Lecture 39

Quantum Coherent State of

Light

The Schrödinger wave equation was a great success. When it was discovered by Schrödinger,
and appplied to a very simple hydrogen atom, its eigensolutions, especially the eigenvalues
En coincide beautifully with spectroscopy experiment of the hydrogen atom. Since the elec-
tron wavefunctions inside a hydrogen atom does not have a classical analog, less was known
about these wavefunctions. But in QED and quantum optics, the wavefunctions have to be
connected with classical electromagnetic oscillations. As seen previously, electromagnetic os-
cillations resemble those of a pendulum. The original eigenstates of the quantum pendulum
were the photon number states also called the Fock states. The connection to the classical
pendulum was tenuous, but required by the correspondence principle–quantum phenomena
resembles classical phenomena in the high energy limit. This connection was finally estab-
lished by the establishment of the coherent state.

39.1 The Quantum Coherent State

We have seen that a photon number states1 of a quantum pendulum do not have a classical
correspondence as the average or expectation values of the position and momentum of the
pendulum are always zero for all time for this state. Therefore, we have to seek a time-
dependent quantum state that has the classical equivalence of a pendulum. This is the
coherent state, which is the contribution of many researchers, most notably, Roy Glauber
(1925–2018) [253] in 1963, and George Sudarshan (1931–2018) [254]. Glauber was awarded
the Nobel prize in 2005.
We like to emphasize again that the modes of an electromagnetic cavity oscillation are
homomorphic to the oscillation of classical pendulum. Hence, we first connect the oscillation
of a quantum pendulum to a classical pendulum. Then we can connect the oscillation of
1 In quantum theory, a “state” is synonymous with a state vector or a function.

403
404 Electromagnetic Field Theory

a quantum electromagnetic mode to the classical electromagnetic mode and then to the
quantum pendulum.

39.1.1 Quantum Harmonic Oscillator Revisited

To this end, we revisit the quantum harmonic oscillator or the quantum pendulum with more
mathematical depth. Rewriting Schrödinger equation as the eigenequation for the photon
number state for the quantum harmonic oscillator, we have
~2 d2
 
1 2 2
Ĥψn (x) = − + mω0 x ψn (x) = En ψn (x). (39.1.1)
2m dx2 2
where ψn (x) is the eigenfunction, and En is the eigenvalue. Thepabove can be changed into
mω0
a dimensionless form first by dividing ~ω0 , and then let ξ = ~ x be a dimensionless
variable. The above then becomes
d2
 
1 2 E
− 2 + ξ ψ(ξ) = ψ(ξ) (39.1.2)
2 dξ ~ω0
d
We can define π̂ = −i dξ and ξˆ = Iξ
ˆ to rewrite the Hamiltonian as

1
Ĥ = ~ω0 (π̂ 2 + ξˆ2 ) (39.1.3)
2
Furthermore, the Hamiltonian in (39.1.2) looks almost like A2 − B 2 , and hence motivates
its factorization. To this end, we first show that
d2
       
1 d 1 d 1 2 1 d d
√ − +ξ √ +ξ = − 2 +ξ − ξ−ξ (39.1.4)
2 dξ 2 dξ 2 dξ 2 dξ dξ
It can be shown easily that as operators (meaning that they will act on a function to their
right), the last term on the right-hand side is an identity operator, namely that
 
d d
ξ−ξ = Iˆ (39.1.5)
dξ dξ
Therefore
d2
     
1 2 1 d 1 d 1
− 2 +ξ =√ − +ξ √ +ξ + (39.1.6)
2 dξ 2 dξ 2 dξ 2
We define the operator
 
1 d
↠= √ − +ξ (39.1.7)
2 dξ
The above is the creations, or raising operator and the reason for its name is obviated later.
Moreover, we define
 
1 d
â = √ +ξ (39.1.8)
2 dξ
Quantum Coherent State of Light 405

which represents the annihilation or lowering operator. With the above definitions of the
raising and lowering operators, it is easy to show that by straightforward substitution that
â, ↠= â↠− ↠â = Iˆ
 
(39.1.9)
Therefore, Schrödinger equation (39.1.2) for quantum harmonic oscillator can be rewritten
more concisely as
 
