 



{{A^F,{X-A)-'A^F))= I 

J V 



|2 



\A+F{pi,p2,p^)[ 



Pi+P2+P3=0 ^ + ^(Pl) + ^{P2) + ^(Ps) 



|2 



< C / ^— , , ' 7'"^^' (3.9) 

where ciS* is the element of surface area on the hyperplane {pi +P2+p-i = 0}. Let F denote 
the region in which at least one of the integration variables, rj or Sj in = 2 or in c? = 1, 
is bounded away from ivr and 0, let us say by 1/8. On F the denominator, 

A + uj{pi) + uj{p2) + uj{p3) >C-^>0 



independent of A. Hence the integration over F in ( p.9[ ) is uniformly bounded in A. We are 
only concerned with terms diverging as A | and hence we can restrict our attention to the 
integration over F*-^. 



10 



We need to divide F*-^ according to whether pi,p3 in c? = 1 and ri,r3, si, S3 in = 2 are 
within 1/8 of — vr = vr or 0. There are four regions in d = 1 and sixteen in d = 2. We have to 
compute each one and add them up. But in fact they are all analogous, and give the same 
result. So for simplicity we only present the region where they are all in [—1/8, 1/8]. 

We call u = pi + p3,v = pi — p^. The integration over corresponding region in (|3^ ) is 
bounded by a constant multiple of 



/ 

J\u\ 



uj(u)\F(u, -u)]"^ \ I . 77^-^ p5 —r^T^du 

d<l/8 ^+ k + ^l + k-^l + , 



|m|<1/8 Wb|<l/8 

in c? = 1 and 

uj{ei ■ u)\F{u, -m) 1^ j / — ■ — 12 , I p 

«.ei|>-e2|<l/8 U|,;-ei|>-e2|<l/8 X + \u + + \u - + \u\ 

in d = 2. Estimating the inside integration, in brackets, in c? = 1, 



/ 

Jm 



— < cix + u^r''\ 



'm<1/8 X + {u + vY+{u- vY + m2 

In = 2, 

/ ^ ,,1] ^^^<C|log(A + n2)|, 

which completes the proof of the lemma. □ 

Lemma 3.3 There exists C < 00 such that 

{fi-l\-l/4 r/ = 1 • 

C-|logA|V2 ^^2! ^'-'^^ 

Proof: d = 1: The Fourier transform of the current w is 

w{p,,p2) = e-'^y2 

From the previous lemma the left hand side of (|3.1CI|) is bounded below by 

We can restrict the integration to the region ^ G [—1/8, 1/8] and replace uj{x) by to have 
a further lower bound, 

'1/8 



This proves Lemma 1.1 for d = 1. 

d = 2: The Fourier transform of the current w is 

w{pi,P2) = e~"^l2 



11 



From the previous lemma the left hand side of (|3.1CI| ) is bounded below by 

^ ' J . J . A + uiv) + uiO + ^(01 log(A + ^(r^) + ^(0)1 ^^'^^^ 

We can restrict to the region ^, r/ G [—1/8, 1/8] to have a further lower bound. In this region, 
we can replace uj{x) by up to a constant factor and use | log(A + + ^^)| < | log(A + r^^)! 
to obtain a further lower bound 

c- r r (3 

J-i/s .l-m >' + ri^+e+e\ log(A + 1,2)1 * ■ ' 

Letting ^ = ^(1 + | log(A + rf)\y^'^ the integration is bounded below by 



J-l/H J -1/8 ^ + V + Z 



Changing to polar coordinates r,9 and restricting to vr/S < ^ < 27r/3 we get another lower 
bound, 

/ TxVlMA + r^)r^/^>C-MlogA 



l/2_ 

□ 



Lemma |TT^ follows immediately from Lemma |3.3| , ( |2.15| ) and Lemma |3.1| in d = 2. How- 
ever m. d = 1 this only gives the slightly weaker result ((w(A - C)-^)) > C-^X~^/^ /\\og\\. 
The log A is purely from the removal of the hard core, and we now show that the correct 
degree 3 lower bound in c? = 1 is of order A"^/^. 

