Fock Space For Fermion-Like Lattices and PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Physica A 387 (2008) 5101–5109

Contents lists available at ScienceDirect

Physica A
journal homepage: www.elsevier.com/locate/physa

Fock space for fermion-like lattices and the linear Glauber model
Érica M. Silva a,c , Paulo T. Muzy b , Ademir E. Santana c,∗
a Núcleo de Física, Universidade Federal do Tocantins, 77020-210, Palmas, TO, Brazil
b Instituto de Física, Universidade de São Paulo, Cidade Universitária, Caixa Postal 66318, 05315-970, São Paulo, SP, Brazil
c Instituto de Física, Universidade de Brasília, 70910-900, Brasília, DF, Brazil

article info a b s t r a c t

Article history: The concept of Fock space representation is developed to deal with stochastic spin lattices
Received 31 December 2007 written in terms of fermion operators. A density operator is introduced in order to follow in
Received in revised form 15 March 2008 parallel the developments of the case of bosons in the literature. Some general conceptual
Available online 8 April 2008
quantities for spin lattices are then derived, including the notion of generating function
and path integral via Grassmann variables. The formalism is used to derive the Liouvillian
PACS:
of the d-dimensional Linear Glauber dynamics in the Fock-space representation. Then the
05.30.Fk
05.50.+q
time evolution equations for the magnetization and the two-point correlation function are
75.10.Pq derived in terms of the number operator.
© 2008 Elsevier B.V. All rights reserved.
Keywords:
Linear Glauber model
Fock space
Spin lattice

1. Introduction

The notion of Fock space, or number representation, was introduced in classical physics by Schönberg [1] to describe
classes of linear equations, such as the Liouville equation in phase space. Later, Doi used this procedure to deal with reaction-
diffusion processes [2], pointing a way to treat stochastic equations with the methods of quantum field theory. In this realm,
the creation and annihilation field operators describe, for instance, the reagents taking place in a chemical reaction.
This theory has been developed in several directions [3], including the work by Martin, Siggia, and Rose [4] deriving
a functional formalism for the nonequilibrium statistical mechanics. A version of path integrals for bosons was proposed
by Peliti [5], and it has been extended and applied to different systems [6,7]. In the trend of developments over the last
decades, approaches for stochastic models based on number representation seem, at first sight, disconnected from each
other, thus raising practical obstructions which have been addressed, for instance, by Grassberger and Scheunert [8] and
Andersen [9] (for a review, see Ref. [3] and references therein). One of these difficulties is that number representation has
been used in quantum theories because of the indistinguishability of subatomic particles, an aspect described by probability
amplitudes. For classical systems, however, probability amplitudes have been defined in association with a generalization
of the Liouville theorem [1,10–25]. The physical and mathematical nature of such a formalism have also been analyzed with
representations of Lie groups [26], where the Fock space is taken as a representative vector space of symmetry. In this sense
there is no “h̄” in the method, resulting in full consistency with classical physics and no ambiguity with quantum theory.
Many of these developments were carried out for bosons [27], leaving the case of fermion-like lattices to be studied in a
broader sense.

∗ Corresponding author. Tel.: +55 61 3307 2900; fax: +55 61 3307 2363.
E-mail address: asantana@fis.unb.br (A.E. Santana).

0378-4371/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.physa.2008.04.015
5102 É.M. Silva et al. / Physica A 387 (2008) 5101–5109

It is worth mentioning, however, that in the context of spin lattices, the number representation method was used by
Shultz, Mattis and Lieb to solve analytically the 2-dimensional Ising model [28]. Since then, in the same algebraic perspective,
several works have been presented [29], dealing, in particular, with adsortion processes [30–32] and the Glauber model [33–
35]. The fermion operator scheme has also been applied in a recent analysis of directed percolation [36] and in a study of
the Boltzmann equation [37]. From such achievements, one can realize that improvements in the antisymmetric structure
of Fock space to stochastic systems should be of interest, and this is one of our goals here. We follow the developments for
the case of bosons as discussed, for example, by Peliti [5], Grassberger and Scheunert [8] and Ali [38]. For that, we introduce
a density operator to describe the state of the system. Using Grassmann variables, conceptual quantities, as the propagator
written in terms of a path integral and generating function, are derived. We apply the method to analyze properties of the
d-dimensional linear Glauber model [39].
The linear Glauber model belongs to the universality class of the voter model, which has been used to study different kinds
of problems, such as critical coarsening without surface tension [40], the kinetics of catalytic reactions [41] and competing
learning [42]. Concerning linear Glauber dynamics, the validity of the fluctuation-dissipation theorem and the occurrence
of an aging regime have been investigated in a recent work [43]; it has also been analyzed as a model for surface catalytic
reaction [44]. In a general perspective, as a non-trivial solvable model, linear Glauber dynamics is useful to verify different
properties and characteristics of nonequilibrium systems [39,43,45–47].
In Fock space we obtain an expression for the Liouvillian of the linear Glauber model in d dimensions, and analyze the time
evolution equation for the magnetization and the two-point correlation function, both defined here via the number (not spin)
operator. The connection between our results and those in the literature [39] is discussed. One can see that the present for-
malism provides a simpler way to treat the linear Glauber model, with some properties useful for practical calculation based
on the rigorous ingredients of an algebraic approach. The main results are derived in a unified way with the case of bosons.
The presentation is organized in the following way. Section 2 addresses the problem of using the Fock space in stochastic
systems, paying attention to the antisymmetric case. Section 3 gives the Liouvillian for the linear Glauber model, and in
Section 4 the magnetization and the correlation function are studied. Section 5 presents our final concluding remarks.

