Quantum Distribution Theory and Partially Coherent Radiation

CHAPTER 3
Quantum distribution
theory and partially
coherent radiation
As we have seen in the previous chapters, there are quantum fluc-

tuations associated with the states corresponding to classically well-
defined electromagnetic fields. The general description of fluctuation
phenomena requires the density operator. However, it is possible to
give an alternative but equivalent description in terms of distribution
functions. In the present chapter, we extend our treatment of quan-
tum statistical phenomena by developing the theory of quasi-classical
distributions. This is of interest for several reasons.
First of all, the extension of the quantum theory of radiation to
involve nonquantum stochastic effects such as thermal fluctuations
is needed. This is an important ingredient in the theory of partial
coherence. Furthermore, the interface between classical and quantum
physics is elucidated by the use of such distributions. The arch type
example being the Wigner distribution.*
In this chapter, we introduce various distribution functions. These
include the coherent state representation or the Glauber-Sudarshan P-
representation. The P -representation is used to evaluate the normally
ordered correlation functions of the field operators. As we shall see in
the next chapter, the P -representation forms a correspondence between
the quantum and the classical coherence theory. This distribution
function does not have all of the properties of the classical distribution
functions for certain states of thefield,e.g., it can be negative. We also
discuss the so-called g-representation associated with the antinormally
The first quasiclassical distribution, Wigner [1932], was written from a wave function perspective.
The later work of Moyal [1949] introduced the characteristic function approach to obtaining the
Wigner distribution. For reviews of the subject see Hillery, O'Connell, Scully, and Wigner [1984],
and Reichl, chapter 7 [1980]. The very readable textbooks by Louisell [1974], Walls and Milburn
[1994], and Cohen [1995] extend the quasiclassical distribution concept and are recommended
reading.
Downloaded from https://www.cambridge.org/core. University of Sussex Library, on 11 Jul 2018 at 08:10:09, subject to the Cambridge Core terms of use,
available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9780511813993.005
3.1 Coherent state representation 73
ordered correlation functions. Other distribution functions and their

properties are also presented.
3.1 Coherent state representation

The study of the interface between quantum and classical physics is a
fascinating subject. Nowhere is this better illustrated than in quantum
optics, where we are often faced with the problem of characterizing
fields which are nearly classical but have important quantum features.
The coherent states are well suited to such studies. In order to see why
this is the case, let us recall that for a fluctuating classical field we are
generally dealing with a probability distribution P{£) for the complex
field amplitude 8 = \S\e^ as indicated in Fig. 3.1.
Now in quantum mechanical problems, a probability distribution
for the system comes from the statistical or density operator which is
defined as follows. Suppose we know that the system is in state \\p),
then an operator 0 has the expectation value
(O)QM = (w\O\ip), (3.1.1)
but we typically do not know that we are in \\p). We only have a

probability Pv for being in this state so we must perform an ensemble
average as well
((O)QM>ensemble = ^ P V > ( v l < % > - (3.1.2)

V
Now using completeness Y^ I")("I = 1
n y>
n ip
. (3.1.3)
Thus the radiation field is, in general, described by the density

operator
5> (3.1-4)
where Pv is the probability of being in the state \xp). The expectation

value of any field operator 0 is then given by
74 Quantum distribution theory and partially coherent radiation
Fig. 3.1
(a) The fluctuating
£(t) classical field as a
function of time for a
field with large
fluctuations (solid
line) and a well
stabilized field
(dashed line), and
(b) associated
probability
distributions.
(a)
P(£)
(0) = Tr(Op), (3.1.5)

where Tr stands for trace. Now the density operator p can be expanded
in terms of the photon occupation number states:
n n
)( \p\m)im\ = (3.1.6)
Likewise the expansion may be made in terms of coherent states as
(3.1.7)
Following Glauber's convention we define the R-representation as
Ml+
W2 \ (3.1.8)
so that the density matrix may be written as
p= f fi^H\a){p\R(u',P)e-^1+^\ (3.1.9)
J J n n
We thus have used two indices n and m or a and /? in order to specify
the density matrix.
We next make contact with P(S"), as discussed earlier, by developing
a diagonal coherent state representation. That is, we express the density
operator p in terms of the diagonal pair |a)(a| in the following.
3.1.1 Definitioffqfthe coherent state representation

Consider an operator ON(a, <fl), which is a function of a and a' in the
normal order (all the creation operators a* on the left-hand side and
all the annihilation operators a on the right-hand side), i.e.,
ON(a, af) = YY cnmWTam- (3.1.10)
It may be noted that any operator involving a and a* can be con-

verted into a normal ordered form by using the commutation relation
[a,a'] = 1. For example a2a^ = afa2 + 2a. The expectation value of
the operator ON(a,a^) can then be written as
As discussed in Appendix 3.A, we define the operator

