Chapter 4

Quantum Light, Atoms and the

Interaction Picture

4.1 The Quantum Harmonic Oscillator

Light was one of the first systems to be studied in the context of the emerging the-
ory of quantum mechanics. The idea that light should be quantised into packets of
energy, “photons”, stemming from Einstein’s reasoning about the photo-electric
effect, was one of the earliest clues that quantum mechanics would have a radi-
cally different structure to classical theories. The quantum description of light is
very closely related to a system which you will have studied previously, the quan-
tum harmonic oscillator (QHO). For that reason, before we quantise light we will
revise the properties of the QHO.
A QHO has the following Hamiltonian,

p̂2 1
Ĥ = + mω 2 x̂2 . (4.1)
2m 2
The eigenstates of the Hamiltonian are a set of energy levels labelled |ni,
where n is an integer, with equally spaced energies En = (n + 1/2)~ω. We can
think of the energy being quantised into equal packets of energy each of energy
~ω. It is convenient to study QHO using ladder operators
r
mω
â = (x̂ + ip̂) ↠= (â)† (4.2)
2~
These operators are called lowering and raising operators (or annihilation and
104 Chapter 4. Quantum Light, Atoms and the Interaction Picture

creation operators) since they act on the states |ni as follows:

√ √
â|ni = n|n − 1i ↠|ni = n + 1|n + 1i (4.3)

and satisfy the commutation relation [â, ↠] = 1. Combining these expressions
we find:
↠â|ni = n|ni (4.4)
The operator ↠â is called the number operator since its eigenvalue is n the number
of excitations in state |ni.
Since
r
~
x̂ = (↠+ â)
r 2mω (4.5)
~
p̂ = i(↠− â)
2mω
we can rewrite the Hamiltonian in the ladder operator picture as

H = ~ω(↠â + 1/2) (4.6)

4.2 Quantum Light

Quantum theory is not derived from classical theory, however one can often obtain
a quantum description of a system for which a classical description is known by
a method known as quantisation, where classical observables e.g. x and p are
replaced by quantum operators x̂ and p̂ satisfying an appropriate commutation
relation. There are standard methods which can be used to do this (e.g. a method
known as “canonical quantisation”) but all such methods are to a certain extent ad
hoc. The replacement of classical values by quantum operators is essentially good
guesswork, which must then be verified by experiment.
In this section we will see how a quantum description of light may be ob-
tained starting with the classical Maxwell equations, casting them in a form in
which simple harmonic motion can be identified, and then replacing the classical
oscillators with quantum oscillators as described in the previous section.
The Maxwell equations in vacuum
∂B ∂E
∇×E =− ∇ × B = µ0 0
∂t ∂t (4.7)
∇·E =0 ∇·B =0

Phas0069 PHASM/G426 Advanced Quantum Theory

4.2. Quantum Light 105

can be combined to derive the electric field wave-equation

1 ∂2
 
2
∇ − 2 2 E=0 (4.8)
c ∂t

where 1/c2 = µ0 0 .
We can solve this equation using the separation of variables technique, writing
E(r, t) = u(r)T (t) where r is the vector position and t is time,

(∇2 u(r))T (t) u(r) ∂ 2 T 1 ω2

= 2 = −
u(r)T (t) c ∂t2 u(r)T (t) c2
(4.9)
(∇2 u(r)) 1 ∂ 2T 1 ω2
= 2 2 =− 2
u(r) c ∂t T (t) c

which gives us two ω dependent equations (−ω/c2 is chosen to be the separation

constant for convenience).
The spatial equation, called the Helmholtz equation is

ω2
∇2 u(r) = − u(r). (4.10)
c2
This is a vector equation, and has many solutions depending on ω and boundary
conditions (i.e. the presence of conducting surfaces such as mirrors).
The temporal equation takes the form of a simple Harmonic oscillator equation