2. Fermionic stochastic lattices

In this section a spin system is presented in a general sense via a Fock space representation. We consider a lattice in
which each site has one or zero particles. This kind of lattice is described by fermion-like operators. Indeed, let us consider
the configurations of a system being characterized by a set of occupation numbers, denoted by {nm , m = 1, . . . , N}, with
nm = 0 or 1, ∀m, describing the mth site empty (nm = 0) or occupied (nm = 1). The state of the system is given by
φn (t) = φ(n1 , n2 , . . . , nN ; t), representing the probability of occurrence of the configuration n = {n1 , n2 , . . . , nN }, and
fulfilling a master equation

∂t φn (t) = (wnm φm (t) − wmn φn (t)),


X
(1)
m

where wnm is the transition rate from state φm to φn .


We assume that the state of the lattice is described by a probability operator b φ, defined in the the Hilbert space (H )
and diagonal in the number representation, that is, we associate n → |ni with b φ|ni = φn |ni. We also consider that the
observables of the theory are Hermitian operators, such that for an arbitrary observable bA, we have bA|ni = An |ni, where An is
A in the basis |ni. We follow with a notation that can be used for bosons if we consider n = 0, 1, 2, . . . ;
a real eigenvalue of b
also, for sake of simplicity, initially we take a one-site lattice.
The normalization of the basis |ni and the completeness relation are, respectively, given by
X 1
hm|ni = n!δnm , 1= |nihn|. (2)
n n!

The trace of an operator b


A is
X 1
Tr b
A= A|ni,
hn|b (3)
n n!

φ is defined by hb
Aiφ = Tr(b
φb
A) = n An φn , corresponding to the usual average of
P
and the average of b
A in a state described by b
a physical quantity An with probability φn .
Introducing the vectors
X 1
|ni, |ni,
X
|I i = |Ii =
n n n!

we see that the probability operator φ̂ leads then to the vectors


X 1
|φi = bφ| I i = φ |ni, n
n n!
| φi = b
φ|Ii = φn |ni,
X

n
É.M. Silva et al. / Physica A 387 (2008) 5101–5109 5103

Table 1
Properties of basic states

hI|Ii = n,m hm|ni = n n!, hI|Ii = n n1! ,


P P P

hn|Ii = 1, hn|Ii = n!, h0|Ii = h0|Ii = 1,


hI|Ii = n,m n1! hm|ni = n δnn , hψ|φi = n n1! ψn φn
P P P

hψ|φi = n n!ψn φn , hψ|φi = n ψn φn = hψ|φi,


P P

φn = n1! hn|φi, φn = hn|φi.

which can properly be used to describe the state of the system. From these definitions we derive the set of properties
presented in Table 1.
Notice that the average of an arbitrary operator hb O|Ii is equivalent to the trace operation for diagonal operators.
Oi = hI|b
Indeed, for instance when b O =bAbφ, we have
φ|Ii = hI|b
A | φi = An φn = Tr bφ.
X
hI|b
Ab Ab
n

φ|Ii = hI|φi = n φn = 1. In particular


P
On the other hand, the normalization of the probability is expressed by hI|b
h0|φi = h0|φ|Ii = n φn h0|ni = h0|φi = φ0 , which is a useful result to fix the initial conditions.
b P

