1141
78. Quantized Field Effects
Quantized Fiel
The electromagnetic field appears almost every-
where in physics. Following the introduction of
Maxwells equations in 1864, Max Planck initi-
ated quantum theory when he discovered h = 2~
in the laws of black-body radiation. In 1905 Al-
bert Einstein explained the photoelectric effect
on the hypothesis of a corpuscular nature of
radiation and in 1917 this paradigm led to a de-
scription of the interaction between atoms and
electromagnetic radiation.
The study of quantized field effects requires
an understanding of the quantization of the
field which leads to the concept of a quantum
of radiation, the photon. Specific nonclassical
features arise when the field is prepared in
particular quantum states, such as squeezed
states. When the radiation field interacts with an
atom, there is an important difference between
a classical field and a quantized field. A classical
field can have zero amplitude, in which case it
does not interact with the atom. On the other hand
a quantized field always interacts with the atom,
even if all the field modes are in their ground
states, due to vacuum fluctuations. These lead to
various effects such as spontaneous emission and
the Lamb shift.
The interaction of an atom with the many
modes of the radiation field can conveniently be
described in an approximate manner by a master
equation where the radiation field is treated as
a reservoir. Such a treatment gives a microscopic
and quantum mechanically consistent description
of damping.
78.1 Field Quantization ............................... 1142
78.2 Field States ......................................... 1142
78.2.1 Number States .......................... 1143
78.2.2 Coherent States......................... 1143
78.2.3 Squeezed States ........................ 1144
78.2.4 Phase States ............................. 1145
78.3 Quantum Coherence Theory .................. 1146
78.3.1 Correlation Functions................. 1146
78.3.2 Photon Correlations ................... 1146
78.3.3 Photon Bunching
and Antibunching ..................... 1147
78.4 Photodetection Theory ......................... 1147
78.4.1 Homodyne
and Heterodyne Detection.......... 1147
78.5 Quasi-Probability Distributions ............. 1148
78.5.1 s-Ordered Operators .................. 1148
78.5.2 The P Function.......................... 1149
78.5.3 The Wigner Function.................. 1149
78.5.4 The Q Function.......................... 1151
78.5.5 Relations
Between Quasi-Probabilities ...... 1151
78.6 Reservoir Theory .................................. 1151
78.6.1 Thermal Reservoir ..................... 1152
78.6.2 Squeezed Reservoir ................... 1152
78.7 Master Equation .................................. 1152
78.7.1 Damped Harmonic Oscillator....... 1153
78.7.2 Damped Two-Level Atom ........... 1153
78.8 Solution of the Master Equation ........... 1154
78.8.1 Damped Harmonic Oscillator....... 1154
78.8.2 Damped Two-Level Atom ........... 1155
78.9 Quantum Regression Hypothesis ........... 1156
78.9.1 Two-Time Correlation Functions
and Master Equation ................. 1156
78.9.2 Two-Time Correlation Functions
Part F 78
and Expectation Values .............. 1156
78.10 Quantum Noise Operators ..................... 1157
78.10.1 Quantum Langevin Equations ..... 1157
78.10.2 Stochastic Differential Equations . 1158
78.11 Quantum Monte Carlo Formalism .......... 1159
78.12 Spontaneous Emission in Free Space ..... 1159
78.13 Resonance Fluorescence ....................... 1160
78.13.1 Equations of Motion .................. 1160
78.13.2 Intensity of Emitted Light ........... 1160
78.13.3 Spectrum
of the Fluorescence Light ........... 1161
78.13.4 Photon Correlations ................... 1161
78.14 Recent Developments........................... 1162
78.14.1 Literature ................................. 1162
78.14.2 Field States .............................. 1162
78.14.3 Reservoir Theory ....................... 1162
References .................................................. 1163
1142 Part F Quantum Optics

78.1 Field Quantization

This section provides the basis for the quantized field The field is quantized by replacing the classical am-
effects discussed in this Chapter [78.1]. We expand the plitude k by the mode annihilation operator ak . The
field in a complete set of normal modes which reduces complex conjugate k is replaced by the mode creation

the problem of field quantization to the quantization of operator ak . They obey the commutation relation
a one dimensional harmonic oscillator corresponding to  

each normal mode. ak , ak  = kk  . (78.8)
The classical free electromagnetic field, i. e., the field The representation of the electric field operator
in a region without charge and current densities, obeys
the Maxwell equations
 ~k 1/2
E(x, t) = i
B=0, (78.1) 20 V
k,
 
D=0, (78.2)
ak
k ei(kxk t) h.c.
B
E+ =0, (78.3) E+ (x, t) + E (x, t) (78.9)
t
D in terms of these operators follows from (78.6) and the
H =0, (78.4)
t operator for the vector potential. Note also the often
where B = 0 H, D = 0 E. The magnetic permeabil- used decomposition of the electric field operator into
ity 0 connects the magnetic induction B with the the positive and negative frequency parts E+ and E
magnetic field H and the electric permittivity 0 of respectively. A similar relation holds for the operator
free space connects the displacement D with the elec- describing the magnetic induction B.
tric field E. In the case of a free field, E and B may be Using the operators for the electric and magnetic
obtained from field, one can transform the field energy
B=A, (78.5) 

1
A H= dV 0 E2 + B2 /0 (78.10)
E= , (78.6) 2
t
where the vector potential A obeys the Coulomb gauge into the form

condition A = 0 and satisfies a wave equation. In H=

~k ak ak + 1/2 , (78.11)
order to solve this wave equation we expand the vector k,
potential
2  1/2 which is a sum of independent harmonic oscillator



Part F 78.2

~ Hamiltonians corresponding to each mode (k, ). The

A(x, t) =
2k0 V
k =1
 
k
k ei(kxk t) + c.c. (78.7)
in a set of normal modes V 1/2 exp(ik x)
k which are
orthonormal in the volume V. Due to the gauge condition
A = 0, we obtain two orthogonal polarization vec-
tors
k1 and
k2 with
k k = 0 for each wave vector k.
The dispersion relation is k = c|k|. The Fourier ampli-
tudes k are complex numbers in the classical theory.
number operator Nk = ak ak represents the number
of photons in the mode (k, ), while ~k /2 is the energy
of the vacuum fluctuations.
Hence each mode of the electromagnetic field is
equivalent to a harmonic oscillator. In the next section
we discuss specific states of a single mode. The general
quantum state of the electromagnetic field consisting of
many modes is given by a superposition of product states
that are composed out of these single mode states.

78.2 Field States

This section summarizes the properties of several im- index (k, ) suppressed, a single mode Hamiltonian is
portant states of the electromagnetic field. From the

~ ~ 2
independence of the normal modes, the discussion may H = ~ a a + 1/2 = p2 + x . (78.12)
be restricted to a single normal mode. With the mode 2 2
 Quantized Field Effects 78.2 Field States 1143

In the second step, the quadrature operators
1
tion of a pure quantum state
x = a + a , (78.13)
2

1
| = cn |n ,
p = a a (78.14)
i 2
are introduced, which are equivalent to scaled position
and momentum operators of a massive particle in a har-
monic potential. The quadratures of a quantized field
are measurable with the help of homodyne detection as
discussed in Sect. 78.4.1
We shall now describe several states of this quantized
field mode: number states, coherent states, squeezed
states, Schrdinger cats, and phase states. A quantized
field in a coherent state shows the most classical be-
havior. A superposition of two coherent states, which is
a Schrdinger cat, already shows nonclassical features.
Number states and squeezed states are further typical
examples of nonclassical states.
78.2.1 Number States
The eigenstates of the Hamiltonian (78.12) are the eigen-
states of the number operator N = a a,
N|n = n|n , (78.15)
where n = 0, 1, 2, . . . denotes the excitations or the
number of photons in the mode. The vacuum state of
the mode |0, is defined by
a|0 = 0 . (78.16)
Number states provide a frequently used representa-
(78.22)
n=0
or a mixed quantum state given by the density operator


= nk |nk| (78.23)
n,k
(Chapt. 7).
78.2.2 Coherent States
The coherent state is a specific superposition of num-
ber states. In contrast to a number state, a coherent
state does not possess a definite number of photons:
the photon distribution is Poissonian. For a large aver-
age photon number, the electric and magnetic fields have
rather well defined amplitudes and phases with vanish-
ing relative quantum fluctuations. Hence the Poissonian
photon distribution frequently serves as a borderline
between classical and nonclassical field states. Non-
classical states show a sub-Poissonian behavior. An
extreme example is a field prepared in a number state.
A parameter which quantifies the deviations from Pois-
sonian behavior is the Q parameter introduced by
Mandel [78.2].
We define the coherent state | as an eigenstate of
the annihilation operator
a| = | (78.24)

Part F 78.2
The ladder of excitations can be climbed up and down
via the application of creation and annihilation operators with the complex amplitude = ||ei . The coherent
state can be represented by
a |n = n + 1|n + 1 , (78.17)
| = ea
a
|0 = D()|0 , (78.25)
a|n = n|n 1 , (78.18)

on a Fock state |n. These number or Fock states form that is, by the action of the displacement operator D()
a complete and orthonormal set of states so that on the vacuum.
The number state representation of | reads


|nn| = 1 , n|k = nk . (78.19)

n
| = e||
2 /2
n=0 |n . (78.26)
n=0
n!
Their quadrature representations are
 1/2 A coherent state |0  that evolves in time according to
Hn (x) ex /2 ,
2
x|n = 2n n! (78.20)
 n 1/2 the free field Hamiltonian (78.12) stays coherent, i. e.,
p2 /2
 p|n = 2 n! (i) Hn ( p) e
n
.
(78.21) |(t) = exp (iHt/~) |0  = eit/2 |(t) ,
(78.27)
The states |x and | p are eigenstates of the quadrature
operators x and p, (78.13) and (78.14). with amplitude (t) = 0 exp[it].
1144 Part F Quantum Optics