1 † 1 E
â â + â↠ψ = ↠â +


ψ= ψ (39.1.10)
2 2 ~ω0
In mathematics, a function is analogous to a vector. So ψ is the implicit representation of a
vector. The operator  
† 1
â â +
2
is an implicit2 representation of an operator, and in this case, a differential operator. So in
the above, (39.1.10), is analogous to the matrix eigenvalue equation A · x = λx.
Consequently, the Hamiltonian operator can be expressed concisely as
 
† 1
Ĥ = ~ω0 â â + (39.1.11)
2
Equation (39.1.10) above is in implicit math notation. In implicit Dirac notation, it is
 
† 1 E
â â + |ψi = |ψi (39.1.12)
2 ~ω0
In the above, ψ(ξ) is a function which is a vector in a functional space. It is denoted as ψ
in math notation and |ψi in Dirac notation. This is also known as the “ket”. The conjugate
transpose of a vector in Dirac notation is called a “bra” which is denoted as hψ|. Hence, the
inner product between two vectors is denoted as hψ1 |ψ2 i in Dirac notation.3
If we denote a photon number state by ψn (x) in explicit notation, ψn in math notation
or |ψn i in Dirac notation, then we have
   
1 En 1
↠â + |ψn i = |ψn i = n + |ψn i (39.1.13)
2 ~ω0 2
where we have used the fact that En = (n + 1/2)~ω0 . Therefore, by comparing terms in the
above, we have
↠â|ψn i = n|ψn i (39.1.14)

and the operator ↠â is also known as the number operator because of the above. It is often
denoted as
n̂ = ↠â (39.1.15)
2A
P
notation like A · x, we will call implicit, while a notation i,j Aij xj , we will call explicit.
3 There is a one-to-one correspondence of Dirac notation to matrix algebra notation. Â|xi ↔ A · x, hx| ↔
x† hx1 |x2 i ↔ x†1 · x2 .
406 Electromagnetic Field Theory

and |ψn i is an eigenvector of n̂ = ↠â operator with eigenvalue n. It can be further shown by
direct substitution that
√ √
â|ψn i = n|ψn−1 i ⇔ â|ni = n|n − 1i (39.1.16)
√ √
↠|ψn i = n + 1|ψn+1 i ⇔ ↠|ni = n + 1|n + 1i (39.1.17)

hence their names as lowering and raising operator.4

39.2 Some Words on Quantum Randomness and Quan-

tum Observables
We saw previously that in classical mechanics, the conjugate variables p and x are determin-
istic variables. But in the quantum world, they become random variables with means and
variance. It was quite easy to see that x is a random variable in the quantum world. But the
momentum p is elevated to become a differential operator p̂, and it is not clear that it is a
random variable anymore.
Quantum theory is a lot richer in content than classical theory. Hence, in quantum theory,
conjugate variables like p and x are observables endowed with the properties of mean and
variance. For them to be endowed with these properties, they are elevated to become quantum
operators, which are the representations of these observables. To be meaningful, a quantum
state |ψi has to be defined for a quantum system, and these operators represent observables
act on the quantum state.
Henceforth, we have to extend the concept of the average of a random variable to the
“average” of a quantum operator. Now that we know Dirac notation, we can write the
expectation value of the operator p̂ with respect to a quantum state ψ as

hp̂i = hψ|p̂|ψi = p̄ (39.2.1)

The above is the elevated way of taking the “average” of an operator which is related to the
mean of the random variable p.
As mentioned before, Dirac notation is homomorphic to matrix algebra notation. The
above is similar to ψ † · P · ψ = p̄. This quantity p̄ is always real if P is a Hermitian matrix.
Hence, in (39.2.1), the expectation value p̄ is always real if p̂ is Hermitian. In fact, it can be
proved that p̂ is Hermitian in the function space that it is defined.
Furthermore, the variance of the random variable p can be derived from the quantum
operator p̂ with respect to to a quantum state |ψi. It is defined as