Proof of Lemma IJ. in d = 1: We will choose a test function / G 7Vl2 in ( p.l5 ) to 
obtain a lower bound. Because of the summation over shifts in the definition ( [L.9| ) of ((■,■)) 
there is really a reduction in dimension from {{xi,X2) G | X2 > to Z+. Hence we 
define for x G Z_|_ 

l{x) = Y,f{y,y+l + x). 



Then we have 



2{{w,f)) = f{Q), 
if,{-S)f)) = \Y.^{x + l)-l{xf 



2 

and if g{xi,X2) = + 1 + Xi,y + 2 + X2 



A+fix,, X2) = 6{x,)[f{x2 - 1) - /(X2)] + 5(x2)[/(xi + 1) - /(xi)], 

and 

{{9,{-S)g))==\ Yl J2^9{{xuX2)+v)-g{x^,X2)r (3.14) 

xi,X26Z-|- V 

where v runs over ±(1,0), ±(0, 1) and ±(—1, 1) with addition intepreted as with reflecting 
boundary conditions at the origin. 



12 



We now make the choice 



We have 



and 



J] [fix + 1) - 7ix)r = x-'/'a - e-'"'') - \"\ 



A+f{xi,X2) = -5(xi)l(x2 > 0)A-^/"(l - e^''')e-^'''^-^ + 5{x2)\~'/\e-^'^' - l)e-^'^'"i 
~ X^/^[6{xi)l{x2 > 0)e-^'^'"^ - 6{x2)e'^'^''^']. 

and the final term {{A+f, (A — S)^^A+f)) can be written 

POO 

/ dte~^^ ^ pti{xi,X2),{yi,y2))A+f{xi,X2)A+f{yi,y2) 

^ xi,x2,yi,y2&'Z+ 

where pt are the transition probabihties of the random walk on with Dirichlet form given 
by (p.l4|) . Using the small A approximation for Af we get 



POO 

X dte-'' Yl e-^'''(^+^)[(l + l(x,2/>0))p,((0,x),(0,2/))-21(x>0)p,((0,a;),(y,0))]. 

(3.15) 

By images we can rewrite this as 

POO 

A / dte-'' e-"'''(™ (l + l(|a;|,|i/| >0)M(0,x),(0,2/)) 



x,yeZ 



21ix>0)pt{{0,x),{y,0)) 



where pt is now the transition density for a continuous time random walk on where the 
particle makes jumps at rate 1/2 of (1,0), (—1,0), (0, 1), (0, —1) and, in the first and third 
quadrants (1, —1) and (—1, 1), and in the second and fourth quadrants (1, 1) and (—1, —1). 
On the axes themselves the rules are changed a bit. For example, on the positive a;- axis the 
particle jumps at rate 1/4 of (0, 1), (0, —1), (—1, 1) and (—1, —1). On the other axes these 
rules are just naturally rotated. The diffusion approximation is 

POO POO POO 

2A / dte-'' / dx / rfye-^'''(l^l+l^l)b,((0,x),(0,i/))-p,((0,x),(y,0))] 



^0 J —oo J —CO 

where pt is now the transition density for a diffusion in with generator A + d'^ where 
d = dy — dx the first and third quadrants and d = dy + va. the second and fourth 
quadrants. The corresponding Dirichlet form is comparable to the standard one and therefore 
we can bound the transition probabilities above and below by those of Brownian motion (see 
^). By change of variables 

POO POO POO o~^~A — 

\ dte" dx dye-''''^\^^y\^^-^ = C\-'/\ 

Jo J-oo J-oo 47rt 

13 



The error 



oo 



A/ dte'' I dx I rfy[e-^'^'(™ - e-^^''(l^-+^l)^ 

Jo J-oo J-oo 47rt 

We can choose a0 for our test function instead of 0. In summary we have 

{{w, (A - C-')w)) > 2a\-'/^ - Co?\-^l^ > C'\-^'^ 
if a is chosen sufficiently smalL □ 



4 Removal of hard core 

In this section we prove Lemma 3.1. Recall the main statement is that ((w, (A — Cn)^^w)) 
can be bounded above and below in terms of {{w, (A — Ln)^^w)) at the expense of constants 
depending on n. Since we only use the bound for n = 3 in this article, the precise dependence 
of the constants on n is not important and in many places is probably not optimal. 