The creation and annihilation operators are given by the anti-commutation relations: {a, a} = {aĎ , aĎ } = 0 and {a, aĎ } = 1,
such that a|ni = n|n − 1i, aĎ |ni = |n + 1i, hn|a = hn + 1| and hn|aĎ = hn − 1|n (regarding the usual definition, observe a
different normalization factor). The number operator is b N = aĎ a, with b
N|ni = n|ni and n = 0, 1.
In terms of the operator aĎ , the vector |Ii can be written as
X 1
(aĎ )n |0i
X
|I i = n!|ni =
m m n!
Ď
= ea |0i.
Ď
This suggests the definition of the following state |αi = eαa |0i, where α is a Grassmann number in the case of fermions,
and in the case of bosons α is a complex c-number. Considering bosons first, the basic operators a and aĎ fulfil commutation
relations. It is simple to show that |αi|α=1 = |Ii, a|αi = α|αi and ∂kα |αi|α=1 = (aĎ )k |Ii.
Defining the generating function, G(φ, α) = hα|φi, the kth binomial (or factorial) momentum, nk (φ) [8,48], is given by

nk (φ) = ∂kα G(φ, α)|α=1 = hI|ak |φi = hn(n − 1) · · · (n − k + 1)i|φ .

Another way to express this result is to observe that nk (φ) = hn|ak |αi|α=1 . In particular, for the number operator b
N = aĎ a we
have,

n1 (φ) = hI|aĎ a|φi = hI|a|φi = hn|a|αi|α=1 .

Notice that for fermions n1 (φ) is the only non zero binomial momentum. However, to derive similar quantities for fermion-
like lattices, we have to introduce in the present context Grassmann variables [49].
A set of Grassmann variables will be denoted by G = {α, β, γ, . . .}, the fermion annihilation and creation operators are,
respectively, a and aĎ , while complex numbers will be denoted by {c}. Grassmann variables fulfil the following properties:
(i) αβ = −βα; (ii) α(β + cγ) = αβ + cαγ , αa = −aα, and αaĎ = −aĎ α. The complex conjugation is an anti-linear mapping
similar to the complex conjugation of complex numbers, such that

(α + caβaĎ γ ∗ )Ď = α∗ + γ aβ∗ aĎ c∗ ,
where α and α∗ are considered as independent quantities.
An arbitrary function of a Grassmann variable, f (α), is given in the form f (α) = f00 + f01 α, where f00 , f01 ∈ C . Integration
is defined by
Z Z
αdα = 1, dα = 0.

Then, for a function f (α), we have f (α)dα = f01 . This result shows that integration, a linear operation, is equivalent to a
R

derivation, in the sense that


∂f (α)
= f01 .
∂α
The scalar product involving two Grassmann functions f (α) and g(α) is defined by
Z
(f , g) = (f (α∗ ))∗ g(α∗ )e−α α dα∗ dα,

where (f (α∗ ))∗ = f00


∗ ∗
+ f01 α.
5104 É.M. Silva et al. / Physica A 387 (2008) 5101–5109

A basis is given by ψ0 = 1 and ψ1 = α∗ such that (ψi , ψj ) = δij . This basis provides a matrix representation for the
operators a and aĎ . To see that, notice that
d d
α∗ ψ0 = ψ1 ; α∗ ψ1 = 0; ψ0 = 0; ψ1 = ψ0 .
dα∗ dα∗
Then we have a = ∂/∂α∗ and aĎ = α∗ . As a consequence, the matrix elements, (ψi , aψj ) = aij and (ψi , aĎ ψj ) = aij are such
Ď

that
0 0 0 1
   
a= , aĎ = .
1 0 0 0

Consider a general operator A(aĎ , a), in the normal order, written as

A(a
Ď
, a) = A00 + A10 a + A01 a + A11 a a =
Ď Ď Ď n
Anm (a ) (a) ,
X
m

n,m=0

where Anm ∈ C . In the representation of Grassmann variables, we can define two functions associated with A. The first one
is the symbol,

A(α , α) = Anm (α ) (α) .


X
∗ ∗ n m

n,m

The other object is the kernel, defined by

A → A(α , α) = An,m (α ) (α) ,


X
∗ ∗ n m
(4)
n,m

where An,m = (ψn , Aψm ), such that

A(α∗ , α) = eα α A(α∗ , α).



(5)

The product of two operators A1 and A2 in the normal order is then given by
Z
(A1 A2 )(α∗ , α) = A1 (α∗ , β)A2 (β∗ , α)e−β β dβ∗ dβ,

where we have used βm β∗ n e−β β dβ∗ dβ = δnm .