S(x - cf)5(a - a)
= \ f exp[-^(oc* - af)] expL3> - a)]d2p, (3.1.12a)
or, in an equivalent form
5(tx" - (f)d(a - a)
= i / expH)8(<x* - af)] expHjT(a - a)]d2p. (3.1.12b)
nz J
We will use (3.1.12a) and (3.1.12b) interchangeably in the text. Equa-
tion (3.1.11) can then be rewritten as
n m
2
= f d aP(a,a)ON(x,a), (3.1.13)
where
P(a, a) = Tr[pd(a - a^)S(a - a)]. (3.1.14)
It is seen from Eq. (3.1.13) that the function P(a, a") can be used to
evaluate the expectation values of any normal ordered function of a
and a? using the methods of classical statistical mechanics. Due to the
Hermiticity of the density operator p, the distribution function P(a, a")
is real. Moreover, since Tr(p) = 1, P(a,a*) is normalized to unity, i.e.,
P(a,a')d2a = 1. (3.1.15)
The function P(<x,a") is referred to as the P-representation or the

coherent state representation. The name coherent state representation
is due to the following representation of the density operator p by
means of a diagonal representation in terms of the coherent states:
= [ (3.1.16)
The equivalence of the definitions of P(ot, a*) as given by Eqs. (3.1.14)

and (3.1.16) can be seen simply by substituting for p from Eq. (3.1.16)
into Eq. (3.1.14). As we shall see in the next chapter, P(a, a") forms a
connection between the classical and quantum coherence theory.
Before considering some examples of the P -representation, we give
a simple procedure to find P(a,a*) from a knowledge of p. Let |/f)
and |—/?) be the coherent states with jS and —ftbeing the eigenvalues
of a, respectively. Then, using Eq. (2.4.7),
= e~W2 /"[ (3.1.17)
At this point we note that if a = xa + iyx and /? = xp + iyp, then

d2a = dxxdya and jSa* — P*a = 2i{ypxa — xpyx), and Eq. (3.1.17)
becomes
= f f[ (3.1.18)
Thus, (—p\p\p)e^ is the two-dimensional Fourier transform of

|a|2
P(a,a*)e~ . This shows the utility of considering the matrix element
(—P\p\P), since the inverse Fourier transform readily gives P(a,a*) in
terms of the density operator p. On taking the Fourier inverse of Eq.

(3.1.17), we obtain
P(a,a') = ^2— / [{-PW^ê^^-X'^dxfsdyfs
e-Va'+srad2p. (3.1.19)
This is the required expression.
3.1.2 Examples of the coherent state representation

As a first example, we calculate P(a,a') for the thermal field. A
field emitted by a source in thermal equilibrium at temperature T is
described by a canonical ensemble
exp(-Jf//c B r)
P (3L20)
=
where &B is the Boltzmann constant and Jf is the free-field Hamil-

tonian, Jf = hv(aâ + 1/2). For simplicity, we restrict ourselves to
a single mode of the field. On substituting this form of J f into Eq.
(3.1.20) we obtain
Correspondingly
x 1
hv ~1
(n) = Tr(a' ap) = exp (3.1.22)
Equation (3.1.21) can therefore be rewritten in terms of (n) as
This leads to the well-known result that the photon distribution in a

thermal field is described by the Bose-Einstein distribution, i.e.,
pnn = (n\p\n)
Next we substitute for p from Eq. (3.1.23) into Eq. (3.1.19). We note
that
expr \-\p\2/ 1+ - - )I, (3.1.25)

1 + ( n ) - " | "-' / V (n
so that
e- |a|2/(n> , (3.1.26)
i.e., the P -representation of the thermal distribution is given by a

Gaussian distribution.
As another example, we consider the P -representation of a coherent
state |<xo). Here p = |ao)(aol so that
(~P\P\P) = {-P\«O)(«O\P)
= exp(-|a o | 2 - li?|2 - «ojS* + /J«S). (3.1.27)
It then follows from Eq. (3.1.19) that
= —2 e|a|2"|o<o12 [ e-P(«'-«Z)+F(<*-
n J
(3.1.28)
i.e., the P -representation of a coherent state is a two-dimensional delta