Another important representation of a coherent state the commutation

  relation [x, p] = i. Their
 uncertainties
is the x representation (x)2 x 2 x2 and ( p)2 p2  p2 fulfill the
 Heisenberg inequality
x| = 1/4 exp [Re()]2

x p 1/2 . (78.33)
exp x 2 /2 + 2x , (78.28)
The coherent state is a special minimum uncertainty
state
where |x denotes again the x quadrature eigenstate. with equal uncertainties x = p = 1/ 2. Squeezed
The photon distribution in a coherent state states comprise a more general class of minimum uncer-
tainty states with reduced uncertainty in one quadrature
||2n e||
2
at the expense of increased uncertainty in the other.
|n||2 = (78.29)
n! These states |,  are obtained by applying the displace-
is a Poisson distribution with average ment operator D() and the unitary squeeze operator
2  photon number
N = ||2 and variance N = N 2 N2 = ||2 . S() = e 2 
1 2 1 2
a 2 a
(78.34)
Hence the relative fluctuations (N )/N = N1/2
vanish for a large average photon number. to the vacuum
The Mandel Q parameter
|,  = D()S()|0 . (78.35)
(N )2 N
Q (78.30) The squeeze operator S() transforms a and a according
N to
vanishes for a field in a coherent state. A nonclassical
field may show sub-Poissonian behavior with Q < 0. As S ()a S() = a cosh r a e2i sinh r , (78.36)

an example, the Schrdinger cat state is a macroscopic S ()a S() = a cosh r a e 2i
sinh r , (78.37)
superposition of two coherent states
where  = r e2i . The rotated quadratures
   1/2
|cat = 2 + 2 cos 2 sin e2 sin (/2)
2 2
X 1 = x cos p sin , (78.38)
    X 2 = p cos + x sin ,
 ei/2 +  ei/2 ,
(78.39)
(78.31)
transform according to
where is assumed to be real. The Q parameter for this
superposition state, shown in Fig. 78.1, takes on negative S ()(X 1 + iX 2 )S() = X 1 er + i X 2 er , (78.40)
values for specific angles . The nonclassical behavior
which yields the uncertainties
of such a |cat-state can be explained [78.3] as a re-
Part F 78.2

sult of quantum interference between the two coherent X 1 = er / 2 , (78.41)

states present in (78.31). The incoherent superposition
X 2 = er / 2 . (78.42)
described by the density operator
1  i/2  i/2   i/2  i/2 

= e e + e e Q
2
(78.32) 2

does not have this nonclassical character: its Q param- 1.5

eter vanishes. Coherent states have a direct physical
significance: the quantum state of a stabilized laser op- 1
erating well above threshold can be approximated by 0.5
a coherent state.
0.1 0.2 0.3 0.4
78.2.3 Squeezed States
0.5

Squeezed states [78.46] minimize the uncertainty prod- 1

uct of the quadrature components of the electromagnetic
field. The quadrature components x and p of the single Fig. 78.1 The Q parameter for a Schrdinger cat state
mode field are defined in (78.13) and (78.14). They obey (78.31) with amplitude = 4
 Quantized Field Effects 78.2 Field States 1145

In particular, for = 0 the squeezed state |, r is a min- larization. Heidmann et al. [78.11] have shown that
imum uncertainty
state for thequadratures x and p with the difference intensity of these twin beams may ex-
x = er / 2 and p = er / 2. The degree of squeez- hibit reduced quantum fluctuations. Pulsed twin beams
ing in the quadrature x is determined by the squeeze also contain reduced noise in the difference of their
factor r. intensities.
The average photon number
78.2.4 Phase States
N = ||2 + sinh2 r (78.43)

of a squeezed state and its photon number variance The problem of a correct quantum mechanical de-
 2 scription of phase has a long history in quantum
(N )2 =  cosh r e2i sinh r  mechanics [78.12]. First attempts to define a quantum
phase are due to London and Dirac. The London phase
+ 2 cosh2 r sinh2 r (78.44)
state is
contain the coherent contribution as well as squeez-
1
in
ing contributions expressed by r and . In particular, | = e |n , (78.47)
for = 0, the Q parameter becomes negative for a large 2 n=0
enough amplitude and r > 0. The photon number dis-
which is an eigenstate of the exponential phase operator
tribution Wn = |, r|n|2 becomes narrower than the
one for the corresponding coherent state with the same .


This sub-Poissonian behavior is one of the nonclassical 
ei = |nn + 1| . (78.48)
features of a squeezed state. Furthermore, Wn shows os- n=0
cillations [78.7] for larger squeezing. The two regimes Since this operator is not unitary, it does not define
with sub-Poissonian and oscillating photon statistics Wn a Hermitian operator for the phase. Nevertheless
are shown in Fig. 78.2. many  treatments of the phase of a quantum state
A second representation of squeezed states has been | = cn |n are based on the London phase distri-
introduced by Yuen [78.8]. In his notation, a squeezed bution
state is an eigenstate of the operator  2
 

1  
b = a + a , (78.45) Pr() = || =
2
 cn ein
 . (78.49)
2  n 
with ||2 ||2 = 1 and eigenvalue . This eigenstate
can be written in the form

Part F 78.2
|,  = S()D()|0 , (78.46)

which connects the squeezing operator S(r e2i ) with Wn

the parameters = cosh r and = e2i sinh r. In con- 0.1
trast to the definition (78.35), the displacement operator
D() and the squeezing operator S() are applied r
now in reversed order. Nevertheless, the two equations 0.05 0
(78.35) and (78.46) define the same state if the relation
= + is fulfilled.
0
Several experiments have demonstrated the gener- 30
ation of squeezed light. Slusher et al. [78.9] obtained 50
70 5
squeezing in the sidemodes of a four-wave mix- 90
ing process. An optical parametric oscillator below n
threshold has been used by Wu et al. [78.10] in or- Fig. 78.2 The photon number distribution Wn of a squeezed
der to generate squeezed light. Nonclassical features state |, r with the coherent amplitude = 7. For a squeez-
can also be found in a down conversion process. ing parameter r = 0, the Poisson distribution of a coherent
This second-order process creates so-called signal and state | = 7 is just visible. When r increases the pho-
idler photons from one pump photon. Signal and ton distribution first becomes sub-Poissonian and then
idler beam are distinguished by frequency or po- oscillatory
1146 Part F Quantum Optics

Later treatments [78.13] rely on the Hermitian oper- Recently [78.14] a Hermitian phase operator was
ators constructed starting from the phase state (78.47),
1 

ei 
 = restricted to a finite Hilbert space. An operational
sin ei , (78.50)
2i phase description has been proposed [78.15] in which

= 1 
cos ei + 

ei , (78.51)
a classical phase measurement is translated to the
2 quantum realm by using an eight-port homodyne
for the sine and cosine function of the phase. detector.

78.3 Quantum Coherence Theory

This section introduces the correlation functions of the In order to analyze the HanburyBrown and Twiss
electromagnetic field. Ideal photon correlation measure- experiment [78.18] it is necessary to define higher order
ments can bring out the phenomenon of photon bunching correlation functions. The general nth order correlation
and antibunching. function is defined by

78.3.1 Correlation Functions G (n) (x1 , . . . , xn , xn+1 , . . . , x2n )


= Tr E (x1 ) E (xn )

Correlation functions were originally introduced to de- E + (xn+1 ) E + (x2n ) , (78.55)
scribe an ideal photodetection process. Glauber [78.16]
has presented a treatment based on an absorption mech- where the field operators are again normal ordered.
anism in the detector which is sensitive to the positive These correlation functions fulfill a generalized
frequency part E + of the electric field evaluated at the Schwartz inequality
detectors space-time position x (x, t). This leads to  2
an average field intensity G (1) (x1 , x1 )G (1) (x2 , x2 ) G (1) (x1 , x2 ) ,
  (78.56)
I(x) = Tr E (x)E + (x) (78.52)
which becomes, for the nth order functions,
at point x. Here the density operator describes the state
of the field. The ordering of the operators, i. e., E E + G (n) (x1 , .., xn , xn , .., x1 )
a a, is known as normal ordering with all annihilation G (n) (xn+1 , .., x2n , x2n , .., xn+1 )
operators to the right of all creation operators.  2
The expression (78.52) now immediately general- G (n) (x1 , .., xn , xn+1 , .., x2n ) . (78.57)
izes to the correlation function of first order
Part F 78.3

  A field is said to be first-order coherent when its

G (1) (x1 , x2 ) = Tr E (x1 )E + (x2 ) , (78.53) normalized correlation function
with x1 = (x1 , t1 ) and x2 = (x2 , t2 ). The classical in- G (1) (x1 , x2 )
terference experiments, such as Youngs double slit g(1) (x1 , x2 ) =  1/2
G (1) (x1 , x1 )G (1) (x2 , x2 )
experiment, can be described in terms of G (1) . Further-
(78.58)
more, the correlation function of first order is connected  
to the power spectrum S() of a quantized field via satisfies g(1) (x1 , x2 ) = 1. In a Young type experiment,
the WienerKhintchine theorem [78.17]. Under the this case gives maximum fringe visibility. A more gen-
assumption of a stationary
 process, i. e., when the au- eral definition of first order coherence is the condition
tocorrelation function E (t)E + (t  ) depends only on that G (1) (x1 , x2 ) factorizes
the time difference = t t  , then
 G (1) (x1 , x2 ) = G (x1 )G(x2 ) , (78.59)
1  
S() = d E ()E + (0) ei + c.c . where G denotes some complex function. This definition
2 can be readily generalized to the nth order case. The nth
0
(78.54) order coherence applies when the relation
This relation between the spectrum and the first-order G (n) (x1 , . . . , x2n )
correlation function is known as WienerKhintchine
theorem. = G (x1 ) G (xn )G(xn+1 ) G(x2n ) (78.60)

holds. A field in a coherent state possesses nth order
coherence.
78.3.2 Photon Correlations
The Young experiment demonstrates the appearance
of first-order correlations. However, experiments that
can distinguish between the classical and quantum
domains have to be based on measurements of
second-order correlations. These experiments are of
the HanburyBrown and Twiss type, and determine
the arrival of a photon at detector position x and
time t and another photon at time t + . Following
the theory of Glauber, the second-order correlation
function
 
G (2) () = E (t)E (t + )E + (t + )E + (t) rive in bunches, an effect known as photon bunching.
(78.61)
is measured. In this formula we have omitted the variable
for the position x. Usually the normalized correlation
function
G (2) ()
g(2) () =  2 (78.62)
G (1) (0)
is introduced. The function g(2) is always positive, which
is true for classical as well as for quantum fields; but
there exists a purely quantum domain given by
0 g(2) (0) < 1 . (78.63)
78.4 Photodetection Theory
So far we have used a very simple theory of photode-
tection: any absorbed photon leads to a photoelectric
emission which can be observed. But in any real ex-
periment, these photons are counted over some time
interval T and the observed photoelectric emissions
are dominated by two statistics: (i) the statistics of
photoelectric emission which is also present for a clas-
sical field and (ii) the specific quantum statistics of
a quantized field. A detailed discussion of the quan-
tum theory of photoelectric detection has been given
by Kelley and Kleiner [78.19]. A central result is the
formula
 In  78.4.1 Homodyne
p(n, t, T) = : exp(I) : (78.66)
n!
for the probability of counting n photoelectrons in the
time interval from t to t + T . This photocounting distri-
Quantized Field Effects 78.4 Photodetection Theory 1147
For example, for a number state |n,
(2)
g|n (0) = 1 1/n , (78.64)
with n 1. In contrast, a coherent state | yields
(2)
g| (0) = 1.
78.3.3 Photon Bunching and Antibunching
In a realistic theory (but not in an oversimplified
one-mode model), the correlation function G (2) () al-
ways factorizes on a sufficiently long time scale, and
g(2) () 1. The photons are then no longer correlated,
and they arrive randomly as in the case of coherent light;
see for example (78.195).
If g(2) (0) > 1, the photons show a tendency to ar-
This effect has been observed for chaotic light. The op-
posite situation with 0 g(2) (0) < 1 demonstrates the
reverse effect, namely photon antibunching. As seen
from (78.63), this is a regime only accessible to non-
classical light. An example is given by the resonance
fluorescence of a two-level atom, treated in Sect. 78.13.
Note that we can rewrite g(2) (0) with the help of Man-
dels Q parameter
 
