Exercise sheets for Introduction to Quantum Optics

Carlos Navarrete-Benlloch⇤
Wilczek Quantum Center - School of Physics and Astronomy - Shanghai Jiao Tong University
(Dated: Spring Semester 2023)
This document contains the exercise sheets of the di↵erent chapters that we have gone through in
the Introduction to Quantum Optics course (PHY3761) at Zhiyuan College of Shanghai Jiao Tong
University (www.carlosnb.com/quantum-optics-course). The notation and tone matches exactly
that of the lecture notes for the course. The sheet for each chapter is divided in two parts. First,
I provide the key questions that I think students should know how to answer, which will form the
backbone of the exam. The goal of these questions is to help students know what they should study.
Second, I provide exercises that can be worked out with the tools developed in the chapter, and
which students should hand in at the agreed times to be announced during the lectures (they will
account for 40% of the final grade).

⇤ [email protected]; www.carlosnb.com
 2

I. BRIEF REVIEW OF QUANTUM MECHANICS

A. Key questions of the chapter

1. The theory of Hilbert spaces is the basic language of quantum mechanics. It’s important that you feel comfortable
with it, for which you should study carefully Appendix A.2 (which hopefully will be just a reminder of concepts
you have already studied in detail before). After that, you should be able to answer confidently and correctly
the following questions:
(a) What is a complex vector space?
(b) What’s an inner product? and an outer product?
(c) What’s a ket? and a bra?
(d) What’s a Euclidean space? is it the same a Hilbert space?
(e) What’s a linear operator? Enumerate the most important types of operators and their properties? How
about the most important operations between or on them?
(f) What’s a basis of a Hilbert space? Is it unique?
(g) Given an infinite-dimensional Euclidean space, what’s the defining property of the vectors contained in the
corresponding Hilbert space? Are the remaining vectors not contained in the Hilbert space still useful?
Can they form a basis of the Hilbert space? What’s a continuous representations?
(h) Do you understand the most characteristic examples of infinite-dimensional spaces, denoted by l2 (1) and
L2 (x) spaces? What do we mean when we say that they are isomorphic to each other? And when we say
that all infinite-dimensional Hilbert spaces are isomorphic to them?
(i) What’s a tensor product?
2. Quantum mechanics is the theory with which we will approach the electromagnetic field and matter in this
course. Hence, you should also be very much acquainted with its basic postulates and how to apply them (in
fact, this course will hopefully serve you to master them even further). In particular, after studying carefully
Chapter I and Appendix A.3 (or even before, if you remember well your quantum mechanics lectures), you
should be able to answer correctly and confidently the following questions:
(a) How do we describe the state of the degrees of freedom of a system in quantum mechanics?
(b) How about measurable quantities (observables)? What are the possible outcomes of a measurement of such
a quantity? Can you predict the measurement outcomes with certainty? How about its statistics?
(c) Can we prepare systems in such a state that we will know with full certainty the outcomes of measurements
of all observables?
(d) What are the canonical commutation relations? What are they useful for?
(e) Why do we care about the so-called expectation value of an observable? And about its variance?
(f) What’s the Schrödinger equation? and the Heisenberg equation? What are they used for in quantum
mechanics? In which sense are they equivalent?
(g) What are composite systems? How do we build their Hilbert space from the Hilbert spaces of the subsystems
that compose them? What are the main properties of the tensor product map? Can you build a basis of
the total Hilbert space form the bases of the subspaces? How about building operators acting on the total
Hilbert space, from operators acting on the subspaces?
(h) What are entangled states? Why can’t the correlations present on these states be present in systems
described by classical physics?
(i) What are the properties of density operators?And their ensemble decomposition? Is the latter unique?
(j) Show that density operators naturally represent the mixed state of noisy systems, for which state prepara-
tion is imperfect.
(k) Given a composite system, show that the reduced state, obtained by tracing out one of the subsystems,
represents the state of the remaining parts of the system.
(l) Show that, even when the state of the full system is pure, the reduced state can be mixed if there is
entanglement between the subsystems. Use this approach to show that ensemble decompositions are not
unique, even though the reduced state is.
 3

(m) Provide an informational interpretation of mixedness by arguing that it represents our ignorance about
information that has ‘leaked out’ of the system. What are maximally-mixed states and how do they
reinforce this interpretation in terms of ignorance? Can mixedness be properly quantified?
(n) When the system is in a given mixed state ⇢ˆ, how can you evaluate expectation values of operators? And
the probability of obtaining a certain outcome when measuring an observable? How about the evolution
of the state of the system? What is the von Neumann equation?
 4

B. Graded exercise 1a: Practical quantum mechanics with pure states

Consider a 3-dimensional Hilbert space a basis {|bj i}j=1,2,3 defined on it. Consider also a vector | i = N (2|b1 i +
2|b2 i + |b3 i), where N is an arbitrary positive number.
1. [2.5] Show that N = 1/3 if we want | i to be a valid quantum state.
Hint: remember that the probabilistic interpretation of quantum mechanics requires states to be normalized to
a specific value.
2. [2.5] Show that the representation of | i in the basis {|bj i}j=1,2,3 is given by
0 1
1 2
~ = @ 2 A. (1)
3 1

Hint: just check the definition of representations in the notes.

3. [7.5] Consider the Hamiltonian Ĥ = ~(|b1 ihb2 | + |b2 ihb1 |). Show that it possess the following eigenvectors and
eigenvalues

|e1 i = p12 (|b1 i + |b2 i), E1 = ~,

|e2 i = p12 (|b1 i |b2 i), E2 = ~, (2)
|e3 i = |b3 i, E3 = 0.

Hint: Use the matrix representation of Ĥ, denoted by H, and your knowledge of linear algebra.
4. [7.5] Use the Schrödinger equation to show that, starting from | (0)i = | i, the state of the system at any other
time is given by
0 1
1 1 2e it
| (t)i = 2e it |b1 i + 2e it |b2 i + |b3 i , ~ (t) = @ 2e it A , (3)
3 3 1

where the representation is in the {|bj i}j=1,2,3 basis.

Hint: Represent the Schrödinger equation in any of the bases that you know, and solve the linear di↵erential
equations with any of the tools that you have learned in your math courses.
5. [10] Show that the unitary evolution operator Û (t) = exp(Ĥt/i~) has the following representation in the
{|bj i}j=1,2,3 basis
0 1
cos t i sin t 0
U (t) = @ i sin t cos t 0 A . (4)
0 0 1

How about its representation in the {|ej i}j=1,2,3 basis?