function.
Even though the P -representation allows us to evaluate the normally
ordered correlation functions of the field operators a and a\ it is not
nonnegative definite and as such cannot be described as a distribution
function for certain field states. This can be readily seen by evaluating
the P-representation of a number state \n), for which p = \n)(n\ and
(~P\P\P) = {-P\n){n\P)
= exp(-|/?| 2 r ;
n ,'
PI
• (3.1.29)
3.2 Q-representation 79
The corresponding P -representation is, therefore, given by
nzn\
Ja| 2 f)2n
2
n n\ B^da'" J '
(3.1.30)
n! da.nda
For n > 0, this is clearly not a nonnegative definite function and, there-
fore, a number state does not have a well-defined P -representation.
As we will discuss in the next chapter, whenever the photon distri-
bution pnn is narrower than the Poisson distribution, as in the case of
number state \n), P(a,a") becomes badly behaved. This is the price we
pay for forcing quantum physics into a classical format, i.e., for using
P(a,a") instead of say, R(x,P').
3.2 Q-representation
Just as the P -representation is associated with the evaluation of nor-

mally ordered correlation functions of the field operators a and a\ we
may define other distribution functions which may be associated with
different orderings of a and d. The distribution function which helps
in determining the antinormally ordered correlation functions is the
so-called Q-representation. It is defined as
6(0, a*) = Tr[p5(a - a)S(u - J)]. (3.2.1)
It follows, on inserting the representation (2.4.6) for unity between

<5(oe - a) and <5(<x* - a f ) and using (2.2.1) that
6(a,a*) = - T r /d2oc'[pd(ot - a +
n J
= -Tr(p|a)(a|)
n
= -<a|p|a>, (3.2.2)
n
i.e., Q(a,a*) is proportional to the diagonal element of the density
operator in the coherent state representation. It follows from the
completeness of the coherent states |a) (Eq. (2.4.6)) and the condition
Tr(p) = 1 that Q(a, a*) is normalized to unity, i.e.,
(a,a)d1a = l. (3.2.3)
In order to see how the antinormally ordered correlation functions

of a and eft are evaluated using the Q-representation, we first define a
function OA{a, cfl) in antinormal order, i.e.,
, m aV) m - (3.2.4)
n m
It then follows that
= J Q(oL,«)0A(a,a)d2ot, (3.2.5)
where, in the third line, we inserted
- [\a.)(a\d2a = L (3.2.6)
71
J
Unlike the P-representation, Q(a,tx*) is nonnegative definite and
bounded. This can be seen by substituting for p from Eq. (3.1.4) into
Eq. (3.2.2). We then obtain
Q(a,a") = - ^ Pv\{ip\a}\2. (3.2.7)
n /-—'
v
Since |(i/>|a)|2 < 1, we have
Q(a,a)<-. (3.2.8)
n
The Q-representation may be related to the P -representation by
taking the coherent state diagonal element of p in Eq. (3.1.16). The
resulting equation is
Q(a, a') = - / P(a', a'> H a - a ' | 2 d 2 a'. (3.2.9)
As an example, Q(ot,a*) for a number state |n) is given by
Q(a,a) = -\(n\a)\2 = * * ^ ", (3.2.10)

which is a well-behaved function. The g-representation of a squeezed
state is given in Section 3.5.
3.3 The Wigner-Weyl distribution 81
3.3 The Wigner-Weyl distribution

So far we have discussed various distribution functions, namely P-
and ^-representations associated with the normal and the antinormal
orderings, respectively, of the operators a and eft. We can similarly
derive distribution functions associated with other orderings.
To summarize, we have introduced
P(a, a*) = Tr[<5(a* - a^S{a - a)p], (3.3.1a)
6(«, a) = Tr[(5(a - a)d(u - ar)p], (3.3.1b)
which we can write in terms of the so-called characteristic functions.
For example, inserting (3.1.12b) into (3.3.1a) we have
P{a.,a) = ^ [<Ppe-V*'-V"C{n)(P,p'), (3.3.2)

%l J
where the characteristic function C(fl)(jS,/r) is defined as
Cl"\p, P*) = Tr ( > a V f l p ) • (3-3.3)
Likewise, we may write (3.3.1b) as
Q(a,a') = 4 j [ d2pe-ilSa'-ils'«C{a)(PJ'), (3.3.4)
nz J
with the characteristic function
C{a)(P,p') = Tr ( e ^ V ^ p ) . (3.3.5)
Another useful distribution, due to Wigner and Weyl, is denned as
W(a,a') = ^ /" d2pe-ifs«'-ift'aC{s)(p,p*), (3.3.6)
where the characteristic function C(s)(/?,j3*) is given by

C(s)(jS, P') = Tr (<$<t+iP'ap\ . (3.3.7)
This distribution function W(a, a*) is associated with symmetric order-

ing. It can be used to evaluate expectation values of any symmetrically
ordered functions of a and eft in a classical fashion. For example,
l
-{acft + aU) = f W{a,a)(xud2o>. (3.3.8)
In Appendix 3.B, we give a procedure to find the c-number func-

tion Os(a, a*) corresponding to the symmetrically ordered form of an
operator O(a,cft).
Historically, the W(a,a*) distribution was introduced in terms of
the position q and momentum p operators in a form equivalent to
W(p,q) = - i - r [do [ dxei{Tp+aq)Tr \e-l(xp+aq) p\. (3.3.9)

