From the sineGordon field theory to the KardarParisiZhang growth equation

Pasquale Calabrese1, Marton Kormos2 and Pierre Le Doussal3

1
Dipartimento di Fisica dellUniversita di Pisa and INFN, 56127 Pisa, Italy
2
2 MTA-BME Momentum Statistical Field Theory Research Group, 1111 Budapest, Budafoki ut 8, Hungary and
3
CNRS-Laboratoire de Physique Theorique de lEcole Normale Superieure, 24 rue Lhomond,75231 Cedex 05, Paris France.

We unveil a remarkable connection between the sineGordon quantum field theory and the
arXiv:1405.2582v1 [cond-mat.stat-mech] 11 May 2014

KardarParisiZhang (KPZ) growth equation. We find that the non-relativistic limit of the two
point correlation function of the sineGordon theory is related to the generating function of the
height distribution of the KPZ field with droplet initial conditions, i.e. the directed polymer free
energy with two endpoints fixed. As shown recently, the latter can be expressed as a Fredholm de-
terminant which in the large time separation limit converges to the GUE TracyWidom cumulative
distribution. Possible applications and extensions are discussed.

The sineGordon (sG) model and its cousin, the sinh Some of these predictions have been tested experimen-
Gordon (shG) model, are paradigmatic integrable quan- tally [25].
tum field theories with countless applications in con- The aim of this paper is to unveil a connection between
densed matter physics (see e.g. [1] as a review). A lot is the KPZ or DP models and the NRL of the sG quan-
known in particular about their spectra in terms of many- tum field theory which arises because, as we mentioned
particle scattering states and about matrix elements of above, in the NRL the sG model reproduces the attrac-
local operators, the so-called form factors [2, 3]. This tive LL model. While m, characterizing the breathers, is
allowed very striking predictions for experimentally rele- a bounded integer for sG, it becomes unbounded in the
vant systems (such as spin chains and ladders [1]) which NRL and reproduces the bound states of the LL model,
in the scaling limit are described by these massive field the so-called string states. Working out the details, we
theories. While the shG model contains only a single show that the connection is remarkably simple and, in
type of particle of mass M , the excitation spectrum of particular, we find that the two point correlation func-
the sG model exhibits solitons as well as breathers Bm tion of exponential fields (vertex operators) in the sG
that can be viewed as bound states of m particles or, al- model encodes the information of the PDF of the height
ternatively, as soliton-antisoliton bound states. The two field in KPZ.
models are also related by an analytic continuation of the KPZ equation and LiebLiniger model: Let us recall
coupling constant. some facts about the KPZ equation in one space dimen-
As a recent experimentally relevant application, the sion. It describes the stochastic growth of an interface of
non-relativistic limit (NRL) of the shG model was con- height h(x, t), as a function of time as
sidered in Ref. [4], where it was shown that in a double
scaling limit (i.e. taking a NRL while taking the shG cou- t h = 2 h + (h)2 + (1)
pling to zero) the LiebLiniger (LL) model [5] is recov-
ered with repulsive interactions. This procedure allowed in dimensionless units, in the presence of white noise
the analytic calculation of some previously unknown local (x, t)(x , t ) = 2c(x x )(t t ). Here we focus on
expectation values [4, 6]. Taking the same double scal- the so-called droplet, or sharp-wedge, initial condition,
ing limit of the sG model instead gives the LiebLiniger where h(x, t = 0) = w|x| with w +. The height
(LL) model [5] with attractive interactions, and indeed field h(x, t) can be written as h = ln Z where Z is the
it is known that the scattering phases of the two mod- partition sum of a fixed endpoint DP, i.e. paths in the
els do coincide [7]. However, this procedure has not yet x, t plane with endpoints fixed at (0, 0) and (x, t), di-
been explored to obtain expectation values of measurable rected along the time direction. In the KPZ context one
observables. introduces the generating function
In an a priori completely different context, there has
been much progress in finding exact solutions to the 1D
X (u)n n
g(u) = Z , (2)
KardarParisiZhang (KPZ) equation [822]. Some of n=0
n!
these approaches used the mapping onto the directed
polymer (DP) model, and from there, using the replica which is a series representation for the Laplace transform
method, onto the LL model of bosons with attractive euZ of the PDF of Z. In the large time limit the (one-
interactions [23]. These bosons form bound states called point) fluctuations of h(x, t) are expected to grow as t1/3 .
strings in the large system size limit [24], and summation To study this limit it is convenient to write h(x =
over these string states has allowed for the calculation of 0, t) vt + 22/3 (t) where = (c2 t/4)1/3 and (t) is
the probability distribution function (PDF) of the KPZ an O(1) random variable. We also define u = es , since
height field for various initial conditions [9, 10, 13, 15, 16]. then g(s) = g(u = es ) Prob((t) < s) for ,
 2