(2) a a aa
g (0) = = 1+ Q . (78.65)
a a2
Hence, a field state with Q < 0 shows the effect of
photon antibunching.
Part F 78.4
bution contains the integrated intensity operator
t+T
I = dt  E (t  )E + (t  ) (78.67)
t
containing the quantum efficiency of the detector. The
notation : : indicates a quantum average where the
operators have to be normally ordered and time or-
dered. This operator ordering reflects the process on
which a photodetector is based. It annihilates or ab-
sorbs photons, one after the other. A good treatment of
photoelectric detection can be found in [78.20].
and Heterodyne Detection
These detection methods allow the extraction of spe-
cific quantum features of a single mode quantum field,
1148 Part F Quantum Optics
the signal field. Figure 78.3 summarizes the principle of
optical homodyning. Two quantum fields described by
the annihilation operators a and b are mixed at a 50/50
beam splitter BS. Both fields have the same frequency.
The mode a represents the signal mode whose quantum
state is given by the density operator . Mode b serves
as a reference field, the local oscillator. The coherent
state | = ||| ei  determines the quantum state of the
local oscillator. Two ideal photodetectors 1 and 2 meas-
ure the number of photons in the output modes of the
beam splitter. For a highly excited coherent state, i. e.,
a classical local oscillator, the statistics of the photocur-
rent difference I can be described by the moments of
the signal mode operator
For example, the photocurrent difference
I X  = Tr(X ) (78.69)
is proportional to the expectation value of X . In par-
ticular, for = 0 and = 2 one is able to measure all
the moments of the two quadratures x and p (78.13) and
(78.14) of the signal mode. In general, the statistics of
the photocurrent I reveal the probability distribution
Pr(X ) = X ||X  (78.70)
of the observable X when the signal mode is in the
state . The states
 
1   |X  = 1/4 exp X 2 /2
X = a ei + a ei . (78.68)


2 1
Hn (X ) ein |n (78.71)
2 n n!
n=0

are eigenstates of the operator X , and are known as

I rotated quadrature states.
2
The heterodyne technique [78.21,22] relies on a sim-
ilar mixing of a signal field with a local oscillator at
BS a beam splitter, but this time the local oscillator fre-
a quency is offset by the intermediate frequency with
1
respect to the frequency 0 of the signal mode. Filters
select the beat frequency components in the photocur-
b rent of the detectors. This photocurrent contains the
quantum statistics of the two quadratures of the signal
Fig. 78.3 The principle of optical homodyning field [78.21, 22].
Part F 78.5

78.5 Quasi-Probability Distributions

 m
Quasi-probability distributions play an important role ordered product like an a , the order of a and a has
in quantum optics for three reasons. First, they changed. A generalized s-ordered product can be defined
are a complete representation of the density oper- as
ator of a quantum field. Second, they allow one     n  
m n m
to calculate expectation values in the spirit of a a
classical statistical physics. Third, they offer the s

possibility of converting a master equation for the 
D(, , s) (78.72)
density operator into an equivalent c-number partial = =0
differential equation. In this section, we relate a spe- with the generalized displacement operator
cific quasi-probability function to a specific operator
 
ordering. 1
D(, , s) exp a a + s . (78.73)
2
78.5.1 s-Ordered Operators
For s = 1 we find again normal ordering. The values
A normally s = 0 and s = 1 produce symmetric and antinormal or-
 ordered
m n product of a and a is a product of
the form a a : the annihilation operators a stand to dered products. As an example we note {a a}s = a a
the right of the creation operators a . In an antinormally (s 1)/2.
 Quantized Field Effects 78.5 Quasi-Probability Distributions 1149

Expectation values of those s-ordered products are generalized functions, such as delta functions and their
easily derived from the characteristic function derivatives, which have a highly singular character.
  The positive P-representation P(, ) for a nondiag-
(, , s) = Tr D(, , s) (78.74)
onal decomposition of a density operator is [78.27, 28]
via differentiation 
 n    | |
m  = d2 d2 P(, ) . (78.82)
(, , s)  |
= =0
The function P(, ) is a direct generalization of the
= {(a )n am }s  . (78.75)
GlauberSudarshan function P(), (78.78). P(, ) ex-
The Fourier transform of yields the quasi-probability ists for any physical density operator [78.27] and is
distribution of Cahill and Glauber [78.23] given by
  
1
W(, s) = 2 d2 (, , s) e , (78.76) 1 1
P(, ) = 2 exp | |2
4 4
   
where d2 = d Re() d Im(). With this distribution one 1  
 1
is able to calculate the expectation value of any s-ordered ( + )  ( + ) . (78.83)
2 2
operator product
     Here the state |1/2( + ) denotes a coherent state
a am
n
= ( )n m W(, s) d2 . (78.77) with the complex amplitude 1/2( + ). Note that
s P(, ) is always positive.
We concentrate now on three important quasi-
probability distributions, namely the cases s = 1, s = 0,
and s = 1, corresponding to the GlauberSudarshan 78.5.3 The Wigner Function
distribution, the Wigner function, and the Q function re-
spectively. A detailed discussion of these three functions This quasi-probability was first introduced by Wig-
can be found in [78.24]. ner [78.29] and may be defined as the distribution
function for a symmetrically ordered operator product
78.5.2 The P Function which is obtained in the case s = 0. The Wigner func-
tion plays an important role in other branches of physics,
The quasi-probability function P() was intro- such as quantum chaology, and in particular in any semi-
duced [78.25, 26] as a diagonal representation of the classical phenomenon when one considers the transition
density operator from quantum mechanics to classical mechanics.
 Consider the Wigner function of a quantum mechan-
= P()|| d2 (78.78) ical particle of position


Part F 78.5
in terms of coherent states |, (78.26). It is related to ~  
x= a + a , (78.84)
the CahillGlauber function W(, s) via 2m
P() = W(, s = 1) . and momentum
(78.79) 
The expectation value of any normally ordered operator m ~  
p=i a a . (78.85)
product 2

 m n  The Wigner function may be written in terms of position
a a = ( )m n P() d2 (78.80) and momentum variables
has a particularly simple form in terms of P(). 
1
From (78.78) the P function of a coherent state |0  W(x, p) = dy e2i py/~ x + y||x y ,
~
becomes
(2) (78.86)
P() = ( 0 ) (78.81)
    where |x y denotes position eigenstates. The s = 0
= Re() Re(0 ) Im() Im(0 ) .
CahillGlauber definition and the above definition of
Quantized fields for which the P function is positive do the Wigner function are related by
not show nonclassical effects such as squeezing and anti-  
1 mx + i p
bunching. For nonclassical states, such as number states W(x, p) = W = , s = 0 . (78.87)
or squeezed states, the P function only exists in terms of 2 2m ~
1150 Part F Quantum Optics

The position and momentum distributions of a particle, and

or equivalently the quadruture distributions in the case  
p V
of a quantized field mode, are + W(x, p)
m x x p
  

1 i~ r1 r V r W
Pr(x) = W(x, p) d p , (78.88) + =0. (78.95)
r! 2 x r pr
r=3,5,..
 The Wigner function has negative parts for most
Pr( p) = W(x, p) dx . (78.89) quantum states. For example, the Wigner function of
a Fock state |n,
Furthermore, the scalar product (1)n x2 p2  2 
 W|n (x, =
p) e L n 2x + 2 p2 ,

|1 |2 | = 2 ~
2
dx d p W|1  (x, p) (78.96)

W|2  (x, p) (78.90) clearly becomes negative due to the oscillating Laguerre
polynomial L n as shown in Fig. 78.4. Note thatwe have
of two quantum states is expressed by the phase space introduced the dimensionless position
x = m/~ x
overlap of the two corresponding Wigner functions.
and momentum p = 1/ m~ p.
Consequently, any Wigner function W(x, p) has to obey On the other hand, the Wigner function
the necessary condition
 1  2
W|,r (x, = exp e2r x 2 Re()
p)
dx d pW(x, p)W| (x, p) 0 (78.91)
 2 
e2r p 2 Im() (78.97)
for all W| representing a pure state. For a normalized
state |, of a squeezed state (78.35) is always positive as shown

 2 1 in Fig. 78.5. It is a long thin ellipse in phase space
dx d p W| (x, p) = . (78.92)
2 ~ (i. e. a Gaussian cigar). Concerning the negative parts
of the Wigner function, the Hudson theorem [78.30]
Instead of solving the Schrdinger equation for the
states that a necessary and sufficient condition for the
dynamics of a massive particle in a potential V(x), we
Wigner function of a pure state | to be nonnegative
can try to solve the equation
is that it can be described by a wave function of the
W p W
1  i~ r1 r V r W
= +
Part F 78.5

t m x r! 2 x r pr
r=1,3,..
(78.93)

for its Wigner function W(x, p, t). Note that here only W(x, p)
the odd derivatives of the potential V enter. This equation
is the quantum analogue of the classical Liouville equa- 0.2
tion, to which it reduces in the limit of ~ 0. However, 0.1
the initial distribution W(x, p, t = 0) has to be a Wigner 0
function in the sense of (78.86). p
Furthermore the Wigner function of an energy eigen-
2
function in the potential V(x) may be obtained from the
equations 2
0
 2 
p ~2 2 ~2 2 V 2 0
+ V(x) W(x, p) 2
2m 8m x 2 8 x 2 p2
 

1 i~ r r V r W
2

+ = EW(x, p) , x
r! 2 x r pr
r=4,6,.. Fig. 78.4 The Wigner function (78.96) of a Fock state
(78.94) |n = 4. The negative parts can be seen clearly
 Quantized Field Effects 78.6 Reservoir Theory 1151