Hint: Use the spectral theorem given your knowledge of the eigensystem of Ĥ (check the spectral theorem in
Appendix A.2.b of the notes if you are not familiar with it).

6. [5] Find Eq. (3) again, this time using | (t)i = Û (t)| (0)i.
Hint: As usual, it comes handy to use representations.
7. [20] Show that the probabilities of obtaining the outcomes Ej when measuring the energy at some time t, denoted
by pj (t), are given by
8 1
p1 (t) = , p2 (t) = 0, p3 (t) = . (5)
9 9
Show also that, at any time, the mean energy is equal to 8~/9, with a variance equal to 8~2 /81.
Is there any special reason why all these statistical objects are time independent, even though the state evolves
nontrivially in time?
Hint: Just use the various definitions given in the notes. Again, representations might come handy.
 5

8. [25] Consider the observable  = i(|b3 ihb2 | |b2 ihb3 |) + |b1 ihb1 |. Show that its eigensystem is given by

|a1 i = |b1 i, a1 = 1,
|a2 i = p12 (|b2 i + i|b3 i), a2 = 1, (6)
|a3 i = p12 (|b2 i i|b3 i), a3 = 1.

Given the state | (t)i, show that the probabilities P±1 (t) of obtaining the outcomes ±1 in a measurement of Â
at time t, are given by
13 2 5 2
P1 (t) = + sin t, P 1 (t) = sin t. (7)
18 9 18 9

Finally, show that the expectation value of  evolves as 4(1+sin t)/9, with a variance given by 1 [4(1+sin t)/9]2 .
Hint: Keep in mind that the outcome can be 1 through two possible eigenvalues, such that you’ll have to add
up their individual probabilities.
9. [10] Probably you used the Schrödinger picture to find the expectation value of  and its variance (otherwise, do
it here). Use the time-evolution operator that you found in (4) to show that the representation of the time-evolved
operator Â(t) is given by
0 i
1
cos2 t 2 sin 2t sin t
A(t) = @ 2i sin 2t sin2 t i cos t A , (8)
sin t i cos t 0

in the {|bj i}j=1,2,3 basis. Evaluate then again the expectation value and variance of the previous section in this
picture.
Hint: Don’t use the Heisenberg equations of motion, unless you really like to work unnecessarily hard. Just
stick to the definitions in terms of the evolution operator and use your preferred representation.
10. [10] Evaluate the commutator [Â, Ĥ] and check that the uncertainty relations are not violated at any time t.
Hint: Just stick to your preferred representation and you’ll be fine.
 6

C. Graded exercise 1b: Practical quantum mechanics with mixed states

Consider a composite system described by the Hilbert space H = A ⌦ B, and the orthonormal bases {|aj i}j=1,2
and {|bl i}l=1,2,3 of A and B, respectively.
1. [5] What are the dimensions of the Hilbert spaces A, B, and H?
2. [10] Suppose that the composite system is in the pure state
p
| i = 2p|'1 i ⌦ | 1 i + |'2 i ⌦ | 2i + |'3 i ⌦ | 3 i, (9)
with
1 1
| 1i = |b1 i, | 2i = p (|b2 i + |b3 i), | 3i = p (|b1 i + |b2 i), (10)
2 2
and
✓p p ◆ ✓p p ◆
1 p 1 p p+ 1 p
|'1 i = p (|a1 i + |a2 i), |'2 i = p |a1 i + p |a2 i = |'3 i, (11)
2 2 2
and where p 2 [0, 1]. Show that | i is normalized to 1.
3. [15] Show that the state | i of Eq. (9) can be written in the form
3
X p
| i= wm | mi ⌦ |bm i, (12)
m=1

with w1 = w3 = 1/2 and w2 = 0, and find the corresponding states {| m i}m=1,2,3 .

4. [15] Show that the representation of the reduced density operator ⇢ˆA = trB {| ih |} in the basis {|aj i}j=1,2 is
given by
✓ ◆
1 1 2p 1
⇢A = . (13)
2 2p 1 1

5. [15] Find the the von Neumann entropy S[ˆ

⇢A ] = trA {ˆ
⇢A log ⇢ˆA }, and discuss the result as a function of p.
6. [10] Show that the state | i of Eqs. (9) or (12) can also be written in the alternative form
2
X p
| i= vm |am i ⌦ | m i, (14)
m=1

with v1 = v2 = 1/2, and states

✓pp ◆ p✓p ◆
p+ 1 p 1 p p
| 1i = p |b1 i + p |b3 i,
2 2
✓p p ◆ ✓p p ◆
p 1 p p+ 1 p
| 2i = p |b1 i + p |b3 i. (15a)
2 2

7. [15] Use expression (14) of the state to show that the representation of the reduced density operator ⇢ˆB =
trA {| ih |} in the basis {|bj i}j=1,2,3 is given by
0 1
1@ 1 0 2p 1
⇢B = 0 0 0 A, (16)
2 2p 1 0 1

and show then that S[ˆ ⇢B ] = S[ˆ

⇢A ]. Do you think that this equality is just a coincidence particular to our choice
of state | i, or is there something deeper to it? Provide physical arguments to support your answer.
8. [15] Can you provide an ensemble decomposition for ⇢ˆA di↵erent than {wm , | m i}m=1,2,3 ?
 7