hence the generating function g(s) is directly the cumu- particles. We will need below the dispersion relation of
lative distribution function (CDF) of the height. The the m-breather. Its total energy and momentum as a
important finding of Refs. [9, 10] is that g(s) can be function of its center of mass rapidity can be obtained
expressed as a Fredholm determinant at all times (see by introducing a = 2a1m
2 i and writing
below) and that it converges for large time to F2 (s), the
m
GUE Tracy Widom CDF [26]. These results were ob- X
Em () = M c2l cosh(a ) = Mm c2l cosh() , (9)
tained exploiting the property that the moments Z n can
a=1
be written as the diagonal propagator of an n-particle m
Bose gas with the Hamiltonian
X
Pm () = M cl sinh(a ) = Mm cl sinh() , (10)
n a=1
1 X X
HLL = x2 2c (x x ) , (3) leading to the same dispersion relation as for 1-breathers
2M =1 < but with a different mass Mm [28]
i.e. the attractive LL model which is integrable via the sin(m /2)
Bethe ansatz [27]. More precisely, Z n reads Mm = M . (11)
sin( /2)
X |h~x0 |n i|2
Z n = h~x0 |etHLL |~x0 i = etEn , (4) Note that the particle content m of the breather is
n
||n ||2 bounded in the relativistic sG model, namely mmax =
1/. Due to its integrability, the sG model sup-
where n are the n-boson eigenstates of HLL with ports diffractionless factorized scattering and the exact
eigenenergies En , ~x0 = (x0 . . . x0 ), so h~x0 |n i is the real 2-particle S-matrix is known [3, 29]. Based on this, ma-
space wave function evaluated at coinciding particle posi- trix elements of exponential vertex operators between
tions in x0 . Evaluating this sum over Bethe states in the scattering states, i.e. form factors have been computed
infinite system size limit is not trivial and was performed exactly [2, 30, 31].
in Refs. [9, 10]. Here we show that the same sum can Double non-relativistic limit and LL model: Here we
be retrieved rather simply, directly in the infinite system are interested in the (double) NRL defined as [4]
limit, from known results in the sG field theory, without
the need to manipulate Bethe states. Besides its interest cl + , 0 , cl = 4 c , (12)
for the LL physics, it also raises hope that other interest-
with c fixed and finite. Hence the renormalized coupling
ing quantities in KPZ could be calculated using the sG
tends to zero as
model.
SineGordon model: The sG model is a relativistically cl 2 2c
invariant integrable quantum field theory in (1 + 1) di- = 0. (13)
8 cl
mension defined by the Euclidean (imaginary time) La-
grangian density In this limit the dispersion relation of the particles be-
comes non-relativistic as
1 1 m20 c2l
LsG = ( )2
+ ( )2
(cos() 1) , (5) 2
2c2l
t
2
x
2 p M cl , E M c2l + , (14)
2M
where (x, t) is a real scalar field, cl is the speed of light, where is the usual (non-relativistic) rapidity, i.e. the
m0 the bare mass, the coupling constant, and we set quasi-momentum. Importantly, this double limit estab-
~ = 1. The renormalized (physical) coupling constant lishes a connection between the exact S-matrices and the
and mass are form factors of the two models. At the level of the fields
cl 2 the correspondence can be written in the double limit as
= , (6)
8 cl 2 1 h 2 2
i
(x, t) = (x, t)em0 cl t + (x, t)em0 cl t ,(15)
sin( ) 2m0
M2 = m20 . (7)

