Escolar Documentos
Profissional Documentos
Cultura Documentos
1, 1303 (2013)
www.sbfisica.org.br
1
Instituto de Fsica Gleb Watahgin, Universidade Estadual de Campinas, Campinas, SP, Brasil
Faculdade de Ciencias de Bauru, Universidade Estadual Paulista Julio de Mesquita Filho, Bauru, SP, Brasil
3
Instituto de Fsica de S
ao Carlos, Universidade de S
ao Paulo, S
ao Carlos, SP, Brasil
Recebido em 2/1/2012; Aceito em 29/8/2012; Publicado em 18/2/2013
We present a derivation of the Lindblad equation - an important tool for the treatment of nonunitary evolutions - that is accessible to undergraduate students in physics or mathematics with a basic background on
quantum mechanics. We consider a specic case, corresponding to a very simple situation, where a primary
system interacts with a bath of harmonic oscillators at zero temperature, with an interaction Hamiltonian that
resembles the Jaynes-Cummings format. We start with the Born-Markov equation and, tracing out the bath
degrees of freedom, we obtain an equation in the Lindblad form. The specic situation is very instructive, for it
makes it easy to realize that the Lindblads represent the eect on the main system caused by the interaction with
the bath, and that the Markov approximation is a fundamental condition for the emergence of the Lindbladian
operator. The formal derivation of the Lindblad equation for a more general case requires the use of quantum
dynamical semi-groups and broader considerations regarding the environment and temperature than we have
considered in the particular case treated here.
Keywords: Lindblad equation, open quantum systems.
Apresentamos uma derivaca
o da equaca
o de Lindblad - uma ferramenta importante no tratamento de
evoluco
es n
ao-unit
arias - acessvel a estudantes de graduaca
o em fsica ou matem
atica com noco
es b
asicas de
mec
anica qu
antica. Consideramos aqui um caso especco, correspondente a uma situaca
o bem simples, onde
o sistema principal interage com um banho de osciladores harm
onicos `
a temperatura nula, com hamiltoniano
de interaca
o que se assemelha ao modelo de Jaynes-Cummings. Iniciamos com a equaca
o de Born-Markov e,
atraves do traco parcial dos graus de liberdade do banho, obtemos uma equaca
o na forma de Lindblad. Essa
situaca
o especca e bem instrutiva, pois permite vericar que os lindblads representam a contribuica
o do sistema
principal ao hamiltoniano de interaca
o com o banho, e que a aproximaca
o markoviana e vital para o surgimento
do lindbladiano. A deduca
o formal da equaca
o de Lindblad para situaco
es gerais requer o uso do formalismo de
semigrupos din
amicos qu
anticos e consideraco
es mais abrangentes sobre o ambiente e a temperatura do que as
utilizadas aqui.
Palavras-chave: equaca
o de Lindblad, sistemas qu
anticos abertos.
1. Introduction
The Lindblad equation [1] is the most general form for a
Markovian master equation, and it is very important for
the treatment of irreversible and non-unitary processes,
from dissipation and decoherence [2] to the quantum
measurement process [3,4]. For the latter, in recent applications [4, 5], the Lindblad equation was used in the
introduction of time in the interaction between the measured system and the measurement apparatus. Then,
the measurement process is no longer treated as instantaneous, but nite, with the duration of that interaction changing the probabilities - diagonal elements of
the density operator - associated to the possible nal
1 E-mail:
carlosbrasil.physics@gmail.com.
1303-2
Brasil et al.
2.
(1)
SB = H
SB .
S + HB + HSB ,
dt
~
(3)
(4)
(5)
and
i
(7)
t[
]
(t ) , (t ) dt .
H
(8)
H (t) ,
H (t ) , (t ) dt .
~2
0
(9)
(10)
(t) depends on H
SB ,
By the denition in Eq. (4), H
{[
t[
] ]}
H (t) ,
H (t ) , (t ) dt .
0
(11)
1303-3
where S is a general operator that acts only on the prin is an operator that acts only on
cipal system S, and B
the bath B. Now, we consider that S commutes with
S , i.e.,
H
S (t) S (t ) =
{[
]}
t [
]
1 2 t
(t ) ,
(t ) , (t ) dt
2
dt trB
H
H
,
~
t
0
which shows that the dierence between S (t) and
S (t ) is of the second order of magnitude in and,
therefore, we can write S (t) in the integrand of
Eq. (11), obtaining a time-local equation for the density
operator, without violating the Born approximation
{[
t[
] ]}
d
1 2
S (t) = 2 trB
H (t) ,
H (t ) , (t) dt .
dt
~
0
(12)
The constant was introduced in Eq. (2) only for
clarifying the order of magnitude of each term in the
iteration and, now, it can be supressed, that is, let us
take = 1 (full interaction). Thus, let us write
{[
t[
] ]}
d
1
S (t) = 2 trB
H (t) ,
H (t ) , (t) dt .
dt
~
0
(13)
For this approximation, we can write (t) = S (t) B
inside the integral and obtain the equation that will be
used in the next calculations (again, for the sake of simplicity, let us ignore the symbol )
d
S (t) =
dt
{[
[
]]}
1
(t) , H
(t ) , S (t) B
2
dt trB H
, (14)
~ 0
where we assume that the integration can be extended
to innity without changing its result. Equation (14) is
the Born-Markov master equation [2].