R ∗

With this representation, let us treat a stochastic equation as given in (1), with formal solution given by |φ(t)i = U (t)|φ0 i,
where

U (t) = etL = lim [A(t)]N ,


N→∞

with A(t) = 1 + τN L, τN = t/N. Using Eqs. (4) and (5) we obtain


N h
Z Y −1
i NY
eβi−1 βi A(β∗i−1 βi ) e−βj βj dβ∗j dβj ,
∗ ∗
U (α∗ , α; t) = lim
N→∞
i=1 j=1

where A(β∗i−1 , βi )
=1+ τN L(β∗i−1 , βi ),
with β0 = α∗ and βN = α. ∗

Let us write A ∼ exp(τN L). Then we have


( )
Z N −1 N −1
U (α, α ; t) = lim [(βi−1 − βi )βi + τN L(βi−1 , βi )] + βN−1 βN + τN L(βN−1 , βN ) dβ∗j dβj .
X Y
∗ ∗ ∗ ∗ ∗
exp
N→∞
i−1 j=1

Using the standard notation for functionals


−1
Z NY Z
dβ∗j dβj . . . → D β∗ D β . . . ,
j=1

N −1 Z t
0 0
τk [. . . fk ] → dt [. . . f (t )]
X
lim
N→∞ 0
k=1

and βi → β(t ), βN = α ≡ β(t), β∗N−1 → β∗ (t), limN→∞ τN L(β∗N−1 , βN ) = 0, we obtain


0

Z Z t 
0 0 0 0 0
U (α, α∗ ; t) = D β∗ D β exp dt [β∗ (t )∂t0 β(t ) + L(β∗ (t ), β(t ))] + β∗ (t)α .
0

This propagator is the fermion counterpart of Peliti’s approach for bosons [5].
Ď
Now we are ready to define a generating function using Grassmann variables. We introduce the state |ᾱi = eαa |0i,
α a
which is a Glauber-like coherent state. Indeed, in the context of Grassmann variables, e works as a displacement operator:
É.M. Silva et al. / Physica A 387 (2008) 5101–5109 5105

Ď
eαa aeαa = a − α. However, we have to be careful to introduce the dual space. The unitary fermion displacement operator is
defined by [50]
Da (α) = exp(aĎ α − α∗ a), (6)
such that Da (α)aDĎa (α) = a − α, with the coherent fermion state given by |αi = Da (α)|0ia and a|αi = α|αi. We can
a (α). As a consequence, hα|βi =
then prove that the dual coherent state is given by hα| = h0|DĎa (α), with DĎa (α) = D−1

exp(α β − 2 α α − 2 β β) and hα|a (α) = hα|α . The projection of |αi on the number basis is
∗ 1 ∗ 1 ∗ Ď ∗

|αi = e−α α/2 (−α)n |ni


∗ X
(7)
n

and then hn|αi = exp(−α∗ α/2)(α)n . Taking α as real Grassmann variables, i.e. α∗ = α, in DĎ (α) we have
Ď −aα Ď
eαa |0i = (1 + αaĎ − aα)|0i = eαa |0i = |ᾱi
and
∂ ∂ Ď
|ᾱi = (1 + αaĎ )|0i = aĎ ea |0i ≡ aĎ |Īi.
∂α ∂α
Thus the generating function can be given in a similar way to that of the case of bosons, that is G(φ, α) = hα|φi.
Let us now review a N-site lattice, where the creating and the annihilating operators anti-commute in the same site but
commute for different sites. This fact results in operators named Paulions, defined by the following mixed rules [8],
{ci , ciĎ } = 1,
{ci , ci } = {ciĎ , ciĎ } = 0,
[ci , cj ] = [ciĎ , cjĎ ] = [ci , cjĎ ] = 0, i 6= j.

These operators can be used to represent the Pauli matrices, that is σ1 = cj + cj , σ2 = i(cj − cj ) and σ3 = 2 − cj cj . The number
j Ď j Ď j Ď

representation for a spin system can be fully accomplished by the association of Paulions with fermionic operators. Such a
relation is given by the Jordan–Wigner formula [3],
! !
cj = aj exp iπ am am = aj exp iπ
Ď
cm cm .
Ď
X X
(8)
m<j m<j

Other useful results can be immediately derived,


cj cj = aj aj ,
Ď Ď

cj±1 cj = aj±1 aj ,
Ď Ď

exp(2πicj cj ) = exp(2πiaj aj ).
Ď Ď

We can write a master equation through the result


φ (t)|ni = b
φ(t)|Ii = |φ(t)i.
X
n
n

Thus we have
∂t |φ(t)i = ∂t φn (t)|ni = {w(n, m)φm (t)|ni − w(m, n)φn (t)|ni}.
X X

n n,m

∂t φn (t)|ni = L|φ(t)i, the master equation would be written as


P
If n

∂t |φ(t)i = L|φ(t)i, (9)


where L is the Liouvillian of the Paulions (or fermions) system. In the next section we apply this approach to the linear
Glauber model.