II. QUANTIZATION OF THE ELECTROMAGNETIC FIELD

AS A COLLECTION OF HARMONIC OSCILLATORS

A. Key questions of the chapter

1. Starting from the Maxwell equations, derive the wave equation for the vector potential. Explain the quasi-
1D approximation and, assuming perfectly conducting boundary conditions, write the corresponding vector
potential in terms of appropriate mode functions. Interpret physically the conditions that such boundaries
impose on the expansion coefficients and the allowed wave vectors.
2. Using the wave equation, prove that the expansion coefficients satisfy the dynamical equations of independent
harmonic oscillators. Show that choosing a proper normalization factor in the vector potential expansion, the
electromagnetic energy turns into the Hamiltonian for the independent oscillators. Quantize the electromagnetic
field, write down the fields in terms of annihilation and creation operators, as well as the Hamiltonian governing
the evolution of the empty cavity, and introduce the concept of photon.
3. Take the Hamiltonian of a harmonic oscillator in terms of its quadratures, Ĥo = ~!(X̂ 2 + P̂ 2 ) with [X̂, P̂ ] =
2i. Use the expansion of the quadratures in terms of annihilation and creation operators to diagonalize the
Hamiltonian, proving that the oscillator’s Hilbert space is infinite dimensional. Discuss the main physical
consequences that quantum physics imposes on the oscillator: energy quantization, zero-point fluctuations, and
absence of well-defined phase-space trajectories.
4. Can we describe quantum mechanics through a probability density function in phase space? Motivated by
this question, introduce physically and mathematically the Wigner function and its main properties. Show in
particular that it has the right marginals for quadrature measurements. Explain why it cannot be interpreted
as a true probability density function and argue that the statistics of quantum mechanics cannot be simulated
with standard noise, and hence quantum physics goes beyond classical physics.
5. Define Gaussian states, discuss their Wigner function, and give meaning to the mean vector and covariance
matrix required to parametrize them. Introduce a general criterion based on coherent states that provides a
necessary and sufficient condition for a state ⇢ˆ to be Gaussian. Introduce also the weaker criterion based on
Gaussian states being connected to other Gaussian states through Hamiltonians quadratic in the quadratures,
and explain what this criterion means for the possibility of generating non-Gaussian states experimentally.
6. Introduce coherent states as displaced vacuum states, and show that they are eigenstates of the annihilation
operator. Argue that they are Gaussian and find the corresponding Wigner function, interpreting its shape in
phase space. Use it to discuss the concepts of amplitude and phase of the oscillator, and how their fluctuations
can be related to a specific pair of quadratures under certain conditions. Find the representation of coherent
states in the Fock-state basis, and discuss the photon-number distribution they lead to. Show also that coherent
states form an overcomplete basis of the Hilbert space. Discuss how do coherent states provide the connection
between quantum and classical physics.
7. Introduce the quantum shot-noise limiting sensing and information encoding, and introduce squeezed states,
explaining how they allow us to go beyond such limit. Discuss the concept of minimum uncertainty states, and
show that coherent states are of this type, but not Fock states with n > 0. Introduce the squeezing operator
and find the transformation that it induces on the mean vector and covariance matrix of general states. Apply
it to the vacuum state to generate the squeezed vacuum state, and argue that the state is Gaussian, finding
the corresponding Wigner function and interpreting it in phase space. How can you transform this state into
amplitude-squeezed or phase-squeezed states? Represent the squeezed vacuum state into the Fock basis, and
discuss the properties of the corresponding distribution of quanta.
8. Explain why the concept of maximally-mixed state must be extended to infinite dimension by adding an energy
constrain, and show that this leads to thermal states. Discuss why we call them thermal states, through their
connection to statistical physics. Apply the definitions to the harmonic oscillator Hamiltonian Ĥ = ~!↠â,
writing the corresponding thermal state in the Fock basis. Prove that the state is Gaussian and compare its
Wigner function with that of vacuum.
 8

B. Graded exercise 2: Cat states

Consider the state |↵cat i = N↵ (|↵i + | ↵i), where | ± ↵i are coherent states, we take ↵ 2 R, and N is a suitable
normalization constant. This state is known as a cat state because it is a quantum superposition of two coherent
states, which are the most classical quantum states in the sense discussed in the lectures.
1. [10] Find the factor N↵ that normalizes the state.
Hint: just operate on h↵cat |↵cat i = 1
2. [20] Find the Fock-state representation of the state and discuss the corresponding photon-number probability
distribution.
Hint: just write the Fock representations of each coherent state and simplify.
3. [20] Find the quantum characteristic function |↵cat i (r) = hD̂(x, p)i and show that it is not Gaussian.
Hint: Use the complex representation of the displacement operator defining = (x + ip)/2, together with the
2 † ⇤
normal form D̂( ) = e | | /2 e â e â
. Use then the fact that coherent states are eigenstates of the annihilation
and creation operators, and you should be able to write the characteristic function as the sum of four exponentials
quadratic in x and p.
4. [20] Find the Wigner function W|↵cat i (r), and try to interpret each of the terms that add up to it. Set x = 0
and discuss the shape and negativities that you can observe along the p axis. R 2
Hint: Just Fourier-transform the characteristic function with the help of the Gaussian integral R dzeBz z /2A =
p 2
2⇡AeAB /2 valid for any A > 0 and B 2 C.
5. [30] Find the probability density functions for measurements
⇥ of the X̂ and P̂ quadratures. Compare
⇤ them with
the ones corresponding to the mixed state ⇢ˆ↵ = N↵2 |↵ih↵| + | ↵ih ↵| + 2 exp( 2↵2 )|0ih0| .
Hint: Find the marginals of the Wigner function with the help of the Gaussian integral of the previous question.
Note that the definition of the Wigner function is linear in the density operator, and hence, the Wigner function
of the mixed state ⇢ˆ↵ is just a mixture of the Wigner function of the states that form ⇢ˆ↵ .
 9

III. QUANTUM THEORY OF ATOMS AND THE TWO-LEVEL APPROXIMATION

A. Key questions of the chapter

1. Explain what we understand by “matter” in quantum optics. Using the example of the hydrogen atom and
atoms with a single-valence electron, show that the spectrum of matter systems is highly non-uniform.
2. Define the parity operator as the unitary transformation that inverts the sign of the relative atomic coordinate.
Show that it also inverts the sign of the relative momentum. Assuming that the atomic Hamiltonian is symmetric
under parity transformations, show that energy eigenstates with the same parity cannot be connected by the
relative-coordinate operator.
3. Justify the two-level approximation using intuitive arguments based on how the atom gets excited when irradi-
ated with a monochromatic field. Introduce the Pauli pseudo-spin operators associated to the two atomic states,
and write the atomic Hamiltonian in terms of ˆz within this approximation.
4. Using the defining properties of density operators, show that a general atomic state can be written as ⇢ˆ =
(Iˆ + bT ˆ )/2, where bj = hˆj i and |b|  1. Argue that b = 0 corresponds to the maximally-mixed state, while
states are pure if and only if |b| = 1. Introduce the Bloch space, and identify the points where the eigenstates
of the Pauli operators are located.
5. Write down a general time-dependent Hamiltonian and the corresponding Bloch equations (both in ordinary
and complex form).
6. Consider an atom evolving freely. Solve the corresponding Bloch equations and interpret the solution as pseudo-
spin precession, explaining how the dynamics looks in Bloch space.
7. Consider now the semiclassical Rabi Hamiltonian. Using the description of light-matter interaction within the
dipole approximation introduced in the next chapter, argue that this corresponds to the interaction of the
two-level atomic system with a classical monochromatic field.
8. Write down the complex Bloch equations associated to the semiclassical Rabi Hamiltonian and introduce the
rotating-wave approximation, justifying it by averaging the equations over an optical cycle.
9. Particularize the solution provided in the notes to the case in which the atom starts in the ground state,
explaining how the dynamics looks in the Bloch sphere. Discuss the frequency and amplitude of the oscillations
performed by the excited-state population. What happens when the detuning between the atomic transition
and the light beam is very large? How does this justify the two-level approximation?
 10