∂ 2 T (t)
2
= −ω 2 T (t) (4.11)
∂t
the solutions of which will have the form exp[±iωt].
Labelling the solutions to these equations Tj (t) and uj (r), where ωj is the
associated frequency, to obtain a general solution we would compute all the solu-
tions to these equations for all values of ω and combine them linearly
X
E(r, t) = αj uj (r)Tj (t) (4.12)
j

where E0j is a unit vector representing the polarisation of the mode and the αj
coefficients depend on the initial conditions. We can think of this classical solution
as representing a sum of independent oscillators, each associated with a particular
106 Chapter 4. Quantum Light, Atoms and the Interaction Picture

spatial function uj (r) and oscillating (according to Tj (t) as an oscillator with a

frequency which we will label ωj 1 .
Each of these oscillators is called a mode, and we call uj (r) for a particular
mode its mode function and ωj its frequency. Analogous equations occur in acous-
tics. There the modes are often called resonant modes, each of which vibrates as
an independent oscillator with a certain frequency.
The quantisation of the EM-field proceeds by replacing each of these classical
oscillators with a quantum oscillator. In other words, each mode will be repre-
sented by a separate quantum harmonic oscillator. The energy eigenstates of each
oscillator |ni, represents a state of n photons in the mode. We shall label the states
and operators for mode j with a subscript j |nij , aj , etc.
Thus each mode has a Hamiltonian

Hj = ~ωj (a†j aj + 1/2) (4.13)

P
and the Hamiltonian for the full field is H = j Hj .
The electric field is now represented by a quantum operator
X
Ê = uj (r)E0j (aj + a†j ) (4.14)
j

where E0j has the units of electric field and other quantities are dimensionless.

Optical Cavities
In free space, light propagates at speed c, however, it is possible to trap light inside
a set of mirrors, called a cavity. The quantum properties of light are more easily
probed and, as we shall see below, interactions with atoms can be enhanced or
suppressed. In particular, it is possible to cause the atom to interact with a single
mode alone. This single mode can be studied as an ideal quantum oscillator,
and such interactions have been an important experimental testing ground for the
predictions of quantum mechanics.
Below we shall consider the interaction between an atom and a single cavity
mode. We shall write the eigenstates of the mode |ni and the mode operators a
and a† . The Hamiltonian for the mode is written

H = ~ω(a† a + 1/2) (4.15)

1
In some cases (e.g. in free space) we would have an integral rather than a sum, but, apart from
some technical subtleties, the interpretation is the same.

Phas0069 PHASM/G426 Advanced Quantum Theory

4.3. Interaction between light and an atom 107

and the electric field operator is

Ê = u(r)E0 (â + ↠) (4.16)

4.3 Interaction between light and an atom

The full theory of the interaction between light and matter would go beyond the
scope of this course. However, in the energy scales of quantum optics exper-
iments, this interaction can be very well approximated by the dipole approxi-
mation. This provides us with a Hamiltonian (known as the Jaynes-Cummings
Hamiltonian) which can be solved to study the interaction of atoms and light. We
shall derive this Hamiltonian in this chapter.
The starting point for the dipole approximation is to picture the atom as a
dipole, consisting of an outer electron of charge −e and the remainder of the atom,
charge +e. A classical dipole has dipole moment D ~ = q~r where ~r is the vector
describing the displacement between positive and negative charges. A classical
dipole in an electric field has interaction energy
~ ·E
U = −D ~ (4.17)

The Jaynes-Cummings interaction is derived from the quantised version of this

interaction energy, which gives an additional term which we add to our Hamil-
tonian (recall that the Hamiltonian represents the energies of the systems under
study).
We have already encountered a quantum mechanical electric field operator
above, so to derive this we require a quantum operator for the dipole moment.