For / a symmetric function on Z"', we denote by pi,i = 1,2,3 the one, two and three 
point functions : 

Pi{x) = ^ Yl [/(^'^i'""" '^"-i)]^ (4-1) 
P2{x,y) = / _^2)!2! ^ [/(x, Xi, ■ ■ ■ , x„_2)]^ (4.2) 

XI,--- ,X„_2 

P3{x,y,z) = / _^3N|3| Yl '^^-3)]^ (4-3) 

^ ' XI,--- ,Xn-3 

for all X 7^ y 7^ z in Z*^. The following is Lemma 4.8 of [^. 

Lemma 4.1 Suppose f is a symmetric function on IT' . Then we have in all dimensions 

(pf ,(-53)pf )<n2(/,(-5)/) 

Proof: By symmetry, 

cr=±, e 

where e are summed over the standard basis of unit vectors in R", and 

{f,{-<S)f)=n ^ [f{x + ae,y,z,xi,...,Xn-3)-f{x,y,z,xi,...,Xn-3)]'^- 

cr=±, e x,y,z,xi,... ,x„-3 

By Schwarz's inequality, {{^j a|)^/^ — 6^)^/^)^ < '^j{aj — bj)"^, we have 

f 1/2/ \ 1/2/ / y,2 

~ (rzT^SyiS! ^ [fix,y,z,xi,...,Xn-3)-f{x',y,z,xi,...,Xn-3)]'^, 



14 



and hence 



x,y,z 



x,y,z,xi,... ,x„-3 



□ 



The following Lemma is a simple extension of Theorem 4.7 of P| or Lemma 3.7 of |T2 
to dimensions d = 1,2. Recall the definition ( p.lOj ) of Si. 



Lemma 4.2 Fix R> 0. There exist C{X) as in ( |3.4| ) such that for f a symmetric function 
on Z" vanishing on , 

J2 {li\x.-x,\^\x,-x,\<R} < C{Xy{f, (A - S)f). 

Proof: By definition of the three point function, the left side is a constant times 

1/2 

By the previous Lemma, we can bound the Dirichlet form of g = p-^ by that of /. Thus we 
only have to prove that 

(I{|xi-x3|+|x3-x2|<ij}/) < C{X){g, (A - S)g) 

for functions g of degree three. 

Recall that for configuration with three particles we have £i = {X3 := (xi,X2,X3) : x-i 7^ 
Xj, for i 7^ j}. We have 5'(x3) = whenever X3 ^ Si and the operator S is the discrete 
Laplacian on Si with Neumann boundary conditions. Define G'(x3) = 5'(x3) if X3 G Si and 
G'(x3) = Avy3g£:^_|y3„x3|<2 ^'(ys) for X3 ^ Si. We claim that for G so defined, 

{Gi\-A)G)<C{g{X-S)g). (4.4) 

Consider ([G(x3) — G{xi + ei,X2,X3)]^) with (xi + ei,X2,X2,) G Si and X3 ^ Si. In this case 
that G'(x3) is the average of gi^ys) with |y3 — X3I < 2 and ys G Si. We can check that 
ys and (xi + ei,X2,X3) G Si can be connected via nearest-neighbor bonds in Si. Thus we 
have ([G'(x3) — G{xi + ei, X2, xs)]^) < C {g{S)g) . A similar inequality can be checked if 
{xi + ei,X2, X3) ^ Si. and this proves ( |4.4|) . 

Since 5'(x3) = G(x3) on Si and otherwise, it is clear that {l{\x-^-x3\+\x3-x2\<R}9'^) ^ 
(l{|xi-x3|+|x'3-x'2|<-R}G'^)- Thus to prove the lemma it will suffice to prove that 

{^{\xi-x3\+\x3-x2\<R}G^) < CW{G, - ^)G) 

We can drop the part of A in the x^ direction, making the right hand side smaller. Hence it 
is enough to prove that 

(1{|,,I+|.,I<^j}G2)<C(A)(G, (A-A)G) 



15 



for functions G{xi,X2). Call V = l{|a;i|+|a:2|<-R}- It is local and bounded. We are in Z^'^ 
and we want to show that there is a C(A) such that — C(A)(A — A) > as operators, or, 
equivalently 

V'/^iX- Ay'V'/^ < C(A). 