(2rc) J J
To cast this into the form first introduced by Wigner we use the
operator identity
eA+B = eAeBe-[A'B]/2,
which holds when the commutator [A, B] commutes with A and B, to

write (3.3.9) as
W(p,q)
= - ^ t da [ dxei{zp+aq)Ti (e~hpe'ê^12p) , (3.3.10)
which by cyclic invariance under the trace may be written as
[ dxei{xp+aq)Tr (e-ixpl2e-iaqpe-ixpl2\ e-
inax/2
. (3.3.11)
Writing the trace in the coordinate representation this becomes
fdq'(q'\e-"tl/2e-"7!>pe-iTm\q'}e-ih'n/2, (3.3.12)
and noting that exp(—hp/2)\q') = \q' — fix/2) etc., we find
'\4 + hx/2\p\q' - hx/2)ehp. (3.3.13)
Finally, we carry out the cr-integration to obtain a delta function d(q —

q'), which allows us to carry out the ^'-integration, and introducing
the notation y = —fix/2, we write W(p, q) in the usual form
W(p, 1) = ^J dye-i2^n(q - y\p\q + y). (3.3.14)
The Wigner function in the form (3.3.14) has been widely used in a
host of problems; and we further elaborate on its connection with the
P- and Q-distributions in the next section.
3.4 Generalized representation of the density operator 83
3.4 Generalized representation of the density operator

and connection between the P-, Q-, and
W -distributions
In the following, we present a generalized representation of the density

operator originally due to Cohen and applied to quantum optics by
Agarwal and Wolf. The P-, Q-, and ^-representations can be derived
as special cases of this generalized representation.
A generalized representation F (n) (a,a*) of the density operator is
given by
p= n [ Fin\a, a*)A(fl)(a - a, a - J)d2o., (3.4.1)
where
(«-a,*-J) = -n 2J f z
x e x p h ^ a * - a*) + j S > - a)]
Here Q(jS,j6*) (such that Q(0,0) = 0) is a function which characterizes

different orderings. For example, when Q(j6,/T) = —1/?|2/2 we have
F<n\a,a) = P(a,a) and when QQ3,jS*) = |j3|2/2 we have F(Q>(o,a*) =
To see these results explicitly, we first consider

I/?!2
Q{P,n = ~ - (3A3)
It follows from Eqs. (2.2.6) and (2.2.7) that
exp (-~+ Paf - fa} = exp(-^'a) exp(/?flt), (3.4.4)
and we obtain
= J_ f feP'C-)
= -|a)(a|. (3.4.5)
n
On substituting this expression for A (n) (a — a, a* — aT) into Eq. (3.4.1)
we recover the definition of the P-representation (Eq. (3.1.16)) with
On the other hand, if we choose 0(0,0*) = \0\2/2,
A<°)(a - a, a* - at) = A /" e-m-*)#(*-«)#$, (3A6)
It follows from Eq. (3.4.1) that
-(al\p\al) = f F^(a,a.*)(oc'\A{il)(a-a,a -a%')d2a. (3.4.7)
However, from Eq. (3.4.6),
= 5(2\a - a')- (3.4.8)

On carrying out the a-integration in Eq. (3.4.7) we recover Eq. (3.2.2)
Another distribution, the Wigner-Weyl distribution, is recovered for

the proper choice of 0, namely, £2(oc,a*) = 0. To that end, we invert
Eq. (3.4.1) by using the function
-a,a-at) = \ [ exp[-fl(j8, 0')]
x exp[j8(a* - a f ) - /T(a - a)]d2p. (3.4.9)

Now, it can be shown that (see Problem 3.3)
Tr [A<n>(a - a, a - flt)A
(Q
V - a, a'* - af)]
_ a '). (3.4.10)
7T
It then follows from Eq. (3.4.1) that