78.5.4 The Q Function

The Q function is defined by the diagonal matrix elem-

ents
Q() = ||/ (78.101)
W(x, p)
of the density operator , where | denotes a coher-
0.4 ent state. The Q() function is always a positive and
0.2
bounded function, which exists for any density oper-
2 ator . The Q function is also known as Husimis
0 function. It allows one to calculate expectation values
0
p of antinormally ordered operator products of the form
2 1     
an a = d2 n ( )m Q() .
m
0 (78.102)
4
1
x Moreover, since the Q function corresponds to the case
Fig. 78.5 The Wigner function of a squeezed state |, r s = 1 of the CahillGlauber distribution,
with coherent amplitude = 1 and squeezing e2r = 0.25. Q() = W(, 1) . (78.103)
For these values the phase space ellipse is oriented along
the x-axis and squeezed in the p-direction. Note that this
function is positive everywhere 78.5.5 Relations Between
Quasi-Probabilities
form
  In general, the relation
1
x| = exp (ax 2 + bx + c) . (78.98)   
2 2 2| |2
W(, s) = d exp 
2
Here a, b and c denote some constants with Re(a) > 0. (s s) s s

Finally, the Kirkwood distribution function W(, s ) (78.104)

 holds between two CahillGlauber distributions with

1
K(x, p) = dy e2i py/~ x||x 2y (78.99) the parameters s > s. In particular, the non-negative
~ Q function

Part F 78.6


is a phase space function that resembles the Wigner 2
Q() = d2 exp[2| |2 ]W(, s = 0)
function. In the case of a pure state = || this
function reduces to (78.105)

p) eix p/~ ,
K(x, p) = (x)( (78.100) turns out to be a smoothed Wigner function W(, s = 0).
It is this smoothing process that washes out possible
p) denotes the Fourier transform of (x).
where ( negative parts in the Wigner function.

78.6 Reservoir Theory

Reservoir theory treats the interaction of one system
with a few degrees of freedom, called the system, with
another system with many degrees of freedom, called
the reservoir. A typical application of reservoir theory
is a microscopic theory of damping: the system inter-
acts with a reservoir, called the heat bath. The system
dissipates energy into the heat bath whereas the heat
bath introduces additional fluctuations to the system.
Since the present chapter focuses on quantized field
effects, the reservoir consists of the many modes of
the radiation field in free space. Such a reservoir is
modeled by a large number of independent harmonic
1152 Part F Quantum Optics
oscillators

   
1 bi  = bi = bi b j  = bi b j = 0 ,
Hr = ~i bi bi + , (78.106)
2  
i bi b j = n i ij .
where bi and bi
are the annihilation and creation
operators for the ith harmonic oscillator of the reser-
voir. For convenience the interaction with the system
is frequently approximated by a Hamiltonian of the
form


is the average number of photons at frequency i , T is
Hint = ~ gi A bi + gi A bi , (78.107)
i
where A is an operator of the small system and
gi is the coupling strength of this system to the
ith oscillator of the reservoir. For example, A may
be an annihilation operator if the system is a har-
monic oscillator or a Pauli spin matrix in the case of
a two-level atom coupled to the free space radiation
field.
Reservoir theory has important applications, and
a detailed discussion can be found in various books,
for example [78.17, 20, 27, 28, 3133].
78.6.1 Thermal Reservoir
The most commonly used reservoir is the thermal reser-
voir or thermal heat bath. Its characteristic properties
78.7 Master Equation
Part F 78.7
In quantum mechanics, density operators are used to
describe mixed states, and are discussed in Chapt. 7.
Here we introduce the concept of the reduced density
operator
s = Trr (sr ) , (78.111)
which is the density operator sr of the complete system
traced over the degrees of freedom of the reservoir. The
equation of motion for s in the Schrdinger picture
is
i the modes of the reservoir are spaced closely to-
s (t) = Trr {[Hsr , sr (t)]} . (78.112)
~ gether, so that the frequency i is a smooth function
In the BornMarkov approximation the trace over the
reservoir can be evaluated and leads to an equation of
motion for s which no longer contains reservoir opera-
tors. This equation of motion is usually called the master
are
 
(78.108)
(78.109)
Here
1
ni = (78.110)
exp (~i /kB T ) 1
the temperature of the reservoir, and kB denotes the
Boltzmann constant.
78.6.2 Squeezed Reservoir
Another example of a reservoir is a squeezed vac-
uum or squeezed reservoir. If, for example, multiwave
mixing is used to squeeze the radiation field, conju-
gate pairs of the reservoir operators b are correlated.
 
Therefore, the expectation values bi b j  and bi b j
may be nonvanishing. Apart from the average num-
ber n i of photons at frequency i , which  take
 into
account nonvanishing expectation values bi bi , addi-
tional complex squeezing parameters are needed to
describe the reservoir [78.28, 33, 34]. The characteri-
zation of a squeezed reservoir based on noise operators
is discussed in Sect. 78.10.
equation. The BornMarkov approximation consists of
two different parts:
1. Born approximation: The coupling to the reservoir
is assumed to be sufficiently weak to allow a per-
turbative treatment of the interaction between the
reservoir and the system.
2. Markov approximation: The correlations of the
reservoir are assumed to decay very rapidly on
a typical time scale of the system, or equiva-
lently, the reservoir has a very broad spectrum.
This approximation involves the assumption that
of i.
Since a general treatment is rather technical, we con-
sider two typical examples. A more general discussion
can be found in [78.17, 20, 27, 28, 3133]
 Quantized Field Effects 78.7 Master Equation 1153

78.7.1 Damped Harmonic Oscillator where g() denotes the coupling strength at fre-
quency . The number of thermal photons at frequency
The universally accepted Hamiltonian in nonrelativistic is
QED for a harmonic oscillator of frequency coupled 1
n= . (78.120)
to a reservoir consisting of a large number of harmonic exp (~/kB T ) 1
oscillators is given by the total Hamiltonian [78.35, 36]
 
  Thus the BornMarkov approximation replaces the dis-
1 1 crete reservoir modes by a continuum of modes with
Hsr = ~ a a + + ~i bi bi + a density D().
2 2
i
+ Hlc + Hsi , (78.113) Harmonic Oscillator in a Squeezed Bath
with the linear coupling term Within the BornMarkov approximation, the reduced





density operator (78.118) in the interaction picture sat-
Hlc = ~ gi a + a bi + bi , (78.114) isfies the master equation
i 1

= (n + 1) 2aa a a a a
and the self-interaction term 2

~ g2
2 1

+ n 2a a a a a a
Hsi = i
a + a . (78.115) 2
i 1

i
m 2a a a a a a
The approach used in quantum optics is to drop the term 2
1
Hsi and to make the rotating-wave approximation, that m (2aa a a a a) . (78.121)

is, to drop the terms a bi and a bi , see also Chapt. 68. 2
Then the approximate total Hamiltonian reads Here is again given by (78.119). The squeezed reser-
 
  voir is characterized by a real number n and a complex
1 1
Hsr = ~ a a + + ~i bi bi + number m. Physically n is the number of photons at
2
i
2 frequency , i. e., similar to the thermal reservoir, it


measures the average energy at frequency . The com-
+~ gi a bi + a bi . (78.116) plex number m determines the amount of squeezing. In
i general, the positivity of the density operator requires
Despite the problems with this approximate Hamiltonian |m|2 n (n + 1) . (78.122)
(see Sect. V.D of [78.35,36] for a discussion) we adopt it
in the present context because it leads to the widely used A more quantitatively definition of n and m in terms of
noise operators is given in Sect. 78.10.

Part F 78.7
master equation for the damped harmonic oscillator. We
consider two reservoirs: a thermal bath and a squeezed
bath. 78.7.2 Damped Two-Level Atom

Harmonic Oscillator in a Thermal Bath The interaction of a two-level atom with a classical
Within the BornMarkov approximation the master electromagnetic field is already discussed in Chapt. 68.
equation is For a quantum mechanical treatment of the field we

only have to replace the classical field by its quan-
1
= (n + 1) 2aa a a a a tum mechanical counterpart (78.9). We then find in the
2 rotating-wave approximation (Chapt. 68), that the dy-
1

namics of a two-level atom with a transition frequency

+ n 2a a a a a a , (78.117)
2 0 coupled to a reservoir consisting of a large number of
where harmonic oscillators is approximately described by the
total Hamiltonian
(t) = eia
a(tt
0) s (t) eia
a(tt
0) (78.118)
 
1 1
Hsr = ~0 z + ~i bi bi +
is the reduced density operator in the interaction picture. 2 2
i
The damping constant is given by

+~ gi bi + gi + bi , (78.123)
= 2D()|g()|2 , (78.119) i
1154 Part F Quantum Optics

where where
   
+ =
0 1
, =
0 0
, (78.124) (t) = ei0 z (tt0 )/2 s (t) ei0 z (tt0 )/2 (78.127)
0 0 1 0
  is the reduced density operator in the interaction picture,
z =
1 0
. (78.125)
and and n are given by (78.119) and (78.120).
0 1
Two-Level Atom in a Squeezed Bath
Again, two reservoirs are considered: a thermal bath and
Within the BornMarkov approximation, the reduced
a squeezed bath.
density operator in the interaction picture, (78.127),
satisfies the master equation
Two-Level Atom in a Thermal Bath
Within the BornMarkov approximation, the master 1
equation is = (n + 1) (2 + + + )
2
1 1
= (n + 1) (2 + + + ) + n (2+ + + )
2 2
1 m+ + m , (78.128)
+ n (2+ + + ) ,
2
(78.126) where , n and m have the same meaning as in (78.121).

78.8 Solution of the Master Equation

78.8.1 Damped Harmonic Oscillator If the system is initially in a superposition


We consider only a thermal reservoir and present the |(t0 ) = ci |i  (78.133)
solution of the master equation (78.117). For n = 0 it i
can be solved in terms of coherent states, see (78.26).
For n = 0 we give solutions in terms of quasi-probability of coherent states, the time evolution is given by
 
1

distributions.