3. Liouvillian for the linear Glauber model

The linear Glauber stochastic dynamics describes an interacting Ising system in such a way that only one spin of the
lattice is flipped at each time. The master equation is given by [39,48]
N
∂t φ(σ, t) = {ωm (σ m )φm (σ m , t) − ωm (σ)φ(σ, t)},
X
(10)
m=1

where σ = (σ1 , σ2 , . . . , σm, , . . . , σN ) stands for the configuration of the system and σ m = (σ1 , σ2 , . . . , −σm, , . . . , σN ). The
transition rate is defined as
 
α λ
ωm (σ) = σm σm+β  ,
X
1− (11)
2 2d β
5106 É.M. Silva et al. / Physica A 387 (2008) 5101–5109

where the summation is over the 2d nearest neighbours of mth site, α is a parameter describing the temporal scale and
λ ∈ (0, 1].
In order to describe Eqs. (10) and (11) in the Fock space representation, let us define the vector |σ1 , σ2 , . . . , σN i =
|σ1 i ⊗ |σ2 i ⊗ · · · ⊗ |σN i, where hσm |σn i = δnm , and a spin operator, b
σm , which is diagonal, that is, for the mth site
σm |σ1 , σ2 , . . . , σm , . . . , σN i = σm |σ1 , σ2 , . . . , σN i.
b
Besides we define a spin-flip operator, b
Fm , by
Fm |σ1 , σ2 , . . . , σm , . . . , σN i = |σ1 , σ2 , . . . , −σm , . . . , σN i.
b

A probability operator φ̂(t) is defined by b


φ(b
σ , t)|σi = φ(σ, t)|σi. Introducing |φ(t)i = b
φ(t)|Ii, with |Ii = σ |σ i, we
P

obtain
|φ(t)i = b
φ(t) |σi = φ(σ, t)|σi.
X X

σ σ
In this basis the transition rate is also assumed to be described by a diagonal operator, i. e.,
bj (b
w σ )|σi = wj (σ)|σi.
Multiplying Eq. (10) by |σ i and summing up over all possible configurations, we write
∂t φ(σ, t)|σi = ∂t |φ(t)i
X

σ
bm (b
σm ) − w
bm (b
σm )}|φ(t)i.
X
= Fm w
{b
m

As a consequence we have the master equation written as ∂t |φ(t)i = L|φ(t)i, where L, the Liouvillian, is

(b
Fm − 1)w
bm (b
σm ),
X
L= (12)
m

with
 
α λ
bm (b
σm ) = σm σm+β  .
X
w 1− b b (13)
2 2d β

Let us treat explicitly the linear Glauber model defined by Eq. (10) in the Fock basis. The relations among the basis vectors
of type |σ i and |ni is given in the following way. Using 1 = n n1! |nihn|
P

X 1
|σ i = |nihn|σ i. (14)
n n!
σ |σi = σ |σ i, with σ = ±1
As b
X 1 X 1
σ
b |nihn|σ i = σ |nihn|σ i. (15)
n n! n n!
Multiplying Eq. (15) by hm|, we have
X 1 X 1
σ
hm|b |nihn|σ i = σ hm|nihn|σ i = σ hn|σ i,
n n! n n!
resulting in hn|σ i = δ2n−1,σ and bσm = 2cmĎ cm − 1. From Eq. (14) we obtain |σ = 1i = |n = 1i, |σ = −1i = |n = 0i and, from
σm = 2b
b nm − 1, it follows that bσ |n = 1i = 1|1i and bσ |n = 0i = −1|0i. Then we have
 
α λ
bm (c, t) = (2cm cm − 1) (2cm+β cm+β − 1)
X Ď
Ď
w 1−
2 2d β
  
α 2 1X
1 − λ b nm+β + 1 ,
X
= nm nm+β − 2b
b nm − b (16)
2 d β d β
Ď
where b
nm = cm cm is the number operator. The spin flip operator b
Fm is given by
Ď
Fm = cm + cm
b . (17)
Taking Eqs. (16) and (17) in Eq. (12), we find the Liouvillian in terms of the number operator,
  
α 2 1
(b
Fm − 1)w (cm + cmĎ − 1) 1 − λ  b nm+β + 1 .
X X X X
L= bm = nm nm+β − 2b
b nm − b (18)
m 2 m d β d β

In next section we use this Liouvillian to derive equations of motion of the magnetization and the two-point correlation
function.
É.M. Silva et al. / Physica A 387 (2008) 5101–5109 5107