F(fi)(a, «*) = Tr [pA(n)(a - a, a* - a f ) ] . (3.4.11)
From Eqs. (3.4.9) and (3.4.11), we obtain
= \z f Tr[p exp(-iSa t + p'a)] expâ* - p'a)d2p, (3.4.12)

n J
which, as expected, is the same as Eq. (3.3.6) with /? replaced by — ifi
and /?* by ifi*. Equations (3.1.14) for the P-representation and (3.2.2)
for the g-representation can be recovered from expression (3.4.11) for
0(0,pm) = -|jS| 2 /2 and 0(0,0') = \p\2/2, respectively.
In the following we derive an explicit expression for the Wigner-
Weyl distribution W(a,a"). First we mention that W(a, a*) is the
Fourier transform of the function Tr[pexp(—ficfi + 0*a)]/n2. We also
note that exp(—2|a|2) is the Fourier transform of exp(—\0\2/2)/2n, i.e.,
3.4 Generalized representation of the density operator 85
exp(-2|a| 2 ) = ± | e x p ( - ^ I P I 2 ) exp(/?a* - /Ta)d2/}.(3.4.13)
It then follows from the convolution theorem that
W(a,a*)exp(-2|a| 2 ) = /"c(/J,/r)exp(jSa' -/Ta)**2/?, (3.4.14)
where C(/?,/T) is the convolution product
S, P') = ^ J Tr{pexp[-(|8 - j5i)a+ + (/T - ^)a]}
/Ji. (3.4.15)
An explicit expression for C(/?,/T) can be obtained by using the

identity (2.2.7) and inserting the resolution of the identity in terms of
coherent states (Eq. (2.4.6)) as follows:
x exp f-I|^ - ^ | 2 - I|^
\h\p\h)kh\h)
(3.4.16)
On carrying out the integrations over ft, Eq. (3.4.16) reduces to
C(PJ') = ^ 4 j J{P}\P\P2){P/2\P3){P2\ ~ P/2)d2p2d2fa

^ (3.4.17)
Finally, on substituting for C(fiJm) from Eq. (3.4.17) into Eq. (3.4.14)
and changing the variables of integration from /?, /T to — 2/J, — 2/T, we
obtain
= ^ exp(2|a|2) J{-p\p\P) exp[-2(/?a* - ^*a)]d2j3. (3.4.18)
This expression, which is very similar to the corresponding expression

for P-representation (Eq. (3.1.19)), can be used to evaluate the Wigner-
Weyl distribution for the given density operator of the field.
3.5 ^-representation for a squeezed coherent state

In this section, we derive the Q-representation for the squeezed coher-
ent state \P, £). According to Eq. (3.2.2)
7C 7T
Now
We therefore need to calculate the function (a\

It follows, on using the properties of the coherent state and the
transformation property (2.7.7) of S(£) that
= - i (a|S(O(a t cosh r - ae~iB sinh r)\0)
= 1 [coshr ( A + ^ ^ _ g-'^sinhr] (a\S(m)-

(3.5.3)
The function (a|S(^)|j8) therefore satisfies the following differential

equation
^ - Pe~ie sinhr + Q/T cosh r - a^\(a\S(O\P) = 0.

(3.5.4)
The solution of this equation is
= X exp( —ijy9j2 + a'^sech r + ê~wp2 t a n h r j . (3.5.5)
The form of K, which may depend upon a,a*,/J*,r, and 0, can be

determined using the unitarity of S(^). It follows from
a\S(i)\P)' = (P\SHt)\*) = ^|S(-^)|«>, (3.5.6)
3.5 Q-representation for a squeezed coherent state 87
that
I J
The form of K is therefore
K(a,a,p',r,6)
= (sech r) 1/2 exp | - ^ M 2 - Uie(a)2 tanhr . (3.5.8)
The coefficient (sech r)1?2 is chosen so that the normalization condition
(3.5.9)
is satisfied.
On substituting this expression for K in Eq. (3.5.5) we obtain
<a|S(OI0> = (sech rf2 exp j - l-(\a\2 + \P\2) + a'j&sech r
-~ [eie(a')2 - e~wp2] tanhrj. (3.5.10)
The Q-representation for the state \P,£) is therefore
Q(a, a*) = exp { - (|a|2 + |j3|2) + ( a ^ + i?*a)sech r
-hew(a2 - P'2) + e-i0(a2 - jS2)] t a n h r j . (3.5.11)
In Fig. 3.2, Q(a,a*) = Q(Xi,X 2 ) (X, = (a+a*)/2, X 2 = ( a - a ' ) / 2 i ) is

plotted as a function of the amplitudes X\,X2 of the two quadratures.
We clearly see the unequal variances in X\ and Xi in the state |a, £). We
can employ expression (3.5.10) for (a|S(^)|j8) to calculate the photon
distribution function of a squeezed coherent state.
The photon distribution function p(n) for the field in state |j8, £) is
given by
p(n) = | ( n | ^ 0 l 2 - (3-5.12)
The quantity (n|j8,£) can be determined by writing
(3.5.13)
n=0 n=0 V n
and expanding the right-hand side of Eq. (3.5.10) in powers of a* by

means of the generating function for the Hermite polynomials Hn(z):
Quantum distribution theory and partially coherent radiation
Fig. 3.2
Q(X,,X2) A plot of
as a function of the
amplitudes Xi and
X2 in a squeezed
coherent state. (From
H. P. Yuen, Phys.
Rev. A 13, 2226
(1976).)
(3.5.14)
H=0
On comparing the resulting expansion with the expansion in Eq.