B. Graded exercise 3: Beyond the rotating-wave approximation.

Consider the semiclassical Rabi Hamiltonian Ĥ(t) = ~"ˆz /2 + ~⌦ cos(!t)ˆx , acting on a two-level atom initially in
the ground state.
1. [15] Writing the state as | (t)i = a(t)ei"t/2 |gi + b(t)e i"t/2 |ei with a(0) = 1 and b(0) = 0, use the Schrödinger
equation to show that the amplitudes obey the evolution equations
i"t
ȧ = i⌦e cos(!t)b, (17a)
ḃ = i⌦ei"t cos(!t)a. (17b)

2. [20] Within the rotating-wave approximation, neglect fast-oscillating terms in these equations (or use the
rotating-wave Hamiltonian Ĥ(t) = ~"ˆz /2 + ~⌦ ei!t ˆ + e i!t ˆ † /2 to find the equations of motion), and solve
them to find the solution
p !
⌦ 2 + ⌦2
b(t) = i p e i t/2 sin t , (18a)
2 + ⌦2 2
" p ! p !#
2 + ⌦2 2 + ⌦2
i t/2
a(t) = e cos t ip sin t , (18b)
2 2 + ⌦2 2

with = ! ".
Hint: Starting from the equations for a and b, you can find a second order di↵erential equation for b with
constant coefficients, which you can easily solve by, e.g., an exponential ansatz.
3. [20] Next you are going to apply a perturbative approach on the original equations (17), under the assumption
that ⌦ is small. Expanding the amplitudes as the power series
1
X 1
X
a(t) = ⌦n a(n) (t), and b(t) = ⌦n b(n) (t), (19)
n=0 n=0

plug these into the evolution equations and match the terms of the same power in ⌦. Prove then that they lead
to the recurrent equations

ȧ(n) = i cos(!t)e i"t (n 1)

b (t), [n = 1, 2, 3, ...] (20a)
(n) i"t (n 1)
ḃ = i cos(!t)e a (t), (20b)

starting at a(0) (t) = 1 and b(0) (t) = 0, arguing as well that the initial conditions of these equations are a(n>0) (0) =
0 = b(n>0) (0).
4. [25] Working on resonance, ! = ", prove that to second order in ⌦ we get

i ⌦
b(t) = ⌦t + 1 e2i"t , (21a)
2 4"
✓ ◆2
⌦ 2 t2 ⌦
a(t) = 1 1 e2i"t + 2i"te 2i"t
. (21b)
8 4"

5. [20] Compare this solution with the rotating-wave solutions (18) for times t ⌧ ⌦ 1 . Use this comparison to
argue that e↵ects beyond the rotating-wave approximation are suppressed by powers of ⌦/".
 11

IV. LIGHT-MATTER INTERACTION

A. Key questions of the chapter

1. Introduce the quantum description of light-matter interactions based on the dipole approximation for matter.
2. Apply this description to the interaction between the electromagnetic field of a cavity and a single atom, and
derive the quantum-Rabi and Jaynes-Cummings Hamiltonians, explaining the conditions under which these are
expected to hold.
3. Introduce the dressed states as the eigenstates of the Jaynes-Cummings Hamiltonian. Discuss the corresponding
energy spectrum, and compare it with that of the non-interacting case. Introduce also the concept of avoided
crossing from the analysis of the energy transitions between the ground state and the single-excitation manifold.
4. Find the evolution of an initial state consisting of an atom in the ground state and the field in an arbitrary
pure state. Evaluate the excited-state population, and introduce the concept of quantum Rabi oscillations by
assuming that the field starts in a Fock state.
5. Discuss the evolution of the excited-state population for a field starting in a coherent state. Make a qualitative
plot showing the relevant physical phenomena: semiclassical Rabi oscillations, collapses, and revivals. Use
approximations and reasonable arguments to find the time scale for each of these phenomena. How do these
results reconcile the quantum and semiclassical pictures?
6. Introduce the dipole model of a dielectric. Use it to show how Maxwell equations inside of them are modified
with respect to the ones in vacuum.
7. Discuss the e↵ect that the linear term of the polarization density has on the field entering the dielectric:
wavelength and amplitude reduction.
8. Use the macroscopic Maxwell equations to derive a wave equation for the electric field where the nonlinear terms
of the polarization density field act as a forcing or source term. Discuss then the frequency conversion phenomena
linked to second-order nonlinearity: second-harmonic generation, sum-frequency generation, di↵erence-frequency
generation, and down-conversion.
9. Write down the general interaction Hamiltonian of the electromagnetic field with the nonlinear dielectric. As-
sume that only two cavity modes with frequencies !0 and !2 ⇡ 2!0 and orthogonal polarization are relevant.
Use energy and momentum conservation arguments to write down the down-conversion Hamiltonian.
10. Perform a classical (parametric) approximation for the pump mode, and find the Hamiltonian of the down-
converted mode within this approximation. Introduce the Bogoliubov mode, and use it to diagonalize the
Hamiltonian in the | | > g region (explaining the meaning of the eigenstates), and to show that the Hamiltonian
is non-diagonalizable in the | | < g limit. What happens when | | = g? Starting from the vacuum state of the
electromagnetic field, find the evolution of the number of photons in the stable (| | > g) and unstable (| | < g)
regions.
 12

B. Graded exercise 4: Third-order (Kerr) nonlinearity

Consider an optical cavity containing a dielectric medium with inversion symmetry, for which P(2) (r, t) vanishes
identically. In these media P(3) (r, t) becomes then the dominant nonlinear term of the polarization density.
1. [10] Neglecting all cavity modes but one with frequency ! polarized along the x direction (you can further assume
that the phase ' of its spatial profile inside the dielectric is zero), show that
(3) ✓ ◆3/2
(3) jxxx ~!
P̂j (r) = i 1/2
sin3 (nx kz)(â ↠)3 . (22)
"0 nx LS