σp2 = hp̂2 i − hp̂i2 (39.2.2)

where σp is the standard deviation of the random variable p and σp2 is its variance [64, 65].
The above implies that the definition of the quantum operators and the quantum states
†
is not unique. One can define a unitary matrix or operator U such that U · U = I. Then
4 The above notation for a vector could appear cryptic or too terse to the uninitiated. To parse it, one can

always down-convert from an abstract notation to a more explicit notation. Namely, |ni → |ψn i → ψn (ξ).
Quantum Coherent State of Light 407

the new quantum state is now given by the unitary transform ψ 0 = U · ψ. With this, we can
easily show that
† †
p̄ = ψ † · P · ψ = ψ † · U · U · P · U · U · ψ
0
= ψ 0† · P · ψ 0 (39.2.3)
0 † 0
where P = U·P·U via unitary transform. Now, P is the new quantum operator representing
the observable p and ψ 0 is the new quantum state vector.
In the previous section, we have elevated the position variable or observable ξ to become an
operator ξˆ = ξ I.
ˆ This operator is clearly Hermitian, and hence, the expectation value of this
position operator is always real. Here, ξˆ is diagonal in the coordinate space representation,
but it need not be in other Hilbert space representations.

39.3 Derivation of the Coherent States

As one cannot see the characteristics of a classical pendulum emerging from the photon
number states, one needs another way of bridging the quantum world with the classical
world. This is the role of the coherent state: It will show the correspondence principle, with
a classical pendulum emerging from a quantum pendulum when the energy of the pendulum
is large. Hence, it will be interesting to see how the coherent state is derived.
The derivation of the coherent state is more math than physics. Nevertheless, the deriva-
tion is interesting. We are going to present it according to the simplest way presented in
the literature. There are deeper mathematical methods to derive this coherent state like
Bogoliubov transform which is outside the scope of this course.
Now, endowed with the needed mathematical tools, we can derive the coherent state. To
say succinctly, the coherent state is the eigenstate of the annihilation operator, namely that

â|αi = α|αi (39.3.1)

Here, we use α as an eigenvalue as well as an index or identifier of the state |αi.5 Since the
number state |ni is complete, the coherent state |αi can be expanded in terms of the number
state |ni. Or that
∞
X
|αi = Cn |ni (39.3.2)
n=0

When the annihilation operator is applied to the above, we have

∞ ∞ ∞
X X X √
â|αi = Cn â|ni = Cn â|ni = Cn n|n − 1i
n=0 n=1 n=1
∞
X √
= Cn+1 n + 1|ni (39.3.3)
n=0
5 This notation is cryptic and terse, but one can always down-convert it as |αi → |f i → f (ξ) to get a
α α
more explicit notation.
408 Electromagnetic Field Theory

Equating the above with α|αi on the right-hand side of (39.3.1), then
∞ ∞
X √ X
Cn+1 n + 1|ni = α Cn |ni (39.3.4)
n=0 n=0

By the orthonormality of the number states |ni and the completeness of the set,
√
Cn+1 = αCn / n + 1 (39.3.5)

Or recursively
√ p √
Cn = Cn−1 α/ n = Cn−2 α2 / n(n − 1) = . . . = C0 αn / n! (39.3.6)

Consequently, the coherent state |αi is

∞
X αn
|αi = C0 √ |ni (39.3.7)
n=0 n!

But due to the probabilistic interpretation of quantum mechanics, the state vector |αi is
normalized to one, or that6

hα|αi = 1 (39.3.8)

Then
∞ 0
X αn αn
hα|αi = C0∗ C0 √ √ hn0 |ni
n,n0
n! n0 !
∞
X |α|2n 2
= |C0 |2 = |C0 |2 e|α| = 1 (39.3.9)
n=0
n!

2
Therefore, C0 = e−|α| /2
, or that
∞
2 X αn
|αi = e−|α| /2
√ |ni (39.3.10)
n=0 n!

In the above, to reduce the double summations into a single summation, we have made use
of hn0 |ni = δn0 n , or that the photon-number states are orthonormal. Also since â is not a
Hermitian operator, its eigenvalue α can be a complex number.
Since the coherent state is a linear superposition of the photon number states, an average
number of photons can be associated with the coherent state. If the average number of
photons embedded in a coherent is N , then it can be shown that N = |α|2 . As shall be
shown, α is related to the amplitude of the quantum oscillation: The more photons there are,
the larger is |α|.