(3.5.13), it follows that
xHn (3.5.15)
The photon distribution function p(n) for an ideal squeezed state is

therefore given by
/ Re~w/2
(3.5.16)
V y/2 cosh r sinh r
3.5 (^-representation for a squeezed coherent state 89
Fig. 3.3 0.16

Comparison of
photon distribution 0.14
function for a
coherent state |a) 0.12
with |a|2 = 60 (solid
line) with the 0.1
squeezed coherent
state |/U> p(n) 0.08
£ = r exp(i0)) with 0.06
\p\2 = 60, r = 0.6,
and <j> = 9/2 (dashed 0.04
line).
0.02
0
0 10 20 30 40 50 60 70 80 90 100
n
Generally, sources of squeezing produce a radiation field in a
squeezed vacuum state |0, £). The detection schemes, however, add
a coherent component to it. The detected state is therefore described
by the distribution (3.5.16). The fluctuations in the mean number of
photons can be found either from Eq. (3.5.16), by using
(3.5.17)
n=0
or through the use of the unitary transformation properties of the

squeeze operator (2.7.6) and (2.7.7). We obtain
(An)2 = |/?|2[cosh4r — cos(0 — 2</>) sinh 4r] + 2 sinh2 r cosh2 r.

(3.5.18)
In the following, we discuss three cases of interest. First, when

|/J | 2 > sinh 2 r, the coherent component is larger than the squeeze
component. Figure 3.3 compares the probability distribution for a
squeezed state with a coherent state. If the squeezing is along the
coherent amplitude, the state has sub-Poissonian photon statistics.
In the second case (Fig. 3.4) when the squeeze component is larger
than the coherent component and squeezing is along the coherent
amplitude, the squeezed state exhibits oscillations. The main peak as
well as the subsequent peaks are narrower than the corresponding Jn
value. But the overall distribution shows super-Poissonian statistics.
0.25 Fig. 3.4

Photon distribution
function p{n) for a
0.2 two-photon coherent
state (Eq.(3.5.16)) for
\pf = 60, r = 1.6.
The squeeze
0.15
component is larger
p(n) than the coherent
component and
o.i squeezing is along the
coherent amplitude.
0.05
0 10 20 30 40 50 60 70 80 90 100
Finally, for zero displacement, i.e., for the squeezed vacuum state, the
distribution function (3.5.16) reduces to
p(2n + 1) = 0. (3.5.19)
In the above equations, a nonzero value for even terms arises due to
squeezing of the vacuum and clearly shows the 'two-photon' nature
of the field. Figure 3.5 shows a plot of the probability distribution
(3.5.19). The distribution peaks sharply at n = 0 and has a very long
tail similar to a thermal distribution.
3.A Verifying equations (3.1.12a, 3.1.12b)

It can be verified that the t\ô-dimensional delta function has the form
(3.1.12a)
d(a - af)<5(a -a) = \ /exp[-)8(a'- a - a)]d2p

nlj
(3.A.1)
by taking the expectation values in a coherent state \y) of both sides

of Eq. (3.A.1). Indeed, o n doing so a n d utilizing the fact that \y) is a n
Appendix 3.A 91
Fig. 3.5
Photon distribution
function p(n) for a
squeezed vacuum
state for r = 1.6. 0.15
p(n) o.i
0.05
10 20
eigenstate of the annihilation operator a with an eigenvalue y, we get
S(a-y')S(a -y) = ^Jexp[-â-y')] exp[/T(a - y)]d2p.

(3.A.2)
If we write a = xx + iya, 0 = xp + iyp, and y = xy + iyy, then

d2fi = dx$dy$ and the right-hand side of Eq. (3.A.2) becomes
~ I exp[-/?(a* - / ) ] exp[/T(a - y)]d2f$

1 /"
= ~2 / exp {2
= (^) / / exp - y y) -
= d(cc - y)d(a - y"), (3.A.3)
where we have replaced 2x/j and 2yp by xp and yp, respectively, in the
second line and used the following expression for the delta function
=~
2 J
f eikxdk. (3.A.4)
Equation (3.1.12b)
S(a - flf)<5(a - a)
f
[-i^*(a - a)]d2p (3.A.5)
can be obtained from (3.A.1) simply by changing the variables /? -> i/J
and /T -+ - i / T
Another formula for the antinormally ordered two-dimensional
delta function, namely,
S(a - a)S(a* - a f )
= \ f exp[P'(a - a)] exp[-/J(a* - ct)]#P, (3.A.6)
which has been used to define the ^-representation (Eq. (3.2.1)), can
be proven by inserting
= 1, (3.A.7)
as follows:
1 /"exp[/T(a - a)] exp[-/?(a* - aVfi
= J_ /" f e^-
= ^J 5{z-y)\y){y\d{a-f)d2y
= 8(a. - a)5(a - a 1 ). (3.A.8)
3.B onumber function correspondence for the