where is the sG field and creates a non-relativistic
The spectrum of the theory contains several kinds of par- particle (i.e. the LL boson). Plugging (15) into (5) leads,
ticles. The fundamental one is called 1-breather which upon expansion and neglecting highly oscillating terms,
has relativistic dispersion relation with energy and mo- to the non-linear Schrodinger Hamiltonian
mentum  
1
Z
E() = M c2l cosh() , p() = M cl sinh() , (8) HLL = dx c , (16)
2M
where is the rapidity. The spectrum also contains soli- which is the second quantized form of the attractive LL
tons as well as m-breathers which are bound states of Hamiltonian (3). This limit procedure was first shown
 3

for the shG model in Ref. [4], leading to the repulsive LL obtained through the NRL from the breather form fac-
model with interaction parameter c = c > 0. Here the tors of the sG model, paralleling the calculation for the
same method shows that c > 0 emerges as the coupling shG model [4, 33]. In the remainder of the paper we will
constant of the attractive LL model. In fact, at the level set M = 1/2 by a choice of units, as customary in the
of the single particle states a lot can be deduced by ana- LL model.
lytical continuation from shG to sG ( i). This tech- From sG correlations to the KPZ/DP model: Let us
nique was used to study a highly excited gas-like state of consider the two-point correlator of the exponential field
the attractive LL model, the super Tonks-Girardeau gas in the Euclidean sG model (i.e. in imaginary time)
[32]. However, the many-particle states in general are
quite different in the repulsive (c < 0) and attractive
G(k, t) = h0|eik(0,t) eik(0,0) |0i , (18)
(c > 0) cases. Consider the energy of the sG m-breather
(9-11) in the double NRL:
as well as the reduced correlation [38]
c2 p2
Em () M mc2l + (m m3 ) + m , (17)
24M 2M G(k, t) = G(k, t)/|heik i|2 , (19)
while the momentum is Pm () = mp, where we have
scaled = p/(M cl ) and neglected terms O(1/cl ). Apart which is defined to equal unity at t . Here and
from the rest energy, these are exactly the total energies below we denote the vacuum expectation value of the
and momenta of the m-string states of the LL model in exponential field, as heik i = h0|eik(0,0) |0i.
the limit of infinite system size [24]. The correspondence The Lehmann formula [13] expresses quite generally
goes further and indeed also the scattering phases coin- such a ground state expectation value (here in the vac-
cide, as pointed out in [7]. Finally, in the NRL the mass uum) in terms of the form factors of the excitations of
of the solitons/antisolitons Ms = 8m20 / 2 [3] diverges, the theory. At this stage we do not yet consider the NRL
hence they disappear from the spectrum and we can ne- but, for simplicity, we ignore the solitons states (which
glect them. We are then left with an infinite number will be justified only in that limit). It thus takes the
of breather modes as 0 corresponding to the LL form of a sum over states with arbitrary number ns of
strings. The form factors of the LL model can also be breathers

ns mX
max Z Pns
1 Y d1 dns
|h0|eik(0,0) |Bm1 (1 ) . . . Bmns (ns )i|2 e j=1 Emj (j )|t| ,
X
G(k, t) ... (20)
n !
n =0 s j=1 m =1
2 2
s j

(n)
where each breather of type mj has rapidity j and par- for n 2, where j is the j-th elementary sym-
ticle content mj . metric polynomial of the variables {e1 , . . . , en }, and
The form factors of the breathers can be obtained from detM1 (k) = 1. The minimal form factor is given by
those of the particles B1 (). Hence let us start with the R t t
dt sinh( 2 ) sinh( 2 (1+))
4 sin2 ( t(i) )
particle states, i.e. we temporarily restrict to mj = 1 Fmin () = N e 0 t sinh(t) cosh( t )
2 ,
2

in (20). Their form factors are known in sG [34] and (23)