Let Gx{x,y) be the kernel of (A — A)^^ and ip : 1?'^ M. Since V is bounded by 1, 
E.,y&'^^y'^'i^)Gx{x,y)V'/\yMxMy) is easily bounded by Gx{0,0)E.v\x). If d = 1, 
we are in and C(A) = Gx{0, 0) < G\ log A|. In = 2, we are in the transient case Z^ and 
G'a(0,0) is bounded uniformly in A (see [Q). □ 

We now divide the complement of £i into two sets. We call a site x an isolated double 
site if 

Xi = Xj = X, \xk — x| > 5 for all k ^ 

We use here the lattice distance for x,y E Z'^: \x — y\ = Yl'j=i k"* ^V^- Denote by S2 the set 
with at most isolated double sites and S3 = [Z'^]" — {Si U S2) the rest. For a configuration 
{xi,Xi, ■ ■ ■ , Xk, Xk, X2k+i, X2k+2 ■ ■ ■ 5 Xn) with k isolatcd double sites, we define F = T / by 

Xi, ... , Xfc, Xk, X2k+l,X2k+2 • • • ,Xn) 

= Av f{xi,yi,...,Xk,yk,X2k+i,---,Xn) (4.5) 

yk-\^i~yi\=^ ioT all i 

If X G S3, then -F(x) = 0, e.g., F{x, x, x + ei, X4, . . . , Xn) = 0. We also define the restriction 
RF by RF(x„) = -F(x„) if x„ G Si and RF(x) = otherwise. Note that T and R are 
not inverse to each other although RT is the identity. 

Lemma 4.3 There is a G{X) as in ( |3.4]) such that for any symmetric function f on Z" 
vanishing on S^ and F = T f 

{l/G{X))n-'{F, (A - A)F) < (/, (A - S)f) < G{F, (A - A)F) (4.6) 

Define F = T HF where R is the restriction. Then in addition, 

{F, (A - A)F) > G{X)n-'{F, (A - A)F) (4.7) 

Proof: By definition, {F,—AF) is given by | y|x-y|=i[-^(^) ~ -^(y)]^- The upper 
bound of ( [4.6|) is immediate, since S has Neumann boundary conditions, corresponding to 
dropping terms in the Dirichlet form with either x or y in Si . We now prove the lower 
bound in (|4.6| ). We decompose x and y into three sets, Si,S2,S3, so that 

(F, -AF) = 

a=l,2,3;/3=l,2,3 

where 

If both X and y satisfy the hard core condition, then the contribution is 

^1,1 = 1 E [/W-/(y)f <(/,-5/). 

x,y:lx-y|=l 

16 



We can estimate terms where either x or y is in 8^ by, 

1*1,3 + '^'2,3 + '^'3,31 ^ {^{\x,-Xk\ + \xk-x,\<R} F^) ■ 

From the definition of F, we can check that 

{'^{\xi-Xk\ + \^k-Xj]<R} F"^) < C ^ {l{\xi-Xk\ + \xk-Xj\<R} P) 

The last term is bounded by C{X)n'^{f, (A — S)f) from Lemma [4 .21 . Thus we have 

1*1,3 + *2,3 + *3,3| < C{Xy{f, (A - S)f) . 

We now bound $1^2- In this case we have, for example, x = {xi,Xi + ei,Xs, ■ ■ ■ ,a;„) and 
y = (xi, Xi, X3, ■ ■ ■ , Xn)- Notice that because x e ^1, y can in fact have at most one double 
site. By assumption of isolated double sites, we have | I > 5 for all j > 3. Thus 

^(y) = , AV f{xi,Z,X3, ■■■ ,Xn) 

\z—xi\=l 

Under the assumption \xj —Xi\ > 5 for all j > 3 we can always connect z to Xi. By Schwarz's 
inequality, we then have 

^ + ei,Z,X3, ■ ■ ■ ,X„) - f{xi,Z,X3, - ■ ■ ,Xn)\^ < C{f,{-S)f) 

Hence 

1*1,2! <C(/, (-5)/). 