(tt0 )
(t) = ci ck exp 1 e |i k |2

Coherent States 2
Part F 78.8

i,k
For n = 0, which is a good approximation for optical fre- 
 
exp i 1 e(tt0 ) Im i k
quencies, if the system is initially in a coherent state |0 
  
with a density operator i e (tt0 )/2 k e (tt0 )/2  . (78.134)
(t0 ) = |0 0 | , (78.129)
For (t t0 )  1, the interference terms |i k |, i = k
then there exists a simple analytical solution of the decay with an effective decay constant |i
master equation (78.117) k |2 /2. Thus the damping constant is modified by
(t) = |0 e (tt0 )/2 0 e (tt0 )/2 | . (78.130) the separation of the two coherent states in phase
A coherent state thus remains a coherent state with space.
an exponentially decaying amplitude 0 e (tt0 )/2 . Ac-
cording to (78.78) a general solution FokkerPlanck Equation
 A widely used procedure for solving the master equa-
  
(t) = d2 0 P(0 )0 e (tt0 )/2 0 e (tt0 )/2  tion for a damped harmonic oscillator, (78.117), or
for similar problems, is to derive an equation of mo-
(78.131) tion for the quasi-probability distributions W(, ; s)
can be constructed for an initial density operator defined in (78.76) from the master equation. The

operators a and a are replaced by appropriate differ-
(t0 ) = d2 0 P(0 )|0 0 | . (78.132)
ential operators. The substitution rules can be derived
 Quantized Field Effects 78.8 Solution of the Master Equation 1155

from (78.73), (78.74) and (78.76) and are The time-dependent solution of this FokkerPlanck
  equation has the form

k s+1 k
a a  
2 W(, , t; s) = G(, , t| ,  , t  ; s)
 
s1 
W, W( ,  , t  ; s) d2  , (78.138)
2

k  
 s+1  where
a a
2
  G(, , t|  ,  , t  ; s)
s1 k    
W,   e (tt  )/2 2
2
  exp   
s1 n s 1 e(tt )
aa =
(78.139)
2 n s 1 e(tt )

 
s1
W,
2 is the Greens function of the FokkerPlanck equation
 
s+1 (78.136). The steady-state solution is
a a
2 1 ||2 /n s
  W(, , t ; s) = e ,
s+1 n s
(78.140)
W,
2
  which is the distribution function of a harmonic os-
s+1
a a cillator in thermal equilibrium with a reservoir of
2 temperature T .
 
s 1
W,
2 78.8.2 Damped Two-Level Atom
 
s1
aa The density operator
2
   
s+1 ee eg
W. (78.135) =
2 ge gg
(78.141)

Part F 78.8
In general, this procedure leads to equations of motion
for a two-level atom can be written as
which involve higher derivatives of W as exemplified
   
by the quantum mechanical Liouville equation (78.93) 1
1 + z   
for the Wigner function. For simple Hamiltonians, how- = 2   . (78.142)
ever, this equation has the form of a FokkerPlanck +  2 1 z 
1

equation which is well known in classical stochastic

Thus, a two-level atom is completely described by the
problems [78.27, 37] (Sect. 78.10.2). In particular, for
expectation values
a damped harmonic oscillator described by the master
equation (78.117), one obtains
z  = ee gg ,
 
W   2 W +  = ge ,
= (W) + W + n s ,
t 2   = eg . (78.143)
(78.136)
Hence the master equation (78.128) can be cast
where into the equations of motions for these expectation
1s 1 1s values
ns = n + = + .  
2 exp (~/kB T ) 1 2 d 1
(78.137) +  = n + +  m   ,
dt 2
1156 Part F Quantum Optics

 
d 1 which can easily be solved for arbitrary ini-
  = n +   m+  , tial conditions. In contrast to a thermal reservoir
dt 2
  (m = 0), a squeezed reservoir results
d 1  in two differ-
z  = 2 n + z  , (78.144) ent
dt 2 n + 12 |m| [78.34].
78.9 Quantum Regression Hypothesis
In the Schrdinger picture, time-dependent expectation
values for system operators A j can be calculated from
the reduced density operator s (t) via
 
A j  = Trs A j s (t) . (78.145)
The reduced density operator, however, is not suffi-
cient to calculate two-time correlation functions such
as A j (t + )Ak (t). For a definition of two-time cor-
relation functions, the Heisenberg picture is more
appropriate. Here, expectation values follow from
 
A j  = Trsr Usr (t, t0 )A j (t0 )Usr (t, t0 )sr (t0 ) ,
(78.146)
where Usr (t, t0 ) describes the unitary time evolution of
the complete system and sr (t0 ) is the density operator in
the Heisenberg picture. Similarly, two-time correlation
functions such as A j (t + )Ak (t) can be defined as
A j (t + )Ak (t)
 pothesis asserts that two-time correlation functions
= Trsr Usr (t + , t0 )A j (t0 )Usr (t + , t0 )

Usr (t, t0 )Ak (t0 )Usr (t, t0 )sr (t0 ) . (78.147)
The quantum regression hypothesis avoids the calcula-
tion of Usr (t, t0 ). Two equivalent formulations exist, one
Part F 78.9
based on the master equation for s and another based on
the equation of motion for the expectation values A j ,
see for example [78.20, 27, 28, 31, 33].
78.9.1 Two-Time Correlation Functions
and Master Equation
It follows from their definition (78.147) in the
Heisenberg picture that two-time correlation functions
A j (t + )Ak (t) for system operators A j and Ak can be
calculated with the help of the operator
 Trs A j Rs (t + , t) is an expectation value for which
Rs (t + , t) = Trr Usr (t + , t)

Ak sr (t)Usr (t + , t) , (78.148)
where Usr (t + , t) describes the unitary time evolution
of the complete system between t and t + . We find
 
A j (t + )Ak (t) = Trs A j Rs (t + , t) . (78.149)
 transverse decay constants n + 12 + |m| and
Note, that in (78.148) and (78.149) we interpret A j and
Ak as operators in the Schrdinger picture and have
omitted the argument t0 . Because the reduced density
operator
 
s (t) = Trr Usr (t, t0 )sr (t0 )Usr (t, t0 ) (78.150)
satisfies the master equation, it is plausible to assume,
that when the time derivative is taken with respect to ,
the operator Rs (t + , t) also satisfies the master equation
for s , subject to the initial condition Rs (t, t) = Ak s (t).
However, this requires the additional assumption that
the approximations made in the derivation of the master
equation for s (t) are also valid for Rs (t + , t).
78.9.2 Two-Time Correlation Functions
and Expectation Values
A second formulation of the quantum regression hy-
A j (t + )Ak (t) obey
A j (t + )Ak (t)


= G j ()A (t + )Ak (t) , (78.151)

provided that the expectation values of a set of system
operators A j satisfy


A j (t) = G j (t)A (t) . (78.152)
t

This is the form of the quantum regression hypothesis
that was first formulated by Lax [78.38].
The equivalence of the two formulations fol-
lows from the interpretation of Rs (t + , t) on the
right side of (78.149)
 as a density operator. Then
we assume that (78.152) is valid; i. e.,
Trs A j Rs (t + , t)


 
= G j ()Trs A Rs (t + , t) . (78.153)

According to (78.149), this is identical to (78.151).
 Quantized Field Effects 78.10 Quantum Noise Operators 1157

78.10 Quantum Noise Operators

The master equation is based on the Schrdinger picture
in quantum mechanics: the state of the system described
by a density operator is time-dependent, whereas opera-
tors corresponding to observables are time independent.
If we use the Heisenberg picture instead and make
similar approximations as in the derivation of the mas-
ter equation, we arrive at equations of motion for the
Heisenberg operators, see for example [78.17,28,31,32].
Due to the interaction with a reservoir these equations
have additional noise terms and damping terms.
78.10.1 Quantum Langevin Equations
Again consider a damped harmonic oscillator. The equa-
tion of motion for the annihilation operator
= ei(tt0 ) a(t)
a(t) (78.154)
in the interaction picture follows from the Heisenberg
equations for the operators a, a , bi and bi and reads

t
da 
= |gi |2
ei(i )(tt ) a(t  ) dt 
dt
i t0

field. Here one can have quality factors of the cavity
i gi ei(i )(tt0 ) bi (t0 ) . (78.155)
i
In general, the noise operator

al frequencies
F(t) = i gi ei(i )(tt0 ) bi (t0 ) (78.156)
i one can have temperatures as low as T 3 mK and
is unphysical
 since
 it does not preserve the commutation
relation a, a = 1. It is the noise term which saves the
commutation relation.
For a thermal reservoir with a sufficiently small
correlation time, the standard derivations [78.32] give
     
F(t) = F (t) = F(t)F(t  ) = F (t)F (t  ) = 0 ,
 
F (t)F(t  ) = n(t t  ) ,
F(t)F (t  ) = (n + 1)(t t  ) , (78.159)
where the averages are taken over the reservoir. The
damping constant and the number of thermal pho-
tons n are given in (78.119) and (78.120). For more
general relations, see [78.35, 36], where it is shown
explicitly that correlation functions involving the fluc-
tuation force do not in fact depend on the oscillator
frequency. The condition of a sufficiently small reser-
voir correlation time requires that c ~/(kB T ) is small
compared with the time scales of the systems. The
only time scale in (78.157) is 1 . The relevant con-
dition is therefore c  1 . For typical applications
in quantum optics, a is the annihilation operator and
a is the creation operator of a single-mode cavity
on the order of Q = / 106 . In terms of the qual-
ity factor, the condition of sufficiently small reservoir
correlation times  requires ~/(kB T )  Q. For optic-
3 1015 Hz) and T 300 K one has
~/(kB T ) 75. In the microwave regime ( 30 GHz)

Part F 78.10
is not delta-correlated, and there are also memory ef- still have ~/(kB T ) 75. Therefore the assumption
fects in (78.155). The noise operator F(t) can be used to of delta-correlated noise is a good approximation for
classify the reservoir: if it is delta-correlated, that is, if typical applications in quantum optics.
the reservoir has a very broad spectrum, one speaks of Similarly, for a squeezed reservoir one has
white noise, see below. If the correlation time is finite  
F(t) = F (t) = 0 ,
so that there are memory effects, one speaks of colored
noise. F(t)F(t  ) = m(t t  ) ,
 