R
can be obtained from the shG ones [35] via analytical 1 t
where N = Fmin (i) = e 0 dt sin(t) / cos 2 . The


continuation to imaginary coupling as
minimal form factor satisfies the exact relation [35]
sin(k ) sinh
Fnk () = h0|eik |1 . . . n i = heik i (2i)n Fmin (i + )Fmin () = . (24)
sin( ) sinh(i) + sinh
 sin( ) n/2 Y Fmin (j l ) We now carefully take the (double) non-relativistic
det Mn (k) , (21)
Fmin (i) ej + el limit of the form factors. To obtain a non-trivial limit
j<l
we need to let k while 0 with k = k fixed
where |1 . . . n i = |B1 (1 ) . . . B1 (n )i. Here we intro- and finite. We also scale the rapidities as j = 2pj /cl (we
duced the matrix (with j, l = 1, . . . n 1) recall M = 1/2). Given that N 1 as 0, using Eq.
(24) we have in the NRL
sin[(j l + k) ] (n)
[Mn (k)]j,l = 2jl , (22) Fmin () . (25)
sin( ) i +
 4

The most complicated term in the form factor (21) is where

det Mn which immensely simplifies in the NRL because
one can neglect the i j k in the matrix Mn to get
Y 4(pi pj )2 + c2 (mi mj )2
[p, m] = , (33)
4(pi pj )2 + c2 (mi + mj )2
h i  sin(k ) n1 1j<lns
(n)
det Mn (k/) det2jl , (26) Y ba 1
sin( ) F [m] = = . (34)
ba+1 m!
1a<bm
(n)
and det2jl 2n(n1)/2 in the NRL. Finally, in the
NRL we also have the simple relation Putting everything together we finally obtain
i
ns X
+ h
sin(k ) sin( 2c k) ik 2
X cnns Y 2 c 2mj
. (27) G(k, t) |he i| sin( k)
sin( ) ns ! j=1 m =1 c 2
n =0
s j

ns Z
All these relations lead for the most complicated parts of Y dpj 2
mj Mc2l t c12 (m3j mj )tmj p2j t
the form factor to the remarkably simple limit e [p, m] . (35)
j=1
2mj
sin(k ) n  sin( ) n/2 Y 1
2 det Mn (k)
Now we can compare this with the expression for the
sin( ) Fmin (i) e + el
j
moments of the partition sum in the KPZ/DP problem.
j<l
 c n/2 h 2 i The calculation of the averaged moments (4) was per-
c n
l
sin( k) . (28) formed in Ref. [9] and found to take exactly the same
2 c 2
expression as above (compare with Eq. (9) of Ref. [9]).
Plugging this in (20) and taking into account the Jaco- Hence we find that the reduced correlation can be written
bian of the variable change from s to ps, we find for as
the mj = 1 contribution i
X 1h 2 c 2n 2
G(k, t) = sin( k) Z n enMcl t , (36)
i n! c 2
X 1h 2 c 2n 2 n=0
G(k, t) |heik i|2 sin( k) enMcl |t|
n=0
n! c 2 showing that there is a relation between the two-point
dp1 dpn Y (pj pl )2 P 2 correlation in sG and the moments in the KPZ/DP prob-
Z
... 2 2
e j pj |t| , (29) lem. Furthermore, the sG correlation takes the same
2 2 c + (pj pl )
j<l form as the KPZ/DP generating function, hence one can
which is a sum of positive contributions, as it should, also write
since it comes from the Lehmann formula.
G(k, t) = g(u) , (37)
Let us now generalize the derivation to arbitrary m-