The two-level atom

To simplify our task, it is common to make a further approximation. Atoms typ-
ically have a complicated level structure, and you will recall from past courses
how even the level structure of a Hydrogen atom is relatively complex. Here we
shall, instead, take a toy model of an atom, the two-level atom, and assume that
the atom’s states consist only of the ground state |gi and a single excited state |ei.
The energy of the ground state will be zero, and the energy of |ei will be ~ωa . The
Hamiltonian for this atom is therefore

H = 0|gihg| + ~ωa |eihe| = ~ωa |eihe| (4.18)

108 Chapter 4. Quantum Light, Atoms and the Interaction Picture

Although very simple, this model is a good approximation for an atom that is
interacting with a light mode at resonance, e.g. when the atomic transition energy
~ωa is equal to the energy of a photon in the mode ~ωj , i.e. when ωa = ωj , but all
transitions to other states are off resonance.
For a two-level atom it is straight-forward to derive a general form for the
dipole moment operator. The dipole operator D̂ for a two-level system can only
have 4 possible matrix elements and we analyse those in turn.
The dipole moment operator is proportional to the displacement operator r̂
for the displacement between electron and positive atomic core. The diagonal
elements of D̂ are thus proportional to expectations of r̂,

hg|D̂|gi = −ehg|r̂|gi
(4.19)
he|D̂|ei = −ehe|r̂|ei

For most atomic states these elements are zero since the electron wave-function is
symmetric about the nucleus (and thus the mean displacement is zero). We will
therefore assume that hg|D̂|gi = he|D̂|ei = 0.
This leaves the off-diagonal elements hg|D̂|ei = he|D̂|gi∗ . We shall leave
these general, except to assume that they are real (this assumption does not change
the physics and simplifies our analysis). The dipole moment is a 3-D vector in
space, and thus the matrix elements have three components. We thus have

hg|D̂|ei = he|D̂|gi = d~ (4.20)

where d~ is a real 3-D vector, which depends upon the atomic states in question.
Thus we obtain a simple form for D̂:
~
D̂ = d(|gihe| + |eihg|). (4.21)

Dipole interaction Hamiltonian

We can now write down a Hamiltonian to describe the interaction between atom
and light in the dipole approximation. The classical expression for the interaction
energy between dipole and field was U = −E ~ · D.
~ To obtain the quantised Hamil-
tonian we simply replace each of these quantities with the quantum operators in
equation (1.16) and (1.21) to achieve

H = (−d~ · u(r)E0 )(â + ↠)(|gihe| + |eihg|)

(4.22)
= ~g(r)(â + ↠)(|gihe| + |eihg|)

Phas0069 PHASM/G426 Advanced Quantum Theory

4.4. Interaction Picture 109

~
where ~g(r) = (−d·u(r)E0 ) represents the strength of the coupling between light
and atom.

Suppressed tensor product notation

In constructing the above operator, we have implicitly combined the state space
of the atom with the state space of the light. The joint space has a tensor product
structure, with basis states

|g, ni = |gi ⊗ |ni |e, ni = |ei ⊗ |ni (4.23)

However, to save space, and since notationally atom states and operators look very
different to light states and operators, it is conventional not to write the tensor
products.
When we write operators â and ↠, we are abbreviating 1 ⊗ â and 1 ⊗ â†
respectively. Likewise we write |gihe| for |gihe| ⊗ 1 etc.
These operators thus act on the joint states as in the following examples:
√ √
â|g, ni = n|g, n − 1i ↠|g, ni = n + 1|g, n + 1i
(4.24)
|gihe||e, ni = |g, ni |gihe||g, ni = 0

We are now able to write down the full Hamiltonian for the atom and light
mode, including their interaction:

H = ~ωj (↠â + 1/2) + ~ωa |eihe| + ~g(r)(â + ↠)(|gihe| + |eihg|) (4.25)

The dynamics of this system can be solved in the Heisenberg or Schrödinger

pictures, however, as we shall see in the next sections, this can be simplified sig-
nificantly by adopting the Interaction picture.

4.4 Interaction Picture

If we know the eigenstates, |φj i, of a constant Hamiltonian H, we can compute
the evolution of a system via the following formula
X
|ψ(t)i = αj e−iEj t/~ |φj i (4.26)
j

where the constants αj are determined by the initial conditions.