Hint: Just take the form of the single-mode field that we introduced in the lectures inside a dielectric medium,
and apply the definition of the third-order nonlinear polarization density.
2. [25] Assume that the dielectric medium has length l and is placed right at the surface of the left mirror, hence
covering the space z 2 [0, l] inside the cavity (as usual, we assume that the transverse dimensions of the dielectric
medium cover the whole cavity). Use the light-matter interaction Hamiltonian within the dipole approximation,
together with the rotating-wave approximation and the physical condition nx kl 1, to show that the dynamics
of the optical mode is ruled by the Hamiltonian

Ĥ = ~ (! + g) N̂ + ~g N̂ 2 , (23)

with
(3)
9 xxxx l~! 2
g= . (24)
4"0 Sn2x L2

Hint: Just apply the definition of the light-matter interaction Hamiltonian within the dipole approximation.
Write the result of the spatial integral along z in terms of sinc functions that fall to zero when their argument is
larger than one. Then, after keeping the energy-conserving terms within the rotating-wave approximation, you’ll
need to use the canonical commutation relations to write the Hamiltonian in terms of the number operator.
3. [20] What are the eigenstates and eigenenergies of this Hamiltonian? Make a graphic representation of the energy
levels for a given |g| ⌧ !, considering both the positive and negative cases, comparing it with the noninteracting
situation g = 0.
4. [45] Move to a picture rotating at frequency !+g, so that the Hamiltonian in the new picture reads as ĤI = ~g N̂ 2 .
Consider a coherent state |↵i as the initial state | (0)i. Find the state | (t)iI in the rotating picture at subsequent
times and use it to show that (take g > 0 from now on): (a) the state is periodic, with period T = 2⇡/g; (b) the
photon-number distribution doesn’t change in time; (c) at times tk = ⇡(2k + 1)/2g with k = 0, 2, 4, ..., the state
is equivalent to the cat superposition
i⇡/4
e
| (tk )iI = p (|↵i + i| ↵i) . (25)
2

Hint: Work in the Fock state basis.

 13

V. QUANTUM OPTICS IN OPEN SYSTEMS

A. Key questions of the chapter

1. Introduce a model for the field outside a cavity with a partially-transmitting mirror (open cavity). For this,
consider an external cavity that shares the partially-transmitting mirror with the main cavity. Write down
the vector potential associated to the external cavity, and show that the infinite-length limit leads to a field
composed of a continuous set of harmonic oscillators.
2. Focusing on a single mode of the main cavity, introduce the Hamiltonian that describes the whole “cavity +
external field” system. In particular, argue that the interaction through the partially-transmitting mirror can be
modeled with a simple photon tunneling Hamiltonian (sometimes referred to as beam splitter ), with a coupling
parameter that can be taken independent of the frequency provided the mirror’s reflectivity is large and slowly-
varying with the frequency. Explain as well why we can introduce fictitious negative-frequency external modes
that simplify enormously the math.
3. Working in the Heisenberg picture, introduce the equations that model the dynamics of the cavity mode (quan-
tum Langevin equation), showing that a damping term appears, together with an input operator that contains
information about the state of the external field. Show that the presence of a coherent component in the external
field (laser injection) can be accounted for with a time-dependent Hamiltonian linear in the intracavity annihi-
lation and creation operators. Evaluate also the type of statistics (expectation values and two-time correlators)
that the input operator has when the external field is in a thermal state (argue why you can assume the same
thermal state for all external modes, that is, a frequency-independent thermal excitation number).
4. Working from the Schrödinger picture, derive an approximate evolution equation for the cavity mode. Start
with the von Neumann equation for the density operator of the whole system, and make two picture changes:
first to a displaced picture where the state of the environment is a thermal state with no displacement, and then
to an interaction picture that discounts the evolution owed to all terms but the weak interaction. Introduce the
Born and non-backaction approximations, and show that they lead to an evolution equation that describes the
dynamics of the cavity mode up to second order in the interaction. Trace out the external field using a thermal
state as its initial state, finding the final Lindblad form for the master equation that rules the dynamics of the
cavity mode. Interpret the meaning of the jump operators, and how they bring irreversibility about.
5. Argue that the state of an empty cavity mode is Gaussian at all times (provided it starts in a Gaussian state).
Find the first and second moments both through the quantum Langevin equations and through the master
equation. Show that they reach an asymptotic value (independent of the initial condition), and interpret the
corresponding state as a displaced thermal state.
6. Consider now the situation in which the !-dependence of the coupling cannot be neglected and the frequency
integrals cannot be extended to 1 from below. Show that the non-Markovianity of the master equation (that
is, how far into the past the evolution of the atomic state sees) is determined by a so-called environmental
correlation function. Introduce then the Markov approximation, explaining its range of applicability. Under
such an approximation, show that, in the long-time limit, the master equation takes the standard Lindblad form,
but including a correction to the cavity frequency (Lamb shift) that can be of the same order as the damping
rate.
7. Argue that the same type of quantum Langevin and master equations can be found for a two-level atom in free
space, just replacing annihilation and creation operators of the cavity by atomic raising and lowering operators.
 14

B. Graded exercise 5: Atomic dephasing

While spontaneous emission has traditionally been the most relevant incoherent process occurring in atoms from
a theoretical point of view, there is another incoherent process which is even more important in practical terms: the
so-called dephasing. Many di↵erent physical processes contribute to it, and in this exercise we will study it with a
generic model: the atomic transition is a↵ected by random fluctuations (coming, for example, from stray random or
thermal magnetic fields in the laboratory).
1. [20] Consider the Hamiltonian Ĥ = ~[" + '(t)]ˆz /2, where '(t) is a random variable that accounts for random
fluctuations in the atomic transition frequency. Use the complex Bloch equations to show that the state, averaged
over stochastic realizations, can be written as
1 hˆ ⇤
(t) ⇤
i
⇢ˆ(t) = I + bz (0)ˆz + 2ei"t b (0)ˆ + 2e i"t (t) b(0)ˆ † , (26)
2
where we have defined
Rt
(t) i dt0 '(t0 )
e =e 0 , (27)
with (t) 2 C in general and where the overbar denotes stochastic average.
Rt 2 2
2. [20] Define (t) = 0 dt0 '(t0 ). Provide arguments in favor of the identity cos (t) + sin (t)  1, and use it
to prove that Re{ (t)} 0. Assuming e Re{ (t)} decays to zero monotonically as a function
p of time, explain the
p
trajectory followed in Bloch space by the initial superposition state | (0)i = p|gi + 1 p|ei with p 2 [0, 1].
Show that the final state is the incoherent mixture
lim ⇢ˆ(t) = p|gihg| + (1 p)|eihe|, (28)
t!1