∞
6 The expression can be written more explicitly as hα|αi = hfα |fα i = −∞ dξfα∗ (ξ)fα (ξ) = 1.
Quantum Coherent State of Light 409

39.3.1 Time Evolution of a Quantum State

The Schrödinger equation can be written concisely as
Ĥ|ψi = i~∂t |ψi (39.3.11)
The above not only entails the form of Schrödinger equation, it is the form of the general
quantum state equation. Since Ĥ is time independent, the formal solution to the above is
|ψ(t)i = e−iĤt/~ |ψ(0)i (39.3.12)
Applying this to the photon number state with Ĥ being that of the quantum pendulum,
then
e−iĤt/~ |ni = e−iωn t |ni (39.3.13)
1


where ωn = n + 2 ω0 . The above simplification follows because |ni an eigenstate of the
Hamiltonian Ĥ for the quantum pendulum, or that
 
1
Ĥ|ni = ~ωn |ni = ~ω0 n + |ni (39.3.14)
2
In other words, |ni is an eigenvector of Ĥ.

Time Evolution of the Coherent State

Using the above time-evolution operator, then the time dependent coherent state evolves in
time as7
∞
2 X αn e−iωn t
|α, ti = e−iĤt/~ |αi = e−|α| /2 √ |ni (39.3.15)
n=0 n!

By letting ωn = ω0 n + 12 , the above can be written as

∞
n
−iω0 t/2 −|α|2 /2
X αe−iω0 t
|α, ti = e e √ |ni (39.3.16)
n=0 n!
Now we see that the last factor in (39.3.16) is similar to the expression for a coherent state
in (39.3.10). Therefore, we can express the above more succinctly by replacing α in (39.3.10)
with α̃ = αe−iω0 t as
|α, ti = e−iω0 t/2 |αe−iω0 t i = e−iω0 t/2 |α̃i (39.3.17)
Consequently,
â|α, ti = e−iω0 t/2 αe−iω0 t |αe−iω0 t i = α̃|α, ti


(39.3.18)
Therefore, |α, ti is the eigenfunction of the â operator with eigenvalue α̃. But now, the
eigenvalue of the annihilation operator â is a complex number which is a function of time t.
It is to be noted that in the coherent state, the photon number states time evolve coherently
together in a manner to result in a phase shift e−iω0 t in the eigenvalue!
7 Note that |α, ti is a shorthand for fα (ξ, t).
410 Electromagnetic Field Theory

39.4 More on the Creation and Annihilation Operator

In order to connect the quantum pendulum to a classical pendulum via the coherent state,
we will introduce some new operators. Since
 
1 d
↠= √ − +ξ (39.4.1)
2 dξ
 
1 d
â = √ +ξ (39.4.2)
2 dξ

We can relate ↠and â, which are non-hermitian, to the momentum operator π̂ and position
operator ξˆ previously defined which are hermitian. Then
1  
↠= √ −iπ̂ + ξˆ (39.4.3)
2
1  
â = √ iπ̂ + ξˆ (39.4.4)
2
We also notice that
1
ξˆ = √ ↠+ â = ξ Iˆ


(39.4.5)
2
i d
π̂ = √ ↠− â = −i


(39.4.6)
2 dξ

Notice that both ξˆ and π̂ are Hermitian operators in the above, with real expectation values.
With this, the average or expectation value of the position of the pendulum in normalized
coordinate, ξ, can be found by taking expectation with respect to the coherent state, or

ˆ = √1 hα|↠+ â|αi
hα|ξ|αi (39.4.7)
2

Since by taking the complex conjugation transpose of (39.3.1)8

hα|↠= hα|α∗ (39.4.8)

and (39.4.7) becomes

√
ξ¯ = hξi ˆ = √1 (α∗ + α) hα|αi = 2<e[α] 6= 0
ˆ = hα|ξ|αi (39.4.9)
2

Repeating the exercise for time-dependent case, when we let α → α̃(t) = αe−iω0 t , then,
letting α = |α|e−iψ yields
√
ˆ
hξ(t)i = 2|α| cos(ω0 t + ψ) (39.4.10)
8 Dirac notation is homomorphic with matrix algebra notation. (a · x)† = x† · (a)† .
Quantum Coherent State of Light 411