If the spectrum of the noise is very broad (as in F (t)F (t  ) = m (t t  ) ,
the derivation of the master equation for the reduced  
F (t)F(t  ) = n(t t  ) ,
satisfies the quantum
density operator), the operator a(t)  
Langevin equation F(t)F (t  ) = (n + 1)(t t  ) , (78.160)

da which gives a quantitative definition of the parameters

= a(t)
+ F(t) , (78.157)
n and m in the master equations (78.121) and (78.128).
dt 2
Again, a detailed discussion in [78.35, 36] shows that
with a damping term a(t)/2
and a noise term F(t).
correlation functions involving the fluctuation forces do
Note that a simple damping equation such as
not depend on the oscillator frequency.
The Langevin equation (78.157) is based on the
= a(t)
a(t) (78.158)
2 use of the approximate Hamiltonian given in (78.116),
1158 Part F Quantum Optics

i. e., it is based on the rotating-wave approximation and Here the last term is a RiemannStieltjes integral defined
the neglect of self-interaction terms. The correspond-
 by
ing Langevin equation for x = ~/(2m) a + a may t
be calculated and, not unexpectedly, it disagrees with  
the AbrahamLorentz equation which Ford and OCon- h x(t  ), t  dW(t  )
nell [78.39] showed could be derived systematically t0
using the exact Hamiltonian (78.113). In fact, Ford

n1
 
and OConnell showed that an improved equation for = lim h[x(i ), i ] W(ti+1 ) W(ti ) ,
n
the radiating electron (improved in the sense that it i=0
is second-order and is not subject to the analyticity (78.164)
problems and the problems with runaway solutions
associated with the AbrahamLorentz equation) may where i is in the interval (ti , ti+1 ).
be obtained by generalizing the Hamiltonian (78.113) There are two different approaches to such problems:
to include electron structure. The implications follow- the Ito approach and the Stratonovich approach. They
ing from these different equations are presently under differ in the definition of stochastic integrals.
study.  In the Stratonovich approach, one evaluates
h x(i ), i at i = (ti + ti+1
 )/2, whereas
 in the Ito
78.10.2 Stochastic Differential Equations approach one evaluates h (x(i ), i ) at i = ti . This
slightly different definition of i leads to different results
In Sect. 78.10.1 we discussed one of the simplest quan- because, as a consequence of the delta-correlated noise
tum systems with dissipation, the damped harmonic term, x(t) is not a continuous path. However, there is a re-
oscillator. For more complicated systems the noise term lation between the solution of a Stratonovich stochastic
can also contain system operators. In such cases there are differential equation and an Ito stochastic differential
two different ways to interpret the Langevin equation. equation. Suppose x(t) is a solution of the Stratonovich
In order to give a feeling for the two possible interpreta- stochastic differential equation
tions, we discuss the one dimensional classical Langevin dx(t) = g(x, t) dt + h(x, t) dW(t) . (78.165)
equation
Then x(t) satisfies the Ito stochastic differential equa-
dx tion
= g(x, t) + h(x, t)F(t) (78.161)  
dt 1 h(x, t)
dx(t) = g(x, t) + h(x, t) dt
for the stochastic variable x(t) with delta-correlated 2 x
noise F(t)F(t  ) = (t t  ). Due to the singular na- + h(x, t) dW(t) . (78.166)
Part F 78.10

ture of delta-correlated noise, such a Langevin equation

does not exist from a strictly mathematical point of Instead of dealing with stochastic differential
view. A mathematically more rigorous treatment is based equations, one can derive a FokkerPlanck equation
on stochastic differential equations [78.27, 28, 37]. The for the conditional probability P(x, t|x0 , t0 ). For the
variable x(t) is said to obey a stochastic differential Stratonovich stochastic differential equation (78.165),
equation the FokkerPlanck equation is
 
P 1 h(x, t)
dx(t) = g(x, t) dt + h(x, t)F(t) dt = g(x, t) + h(x, t) P
t x 2 x
= g(x, t) dt + h(x, t) dW(t) , (78.162)
1 2 2
+ h (x, t)P , (78.167)
if, for all times t and t0 , x(t) is given by 2 x 2
which takes the form
t
x(t) = x(t0 ) + g(x(t  ), t  ) dt  P 1 2 2
= g(x, t)P + h (x, t)P , (78.168)
t0
t x 2 x 2
t if (78.165) is interpreted as a stochastic differential
equation in the Ito sense.
+ h[x(t  ), t  ] dW(t  ) . (78.163)
The two approaches have the following properties:
t0 (i) in most of the models used in physics the Stratonovich
 Quantized Field Effects 78.12 Spontaneous Emission in Free Space 1159

definition of a stochastic integral is needed to give Langevin equations with h(x, t) = const, as in (78.157),
correct results, (ii) rules from ordinary calculus are ap- the Stratonovich interpretation and the Ito interpretation
plicable only in the Stratonovich approach, and (iii) for of stochastic integrals are equivalent.

78.11 Quantum Monte Carlo Formalism

The quantum Monte Carlo formalism was developed 2. The system makes a jump; i. e.
to solve numerically master equations of the Lindblad
type [78.20, 4044] C|(t)
|(t + dt) = . (78.172)
i (t)|C C|(t)
s = [Hs , s ]
~
1

Since both possibilities describe a nonunitary time evo-

+ 2C j s C j C j C j s s C j C j .
2
j
lution, | must be normalized after each step. For each
(78.169) time interval dt one of these two possibilities is randomly
Here C j are arbitrary system operators. chosen according to the probability
As an illustrative example, consider
i 1
P(t)dt = (t)|C C|(t) dt (78.173)
s = [Hs ,s ]+ 2Cs C C Cs s C C ,
~ 2 to make a jump between t and t + dt. We can now define
(78.170)
a density operator
where C is an arbitrary system operator. Instead of solv-
ing the master equation, one defines quantum trajectories s (t) = |(t)(t)| (78.174)
or stochastic wave functions as follows. Starting from for a specific quantum trajectory |(t). The density
|(t), there are two possibilities for the time evolution operator
during the interval dt:
1. The system evolves according to the non-Hermitian s (t) = |(t)(t)| (78.175)
Hamiltonian Hs i2~ C C; i. e.
   
1i dt Hs i2~ C C /~ |(t)
|(t+ dt) = .
1 (t)|C C|(t) dt
(78.171)
78.12 Spontaneous Emission in Free Space
Consider an atom which is initially in one of its excited
states and which interacts with the quantized electro-
magnetic field of free space. Even if none of the modes
of the electromagnetic field is excited, there are still the
vacuum fluctuations which interact with the atom and
give rise to important effects:
1. Spontaneous emission: the atom spontaneously
emits a photon and decays from the excited state.
2. Natural linewidth: due to the finite lifetime of the
atomic levels, the radiation from an atomic transition
has a finite linewidth, called the natural linewidth.
3. Lamb shift: the energy levels of the atom are shifted.
The standard theory of spontaneous emission is the
WignerWeisskopf theory [78.17, 32]. Here an initially
averaged over all trajectories (indicated by the bar) is
then a solution of the master equation (78.170).
This method can easily be generalized to master
equations of the form (78.169).
Part F 78.12
excited atomic state | decays exponentially according
to
|c (t)|2 = e t , (78.176)
where the decay constant  is given by

3 |di |2
i
 = , (78.177)
30 ~c3
i
and the sum is over all atomic states with an en-
ergy E i lower than the energy E  of the state |.
i = (E  E i )/~ is the transition frequency for the
transition | |i, and di = e|r|i is the correspond-
ing dipole moment.
The same decay constant  is also observed as
a linewidth in the spectrum of the radiation scattered
1160 Part F Quantum Optics

by an atom when the incoming photon excites the atom
to the level |.
The energy level shift is more troublesome and needs
the concept of mass renormalization, a standard problem
in quantum electrodynamics. The theory and results are
discussed in Chapt. 27 and Chapt. 28.
Recent calculations of Pachucki [78.45] based on
fully relativistic quantum electrodynamics and includ-
ing two-loop corrections predict 1 057 838(6) kHz
for the energy difference between the 2s 1/2 -
state and the 2p 1/2 -state which is in excel-
lent agreement with the experimental result of
1 057 839(12) kHz [78.46].
For a discussion of energy levels and transition fre-
quencies in hydrogen and deuterium atoms see also
Sect. 28.3.

78.13 Resonance Fluorescence

Consider a two-level atom driven by a continuous where a resonant driving term has been added to the
monochromatic wave which is treated classically. The Hamiltonian (78.123). Here d is the projection of the
excited state of the atom can decay by spontaneous emis- dipole matrix element eg|r|e onto the polarization
sion into vacuum modes of the electromagnetic field. vector of the driving field with an amplitude E. The cor-
This emission is called resonance fluorescence. Of par- responding master equation in the interaction picture is
ticular interest are the properties of the emitted light. For 1  
a detailed discussion of resonance fluorescence, see for = i 1 + + ,
2
example [78.20, 3133]. 1
The far field at position R emitted by an atom at the + (2 + + + ) , (78.181)
2
origin is proportional to its dipole moment and can be where 1 = E d /~ is the Rabi frequency associated
expressed in terms of the dipole operators + and with the driving field. The vacuum modes of the field
according to the relation [78.20] are described by a thermal reservoir at zero temperature.
2 (d R) R The equations of motion for the expectation val-
E+ (R, t) = 0 (t r/c) , ues + ,  , and z  are the optical Bloch equations
40 c2 R3
with radiative damping (Chapt. 68) and are
2 (d R) R 1
E (R, t) = 0 + (t r/c) , d
+  = +  i z  ,
40 c2 R3 dt 2 2
(78.178) d 1
where   =   + i z  ,
dt 2 2
d = eg|r|e (78.179) d
z  = (z  + 1) i1 (+   ) .
Part F 78.13

is the atomic dipole matrix element and the field op- dt

(78.182)
erators E+ (R, t) and E (R, t) as well as the dipole These expectation values determine the density opera-
operators + (t) and (t) are in the Heisenberg picture. tor (78.142) of the two-level atom. Because (78.182)
Knowledge of the operators + (t) and (t) is therefore are a system of linear differential equations for + ,
sufficient to study the properties of the emitted light in  , and z , they can be solved analytically. Further-
the far field. more, the quantum regression hypothesis allows one
to calculate two-time correlation functions as shown
78.13.1 Equations of Motion in Sect. 78.9.
The total Hamiltonian for the system reads

  78.13.2 Intensity of Emitted Light
1 1
Hsr = ~0 z + ~i bi bi +
2 2 According to (78.52)) and (78.178), the intensity of the
i



fluorescence light at position R is given by
+~ gi bi + gi + bi I = E (R, t)E + (R, t) +  , (78.183)
i
1
and can be decomposed into two parts: the coherent
d E ei0 t + d E + ei0 t , intensity
2
(78.180) Icoh +   (78.184)
 Quantized Field Effects 78.13 Resonance Fluorescence 1161

originating from the mean motion of the dipole, and the The central peak at = 0 has a width of /2 whereas
incoherent intensity the width of the two side peaks at = 0 1 is 3/4.
Iinc +  +   , (78.185)
Their heights are one third of the height of the central
peak. This spectrum was predicted by Burshtein [78.47]
which is due to fluctuations of the dipole motion around and Mollow [78.48] and experimentally confirmed by
its average value. The steady state intensities are Schuda et al. [78.49], Wu et al. [78.50], and Hartig
12 2 et al. [78.51].
Icoh  2 , (78.186) This triplet can be explained in terms of the dressed
2 + 212 states |1, n and |2, n introduced in Chapt. 68 (78.50).
and If the driving field is resonant with the atomic transition,
214 these states have the energies
Iinc  2 . (78.187)  
2 + 212 1
E 1,n = ~ n + 0 ~ Rn ,
For weak laser intensities (1 small) the intensity of 2
 
the fluorescence light is dominated by the coherent part 1
E 2,n = ~ n + 0 ~ Rn , (78.191)
whereas for high intensities (1 large) it is dominated 2
by the incoherent part.
(78.52). The energy differences between the allowed
78.13.3 Spectrum of the Fluorescence Light transitions are