Finally we bound $2,2- The typical case looks like x = (xi,Xi,X3,X3 + ei,X5, ■ ■ ■ and 
y = (xi, xi, X3, X3, X5, ■ ■ ■ , Xn). Then 

F(xi,Xi,X3,X3 + ei,X5 , Xi, X3, X3, X5, ■ ■ ■ , Xn) 

= Av Av [/(Xi,2;, X3,X3 + ei,X5, ■ ■ ■ ,X„) - /(xi,2;,X3,ti;,X5, ■ ■ ■ ,x„)] 
|2— 2:i|=l a;3|=l 

Using Jensen's inequality and the same arguments as in the estimate of $1^2 above we obtain 
1*2,2! — C{f, (—5)/). Putting all these estimates together, we have the lower bound of (|4.6|) . 

To prove call / = RF so that F = T f. From (gj) we have (/, (A - S)f) > 

C{X)n^'^{F, (A — A)F). Since S is an operator with Neumann boundary condition, for any 
F with RF = / we have (F, (A - A)F) > C{f, (A - S)f) and this proves Q. □ 

Lemma 4.4 Suppose that f,g & Cn- Let F,G = T f, T g. There is a C{X) as in (|3.4|) such 
that 

I {g, Af) - {G, AF) I < C{X) n^'^ {G, (A - A)G) (f, {-A)Fy/\ (4.8) 
Recall F = TRF. We have 

{AF, G) - {AF, G) 

where G is independent of X. 

17 



< G{Xy/^G, (A - A)Gy/^F, (-A)F)^/2 ^4_9) 



Proof: We first prove ( [4 .81 ). Note that it suffices to prove it with A replaced by A+, 
since it then follows for A\ and thus for A = A+ — A'^. Suppose first of all that / G A4n- 
Recall the definition of A+ in ( |2.8| ). It suffices to consider one term in the summation, say 
A+f = S{x2 - xi - ei) nfc^2(l ~ '^(^2 - Xfc))V+V and A'lf = 6{x2 - xi - ei)V+V- We 
divide Z*^""*"^-''^ into J\f = {\xj — xi\ > 5 for all j > 3}. and its complement. If x„_|_i G J\f, 
then {A^F — A^ f){'Xn-f-i) = so we can estimate 

\{g, lj^cA^^f)\^ < II f{xi + ei,X3--- ,Xn+i) - f{xi,x^-- ■ ,Xn+i) If IbWc^ll^ 
Clearly, Ij^c < Ylij>?,{\^j ^il — ^l- By Lemma |^, 

\\glj^cr<C{\)n^g,{\~S)g). (4.10) 
Replacing 1,2 by i, j and summing over all j, by the permutation symmetry of / we have 
|(^, VcA/)p < C{\)n^g, (A - S)g)n-\f, {-S)f) (4.11) 



A similar bound holds for F on A/"^. Combining these estimates and using Lemma [4.3| , we 
obtain (U) for A+. 

If / e C„, write / = (/i, . . . , /„, 0, 0, . . . ). From ( pi]) we have 



|((7(fc+i), < C7(A)nV2e(^(,+i), (A - S)g(^k+i)) + Cn'/'e'\f^k), {-S)fik)) 

Summing over k and optimizing e, we get ([4.8|) . Repeating the proof, i.e., noting that 
A^^F — A^F = on A/" and using Lemma 0|, to bound the term on A/"*", gives ([4.91). □ 



Lemma 4.5 T/iere is a C(A) as m ( P^ ) swc/i that for all f E Cn, if F = T f 

C-\AF, (A - A)-MF) - C2(A)r2(F, (A - A)F) 
< {Af,{X-S)-Uf) 

< Ci\)n^{AF, (A - A)-'AF) + C^{\)n^F, (A - A)F) . (4.12) 