4. The magnetization and the two-point correlation function

Since we have bnm = cmĎ


cm = 12 (b
σm + 1), relevant physical quantities can be derived by the average of b
nm and its products.
nm i,which is written as
Let us then consider hb
nm |φ(t)i,
nm i = hI|b
hb
and from Eq. (9), we find
∂ ∂
nm |φ(t)i = hI|cm
nm i = hI|b
hb Ď
cm L|φ(t)i
∂t ∂t X
Ď
= hI|cm cm (b
Fn − 1)w
bn |φ(t)i,
n

such that for m 6= n, we have the following null commutation relation


Ď
[cm cm , (b
Fn − 1)w
bn ] = 0. (19)
Then we can write
∂ Ď
nm i = hI|cm
hb cm (b
Fm − 1)w
bm |φ(t)i
∂t
Ď
= hI|cm bm |φ(t)i − hI|cm
w Ď
bm |φ(t)i.
cm w (20)
Using the properties
Ď
hI|cm = (h0| + h1|)cm
Ď
= h0|,
Ď
h0|cm cm = 0,
Ď
h1|cm cm = h1|,
Ď
hI|cm cm = (h0| + h1|)cm
Ď
cm = h1|, (21)
we rewrite the rhs-terms of Eq. (20) as
Ď
hI|cm bm |φ(t)i = h0|w
w bm |φ(t)i
  
α 1X
= h0| 1 + λ nm+β − 1 |φ(t)i
b
2 d β

and
Ď
hI|cm bm |φ(t)i = h1|w
cm w bm |φ(t)i
  
α 1X
= h1| 1 − λ nm+β − 1 |φ(t)i.
b
2 d β

Therefore, Eq. (20) reads


   
2 ∂ 1X 1X
nm i = h0|1 + λ 
hb nm+β − 1 |φ(t)i − h1|1 − λ  nm+β − 1 |φ(t)i
α ∂t
b b
d β d β
  
λ X
= (h0| − h1|)|φ(t)i + (h0| + h1|)   nm+β − d |φ(t)i
b
d β
or
2 ∂ λ
nm − 1|φ(t)i + nm+β − 1|φ(t)i,
X
nm i = −hI|2b
hb hI|
α ∂t
b
d β

where we have used h0| − h1| = −hI|(2b


nm − 1). Finally we obtain

1 ∂ λ X 1−λ
nm i = −hb
hb nm i + nm+β i +
hb . (22)
α ∂t 2d β 2

σm i, is derived considering that


The connection of this equation with the evolution equation for the magnetization, hb
σm = 2b
b nm − 1. Hence we get
1 ∂ 1 λ X 1−λ
σm i = − hb
hb σm + 1i + σm+β + 1i +
hb
2α ∂ t 2 4d β 2

1 λ X
σm i +
= − hb σm+β i,
hb (23)
2 4d β
5108 É.M. Silva et al. / Physica A 387 (2008) 5101–5109

leading to the known equation for the magnetization for the d-dimensional linear Glauber model [39]
1 ∂ λ X
σm i = −hb
hb σm i + σm+β i.
hb (24)
α ∂t 2d β

A similar procedure can be used to derive the evolution equation for the pair correlation function, hb nn i. We start with
nmb
∂ ∂
nmb
hb nn i = hI|b
nmbnn |φ(t)i
∂t ∂t X
Ď
= hI|cm cm cnĎ cn (b
Fr − 1)w
br |φ(t)i
r

(b
Fr − 1)w Ď
cm cnĎ cn |φ(t)i.
X
= hI| b r cm
r

For r 6= m, n, we have
(b
Fr − 1) = (h0| + h1|)(crĎ + cr − 1) = 0,
X X
hI|
r6=m,n r6=m,n

and so

nmb
hb Ď
nn i = hI|cm cm cnĎ cn (b
Fm − 1)w
bm |φ(t)i + hI|cm
Ď
cm cnĎ cn (b
Fn − 1)w
bn |φ(t)i. (25)
∂t
Taking into account Eq. (19) and the commutation and anticommutation relations for the Paulions, the rhs-terms of Eq. (25)
read
hI|cĎ cm cĎ cn (b
m n Fm − 1)w
bm |φ(t)i = hI|cĎ cm (cĎ + cm − 1)w
bm cĎ cn |φ(t)i
m m n
Ď
= hI|cm w nn |φ(t)i + hI|cm
bmb Ď Ď
cm cm w nn |φ(t)i − hI|cm
bmb Ď
cm w nn |φ(t)i
bmb
= h0|w nn |φ(t)i − h1|w
bmb nn |φ(t)i
bmb (26)
and
Ď
hI|cm cm cnĎ cn (b
Fn − 1)w
bn |φ(t)i = hI|cnĎ cn (cnĎ + cn − 1)w Ď
b n cm cm |φ(t)i
= hI|cnĎ w nm |φ(t)i − hI|cnĎ cnĎ cn w
bnb nm |φ(t)i − hI|cnĎ cn w
bnb nm |φ(t)i
bnb
= h0|w nm |φ(t)i − h1|w
bnb nm |φ(t)i.
bnb (27)
Using the properties given in Eq. (21), the rhs-terms of Eq. (26) are written as
  