By the same token,

i √
π̄ = hπ̂i = hα|π̂|αi = √ (α∗ − α) hα|αi = 2=m[α] 6= 0 (39.4.11)
2

For the time-dependent case, we let α → α̃(t) = αe−iω0 t ,

√
hπ̂(t)i = − 2|α| sin(ω0 t + ψ) (39.4.12)

Hence, we see that the expectation values of the normalized coordinate and momentum just
behave like a classical pendulum. There is however a marked difference: These values have
standard deviations or variances that are non-zero. Thus, they have quantum fluctuation or
quantum noise associated with them. Since the quantum pendulum is homomorphic with the
oscillation of a quantum electromagnetic mode, the amplitude of a quantum electromagnetic
mode will have a mean and a fluctuation as well.

Figure 39.1: The time evolution of the coherent state. It follows the motion of a classical
pendulum or harmonic oscillator (courtesy of Gerry and Knight [255]).
.
412 Electromagnetic Field Theory

Figure 39.2: The time evolution of the coherent state for different α’s. The left figure is for
α = 5 while the right figure is for α = 10. Recall that N = |α|2 .
.

39.4.1 Connecting Quantum Pendulum to Electromagnetic Oscilla-

tor
We see that the electromagnetic oscillator in a cavity is similar or homomorphic to a pendu-
lum. The classical Hamiltonian is

p2 1 1 2
+ mω02 x2 = P (t) + Q2 (t) = E

H =T +V = (39.4.13)
2m 2 2
In the above, P is a normalized momentum and Q is a normalized coordinate, and their
squares have the unit of energy. We have also shown that when the classical pendulum
is elevated to be a quantum pendulum, then H → Ĥ, where Ĥ = ~ωl ↠â + 12 . Then

Schrödinger equation becomes

 
1
Ĥ|ψ, ti = ~ωl ↠â + |ψ, ti = i~∂t |ψ, ti (39.4.14)
2

Our next task is to connect the electromagnetic oscillator to this pendulum. In general,
the total energy or the Hamiltonian of an electromagnetic system is
  
1 2 1 2
H= dr εE (r, t) + B (r, t) . (39.4.15)
2 µ
V

It is customary to write this Hamiltonian in terms of scalar and vector potentials. For
simplicity, we use a 1D cavity, and let A = x̂Ax , ∇ · A = 0 so that ∂x Ax = 0, and letting
.
Φ = 0. Then B = ∇ × A and E = −A, and the classical Hamiltonian from (39.4.15) for a
Quantum Coherent State of Light 413

Maxwellian system becomes

  
1 . 1
2 2
H= dr εA (r, t) + (∇ × A(r, t)) . (39.4.16)
2 µ
V
.
For the 1D case, the above implies that By = ∂z Ax , and Ex = −∂t Ax = −Ax . Hence, we
let

Ax = A0 (t) sin(kl z) (39.4.17)

.
Ex = −A0 (t) sin(kl z) = E0 (t) sin(kl z) (39.4.18)
By = kl A0 (t) cos(kl z). (39.4.19)
.

L
where E0 (t) = −A0 (t). After integrating over the volume such that V
dr = A 0
dz, the
Hamiltonian (39.4.16) then becomes
V0 ε  . 2 V
0 2 2
H= A0 (t) + k A (t). (39.4.20)
4 4µ l 0
where V0 = AL, is the mode volume. The form of (39.4.20) now resembles the pendulum
Hamiltonian. We can think of A0 (t) as being related to the displacement of the pendulum.
Hence, the second term resembles the potential energy. The first term has the time derivative
of A0 (t), and hence, can be connected to the kinetic energy of the system. Therefore, we can
rewrite the Hamiltonian as
1 2
P (t) + Q2 (t)

H= (39.4.21)
2
where r r s
V0 ε . V0 ε V0
P (t) = A0 (t) = − E0 (t), Q(t) = kl A0 (t) (39.4.22)
2 2 2µ
By elevating P and Q to be quantum operators,
p p
P (t) → P̂ = ~ωl π̂(t), ˆ
Q(t) → Q̂ = ~ωl ξ(t) (39.4.23)

so that the quantum Hamiltonian now is

1h 2 i 1  
Ĥ = P̂ + Q̂2 = ~ωl π̂ 2 + ξˆ2 (39.4.24)
2 2
similar to (39.1.3) as before except now that the resonant frequency of this mode is ωl instead
of ω0 because these are the cavity modes, each of which is homomorphic to a quantum
pendulum of frequency ωl . An equation of motion for the state of the quantum system can
be associated with the quantum Hamiltonian just as in the quantum pendulum case.
We have shown previously that
√
↠+ â = 2ξˆ (39.4.25)
†
√
â − â = − 2iπ̂ (39.4.26)
414 Electromagnetic Field Theory