Recall F = T HF. Then we have 

{AF, (A - A)-^AF) > C-\AF{X - A)-^AF) - C\X)n{F, (A - A)F) (4.13) 
Proof: Recall the variational formula 

{Af, (A - Sy'Af) = sup {2{g, Af) - {g, (A - S)g)} 

9 



By Lemmas and we can bound the right side from above by 
2{g,Af)-{g,{X-S)g) 

< 2{G, AF) - {l/CiX))n-^G, (A - A)^) + C(A)n^/='(G, (A - A)Gy/^F, (A - A)F)^/^ 

< C{X)n^{AF, (A - A)-^AF) + C'\X)n^F, (A - A)F) 

which give the upper bound in (|4.12| ). Alternatively we can bound the right side below by 
2{g,Af)-{g,{X-S)g) 

> 2{G, AF) - C(A)r2i/2(G, (A - A)G)^/^F, (A - A)^)^/^ - G{G, (A - A)^) 

> 2{G, AF) - G{G, (A - A)^) - G\X)n{F, (A - A)F) 



18 



Optimizing over G, 

2(^7, Af) - {g, (A - S)g) > C{AF, (A - A)-'AF) - C\\)n{F, (A - A)F) 
This proves the lower bound in ( [4.121) . By definition, 

{AF, (A - /\)-^AF) = sup {2{AF, G) - {G{X - A)G)} 



G 



(4.14) 



By Lemma [4.4] , the right hand side of ( [4.14| ) is bounded by 
2{AF,G) - (G(A- A)G) 

> 2{AF, G) - G{\)n^'^{G, (A - A)G)^/2(F, (A - A)FY'^ - {G, (A - A)G) 

> 2{AF, G) - G{G, (A - A)^) - G\X)n{F, (A - A)F). 

Taking the sup over G, we prove (|4.13| ). 



□ 



Our interest is in the inner product ((,)). It is easy to check that all previous Lemmas 
hold for the inner product ((, )) as well. Since extensions from (, ) to ((, )) were carried out in 
detail in f^, we shall only outline the basic procedures to prove these Lemmas for the inner 
product ((,)). 

We now prove the analogue of Lemma [4.3| for the inner product ((,)). For any local 
functions we rewrite the inner product as 



lim(2fc + l)-2(V r,.(7; V r,A) 

\x\<k \x\<k 



(4.15) 



Similarly, we have 



U {-s)h)) 



\x\<k 



(4.16) 



\x\<k 



Recall the definition of F = T / is linear in /. Furthermore, the intersection properties 
of ( independent of the translation, i.e. x G iff r^x G Si. Thus we have 

\x\<k |2;|<fc 



For each fixed k everything is still local so by Lemma |4.3| , 

{2k + l)-2( rxF, {\-A)Y ^^F) < C{\)n\2k + l)-2( r./, (A - 5) J] rj) 

\x\<k \x\<k \x\<k \x\<k 



Lettting k ^ oo limit and using that the limits exist on both side we obtain Lemma [4.3| for 
((■,■)). To prove lemmas ^.4| and for ((■,■)), we only have to use that A commutes with 
translations. 

Proof of Lemma 3.1 Upper Bound: We start with the variational formula (see (p.l6|) ), 
{{w, (A - Cnr'w)) = sup {2{{w, /)) - (((A - £„)/, (A - S)-\X - £„)/))} (4-17) 



19 



where C„ denotes the functions of degree less than or equal to n. By definition, 

(((A - (A - S)-\X - £„)/)) = ((/, (A - S)f)) + {{Af, (A - SyUf)) 



Here we have set = on since we are considering only £„. By Lemmas [4.3| and ^ 



((/, (A - S)f)) + {{Af, (A - Sr'Af)) >(n-VC(A))((F, (A - A)F)) 

+ n-^C{X){{AF,{X - Ay'AF)). 