α 1X
h0|w
bmbnn |φ(t)i = h0| 1 + λ nm+β − 1 b
b nn |φ(t)i
2 d β

and
  
α 1X
h1|w nn |φ(t)i = h1|
bmb 1 − λ nn |φ(t)i.
nm+β − 1 b
b
2 d β

Similar results follow from each rhs-term in Eq. (27). Considering that h0| − h1| = −hI|(2b
nm − 1), Eq. (25) results in
1 ∂ λ X 1−λ
hb nn i = −2hb
nmb nmb
nn i + nmb
[hb nn+β i + hb
nnb
nm+β i] + (hb nn i).
nm i + hb (28)
α ∂t 2d β 2
σm = 2b
Since b nm − 1, and from the fact that
σm b
hb σn i = h(2b
nm − 1)(2b
nn − 1)i
= 4hb nn i − 2hb
nmb nn i + 1,
nm i − 2hb
∂ b b
we calculate ∂t hσm σn i. From Eq. (22), we write
4 ∂ 2 ∂ 2 ∂
nmb
hb nn i = nm i +
hb nn i − 8hb
hb nn i + 4hb
nmb nm i + 4hb
nn i − 2
α ∂t α ∂t α ∂t
λ X
+ [4hb nn+β i + 4hb
nmb nm+β i − 2hb
nnb nn+β i − 2hb
nm+β i − 2hb nn i + 2],
nm i − 2hb
2d β

leading to [39]
1 ∂ λ X
σm b
hb σn i = −2hb
σmb
σn i + σm b
[hb σn+β i + hb
σnb
σm+β i]. (29)
α ∂t 2d β
Notice that Eqs. (22) and (28) provide an alternative way to study the physical properties in the dynamical linear (or non
linear) Glauber model, by exploring the content of Fock space. In particular, the averages hb
nm i and hb nn i can be associated
nmb
directly with the propagator in the continuous limit.
É.M. Silva et al. / Physica A 387 (2008) 5101–5109 5109

5. Final concluding remarks

In this paper we have developed a formalism based on Fock space to treat stochastic spin systems. The general results
were derived in parallel with the case of bosons [5,8,38], except by some peculiarities as, for example, the definition of a
propagator or the generating function, where we used Grassmann variables. Some results about path integrals for fermions,
although known in the literature of quantum field theory, were presented here in order to specify the characteristics of the
lattice. The central ingredient in the development was the definition of a density probability operator, giving rise to a general
approach for stochastic equations in the context of number representation for bosons and fermions.
We have applied the formalism to the study of the d-dimensional linear Glauber model, deriving equations for the
magnetization and the two-point correlation function, in terms of creation and annihilation operators. That is, we obtained
the time evolution equation for hb nm i and hb nn i, in order to study the evolution of the magnetization and the two-point
nmb
correlation function. This analysis points, in particular, to the fact that one can use this procedure to study the non-linear
Glauber model. Applications of this approach to other stochastic spin systems, such as the use of the path integral formalism
to treat the problem of renormalization in directed percolation, are in progress and will be presented in another work.

Acknowledgements

The authors thank S.R. Salinas and Tânia Tomé for helpful discussions. This work was supported by CAPES and CNPq of
Brazil.

References

[1] M. Schönberg, N. Cimento 9 (1952) 1139; 10 (1953) 419; 10 (1953) 697.