Then we can let r

V0 ε p
P̂ = − Ê0 = ~ωl π̂ (39.4.27)
2
Finally, we arrive at
r r
2~ωl 1 ~ωl †
Eˆ0 = −


π̂ = â − â (39.4.28)
εV0 i εV0

Now that E0 has been elevated to be a quantum operator Ê0 , from (39.4.18), we can put in
the space dependence to get

Êx (z) = Ê0 sin(kl z) (39.4.29)

Consequently,
r
1 ~ωl †

Êx (z) = â − â sin(kl z) (39.4.30)
i εV0

Notice that in the above, Ê0 , and Êx (z) are all Hermitian operators and they correspond to
quantum observables that have randomness associated with them. Also, the operators are
independent of time because they are in the Schrödinger picture.
In the Schrödinger picture, to get time dependence fields, one has to take the expectation
value of the operators with respect to time-varying quantum state vector like the time-varying
coherent state.
To let Eˆx have any meaning, it should act on a quantum state. For example,

|ψE i = Eˆx |ψi (39.4.31)

Notice that thus far, all the operators derived are independent of time. To derive time
dependence of these operators, one needs to find their expectation value with respect to
time-dependent state vectors.9
To illustrate this, we can take expectation value of the quantum operator Êx (z) with
respect to a time dependent state vector, like the time-dependent coherent state, Thus
r
1 ~ωl
hEx (z, t)i = hα, t|Êx (z)|α, ti = hα, t|↠− â|α, ti
i εV0
r r
1 ~ωl ∗ ~ωl
= (α̃ (t) − α̃(t)) hα, t|α, ti = −2 =m(α̃) (39.4.32)
i εV0 εV0

Using the time-dependent α̃(t) = αe−iωl t = |α|e−i(ωl t+ψ) in the above, we have
r
~ωl
hEx (z, t)i = 2 |α| sin(ωl t + ψ) (39.4.33)
εV0
where α̃(t) = αe−iωl t . The expectation value of the operator with respect to a time-varying
quantum state, in fact, gives rise to a time-varying quantity. The above, which is the average
9 This is known as the Schrödinger picture.
Quantum Coherent State of Light 415

of a random field, resembles a classical field. But since it is rooted in a random variable, it
has a standard deviation in addition to having a mean.
We can also show that
r r
2µ~ωl ˆ µ~ωl †
B̂y (z) = kl Â0 cos(kl z) = ξ= (â + â) (39.4.34)
V0 V0
Again, these are time-independent operators in the Schrödinger picture. To get time-dependent
quantities, we have to take the expectation value of the above operator with respect to to a
time-varying quantum state.

39.5 Epilogue
In conclusion, the quantum theory of light is a rather complex subject. It cannot be taught
in just two lectures, but what we wish is to give you a peek into this theory. It takes much
longer to learn this subject well: after all, it is the by product of a century of intellectual
exercise. This knowledge is still very much in its infancy. Hopefully, the more we teach
this subject, the better we can articulate, understand, and explain this subject. When James
Clerk Maxwell completed the theory of electromagnetics over 150 years ago, and wrote a book
on the topic, rumor has it that most people could not read beyond the first 50 pages of his
book [256]. But after over a century and a half of regurgitation, we can now teach the subject
to undergraduate students! When Maxwell put his final stroke to the equations named after
him, he could never have foreseen that these equations are valid from nano-meter lengthscales
to galactic lengthscales, from static to ultra-violet frequencies. Now, these equations are even
valid from classical to the quantum world as well!
Hopefully, by introducing these frontier knowledge in this course, it will pique your inter-
est enough in electromagnetic field theory, so that you will take this as a life-long learning
experience.
416 Electromagnetic Field Theory