Wigner-Weyl distribution
Given an operator 0{a,a*) and the Wigner-Weyl distribution W{a,a"),
we calculate the c-number function O s (a,a*) such that
(O(a,fl+J)= Tr{Op) = Id2a0s{0L,a)W{a,a). (3.B.1)
Recall that the Wigner-Weyl distribution is defined as (Eqs. (3.3.6)

and (3.3.7))
Appendix 3.B 93
W(a,a) = \ fd2pTr(eils'a+iliatp)e-ils'*-ilSa' (3.B.2)
with the characteristic function
9,y9*) = f eip'a+illa'W(a,!x')
= Tr(eiêiâe-^2p). (3.B.3)
where, in the last line, we use the Baker-Hausdorff formula (2.2.7).

Now for any normally ordered operator O(a,ai), one can write
(3.B.4)
It can be easily found that
•=8=0'
and
Then we have
(O(a,cJ
= f d2aOs(a,*)W(oi,z), (3.B.5)
which yields
.(3.B.6)
(W) 2i r=j?=o
Equation (3.B.6) is our desired result. Consider some examples:
(a)
Os(«,*)
Jft'a+ifa'
2i W) ?*=fl=O
2i J*=fl=O
. 1
(3.B.7)
(b)
Os(a,a)
d , /TV
?*=8=0
+
W) 2i
= a*2a — a*. (3.B.8)
The operator corresponding to the Wigner distribution func-

tion in the coordinate-momentum representation is given by
Cohen (1986).
Problems
3.1 Show that
1
•aU)=
2<flfl'+. [w(aaM2d2«
where W(a,a*) is the Wigner-Weyl distribution.
Problems 95
3.2 Show that
Tr[D(a)] = nd
Tr[D(oc)I>y)] = 7t<5(2)(oc - a'),
where D(a) is the displacement operator. Using these results,

show that
Tr[A(Q)(a - a, a* - a ^ A ^ V - a, a" - a+)]
= -<5<2>(a - a'),
n
The operators A' n) and A'Q) are defined in Eqs. (3.4.2) and
(3.4.9), respectively.
3.3 Show that the Wigner-Weyl distribution W(a,a*) can be ex-

pressed in terms of the P-representation P(a,a') via the rela-
tion
3.4 Determine Q(a,a") and W(u,a*) for a coherent state and a

thermal state.
References and bibliography
P -representation
R. J. Glauber, Phys. Rev. 131, 2766 (1963).
E. C. G. Sudarshan, Phys. Rev. Lett. 10, 277 (1963).
D. F. Walls and G. J. Milburn, Quantum Optics, (Springer-Verlag, Berlin 1994).
Q-, Wigner-Weyl and other distributions

E. Wigner, Phys. Rev. 40, 749 (1932).
J. Moyal, Proc. Cambridge Phil. Soc. 45, 99 (1949).
Y. Kano, J. Math. Phys. 6, 1913 (1965).
J. R. Klauder, J. McKenna, and D. G. Currie, J. Math. Phys. 6, 734 (1965).
C. L. Mehta and E. C. G. Sudarshan, Phys. Rev. 138, B274 (1965).
L. Cohen, J. Math. Phys. 7, 781 (1966).
M. Lax, 1968 Brandis Lectures on Fluctuation and Coherence Phenomena (Gor-
don and Breach, New York 1968).
K. E. Cahill and R. J. Glauber, Phys. Rev. 177, 1857 (1969).
G. S. Agarwal and E. Wolf, Phys. Rev. D 2, 2161, 2187, 2206 (1970).
W. Louisell, Quantum Statistical Properties of Radiation, (Wiley, New York
1974).
P. Carmthers and F. Zachariaser, Phys. Rev. D 13, 950 (1976).
L. A. Reichl, A Modern Course in Statistcal Physics, (University of Texas Press
1980).
M. Hillery, R. F. O'Connell, M. O. Scully, and E. P. Wigner, Physics Reports,
106, 123 (1984).
L. Cohen, in Frontiers of Nonequilibrium Statistical Physics, ed. G. T. Moore
and M. O. Scully (Plenum, New York 1986), p. 97.
L. Cohen, Time Frequency Analysis, (Prentice Hall, New York 1995).
W. Vogel and D.-G. Welsch, Lectures on Quantum Optics, (Akademe Verlag,
Berlin 1994).