[2] M. Doi, J. Phys. A: Math. Gen. 9 (1976) 1465; 9 (1976) 1479.
[3] D.C. Mattis, M.L. Glasse, Rev. Mod. Phys. 70 (1998) 979.
[4] C. Martin, E.D. Siggia, H.A. Rose, Phys. Rev. A 8 (1973) 423.
[5] L. Peliti, J. Physique 46 (1985) 1469.
[6] R. Dickman, R.R. Vidigal, Braz. J. Phys. 33 (2003) 73.
[7] J. Cardy, Int. J. Mod. Phys. B 8 (1994) 3463.
[8] P. Grassberger, M. Scheunert, Fortschr. Phys. 28 (1980) 547.
[9] H.C. Andersen, J. Math. Phys. 41 (2000) 1979.
[10] A. Loinger, Ann. Phys. (NY) 20 (1962) 132.
[11] G. Della Riccia, N. Wiener, J. Math. Phys. 7 (1966) 1372.
[12] G. Lugarini, M. Pauri, Ann. Phys. (NY) 44 (1967) 226.
[13] J.J. Hopfield, A.J.F. Bastini, Phys. Rev. 168 (1968) 193.
[14] E.C.G. Sudarshan, N. Mukunda, Classical Dynamics: A Modern Perspective, John Wiley and Sons, New York, 1974.
[15] T. Ali, E. Prugovečki, Physica A 89 (1977) 501.
[16] T.N. Sherry, E.C.G. Sudarshan, Phys. Rev. D 18 (1978) 4580.
[17] B. Misra, Proc. Natl. Acad. Sciences USA 17 (1978) 315.
[18] C. George, I. Prigogine, Physica A 99 (1979) 369.
[19] D. Bohm, B.J. Hiley, Found. Phys. 11 (1981) 179.
[20] R. Paul, Field Theoretical Methods in Chemical Physics, Elsevier, Amsterdam, 1982.
[21] B. Misra, I. Prigogine, Lett. Math. Phys. 7 (1983) 421.
[22] A. Matos Neto, J.D.M. Vianna, N. Cimento B 86 (1985) 117.
[23] P.R. Holland, Found. Phys. 16 (1986) 701.
[24] M.C.B. Fernandes, J.D.M. Vianna, Found. Phys. 29 (1999) 201.
[25] L.M. Abreu, A.E. Santana, A. Ribeiro Filho, Ann. Phys. (NY) 297 (2002) 396.
[26] A.E. Santana, F.C. Khanna, H. Chu, Y.C. Chang, Ann. Phys. (NY) 246 (1996) 481.
[27] U.C. Täuber, Field-theory approaches to nonequilibrium dynamics, in: Ageing and the Glass Transition, in: M. Henkel, M. Pleimling, R. Sanctuary (Eds.),
Lecture Notes in Physics, vol. 716, Springer, Heidelberg, 2007, pp. 295–348.
[28] T.D. Schultz, D.C. Mattis, E.H. Lieb, Rev. Mod. Phys. 36 (1964) 856.
[29] F.C. Alcaraz, M. Droz, M. Henkel, V. Rittenberg, Ann. Phys. (NY) 230 (1994) 250.
[30] T. Tomé, M.J. de Oliveira, Phys. Rev. E 58 (1998) 4242.
[31] R. Dickman, J. Stat. Phys. 55 (1989) 997.
[32] M.J. de Oliveira, T. Tomé, R. Dickman, Phys. Rev. A 46 (1992) 273.
[33] C. Godrèche, J.M. Luck, J. Phys. A: Math. Gen. 33 (2000) 1151.
[34] M.D. Grynberg, R.B. Stinchcombe, Phys. Rev. E 71 (2005) 066104.
[35] T. Michael, S. Trimper, M. Schulz, Phys. Rev. E 73 (2006) 062101.
[36] V. Brunel, K. Oerding, F. van Wijland, J. Phys. A: Math. Gen. 33 (2000) 1085.
[37] L. Erdős, M. Salmhofer, H.T. Yau, J. Stat. Phys. 116 (2004) 367.
[38] A.H. Ali, J. Math. Phys. 47 (2006) 083302.
[39] M.J. de Oliveira, Phys. Rev. E 67 (2003) 066101.
[40] I. Dornic, H. Chaté, J. Chave, H. Hinrichsen, Phys. Rev. Lett. 87 (2001) 045701.
[41] L. Franchebourg, P.L. Krapivsky, Phys. Rev. E 53 (1996) R3009.
[42] A. Mehta, J.M. Luck, Phys. Rev. E 60 (1999) 5218.
[43] M.O. Hase, S.R. Salinas, T. Tomé, M.J. de Oliveira, Phys. Rev. E 73 (2006) 056117.
[44] P.L. Krapivsky, Phys. Rev. A 45 (1992) 1067.
[45] M.J. de Oliveira, J.F.F. Mendes, M.A. Santos, J. Phys. A: Math. Gen. 26 (1993) 2317.
[46] M.-J. Drouffe, C. Godrèche, J. Phys. A: Math. Gen. 32 (1999) 249.
[47] P.T. Muzy, S.R. Salinas, A.E. Santana, T. Tomé, Rev. Bras. Ens. Fís. 27 (2005) 447.
[48] T. Tomé, M.J. de Oliveira, Dinâmica Estocástica e Irreversibilidade, EdUSP, São Paulo, 2001.
[49] C. Itzykson, J.-B. Zuber, Quantum Field Theory, McGraw-Hill, New York, 1980.
[50] K.E. Cahill, R.J. Glauber, Phys. Rev. A 59 (1999) 1538.

You might also like