110 Chapter 4. Quantum Light, Atoms and the Interaction Picture

In many examples in quantum physics, in particular when we are studying the

interactions between systems, the Hamiltonian will be of the form

H = H0 + V (4.27)
where H0 is a solved Hamiltonian (i.e. whose eigenstates and energies are known).
Typically H0 will represent the Hamiltonian for the system if it were not interact-
ing, and V will represent extra terms introduced due to the interaction. We can
see that equation (1.25) is of this form, with

H0 = ~ωj (↠â + 1/2) + ~ωa |eihe| (4.28)

V = ~g(r)(â + ↠)(|gihe| + |eihg|) (4.29)

The eigenstates of H0 are known (|g, ni and |e, ni), and if there were no inter-
action, we could write down the evolution, without further calculation, via equa-
tion (1.26). The Interaction picture (IP) allows us to exploit the fact that part of
the Hamiltonian is solved to simplify our calculations. In the Heisenberg picture,
all dynamics are carried by evolution of observables, and in the Schrödinger pic-
ture, all dynamics are carried by the state. In the Interaction picture, we map the
dynamics associated with the solved part of the Hamiltonian (H0 ) onto operators,
and thus the evolution of the states reflects the interaction alone. This approach
usually simplifies the calculations, and is analogous to working in an appropriate
rotating frame when studying classical mechanics problems.
Our starting point is to define the evolution operator U0 (t)

U0 (t) = exp[−iH0 t/~]. (4.30)

This is sometimes called the free-evolution operator, and represents the dynamics
we would see if no interaction were present.
Let |ψiS (t) represent the time-evolving state in the Schrödinger picture. In
the interaction picture, we “remove” from this state the dynamics due to H0 , by
defining the interaction picture state |ψiI (t) as follows:

|ψiI (t) = U0 (t)† |ψiS (t) (4.31)

Since U0 (t)† is the inverse of U0 (t), this is equivalent to “unwinding” the evolution
in the state just due to H0 alone. It is usually easier to calculate |ψiI (t) (we shall
see below how to do this), and then apply U0 (t) to recover the Schrödinger picture.

Phas0069 PHASM/G426 Advanced Quantum Theory

4.4. Interaction Picture 111

Expectation values must be the same in all pictures, which mean that observ-
ables ÔI (t) in the interaction picture must evolve as follows

ÔI (t) = U0† (t)ÔS U0 (t) (4.32)

where ÔS is the Schrödinger picture observable. You can see that in the Interaction
picture, similar to the Heisenberg picture, observables evolve in time, however,
they evolve under U0 (t) rather than the full evolution operator U (t). The following
table summarises the similarities and differences between all three pictures.
Picture Observables States
Schrödinger ÔS |ψ(t)iS = U (t)|ψ(0)iS
Heisenberg ÔH = U † (t)ÔS U (t) |ψiH = |ψ(0)iS
Interaction ÔI = U0† (t)ÔS U0 (t) |ψ(t)iI = U0† (t)|ψ(t)iS
All three pictures coincide at t = 0.
As an indication of why the Interaction picture will simplify calculations, con-
sider the following example. Let |φk i represent an orthonormal basis. We can
therefore represent an Interaction picture state as
X
|ψ(t)iI = cj (t)|φk i (4.33)
j

Transforming back to the Schrödinger picture, the state will be

X
|ψ(t)iS = U0 (t)|ψ(t)iI = cj (t) exp[−iEj t/~]|φk i (4.34)
j

From this we see that the interaction picture focuses on the evolution of co-
efficients cj (t), ignoring the phases exp[−iEj t/~] which are due to the non-
interaction part of the Hamiltonian.

Interaction Picture Schrödinger picture

The main advantage of the Interaction picture is that it simplifies the form of
the Schrödinger equation. To derive the Schrödinger equation in the Interaction
picture let us start with the standard Schrödinger equation:

∂
i~ |ψS (t)i = H|ψS (t)i (4.35)
∂t
112 Chapter 4. Quantum Light, Atoms and the Interaction Picture

and substitute in the defining identity |ψ(t)iS = U0 (t)|ψ(t)iI . We then use the
product rule to transform the left-hand-side of the above equation:
   
∂ ∂|ψ(t)iI ∂U0 (t)
i~ U0 (t)|ψ(t)iI = i~ U0 (t) + |ψ(t)iI
∂t ∂t ∂t
  (4.36)
∂|ψ(t)iI −iH0
= i~ U0 (t) + U0 (t)|ψ(t)iI
∂t ~