H
S
S,
3.1.
S (t)
(16)
B = ~ k a
H
k a
k
(17)
where a
k e a
k are the annihilation and creation bath
operators, the k are the characteristic frequencies of
operator on Eq. (15) dened by
each mode, and the B
gk a
k ,
(18)
~ HB t .
e ~ HB t Be
i
(19)
gk a
k eik t .
(20)
0,
(S is not aected by the interaction-picture transformation). Let us consider the bath hamiltonian dened
by a bath of bosons,
Lindblad equation
SB
H
resulting in
(t) =
B
3.
(15)
[
]]
(t) , H
(t ) , B S (t)
H
[
[
]]
(t) + S B
(t) , H
(t ) , B S (t)
= ~ SB
[
[
]]
(t) , H
(t ) , B S (t)
= ~ SB
[
[
]]
(t) , H
(t ) , B S (t) .
+ ~ S B
(21)
1303-4
Brasil et al.
[
]]
[
[
]]
(t) , H
(t ) , B S (t)
(t) , SB
(t ) + S B
(t ) , B S (t)
SB
= ~ SB
[
]
(t) SB
(t ) + S B
(t ) B S (t)
= ~SB
[
]
(t) B S (t) SB
(t ) + S B
(t )
~SB
[
]
(t ) + S B
(t ) B S (t) SB
(t)
~ SB
]
[
(t)
(t ) + S B
(t ) SB
+ ~
B S (t) SB
(22)
[
]]
[
[
]]
(t) , H
(t ) , B S (t)
(t) , SB
(t ) + S B
(t ) , B S (t)
S B
= ~ S B
[
]
(t) SB
(t ) + S B
(t ) B S (t)
= ~S B
[
]
(t) B S (t) SB
(t ) + S B
(t )
~S B
[
]
(t ) + S B
(t ) B S (t) S B
(t)
~ SB
[
]
(t ) + S B
(t ) S B
(t) ,
+ ~
B S (t) SB
(23)
and
[
or, expanding Eqs. (22) and (23) and grouping the similar terms in S and B, we have
[
[
]]
(t) , H
(t ) , B S (t)
SB
(t) B
(t ) B + ~SS S (t) B
(t) B
(t ) B
= ~SSS (t) B
(t) B B
(t )
(t) B B
(t ) ~SS (t) S B
~SS (t) SB
(t ) B B
(t) ~S S (t) SB
(t ) B B
(t)
~SS (t) SB
(t ) B
(t)
(t ) B
(t) + ~
+ ~
S (t) SSB B
S (t) S SB B
(24)
and
[
[
]]
(t) , H
(t ) , B S (t)
(t) B
(t ) B + ~S S S (t) B
(t) B
(t ) B
S B
= ~S SS (t) B
(t) B B
(t )
(t) B B
(t ) ~S S (t) S B
~S S (t) SB
(t ) B B
(t) ~S S (t) S B
(t ) B B
(t)
~2 SS (t) S B
(t ) B
(t) .
(t ) B
(t) + ~
+ ~
S (t) SS B B
S (t) S S B B
3.2.
(25)
Now we are in a position to trace out the bath degrees of freedom in Eqs. (24) and (25). As we can verify with
Eq. (20),
{
}
{
}
(t) B
(t ) B
(t) B
(t ) B = 0, t, t .
trB B
= trB B
With this, then,
{[
trB
[
]]}
{
}
(t) , H
(t ) , B S (t)
(t) B
(t ) B
SB
= ~SS S (t) trB B
{
}
(t) B B
(t )
~SS (t) S trB B
{
}
B B
(t ) B B
(t)
~S S (t) Str
{
}
B B B
(t ) B
(t)
+ ~
S (t) S Str
(26)
and
trB
1303-5
[
]]}
{[
(t) , H
(t ) , B S (t)
S B
{
}
(t) B
(t ) B
= ~S SS (t) trB B
}
{
B B
(t) B B
(t )
~S S (t) Str
{
}
(t ) B B
(t)
~SS (t) S trB B
{
}
(t ) B
(t) ,
+ ~
S (t) SS trB B B
(27)
[
]]}
[
]
{
}
(t) , H
(t ) , B S (t)
(t) B
(t ) B
SB
= ~ SS S (t) S S (t) S trB B
[
]
{
}
(t ) B
(t) B
+ ~ S (t) S S SS (t) S trB B
(28)
[
]]}
[
]
{
}
(t) , H
(t ) , B S (t)
(t) B
(t ) B
S B
= ~ S SS (t) SS (t) S trB B
{
}
[
]
(t ) B
(t) B .