The WienerKhintchine theorem (78.54) allows one to E 2,n E 2,n1 = ~0 + ~ Rn ~ Rn1 ~0 ,

express the steady state spectrum of the fluorescence E 2,n E 1,n1 = ~0 + ~ Rn + ~ Rn1 ~0 + ~1 ,
light as the Fourier transform of the correlation function
E 1,n E 2,n1 = ~0 ~ Rn ~ Rn1 ~0 ~1 ,
+ () (0)ss in the form
 E 1,n E 1,n1 = ~0 + ~ Rn ~ Rn1 ~0 ,
1 (78.192)
S() = ei E ()E + (0)ss d + c.c.
2
0 where we have made the approximations

1 Rn Rn1 0 ,
ei(0 ) + () (0)ss d + c.c.
2 Rn + Rn+1 2g n + 1 1 . (78.193)
0
(78.188) This is a good approximation for an intense driv-

Part F 78.13
Again it consists of two contributions: a coherent part ing field which can approximated by a highly
Scoh (), and an incoherent part Sinc (). The coherent excited coherent state with an average photon num-
part is
ber n.
Figure 78.6 shows these energy levels and the al-
12 2
Scoh ()  2 ( 0 ) . (78.189) lowed transition. Obviously, the transitions correspond
2 + 212 to frequencies 0 , 0 1 and 0 + 1 . The dressed
The incoherent part of the fluorescence light has two
qualitatively different spectra. For 1 < /4, it has a sin- 2, n
gle peak at 0 , whereas it consists of three peaks for 1
1 > /4. For 1  /4 it is given by 1, n

1 (/2)2
Sinc () 0 0 0 1 0 + 1
2 ( 0 )2 + (/2)2
1 (3/4)2 2, n 1
+
3 ( 0 + 1 )2 + (3/4)2 1

1 (3/4)2 1, n 1
+ .
3 ( 0 1 )2 + (3/4)2 Fig. 78.6 Energy level diagram of dressed states. The tran-
(78.190) sition frequencies are 0 , 0 1 and 0 + 1
1162 Part F Quantum Optics
state picture also explains the 2:1 ratio for the inte-
grated intensities of the central peak and the side peak
in (78.190).
78.13.4 Photon Correlations
In addition to the spectrum which is based on the cor-
relation function E ()E + (0)ss in Sect. 78.13.3, the
second-order correlation function
G (2) + +
ss () = E (0)E ()E ()E (0)ss For = 0, g(2) (0) = 0, indicating a tendency of photons
+ (0)+ () () (0)ss (78.194) to be separated. This tendency is known as photon an-
can be measured to gain more insight into the flu-
orescence light, see also Sect. 78.3.2. Experimentally
this is done by measuring the joint probability for
detecting a photon at time t = 0 and a subsequent
photon at time t = . Again the result can be ob-
tained from the quantum regression hypothesis and
78.14 Recent Developments
This chapter has discussed the fundamentals of the quan-
tized electromagnetic field and applications to the broad
area of quantum optics. However, in the last eight years,
quantum optics has blossomed in several new direc-
tions particularly in the key role it is playing in recent
investigations of the fundamentals of quantum theory
and related applications. In particular, the superposition
principle (the bedrock of quantum mechanics), entan-
Part F 78.14
glement, the quantum-classical interface, and precision
measurements have become very topical research ar-
eas, especially in respect to their relevance to quantum
information processing.
78.14.1 Literature
During the last eight years, several books on quantum
optics [78.5662] have been published. These books
cover the topics of this chapter to some extent and take
into account recent developments. For an introduction
to the rapidly evolving fields of quantum information
processing, we refer the reader to Chapt. 81 and [78.63
65].
78.14.2 Field States
Recently, number states of the radiation field were ob-
served in a cavity-QED experiment [78.66].
reads
G (2)
ss ()
g(2) () =
|G (1)
ss (0)|
2
 
3
= 1 e3/4 cos + sin ,
4
(78.195)
where is given by
!
= 12 2 /4 . (78.196)
tibunching and was first predicted by Carmichael and
Walls [78.52, 53] and experimentally verified by Kim-
ble et al. [78.54, 55]. Photon antibunching of radiation
emitted from a two-level atom has a simple explanation:
After the atom has emitted a photon it is in the ground
state and must first be excited again before it can emit
another photon.
78.14.3 Reservoir Theory
New research topics, such as quantum information
processing, rely on the superposition principle and en-
tangled quantum states. Since these states are very
sensitive to decoherence, reservoir theory has attracted
a lot of interest in recent years. Furthermore, as discussed
in [78.6772], decoherence is the physical process by
which the classical world emerges from its quantum
underpinning.
Many investigations in this area involve the presence
of a reservoir (heat-bath/environment) and master equa-
tions are a ubiquitous tool. The familiar master equations
of quantum optics are in Lindblad form [78.73], which
guarantees that the density matrix is always positive
definite during time evolution. In the derivation of this
equation [78.74, 75], rapidly oscillating terms are omit-
ted by the method of coarse-graining in time; the high
frequencies correspond to the oscillator frequency 0
and, in the usual weak coupling limit, 0  , where
is a typical decay constant. This is the rotating wave
approximation (Sect. 66.3.2).
We have referred to the equations obtained prior to
coarse-graining in time as pre-master (or pre-Lindblad)
equations [78.74,76], and such equations have been used
extensively in other areas of physics [78.77, 78]; other
authors have simply referred to them as master equations

but, to avoid confusion, we reserve the latter term for
equations in Lindblad form. Pre-master equations, like
the master equations, describe an approach to the equi-
librium state. This equilibrium state is the same in either
case [78.76], but with pre-master (non-Lindblad) equa-
tions the approach can be through non-physical states of
negative probability. However, as recently demonstrated,
pre-master equations have other advantages vis a` vis
master equations:
(a) they lead to the exact expression for the mean value
of x(t) (as obtained from the exact Langevin equation
for the problem);
(b) they lead, in the classical limit (~ 0), to the famil-
iar FokkerPlanck equation of classical probability; and
(c) the exact master equation [78.7983] is for long
times of pre-master form. However, the general expec-
tation (based on the time dependence of the coefficients)
that the exact master equation preserves positivity for
all times has not been realized since Ford and OCon-
nell have recently shown that, even in high temperature
regime, the density matrix is not necessarily posi-
tive [78.84].
In traditional
Quantized Field Effects References 1163
particular by the desire to study decoherence and other
short time phenomena, an exact master equation was de-
rived [78.85] to study the non-Markovian dynamics of
a two-level atom interacting with the electromagnetic
field.
In addition, motivated by the desire to study a driven
oscillator, the usual two-level atom master equation was
generalized to include the case of an external force
field [78.86, 87]. This generalized equation was then
used not only to obtain the familiar zero-temperature
BurshteinMollow spectrum, but also the corresponding
high temperature results. For strong resonant driving at
high temperature, the same three-peaked structure was
observed in the zero temperature case, but a much larger
width was found. The analysis, following other investi-
gations, used the Lax formula for calculating two-time
correlation functions. This formula is not a quan-
tum regression theorem as it is often designated (see
also Sect. 78.9), but simply an approximation (which
more resembles an Onsager classical regression theo-
rem [78.88]) which works very well in the case of weak
coupling and for frequencies near a resonant frequency,
 quantum optics, the emphasis has been but not otherwise [78.86, 87].
on long-time t  1 phenomena, for which the use of In Sect. 78.6, we stressed the usefulness of quasi-
either master or pre-master equations is justified. How- probability distributions instead of the density matrix,
ever,
 they are both inadequate for dealing with short-time
t  1 phenomena (as can be shown most simply by
calculating the mean-square displacement, a key ingre-
dient in decoherence calculations), which are of much
recent interest. Thus, it is desirable to use exact mas-
ter equations. In that respect, the exact master equation
of Hu et al. [78.79, 80] for an oscillator is an arbitrary
dissipative environment has proved to be a popular and
useful tool for which an exact solution has now been
obtained [78.83]. However, it should also be mentioned
that the solution of the initial value quantum Langevin
equation gives all the same information as the exact mas-
ter equation, and in fact, the solutions of the former were
used to obtain the solutions of the latter [78.83].
The familiar two-level atom master equation is, of
course, similar in form to the usual Lindblad-type mas-
ter equation for the oscillator. However, motivated in
References
78.1 C. Cohen-Tannoudji, J. Dupont-Roc, G. Grynberg:
Photons, and Atoms. An Introduction to Quantum
Electrodynamics (Wiley, New York 1989)
78.2 L. Mandel: Opt. Lett. 4, 205 (1979)
with particular attention to the Wigner distribution. In
particular, for simple Hamiltonians, we pointed out that
the equation for the corresponding Wigner function has
the form of a FokkerPlanck equation and we con-
sidered the explicit form describing the usual master
equation. The more general equations associated with
an exact master equation and their solution was the sub-
Part F 78
ject of [78.83] and interesting limits of that equation,
including the pre-master equation for both momentum
coupling and coordinate coupling were discussed at
length in [78.89, 90]. In the case of two-level systems,
it is not convenient to use quasi-probability distribu-
tions; instead, it is found that the preferred tool is the
polarization vector [78.91]. Surprisingly, it has not been
generally adopted by the quantum optics community
although its usefulness in that context has been demon-
strated recently in [78.86, 87].
78.3 W. Schleich, M. Pernigo, Fam Le Kien: Phys. Rev. A
44, 2172 (1991)
78.4 H. J. Kimble, D. F. Walls (Eds.): J. Opt. Soc. Am. B
4(10), 14531737 (1987)
1164 Part F Quantum Optics