Hence,
∂|ψ(t)iI
i~U0 (t) + H0 U0 (t)|ψ(t)iI = HU0 (t)|ψ(t)iI (4.37)
∂t
and multiplying from the left by U0† (t) and rearranging, we obtain

∂|ψ(t)iI
i~ = U0† (t)(H − H0 )U0 (t)|ψ(t)iI = U0† (t)V U0 (t)|ψ(t)iI (4.38)
∂t
This has a very similar form to the standard Schrödinger equation, and indeed
if we define an Interaction Picture Hamiltonian HI = U0† (t)V U0 (t), we have
derived the Interaction Picture Schrödinger equation,

∂|ψ(t)iI
i~ = HI |ψ(t)iI (4.39)
∂t
This has exactly the same form as the standard SE and we can use the same tech-
niques to solve it. The key difference is that the Interaction Picture Hamiltonian
has few terms in it, simplifying the calculations.

Example: Atom-cavity coupling

In the past section, we derived a Hamiltonian for the interaction between an atom
and a single mode of light (e.g. inside an optical cavity). This Hamiltonian was
of the form H = H0 + V with

H0 = ~ωj (↠â + 1/2) + ~ωa |eihe| (4.40)

and
V = ~g(r)(â + ↠)(|gihe| + |eihg|) (4.41)
We will study this evolution in the interaction picture. To do so we need
to compute the Interaction Picture Hamiltonian HI = U0† (t)V U0 (t). This seems

Phas0069 PHASM/G426 Advanced Quantum Theory

4.4. Interaction Picture 113

daunting at first sight, but we can break the calculation down. The first observation
is that ↠â and |eihe| commute, so we can write
U0 (t) = exp[−iωj (↠â + 1/2)t] exp[−iωa |eihe|t] (4.42)
Since the mode operators â and ↠commute with the atomic operators |gihe|, etc.,
we can group the two different types of operator in HI together
HI = U0† (t)V U0 (t)
= ~g(r) exp[+iωj (↠â + 1/2)t](â + ↠) exp[−iωj (↠â + 1/2)t] (4.43)
exp[+iωa |eihe|t](|gihe| + |eihg|) exp[−iωa |eihe|t]
We then proceed term by term
exp[+iωj (↠â + 1/2)t]â exp[−iωj (↠â + 1/2)t]
(4.44)
= exp[+iωj ↠ât]â exp[−iωj ↠ât]
We can simplify this by using
exp[+iωj ↠ât]|ni = exp[+iωj nt]|ni (4.45)
which holds because |ni is an eigenstate of ↠â with eigenvalue n, and can be
verified using the power series for the exponential, and also
X√
â = n + 1|nihn + 1| (4.46)
n

to derive
exp[+iωj ↠ât]â exp[−iωj ↠ât]
X√
= n + 1 exp[+iωj nt]|nihn + 1| exp[−iωj (n + 1)t]
n (4.47)
X√
= exp[−iωj t] n + 1|nihn + 1| = exp[−iωj t]â
n

and taking the Hermitian conjugate of this we obtain:

exp[+iωj ↠ât]↠exp[−iωj ↠ât] = exp[+iωj t]↠. (4.48)
Hence
exp[+iωj (↠â+1/2)t](â+↠) exp[−iωj (↠â+1/2)t] = exp[−iωj t]â+exp[+iωj t]â†
(4.49)
114 Chapter 4. Quantum Light, Atoms and the Interaction Picture

Similarly, we can show that (this is left as an exercise)

exp[+iωa |eihe|t](|gihe|+|eihg|) exp[−iωa |eihe|t] = exp[−iωa t]|gihe|+exp[+iωa t]|eihg|

(4.50)
Thus, the IP Hamiltonian is

HI = ~g(r)(exp[−iωj t]â + exp[+iωj t]↠)(exp[−iωa t]|gihe| + exp[+iωa t]|eihg|)