breather states. Fusion relations relate the form factors c 2 Mc2l t
where u = [ 2c sin( 2 k)] e . Interestingly, by an-
of breathers to the ones of n particles Fn as follows. Let
us recall that the rapidities can be written as ja = j alytic continuation ik k one also finds
2a1mj P
2 i . Then, recalling n = j mj , we have [34]
h0|ek((0,t)(0,0)) |0i
=
h0|eik |Bm1 (1 ) . . . Bmns (ns )i = hek i2
i
ns X (1)n h 2 c 2n 2
Y a
mj Fn ({1a1 }a1 =1,...m1 , . . . {nns s }ans =1,...mns ) , (30) sinh( k) Z n enMcl t = g(u) , (38)
n=0
n! c 2
j=1
2
where the NRL of m given in [34] is where now u = [ 2c sinh( 2c k)]2 eMcl t is a positive num-
ber, thereby making the connection closer. Since the
m = ( )(m1)/2 [m] m . (31) KPZ generating function obeys 0 < g(u) < 1, it implies
that both sides of the above equation are now positive
Note that ja c [pj ic
2 (2a 1 mj )] which coincide numbers in the interval [0, 1], increasing with k. Note
with the string rapidities [24]. that although the operator in the left hand side of Eq.
In taking the NRL of the form factor Fn , only the (38) may not be formally defined in the sG field theory, in
term with Fmin changes compared to the previous case, the Euclidean version it takes the meaning of a canonical
leading to statistical mechanics average, with a discretization and
Y ns
Y regularization at small and large scale (as one would do
|Fmin (j l )|2 = [p, m] |F [mj ]|2 , (32) in a numerical simulation). While the two correlations
1j<l<n j=1 in Eq. (38) may be singular as the regularizations are
 5

removed, their ratio should be a well defined number in whose solution was obtained in [13] (ii) the imaginary-
the interval [0, 1]. time propagator for arbitrary positions [36, 37]. Here we
Having shown that the NRL of the sG correlation con- have shown that in the NRL the diagonal propagator can
tains information about the PDF of the KPZ field, we be retrieved as [39]
can ask whether a tighter physical connection exists. Let i
1h 2 c 2n
us recall the expression of the generating function as a G(k, t)| nM c2l t = sin( k) G(|~x0 i, |~x0 i; t) .
Fredholm determinant (FD) obtained in [9]: e n! c 2
(42)
g(s) = Det[1 + P0 Ks P0 ] , (39) It would be interesting to obtain more general overlaps
via this sG correspondence which we leave for future in-
where P0 is the projector on [0, +[ and the kernel can vestigations.
be written as Relation (42) can be understood in the following
n