78.5 P. Knight, R. London (Eds.): J. Mod. Opt. 34(6) (1987) D. Han, Y.-S. Kim, V. I. Manko (NASA, Maryland,
78.6 E. Giacobino, C. Fabre (Eds.): Appl. Phys. B 55(3) USA 1993)
(1992) 78.37 H. Risken: The Fokker-Planck Equation (Springer,
78.7 W. Schleich, J. A. Wheeler: J. Opt. Soc. Am. B 4, 1715 Berlin, Heidelberg 1989)
(1987) 78.38 M. Lax: Fluctuations and Coherence Phenomena
78.8 H. P. Yuen: Phys. Rev. A 13, 2226 (1976) in Classical and Quantum Physics. In: Brandeis
78.9 R. E. Slusher, L. W. Hollberg, B. Yurke, J. C. Mertz, University Summer Institute Lectures, Vol. 2, ed.
J. F. Valley: Phys. Rev. Lett. 55, 2409 (1985) by M. Chretin, E. P. Gross, S. Deser (Gordon and
78.10 L. A. Wu, M. Xiao, H. J. Kimble: J. Opt. Soc. Am. B 4, Breach, New York 1966)
1465 (1987) 78.39 G. W. Ford, R. F. OConnell: Phys. Lett. A 157, 217
78.11 A. Heidmann, R. J. Horowicz, S. Reynaud, E. Gia- (1991)
cobino, C. Fabre: Phys. Rev. Lett. 59, 2555 (1987) 78.40 J. Dalibard, Y. Castin, K. Mlmer: Phys. Rev. Lett.
78.12 W. P. Schleich, S. M. Burnett (Eds.): Special issue on 68, 580 (1992)
Quantum Phase and Phase Dependent Measure- 78.41 K. Mlmer, Y. Castin, J. Dalibard: J. Opt. Soc. Am B
ments, Phys. Scr.T. T48 (1993) 10, 524 (1993)
78.13 P. Carruthers, M. M. Nieto: Rev. Mod. Phys. 40, 411 78.42 R. Dum, P. Zoller, H. Ritsch: Phys. Rev. A 45, 4879
(1968) (1992)
78.14 D. T. Pegg, S. M. Barnett: Phys. Rev. A 39, 1665 78.43 C. W. Gardiner, A. S. Parkins, P. Zoller: Phys. Rev. A
(1989) 46, 4363 (1992)
78.15 J. W. Noh, A. Fougres, L. Mandel: Phys. Rev. A 45, 78.44 R. Dum, A. S. Parkins, P. Zoller, C. W. Gardiner:
424 (1992) Phys. Rev. A 46, 4382 (1992)
78.16 R. J. Glauber: Phys. Rev. 130, 2529 (1963) 78.45 K. Pachucki: Phys. Rev. Lett. 72, 3154 (1994)
78.17 W. H. Louisell: Quantum Statistical Properties of 78.46 E. W. Hagley, F. M. Pipkin: Phys. Rev. Lett. 72, 1172
Radiation (Wiley, New York 1973) (1994)
78.18 R. Hanbury Brown, R. Q. Twiss: Nature 177, 27 (1956) 78.47 A. I. Burshtein: Sov. Phys. JETP 22, 939 (1966)
78.19 P. L. Kelley, W. H. Kleiner: Phys. Rev. 136, 316 (1964) 78.48 B. R. Mollow: Phys. Rev. 188, 1969 (1969)
78.20 H. Carmichael: An Open Systems Approach to Quan- 78.49 F. Schuda, C. R. Straud, Jr., M. Hercher: J. Phys. B
tum Optics (Springer, Berlin, Heidelberg 1993) 7, L198 (1974)
78.21 H. Yuen, H. P. Shapiro: IEEE Trans. Inf. Theory 26, 78.50 F. Y. Wu, R. E. Grove, S. Ezekiel: Phys. Rev. Lett. 35,
78 (1980) 1426 (1975)
78.22 J. H. Shapiro, S. S. Wagner: IEEE J. Quantum Elec- 78.51 W. Hartig, W. Rasmussen, R. Schieder, H. Walther:
tron. 20, 803 (1984) Z. Phys. A 278, 205 (1976)
78.23 K. E. Cahill, R. J. Glauber: Phys. Rev. A 177, 1882 78.52 H. J. Carmichael, D. F. Walls: J. Phys. B 9, L43 (1976)
(1969) 78.53 H. J Carmichael, D. F. Walls: J. B. Phys 9, 1199 (1976)
78.24 M. Hillery, R. F. OConnell, M. O. Scully, E. P. Wigner: 78.54 H. J. Kimble, M. Dagenais, L. Mandel: Phys. Rev.
Phys. Rep. 106, 121 (1984) Lett. 39, 691 (1977)
78.25 R. J. Glauber: Phys. Rev. 131, 2766 (1963) 78.55 M. Dagenais, L. Mandel: Phys. Rev. A 18, 201 (1978)
Part F 78

78.26 E. C. G. Sudarshan: Phys. Rev. Lett. 10, 277 (1963)
78.27 C. W. Gardiner: Handbook of Stochastic Methods
(Springer, Berlin, Heidelberg 1985)
78.28 C. W. Gardiner: Quantum Noise (Springer, Berlin,
Heidelberg 1991)
78.29 E. Wigner: Phys. Rev. 40, 749 (1932)
78.30 R. L. Hudson: Rep. Math. Phys. 6, 249 (1974)
78.31 P. Meystre, M. Sargent III: Elements of Quantum
Optics (Springer, Berlin, Heidelberg 1991)
78.32 C. Cohen-Tannoudji, J. Dupont-Roc, G. Grynberg:
Atom-Photon Interactions (Wiley, New York 1992)
78.33 D. F. Walls, G. J. Milburn: Quantum Optics (Springer,
Berlin, Heidelberg 1994)
78.34 C. W. Gardiner: Phys. Rev. Lett. 56, 1917 (1986)
78.35 G. W. Ford, J. T. Lewis, R. F. OConnell: Phys. Rev. A
37, 4419 (1988)
78.36 R. F. OConnell: Dissipation in a squeezed-state
environment, Proceedings of the Second In-
ternational Workshop on Squeezed States and
Uncertainty Relations, Moscow, Russia 1992, ed. by
78.56 M. O. Scully, M. S. Zubairy: Quantum Optics (Cam-
bridge Univ. Press, Cambridge 1996)
78.57 H.-A. Bachor: A Guide to Experiments in Quantum
Optics (Wiley-VCH, Weinheim 1998)
78.58 H. J. Carmichael: Statistical Methods in Quan-
tum Optics 1: Master Equations and Fokker-Planck
Equations (Springer, Berlin, Heidelberg 1999)
78.59 P. Meystre, M. Sargent III: Elements of Quantum
Optics (Springer, Berlin, Heidelberg 1999)
78.60 C. W. Gardiner, P. Zoller: Quantum Noise (Springer,
Berlin, Heidelberg 2000)
78.61 M. Orszag: Quantum Optics (Springer, Berlin, Hei-
delberg 2000)
78.62 W. P. Schleich: Quantum Optics in Phase Space
(VCH-Wiley, Weinheim 2001)
78.63 D. Bouwmeester, A. Ekert, A. Zeilinger (Eds.):
The Physics of Quantum Information: Quan-
tum Cryptography, Quantum Teleportation, Quan-
tum Information (Springer, Berlin, Heidelberg
2000)
 Quantized Field Effects References 1165

78.64 M. A. Nielsen, I. L. Chuang: Quantum Computation 78.76 G. W. Ford, R. F. OConnell: Phys. Rev. Lett. 82, 3376
and Quantum Information (Cambridge Univ. Press, (1999)
Cambridge 2000) 78.77 A. O. Caldeira, A. J. Leggett: Physica A 121, 587 (1983)
78.65 G. Alber, T. Beth, M. Horodecki, P. Horodecki, 78.78 R. Dekker: Phys. Rep. 80, 1 (1981)
R. Horodecki, M. Rtteler, H. Weinfurter, R. Werner, 78.79 B. L. Hu, J. P. Paz, Y. Z. Zhang: Phys. Rev. D 45, 2843
A. Zeilinger: Quantum Information: An Introduc- (1992)
tion to Basic Theoretical Concepts and Experiments, 78.80 J. J. Halliwell, T. Yu: Phys. Rev. D 53, 2012 (1996)
Springer Tracts in Modern Physics, Vol. 173 (Springer, 78.81 F. Haake, R. Reibold: Phys. Rev. A 32, 2462 (1985)
Berlin, Heidelberg 2001) 78.82 R. Karrlein, H. Grabert: Phys. Rev. E 55, 153 (1997)
78.66 B. T. H. Varcoe, S. Brattke, M. Weidinger, H. Walther: 78.83 G. W. Ford, R. F. OConnell: Phys. Rev. D 64, 105020
Nature 403, 743 (2000) (2001)
78.67 H. D. Zeh: Found. Phys. 1, 69 (1970) 78.84 G. W. Ford, R. F. OConnell: Limitations on the utility
78.68 A. J. Leggett: Suppl. Prog. Theor. Phys. 69, 80 (1980) of exact master equations, Ann. Phys (NY) 319, in
78.69 E. P. Wigner: In: Quantum Optics, Experimen- press
tal Gravity and Measurement Theory, ed. by 78.85 C. Anastopoulos, B. L. Hu: Phys. Rev. A 62, 033821
P. Meystre, M. O. Scully (Plenum, New York 1983) (2000)
p. 43 78.86 R. F. OConnell: Optics Comm. 179, 451 (2000)
78.70 D. F. Walls, G. J. Milburn: Phys. Rev. A 31, 2403 (1985) Reprinted in Ode to a Quantum Physicist edited
78.71 E. Joos, H. D. Zeh, C. Kiefer, D. Giulini, J. Kupsch, by W. Schleich, H. Walther, W. E. Lamb (Else-
I.-O. Stamatescu: Decoherence and the Appearence vier,Amsterdam, 2000)
of a Classical World in Quantum Theory, 2nd edn. 78.87 R. F. OConnell: Optics Comm. 179, 477 (2000)
(Springer, Berlin, Heidelberg 2003) 78.88 G. W. Ford, R. F. OConnell: Phys. Rev. Lett. 77, 798
78.72 W. H. Zurek: Rev. Mod. Phys. 75, 715 (2003) (1996)
78.73 G. Lindblad: Commun. Math. Phys. 48, 119 (1976) 78.89 G. W. Ford, R. F. OConnell: Acta Phys. Hung. B 20,
78.74 G. W. Ford, J. T. Lewis, R. F. OConnell: Ann. Phys. 91 (2004)
(NY) 252, 362 (1996) 78.90 R. F. OConnell: J. Optics B 5, S349 (2003)
78.75 J. T. Lewis, R. F. OConnell: Ann. Phys. (NY) 269, 51 78.91 E. Merzbacher: Quantum Mechanics, 3rd edn. (Wi-
(1998) ley, New York 1998) p. 394

Part F 78
1166