so that the populations are left untouched, but the coherence has been destroyed.
Hint: Just evaluate the Bloch vector components at t = 0 using | (0)i, plug them in (26), and infer the Bloch
vector components at any other time from it. In order to visualize the trajectory, it is useful to first forget about
the fast z-precession at frequency "; the remaining trajectory should just be a straight line (which becomes a
spiral after including the precession).
3. [20] Show that the state (26) obeys the master equation
n o
 ⇣ n o⌘ Re ˙ (t)
i
@t ⇢ˆ = " + Im ˙ (t) ˆz , ⇢ˆ + (ˆz ⇢ˆˆz ⇢ˆ) , (29)
2 2
where the overdot denotes time derivative.
Hint: Just evaluate the left- and right-hand sides independently, and show that they match.
4. [20] Consider the case in which '(t) comes from a Gaussian stochastic process and has zero mean, '(t) = 0.
Rt
Defining (t) = 0 dt0 '(t0 ), use the Gaussian-moment formula
(
0 n 2 odd
n (t) = , (30)
n/2
2
(n 1)!! (t) n 2 even
to show that
2 (t)/2
e i (t) =e . (31)

Hint: Simply expand the exponential on the left-hand side in Taylor series, use the formula, and simplify the
expression to turn the elements of the sum into the ones required to obtain the right-hand-side’s exponential.
5. [20] Consider the white-noise limit, that is, '(t)'(t0 ) = ' (t t0 ), where ' is some constant rate. Show that
(t) = ' t/2 in this case, leading to a master equation

dˆ
⇢ i '
= "ˆz , ⇢ˆ + (ˆz ⇢ˆˆz ⇢ˆ) , (32)
dt 2 4
which is the usual master equation commonly used in quantum optics and quantum information to account for
dephasing e↵ects.
Hint: Just use the definition of (t) and (t), together with (31).
 15

VI. ANALYZING THE EMISSION OF OPEN SYSTEMS

A. Key questions of the chapter

1. Working in the Heisenberg picture, and using reasonable approximations such as setting slowly-varying functions
of the frequency to their value at the cavity resonance or ignoring retardation e↵ects, write down the field coming
out of an open cavity as a function of the intracavity and input annihilation operators (input-output relations).
2. Using the model of the previous chapter for the interaction between the cavity mode and the external modes,
derive a formal solution for the external annihilation operators in terms of their final conditions at some future
late time and the cavity annihilation operator. Use this solution to find an alternative form of the output field’s
annihilation operator as a Fourier transform of the external field’s late-time condition. Use it to show that
the output annihilation and creation operators satisfy canonical commutation relations in time. Prove that
the structure of the forward and backward quantum Langevin equations make the open-cavity model satisfy
causality.
3. Introduce the qualitative model of a photodetector, explaining how a single photon is capable of creating a
measurable electronic pulse. Explain that the combination of all the pulses generated by di↵erent photons
creates a macroscopic stochastic current, and write down the relation between its moments and the quantum
moments of the number operator associated to the output field coming from a source. Inspired by this relation,
define the correlation functions of the output field, and use the input-output relations to show that, at zero
temperature, they are directly proportional to correlation functions of the source.
4. Introduce, without proving but explaining them carefully, the quantum regression theorem and the quantum
regression formula for two-time correlators. Using the example of a radiating atom, provide a physical in-
terpretation for the two-time correlation function associated to direct photodetection (coincidence correlation
function) as the probability of getting two consecutive photodetection events separated by a given time interval.
5. Taking the master equation of the monochromatically-driven cavity as the starting point (in a picture rotating
at the laser frequency), evaluate the (normalized) coincidence correlation function for two situations: zero
temperature (n̄ = 0) and zero drive (A = 0). You may use the properties of coherent states for the first one,
and the quantum regression formula for the second. Interpret the results as photons arriving to the detector at
random in the first case, and photons arriving in bunches in the second case.
6. Consider an atom subject to resonant driving and spontaneous emission (at zero temperature). Using the master
equation in a picture rotating at the laser frequency, write down the Bloch equations. Turn them into a second-
order di↵erential equation for the excited-state population and find its value in the asymptotic, stationary limit.
Show that it is never possible to obtain population inversion (that is, more population in the excited state than in
the ground state), reaching a limit of 1/2 for infinite driving. Find a expression of the (normalized) coincidence
correlation function as a function of the excited-state population and solve its equation of motion to find the
full time-dependence of the correlation function. Use the result to introduce the concept of antibunching, and
to show that (damped) Rabi oscillations appear in the correlation function.
7. Introduce homodyne detection for the field coming out of a source, and show that the first and second moments
of the corresponding photocurrent are proportional to normally-ordered moments of the quadratures. Introduce
also the quadrature noise spectrum and the definition of squeezing for the fields coming out of an open source.
8. Write down the quantum Langevin equations of an open cavity containing a second-order nonlinear medium
that provides down-conversion for the cavity mode (use the parametric approximation). Assuming zero detuning
for the down-conversion process and working below threshold (damping rate larger than the down-conversion
rate), write down decoupled equations for two orthogonal quadratures and solve them. Argue that the state
is Gaussian, and study the asymptotic state by evaluating the mean vector and covariance matrix. How much
squeezing can you get inside the cavity? Is the state a minimum-uncertainty one? How about the state of the
output field? Evaluate the noise spectrum of the independent quadratures and answer this question.
 16

B. Graded exercise 6: Coincidence correlation function of the below-threshold OPO

We have studied in class the squeezing properties of the light coming out of an optical parametric oscillator (OPO)
below threshold and at resonance. Here we will now evaluate the coincidence correlation function, which is essential
to understand the photon statistics of the field emitted by any system.
1. Can you guess, based on physical arguments, whether he photons emitted by the OPO will arrive at the photode-
tector bunched, antibunched, or randomly?
2. Let us remind the solution for the slowly-varying position and momentum quadratures,
Z t h
p 0 ' ⇡i
'
lim X̃ (t) = 2 lim d⌧ 0 e ' ⌧ X̃in (t ⌧ 0 ), ' = 0, , (33)
t!1 t!1 0 2

with 0 = g, ⇡/2 = + g, g 2 [0, [, and where the input operators satisfy the statistical properties
0 0
hX̃in (t)i = 0 and hX̃in (t)X̃in (t0 )i = ei( t0 ). Use it to show that
)
(t
⇢
' '0 2 i('0 ') e '0 ⌧ for ⌧ > 0
lim hX̃ (t)X̃ (t + ⌧ )i = e . (34)
t!1 ' + '0 e ' ⌧ for ⌧ < 0