= ~g(r) exp[−i(ωj − ωa )t]â|eihg| + exp[i(ωj − ωa )t]↠|gihe|
+ exp[−i(ωj + ωa )t]â|gihe| + exp[i(ωj + ωa )t]↠|eihg|

(4.51)

We mentioned above that an atom and cavity mode have a particularly strong
interaction when they are at resonance, i.e. when the transition energy ~ωa is close
to photon energy ~ωj . Here we shall assume exact resonance ωa = ωj = ω where
we shall use ω to stand for either quantity.
At resonance, the interaction picture Hamiltonian is

HI = ~g(r) â|eihg| + ↠|gihe|

(4.52)
+ exp[−i2ωt]â|gihe| + exp[i2ωt]↠|eihg|

Rotating wave approximation

In equation (1.52) we see that 2 of the terms are time-independent, while 2 terms
are oscillating at angular frequency 2ω. In the Rotating Wave Approximation, we
neglect those fast oscillating terms. This is justified when the period of oscillation
is much shorter than the time-scales of the most significant physical processes.
Loosely speaking, the fast oscillating terms have little effect on the time-evolution
at longer time-scales since they average to zero. The RWA can be shown to be
a very good approximation in cavity QED experiments. Making the RWA, and
dropping the final two terms we obtain a very simple form for HI

HI = ~g â|eihg| + ↠|gihe|

(4.53)

A good way to remember which terms are retained in the RWA is to note that
the terms we kept correspond to an energy conserving process (absorbtion of a
photon + excitation of atom, or emission of photon plus deexcitation of the atom),
whereas the terms we dropped describe a non-energy conserving process (photon
absorbed plus atom excited, etc.).

Phas0069 PHASM/G426 Advanced Quantum Theory

4.4. Interaction Picture 115

Example: Single photon in a cavity

We shall now solve the IP Schrödinger equation for the case where initially, there
is a single photon in the cavity and the atom is in its ground state, e.g. |ψ(0)iI =
|g, 1i.
Notice that
HI |g, 1i = ~g|e, 0i HI |e, 0i = ~g|g, 1i (4.54)
This means that HI only couples |e, 0i and |g, 1i, which means that the general
solution to the IP Schrödinger equation with this initial condition will have the
form
|ψ(t)iI = α(t)|g, 1i + β(t)|e, 0i (4.55)
To solve it, we need to compute α(t) and β(t). Substituting equation (1.55)
into the IP Schrödinger equation, and multiplying from the left with hg, 1| or he, 0|
respectively we obtain two coupled differential equations
∂α(t)
= −igβ(t)
∂t (4.56)
∂β(t)
= −igα(t)
∂t
Applying the boundary conditions α(0) = 1 and β(0) = 0 we can solve these by
standard methods to obtain the solution
α(t) = cos(gt) β(t) = −i sin(gt) (4.57)
and hence
|ψ(t)iI = cos(gt)|g, 1i − i sin(gt)|e, 0i. (4.58)
We see that the system oscillates back and forward between these two states. This
phenomenon is a signature of a strong atom cavity coupling and is called Rabi
oscillation.
To transform the state back to the Schrödinger picture, we apply the operator
U0 (t), which adds the phases exp[−iEj t/~] to each term, where Ej is the energy
of that eigenstate. Both states above have the same energy (3/2)~ω, and hence
the Schrödinger depiction of the state is
|ψ(t)iS = U0 (t)|ψ(t)iI = exp[−i3ωt/2](cos(gt)|g, 1i − i sin(gt)|e, 0i). (4.59)
Notice that the interaction picture contained all the important dynamics of the
interaction, and transforming back has just added (less important) phases to the
evolution.
116 Chapter 4. Quantum Light, Atoms and the Interaction Picture

Rabi oscillation occurs when the cavity is prepared in any Fock state |ni, and
via similar methods to those used here one can show that a similar oscillation
occurs when the system is prepared in |g, ni. Indeed we find
√ √
|ψ(t)iI = cos( ngt)|g, ni − i sin( ngt)|e, n − 1i. (4.60)

Phas0069 PHASM/G426 Advanced Quantum Theory