way. The diagonal propagator, hn (t) (0)i can be ex-
dk eyik(vv )
Z
Ks (v, v ) = dyAi(y + k 2 + s+ v + v ) , tracted from the NRL of the sG correlator hn (t)n (0)i
2 1 + ey where the operator n can be obtained naively from the
(40) nth order of the series expansion of eik in powers of
where t1/3 was defined above. Let us also recall that k. However, it turns out that the n operator obtained
at large time this generating function g(s) converges to in this way has non-zero matrix elements between states
F2 (s), the GUE Tracy Widom CDF [26]. So it is interest- having any number of particles. These form factors sur-
ing that this FD contains the information about the pre- vive in the NRL meaning that in the limit one cannot
cise time dependence of the coefficients of each power of recover the n operator. It turns out that certain linear
2
eMcl t in the decay of the sG correlation function, these combinations of different powers of lead to n in the
coefficients being proportional to the KPZ/DP moments NRL. Quite interestingly, these combinations in terms of
Z n . Note however, that because of these fast decaying ex- k give just the sine factor in Eq. (42), so this formula,
ponentials, there is no point-wise convergence as cl quite miraculously, automatically takes care of this oper-
in (36) and (38) and, at this stage, there is no direct ator mixing.
correspondence between a sG observable and KPZ gen- Conclusion. In this paper we have shown yet another
erating function. An outstanding question is thus how method to calculate the PDF for the KPZ growth equa-
far can this correspondence be pushed and whether one tion with the narrow wedge initial condition exploiting
can construct sG observables with an even closer relation the NRL limit of sG field theory. The obtained result
to KPZ. fully agrees with previous derivations [912]. On the one
One last interesting point relates to the so-called mo- hand this result is a useful check of our new method and
ment problem in the continuum KPZ/DP problem, i.e. of present and previous results, on the other hand this
to the fact that the growth of the integer moments Z m as new correspondence raises the hope that it can be used
a function of m at fixed t is too rapid (i.e. ln Z m m3 t) to obtain yet unknown observables for the KPZ growth
to guarantee a unique solution for the PDF for h = ln Z. equation, as well as provide a new interesting regulariza-
While this is still an open question in the mathemat- tion of the problem.
ics community, it is usually circumvented by resorting to Acknowledgements: MK thanks Gabor Takacs for use-
discrete models (such as TASEP, see e.g. [11]) which do ful discussions. PC acknowledges the ERC for financial
reproduce continuum KPZ in some limit and do not suf- support under Starting Grant 279391 EDEQS. MK ac-
fer from the same problem. We point out here that the knowledges financial support from the Marie Curie IIF
relativistic sG theory can provide yet another interesting Grant PIIF-GA-2012- 330076. PLD acknowledges the
regularization of the moment problem because the num- hospitality of the Dipartimento di Fisica dellUniversita
ber of breather types 1/ is bounded until the NRL is di Pisa where much of this work was completed.
taken.
Overlaps and propagators in the LL model: One can
ask more precisely how can the sG theory retrieve
more detailed information about the attractive LL model
hereby helping solve KPZ problems. For example a re- [1] F. H. L. Essler and R.M. Konik, Applications of Mas-
lated fundamental quantity (both for the LL model and sive Integrable Quantum Field Theories to Problems
in Condensed Matter Physics, in Ian Kogan Memo-
the KPZ growth) is the overlap
rial Collection From Fields to Strings: Circumnavi-
gating Theoretical Physics, eds M. Shifman, A. Vain-
G(|1 i, |0 i; t) = h1 |etHLL |0 i (41) shtein and J. Wheater, World Scientific Singapore 2005
[cond-mat/0412421].
between more general initial |0 i and final states |1 i. [2] F. A. Smirnov, Form factors in completely integrable
Examples are (i) KPZ with flat initial condition, which models of quantum field theory (World Scientific, Sin-
requires the overlap with an initial uniform state and gapore, 1992).
 6