Since {{f,w)) and {{F,w)) vanish except for functions of degree 2, it is easy to check that 
((/, w)) = {{F, w)) . Thus we have 

{{w,{\-C^)-'w)) 

< sup {2((F, w)) - il/CiX))n-'{{AF, (A - A)-'AF)) - C{\)n-^{F, (A - A)F))} 

F=Tf 

<C{\)n'{{w,{\-Lr,)-^w)). 
Lower Bound: We have 

{{w, (A - L^r'w)) = sup {2((F, w)) - {{AF, (A - A)-'AF)) - {{F, (A - A)F)) } 

F 

Since ((F, w)) = {{F, w)), from Lemma we have 

{{w, (A - L^r'w)) < sup |2((F,^)) - C-'{{F, (A - A)F)) - ^P^, (A - A)-MF))| . 

By Lemma Q ((F, (A - A)F)) > C(A)n-2((F, (A - A)F)). Recall RF = /, F = T /. Thus 
the previous line is bounded above by 

^sup^ |2(( T /, w)) - T /, (A - A) T /)) - ^^n'' {{A T /, (A - A)-^ T /)) | 

Clearly, {{T f,w)) = {{f,w)). By Lemma g^, we have (( T /, (A - A) T /)) > (A -5)/)). 
By Lemma [4.5| , we have 



((^, (A - L^r'w)) < 2{{f,w)) - C-'n-'{{f, (A - S)f)) - il/C{\))n{{Af, (A - Sy'Af)) 
< C{X)n^{w,{X- CNy^w)) 

This proves the Lemma. □ 

Acknowledgement: H.T.Yau would like to thank P. Deift, J. Baik and H. Spohn for 
explaining their results. In particular, Spohn has pointed out the relation (LI) so that 
the connection between the current across the zero and the diffusion coefficient becomes 
transparent. H.T.Yau would also like to thank A. Sznitman for his hospitality and invitation 
to lecture on this subject at ETH. 



20 



References 

[1] J. Baik, P.A. Deift, K. Johansson, On the distribution of the length of the longest in- 
creasing subsequence in a random permutation. J. Amer. Math. Soc. 12, 1119-1178 
(1999) 

[2] J. Baik, E.M. Rains: Limiting distributions for a polynuclear growth model with external 
sources. J. Stat. Phys. 100, 523-542 (2000) 

[3] H. van Beijeren, R. Kutner, H. Spohn: Excess noise for driven diffusive systems. Phys. 
Rev. Lett. 54, 2026-2029 (1985) 

[4] E.B. Davies, Heat Kernels and Spectral Theory, Cambridge U. Press, 1989. 

[5] K. Johansson: Shape fluctuations and random matrices. Comm. Math. Phys. 209, 437- 
476 (2000) 

[6] M. Kardar, G. Parisi, Y.C.Zhang, Dynamical scaling of growing interfaces. Phys. Rev. 
Lett. 56 , 889-892. 

[7] Landim, C; 011a, S.; Yau, H. T: Some properties of the diffusion coefficient for asym- 
metric simple exclusion processes. Ann. Probab. 24 (1996), no. 4, 1779-1808. 

[8] Landim, C; Yau, H. T. Fluctuation- dissipation equation of asymmetric simple exclusion 
processes. Probab. Theory Related Fields 108 (1997), no. 3, 321-356. 

[9] T. Liggett, Interacting Particle Systems, Springer- Verlag, New York, 1985. 

[10] M. Prahofer, H. Spohn: Current Fluctuations for the Totally Asymmetric Simple Ex- 
clusion Process, preprint |cond-mat/0101200| v2. 

[11] M. Prahofer, H. Spohn, Universal distribution for growth processes in 1+1 dimensions 
and random matrices. Phys. Rev. Lett. 84, 4882-4885 (2000) 

[12] Sethuraman, S.; Varadhan, S. R. S.; Yau, H.T. : Diffusive limit of a tagged particle 
in asymmetric simple exclusion processes. Comm. Pure Appl. Math. 53 (2000), no. 8, 
972-1006. 

[13] F. Spitzer, Principles of Random Walks, Springer- Verlag, New York, 1976. 

[14] Spohn, H: "Large Scale Dynamics of Interacting Particles" , Springer- Verlag, New York, 
1991. 



21