+ ~ S (t) SS S S (t) S trB B
(29)
and
{[
trB
The terms represented by Eqs. (28) and (29) allow us to return to the Eq. (21)
trB
{[
[
]]}
(t) , H
(t ) , B S (t)
H
=
[
]
{
}
(t) B
(t ) B
~2 SS S (t) S S (t) S trB B
]
{
}
[
(t ) B
(t) B
~2 S (t) S S SS (t) S trB B
[
]
{
}
(t) B
(t ) B
~2 S SS (t) SS (t) S trB B
[
]
{
}
(t ) B
(t) B .
~2 S (t) SS S S (t) S trB B
+
+
+
3.3.
(30)
With the results of the preceding paragraphs, the integrand in Eq. (14) becomes
trB
{[
[
]]}
(t) , H
(t ) , B S (t)
H
=
[
]
{
}
(t) B
(t ) B
~2 SS S (t) S S (t) S trB B
[
]
{
}
(t ) B
(t) B
~2 S (t) S S SS (t) S trB B
]
{
}
[
(t) B
(t ) B
~2 S SS (t) SS (t) S trB B
[
]
{
}
(t ) B
(t) B .
~2 S (t) SS S S (t) S trB B
+
+
+
(31)
F (t)
=
0
G (t)
{
}
(t) B
(t ) B ,
dt trB B
{
}
(t ) B
(t) B .
dt trB B
Then,
F (t)
=
0
G (t)
{
}
(t ) B
(t) B ,
dt trB B
{
}
(t) B
(t ) B .
dt trB B
(32)
1303-6
Brasil et al.
[
]
[
]
SS S (t) S S (t) S G (t) S (t) S S SS (t) S F (t)
[
]
[
]
S SS (t) SS (t) S F (t) S (t) SS S S (t) S G (t) .
(33)
Actually, the usual Lindblad equation emerges when G(t) = 0 and F (t) = F (t). In the following, we make
some specications about the environment to discuss these approximations in detail.
3.4.
Furthermore, for the initial state of the thermal bath, we consider the vacuum state
B = (|0 |0 ...) (0| 0| ...) .
(34)
(t) and B
The evaluation of the F (t) and G (t) functions dened in Eq. (32) are done considering the B
(t) =
B
gk a
k eik t .
(35)
k
(36)
{
}
(t ) B
(t) B
trB B
(37)
and
{
}
(t ) B
(t) (|0 |0 ...) (0| 0| ...) .
= trB B
If we use some bath state basis{|b}, Eqs. (36) and (37) become
{
}
(t) B
(t ) B
trB B
(t) B
(t ) (|0 |0 ...) (0| 0| ...) |b
b| B
(t) B
(t ) (|0 |0 ...)
(0| 0| ...) |b b| B
= (0| 0| ...)
(t) B
(t ) (|0 |0 ...)
|b b| B
(t) B
(t ) (|0 |0 ...)
= (0| 0| ...) B
(38)
and
{
}
(t ) B
(t) B
trB B
(t ) B
(t) (|0 |0 ...) (0| 0| ...) |b
b| B
(t ) B
(t) (|0 |0 ...)
(0| 0| ...) |b b| B
= (0| 0| ...)
(t ) B
(t) (|0 |0 ...)
|b b| B
(t ) B
(t) (|0 |0 ...) .
= (0| 0| ...) B
(39)
(t) and B
(t) using Eqs. (20) and (35)
Let us expand B
{
}
(t) B
(t ) B
trB B
= (0| 0| ...)
gk a
k eik t
gk a
k eik t (|0 |0 ...)
=
k,k
(40)
1303-7
and
{
}
(t ) B
(t) B
trB B
= (0| 0| ...)
gk a
k eik t
gk a
k eik t (|0 |0 ...)
=
gk gk ei(k tk t )
(0| 0| ...) a
k a
k (|0 |0 ...) = 0
(41)
k,k
(42)
Then
{
}
(t) B
(t ) B
gk gk ei(k tk t ) (0| 0| ...) a
k a
k (|0 |0 ...)
gk gk ei(k tk t ) k,k +
trB B
=
k,k
k,k
(43)
|gk |
dt eik (tt ) ,
(44)
G (t) = 0.
3.5.
In the expression of F (t) in Eq. (44), if we adopt the general denition of the density of states as
2
J () =
|gl | ( l ) ,
(45)
then the sum over k can be replaced by an integral over a continuum of frequencies
t
F (t) =
dJ ()
dt ei(tt ) .