Hint: In class we evaluated limt!1 hX̃ ' (t)X̃ ' (t + ⌧ )i. Just follow similar steps.
3. In class we saw the formula

h L̂1 L̂2 L̂3 L̂4 i = h L̂1 L̂2 ih L̂3 L̂4 i + h L̂1 L̂3 ih L̂2 L̂4 i + h L̂1 L̂4 ih L̂2 L̂3 i, (35)

valid for Gaussian states when the operators L̂j are all linear in position and momentum, and L̂j = L̂j hL̂j i.
Argue that for the OPO below threshold (and any other problem described by linear quantum Langevin equations),
this formula applies even when considering Heisenberg-picture operators at di↵erent times, that is, for the set
{L̂1 (t1 ), L̂2 (t2 ), L̂3 (t3 ), L̂4 (t4 )}.
4. Use the previous formula to show that the coincidence correlation function of the OPO below threshold can be
written as
h⌦ ↵2 ⌦ ↵2 ⌦ ↵2 i
Ḡ(2) (⌧ ) = lim ↠(t)↠(t + ⌧ ) + ↠(t)â(t + ⌧ ) + ↠(t)â(t) . (36)
t!1

Use then (34) to obtain


⌦ † † 1 ↵ (1 ) ⌧ (1+ ) ⌧
lim â (t)â (t + ⌧ ) = e + e , (37a)
t!1 4 1 1+

⌦ † ↵ 1 (1 ) ⌧ (1+ ) ⌧
lim â (t)â(t + ⌧ ) = e e , (37b)
t!1 4 1 1+

where = g/ 2 [0, 1[. Show from these expressions that Ḡ(2) (⌧ ) is a monotonically decreasing function of time.
Does this match your answer in the first question of the exercise?
Hint: For the first step, just apply the identity hÂB̂i = hB̂ † † i⇤ . For the second part, just write the annihilation
and creation operators as a function of the position and momentum quadratures, and use (34).
 17

VII. EFFECTIVE MODELS: ELIMINATION OF SPURIOUS DEGREES OF FREEDOM

A. Key questions of the chapter

1. What is an e↵ective model? Can you explain in which situations we might expect them to appear?
2. Consider a closed system described by a time-independent Hamiltonian. Define projector operators that divide
the Hilbert space into relevant and irrelevant sectors, and use them to find an equation of motion for the state
projected on the relevant sector. Define formally an e↵ective Hamiltonian from these equations, showing that
in general we obtain a non-Hermitian and time-dependent expression for it.
3. While for some problems the e↵ective Hamiltonian becomes Hermitian under suitable conditions, for other it
does not. For the latter, how do you interpret the fact that we cannot find a Hermitian e↵ective Hamiltonian?
4. Show that the original Hamiltonian of the system can be split into a term that doesn’t connect the relevant
and irrelevant subspaces (free Hamiltonian), and another that does (interaction Hamiltonian). Find a simple
expression for the e↵ective Hamiltonian up to second order in the interaction. Use it to show that a far-detuned
monochromatic optical beam can be used to generate a motional potential on the center of mass motion of an
atom.
5. Consider now an open system described by a master equation with time-independent Liouvillian. Define projec-
tor superoperators that divide the space of operators into relevant and irrelevant sectors, and use them to find
an e↵ective (time-nonlocal) master equation for the relevant part of the state. Show that the original Liouvillian
can be split into a term that does not connect the relevant and irrelevant subspaces (free Liouvillian), and
another one that does (interaction Liouvillian). Approximate the e↵ective master equation to second order in
the latter.
6. Consider a bipartite Hilbert space structure, consisting in the tensor product of two degrees of freedom that
we call system and environment. Consider physical situations in which, from the point of view of the system,
the environment remains approximately frozen in its free-Liouvillian equilibrium state. Define a projector
superoperator adapted to such situation. Particularize the previous e↵ective master equation to this choice of
projector, and show how it can be approximated by a time-local e↵ective master equation provided that the
correlation functions of the environment decay fast enough compared to all incoherent rates of the system’s
evolution.
7. Consider an atom at high temperature emitting in all directions. Consider as well one mode of an open cavity
at zero temperature. Argue qualitatively how the atom can be e↵ectively cool down and its emission directed
by coupling it to the cavity via a Jaynes-Cummings interaction, provided that the coupling is strong (i.e., larger
than the atomic spontaneous emission rate) but still much smaller than the cavity decay rate. Use the e↵ective
master equation derived above (taking the atom as the system and the cavity as the environment) to make this
statement mathematically rigorous.
 18

B. Graded exercise 7a: Some important e↵ective Hamiltonians

1. Kerr from down-conversion. Consider the down-conversion Hamiltonian

g0
Ĥ = ~!0 ↠â + ~!2 b̂† b̂ + i~ (b̂â†2 b̂† â2 ), (38)
2
in the large-detuning limit, | | g0 , with = !0 !2 /2. Find an e↵ective Hamiltonian for the down-converted
mode, under the assumption that the pump mode is not populated initially. In particular, argue that P̂ = |0ib h0|
is the appropriate projector in this case, with b̂|0ib = 0, and show that the e↵ective Hamiltonian has the Kerr
form

Ĥe↵ = ~ (! g) n̂ + ~gn̂2 , (39)

with n̂ = ↠â and g = g02 /8 .