Quantum Distribution Theory and Partially Coherent Radiation

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Quantum Distribution Theory and Partially Coherent Radiation

Uploaded by

Copyright:

Available Formats

CHAPTER 3

As we have seen in the previous chapters, there are quantum fluc-

ordered correlation functions. Other distribution functions and their

3.1 Coherent state representation

(O)QM = (w\O\ip), (3.1.1)

but we typically do not know that we are in \\p). We only have a

((O)QM>ensemble = ^ P V > ( v l < % > - (3.1.2)

Now using completeness Y^ I")("I = 1

Thus the radiation field is, in general, described by the density

where Pv is the probability of being in the state \xp). The expectation

(0) = Tr(Op), (3.1.5)

Likewise the expansion may be made in terms of coherent states as

Following Glauber's convention we define the R-representation as

3.1.1 Definitioffqfthe coherent state representation

It may be noted that any operator involving a and a* can be con-

As discussed in Appendix 3.A, we define the operator

P(a, a) = Tr[pd(a - a^)S(a - a)]. (3.1.14)

The function P(<x,a") is referred to as the P-representation or the

The equivalence of the definitions of P(ot, a*) as given by Eqs. (3.1.14)

= e~W2 /"[ (3.1.17)

At this point we note that if a = xa + iyx and /? = xp + iyp, then

Thus, (—p\p\p)e^ is the two-dimensional Fourier transform of

terms of the density operator p. On taking the Fourier inverse of Eq.

P(a,a') = ^2— / [{-PW^^e^^-X'^dxfsdyfs

This is the required expression.

3.1.2 Examples of the coherent state representation

where &B is the Boltzmann constant and Jf is the free-field Hamil-

Equation (3.1.21) can therefore be rewritten in terms of (n) as

This leads to the well-known result that the photon distribution in a

expr \-\p\2/ 1+ - - )I, (3.1.25)

i.e., the P -representation of the thermal distribution is given by a

It then follows from Eq. (3.1.19) that

i.e., the P -representation of a coherent state is a two-dimensional delta

The corresponding P -representation is, therefore, given by

Just as the P -representation is associated with the evaluation of nor-

6(0, a*) = Tr[p5(a - a)S(u - J)]. (3.2.1)

It follows, on inserting the representation (2.4.6) for unity between

In order to see how the antinormally ordered correlation functions

It then follows that

where, in the third line, we inserted

As an example, Q(ot,a*) for a number state |n) is given by

Q(a,a) = -\(n\a)\2 = * * ^ ", (3.2.10)

3.3 The Wigner-Weyl distribution

P{a.,a) = ^ [<Ppe-V*'-V"C{n)(P,p'), (3.3.2)

W(a,a') = ^ /" d2pe-ifs«'-ift'aC{s)(p,p*), (3.3.6)

where the characteristic function C(s)(/?,j3*) is given by

This distribution function W(a, a*) is associated with symmetric order-

In Appendix 3.B, we give a procedure to find the c-number func-

W(p,q) = - i - r [do [ dxei{Tp+aq)Tr \e-l(xp+aq) p\. (3.3.9)

which holds when the commutator [A, B] commutes with A and B, to

= - ^ t da [ dxei{zp+aq)Ti (e~hpe'^e^12p) , (3.3.10)

which by cyclic invariance under the trace may be written as

Writing the trace in the coordinate representation this becomes

and noting that exp(—hp/2)\q') = \q' — fix/2) etc., we find

'\4 + hx/2\p\q' - hx/2)ehp. (3.3.13)

Finally, we carry out the cr-integration to obtain a delta function d(q —

W(p, 1) = ^J dye-i2^n(q - y\p\q + y). (3.3.14)

3.4 Generalized representation of the density operator

In the following, we present a generalized representation of the density

p= n [ Fin\a, a*)A(fl)(a - a, a - J)d2o., (3.4.1)

x e x p h ^ a * - a*) + j S > - a)]

Here Q(jS,j6*) (such that Q(0,0) = 0) is a function which characterizes

A<°)(a - a, a* - at) = A /" e-m-)#(-«)#$, (3A6)

The form of K, which may depend upon a,a,/J,r, and 0, can be

Q(a, a) = exp { - (|a|2 + |j3|2) + ( a ^ + i?a)sech r

In Fig. 3.2, Q(a,a) = Q(Xi,X 2 ) (X, = (a+a)/2, X 2 = ( a - a ' ) / 2 i ) is

= a2a — a. (3.B.8)