[3] G. Mussardo, Statistical Field Theory, An Introduction [21] P. Le Doussal, J. Stat. Mech. (2014) P04018.
to Exactly Solved Models in Statistical Physics (Oxford [22] P. Calabrese and P. Le Doussal, arXiv:1402.1278.
University Press, Oxford 2009). [23] M. Kardar, Nucl. Phys. B 290, 582 (1987).
[4] M. Kormos, G. Mussardo and A. Trombettoni, Phys. [24] J. B. McGuire, J. Math. Phys. 5, 622 (1964).
Rev. Lett. 103, 210404 (2009); M. Kormos, G. Mussardo [25] K. A. Takeuchi and M. Sano, Phys. Rev. Lett. 104,
and A. Trombettoni, Phys. Rev. A 81, 043606 (2010). 230601 (2010); K. A. Takeuchi, M. Sano, T. Sasamoto,
[5] E. H. Lieb and W. Liniger, Phys. Rev. 130, 1605 (1963). and H. Spohn, Sci. Rep. (Nature) 1, 34 (2011).
[6] M. Kormos, Y.-Z. Chou, and A. Imambekov, Phys. Rev. [26] C.A. Tracy and H. Widom, Comm. Math. Phys. 159, 151
Lett. 107, 230405 (2011). (1994).
[7] P. Calabrese and J.-S. Caux, Phys. Rev. Lett. 98, 150403 [27] V.E. Korepin, N.M. Bogoliubov, and A.G. Izergin, Quan-
(2007); P. Calabrese and J.-S. Caux, J. Stat. Mech. tum Inverse Scattering Method and Correlation Func-
P08032 (2007). tions (Cambridge University Press, Cambridge, 1993).
[8] M. Kardar, G. Parisi and Y.C. Zhang, Phys. Rev. Lett. [28] R. Dashen, B. Hasslacher, and A. Neveu, Phys. Rev. D
56, 889 (1986). 11, 3424 (1975).
[9] P. Calabrese, P. Le Doussal and A. Rosso, EPL 90, 20002 [29] A. B. Zamolodchikov, Al. B Zamolodchikov, Annals
(2010). Phys. 120, 253 (1979).
[10] V. Dotsenko, EPL 90, 20003 (2010); J. Stat. Mech. [30] M. Karowski and P. Weisz, Nucl. Phys. B 139, 455
P07010 (2010); V. Dotsenko and B. Klumov, J. Stat. (1978); H. M. Babujian, A. Fring, M. Karowski, and A.
Mech. (2010) P03022. Zapletal, Nucl. Phys. B 538, 535 (1999); H. Babujian
[11] T. Sasamoto and H. Spohn, Phys. Rev. Lett. 104, 230602 and M. Karowski, Nucl. Phys. B 620, 407 (2002).
(2010), Nucl. Phys. B 834, 523 (2010), J. Stat. Phys. [31] S. L. Lukyanov, Mod. Phys. Lett. A 12, 2543 (1997).
140, 209 (2010). [32] M. Kormos, G. Mussardo, and A. Trombettoni, Phys.
[12] G. Amir, I. Corwin, J. Quastel, Comm. Pure Appl. Math Rev. A 83, 013617 (2011).
64, 466 (2011); I. Corwin, arXiv:1106.1596. [33] M. Kormos, G. Mussardo, and B. Pozsgay, J. Stat. Mech.
[13] P. Calabrese and P. Le Doussal, Phys. Rev. Lett. 106, P05014 (2010).
250603 (2011); P. Le Doussal and P. Calabrese, J. Stat. [34] G. Takacs, Nucl. Phys. B 825, 466 (2010).
Mech. P06001 (2012). [35] A. Koubek and G. Mussardo, Phys. Lett. B 311 (1993)
[14] J. Quastel, J. Ortmann, and D. Remenik in preparation. 193.
[15] T. Imamura and T. Sasamoto, Phys. Rev. Lett. 108, [36] S. Prolhac and H. Spohn, J. Math. Phys. 52, 122106
190603 (2012); T. Imamura and T. Sasamoto, J. Phys. A (2011).
44, 385001 (2011). [37] Cr. A. Tracy and H. Widom, J. Phys. A 41, 485204
[16] T. Gueudre and P. Le Doussal, EPL 100, 26006 (2012). (2008).
[17] S. Prolhac and H. Spohn, Phys. Rev. E 84, 011119 [38] In our normalization indeed |heik i|2 = 1 and taking the
(2011). ratio (18) is superfluous. However, this ratio is indepen-
[18] T. Gueudre, P. Le Doussal, A. Rosso, A. Henry, P. Cal- dent of normalizations and conventions and the results
abrese, Phys. Rev. E 86, 041151 (2012). for this quantity are fully general.
[19] V. Dotsenko, J. Phys. A 46, 355001 (2013); J. Stat. Mech. [39] The case G(|~xi, |~ y i; t) with ~x = (x0 . . . x0 ) and ~
y =
(2013) P06017; J. Stat. Mech. (2013) P02012. (y0 . . . y0 ) is a trivial generalization of the present for-
[20] T Imamura, T. Sasamoto, and H. Spohn, J. Phys. A 46, mula, introducing the total momentum in Eq. (42), see
355002 (2013); S. Prolhac and H. Spohn, J. Stat. Mech. e.g. [13].
(2011) P01031.