0
= t t ,
= dt ,
d ei .
0
As the integrand oscillates, we will use the device
d ei = lim+
d ei
with
dt
d =
d,
yielding
F (t) =
dJ ()
0
3.6.
1
+ i
i
= lim+ 2
0 + 2
i
= lim+ 2
lim+ 2
2
0 +
0 + 2
1
= () iP ,
0
0
=
d ei .
lim
0+
1303-8
3.7.
Brasil et al.
dJ () () ,
0
F =
+ i
,
2
(46)
2P
d
0
where
J ()
.
(47)
(48)
[
] i [
] + i
S (t) S S SS (t) S
S SS (t) SS (t) S
2
2
]
[
= S (t) S S SS (t) S + S SS (t) SS (t) S
2
]
[
+ i S (t) S S SS (t) S S SS (t) SS (t) S .
2
d
S (t) =
dt
If the density of states is chosen to yield = 0 (a Lorentzian, for example, where we can extend the lower limit of
integration to ), the nal result is
[
}]
d
1 {
S (t) = SS (t) S
S S, S (t) .
(49)
dt
2
Let us, then, return to the original picture. Since
S (t) = e ~ HS t S e ~ HS t ,
i
(50)
then
d
S (t) =
dt
i
i
i i H S t
d
S i H S t i i H S t i H S t
HS S e ~ HS t + e ~ HS t
e~
e ~
e~
S HS e ~
~
dt
~
] i
i
d
S i H S t i i H S t [
= e ~ HS t
+ e~
HS , S e ~ HS t .
e ~
dt
~
Performing the same operation on the right-hand side of Eq. (49) gives
[
[
}]
}] i
i
1 {
1 {
HS t
~
S S S
S S (t) S
S S, S (t) = e
S S, S e ~ HS t .
2
2
(51)
(52)
= HS , S + S S S
S S, S .
dt
~
2
(53)
4.
Conclusion
] [
}]
1 {
d
S
i [
Lj S Lj
= HS , S +
Lj Lj , S .
dt
~
2
j
(54)
If we consider only the rst term on the right hand
side of Eq. (54), we obtain the Liouville-von Neumann
equation. This term is the Liouvillian and describes the
unitary evolution of the density operator. The second
term on the right hand side of the Eq. (53) is the Lindbladian and it emerges when we take the partial trace
- a non-unitary operation - of the degrees of freedom of
system B. The Lindbladian describes the non-unitary
evolution of the density operator. By the interaction
form adopted here, Eq. (15), the physical meaning of
the Lindblad operators can be understood: they represent the system S contribution to the S B interaction
- remembering once more that the Lindblad equation
was derived from the Liouville-von Neumann one by
tracing the bath degrees of freedom. This conclusion is
also achieved with the more general derivation [1, 2]. It
is important to emphasize that, due to our simplifying
assumptions, the summation appearing in Eq. (54) was
not obtained in our derivation of Eq. (53).
j are Hermitian (obIf the Lindblad operators L
servables), the Lindblad equation can be used to treat
the measurement process. A simple application in this
S
sense is the Hamiltonian H
z (
z is the 2-level z Pauli mattrix) when we want to measure one specic
same Hamiltonian H
z with the Lindblad L
x i
y
(
=
),
where
will
be
the
spontaneous
emis2
sion rate [6].
1303-9
References
[1] G
oran Lindblad, Commun. Math. Phys. 48, 119
(1976).
[2] Heinz-Peter Breuer and Francesco Petruccione, The
Theory of Open Quantum Systems (Oxford University
Press, Oxford, 2002).
[3] Ian Percival, Quantum State Diusion (Cambridge
University Press, Cambridge, 1998).
[4] Carlos A. Brasil and Reginaldo de J. Napolitano, Eur.
Phys. J. Plus 126, 91 (2011).
[5] Carlos A. Brasil, Leonardo A. de Castro and Reginaldo
de J. Napolitano, Phys. Rev. A 84, 022112 (2011).
[6] Michael A. Nielsen and Isaac L. Chuang, Computaca
o
Qu
antica e Informac
ao Qu
antica (Bookman, Porto
Alegre, 2005).
[7] Amir O. Caldeira and Anthony J. Leggett, Physica A
121, 587 (1983).
[8] Edward Brian Davies, Quantum Theory of Open Systems (Academic Press, London, 1976).
[9] Robert Alicki and Karl Lendi, Quantum Dynamical
Semigroups and Applications (Springer-Verlag, Berlin,
2007).
[10] Stephen L. Adler, Phys. Lett. A 265, 58 (2000).
[11] Stephen L. Adler, Phys. Lett. A 267, 212 (2000).
[12] Asher Peres, Phys. Rev. A 61, 022116 (2000).
Acknowledgments