2. Raman driving. Consider the following problem: we would like to couple two atomic energy eigenstates whose
transition does not couple to a laser (e.g., because they have the same parity or their transition frequency is not
in the optical domain, but in the microwave domain, as usually happens with hyperfine states). You are going to
show that there is an e↵ective way of connecting them by using an additional atomic state and two far-detuned
laser beams that couple it to the original states. Consider specifically the situation depicted in Fig. 1, described
by the Hamiltonian

Ĥ(t) = ~"g |gihg| ~"e |eihe| + ~ ⌦g e i!g t

|aihg| + ⌦⇤g ei!g t |giha| + ⌦e e i!e t
|aihe| + ⌦⇤e ei!e t |eiha| , (40)

where we have taken the energy origin in the auxiliary level |ai, and chosen the signs so that "j > 0.
Show that in a new picture defined by the transformation Ûc (t) = eĤc t/i~ with Ĥc = ~!g |gihg| ~!e |eihe|, the
Hamiltonian reads

H̃ = ~ g |gihg| +~ e |eihe| + ~ ⌦g |aihg| + ⌦⇤g |giha| + ⌦e |aihe| + ⌦⇤e |eiha| , (41)

with j = !j "j .
In the far-detuned limit | j | |⌦j |, argue that starting from an unpopulated auxiliary level, the relevant sector
of the Hilbert space is defined by the projector P̂ = |gihg| + |eihe|, leading to the e↵ective Hamiltonian

H̃e↵ ⇡ ~ g |gihg| +~ e |eihe| + ~⌦e↵ |eihg| + ~⌦⇤e↵ |gihe|, (42)

with ⌦e↵ = ⌦g ⌦⇤e / g .

Come back to the original picture and show that the Hamiltonian can be written in the common form (you’ll
need to make an energy shift)
~" i!eff t †
Ĥe↵ = ˆz + ~ ⌦e↵ e ˆ + ⌦⇤e↵ ei!eff t ˆ , (43)
2
where !e↵ = !e !g , " = "g "e , and the Pauli operators are defined in the usual way.
Hint: Just follow the steps we have developed in class, and be careful that at the end you will need to justify
the approximation ⌦g ⌦⇤e / g ⇡ ⌦g ⌦⇤e / e to get a Hermitian Hamiltonian.

Raman driving Optomechanical cooling

moving
|ai mirror cavity mode
<latexit sha1_base64="cSvU5YNRwdRWTIqBmXlga07XdJ4=">AAAB73icdVBNTwIxEO3iF+IX6tFLIzHxtOkiCNxIvHjExBUMbEi3FGhou5u2a0JWfoUXD2q8+ne8+W8sCyZq9CWTvLw3k5l5YcyZNgh9OLmV1bX1jfxmYWt7Z3evuH9wo6NEEeqTiEeqE2JNOZPUN8xw2okVxSLktB1OLuZ++44qzSJ5baYxDQQeSTZkBBsr3d7jnsJyxGm/WEJutX7uNTyI3LJXqXoZQbXyWRV6LspQAku0+sX33iAiiaDSEI617nooNkGKlWGE01mhl2gaYzLBI9q1VGJBdZBmB8/giVUGcBgpW9LATP0+kWKh9VSEtlNgM9a/vbn4l9dNzLAepEzGiaGSLBYNEw5NBOffwwFTlBg+tQQTxeytkIyxwsTYjAo2hK9P4f/EL7sNF11VSs36Mo08OALH4BR4oAaa4BK0gA8IEOABPIFnRzmPzovzumjNOcuZQ/ADztsnw+mQjg==</latexit>

<latexit sha1_base64="69bjyJ/GsAfQekOUg5O+oUxJIsg=">AAAB7nicdVDJSgNBEK2JW4xb1KOXxiB4GnqSkOUW0IPHCI4JJEPo6fQkTXoWu3uEMOQnvHhQ8er3ePNv7CyCij4oeLxXRVU9PxFcaYw/rNza+sbmVn67sLO7t39QPDy6VXEqKXNpLGLZ9YligkfM1VwL1k0kI6EvWMefXMz9zj2TisfRjZ4mzAvJKOIBp0Qbqdu/ZEKTwWhQLGG75tRq5QrCdr1SxU1sSMNxnGoDOTZeoAQrtAfF9/4wpmnIIk0FUarn4ER7GZGaU8FmhX6qWELohIxYz9CIhEx52eLeGTozyhAFsTQVabRQv09kJFRqGvqmMyR6rH57c/Evr5fqoOFlPEpSzSK6XBSkAukYzZ9HQy4Z1WJqCKGSm1sRHRNJqDYRFUwIX5+i/4lbtps2vq6WWuVVGnk4gVM4Bwfq0IIraIMLFAQ8wBM8W3fWo/VivS5bc9Zq5hh+wHr7BKdDj98=</latexit>
g <latexit sha1_base64="TYBcRaHiSwOP28EX3gevx5gb5xI=">AAAB7nicdVDJSgNBEO2JW4xb1KOXxiB4GmYm+y2gB48RHBNIhtDTqUma9Cx29whhyE948aDi1e/x5t/YWQQVfVDweK+Kqnp+wplUlvVh5NbWNza38tuFnd29/YPi4dGtjFNBwaUxj0XXJxI4i8BVTHHoJgJI6HPo+JOLud+5ByFZHN2oaQJeSEYRCxglSkvd/iVwRQYwKJYss+bYZaeGLbNesRy7rkm13KhWmtg2rQVKaIX2oPjeH8Y0DSFSlBMpe7aVKC8jQjHKYVbopxISQidkBD1NIxKC9LLFvTN8ppUhDmKhK1J4oX6fyEgo5TT0dWdI1Fj+9ubiX14vVUHDy1iUpAoiulwUpByrGM+fx0MmgCo+1YRQwfStmI6JIFTpiAo6hK9P8f/EdcymaV1XSi1nlUYenaBTdI5sVEctdIXayEUUcfSAntCzcWc8Gi/G67I1Z6xmjtEPGG+fqtqP4g==</latexit>
e

⌦e
⌦g
<latexit sha1_base64="bJ771CrmSd9XLVQLQnj43NNmmcs=">AAAB7nicdVDLSgMxFM3UV62vqks3wSK4Gmba2tZdwY07Kzi20A4lk95pQ5OZMckIZehPuHGh4tbvceffmD4EFT0QcjjnXu69J0g4U9pxPqzcyura+kZ+s7C1vbO7V9w/uFVxKil4NOax7AREAWcReJppDp1EAhEBh3Ywvpj57XuQisXRjZ4k4AsyjFjIKNFG6vSuBAxJH/rFkmNXnEa55mJDzO+UDXHPavVKFbu2M0cJLdHqF997g5imAiJNOVGq6zqJ9jMiNaMcpoVeqiAhdEyG0DU0IgKUn833neITowxwGEvzIo3n6veOjAilJiIwlYLokfrtzcS/vG6qw4afsShJNUR0MShMOdYxnh2PB0wC1XxiCKGSmV0xHRFJqDYRFUwIX5fi/4lXts9t57paajaWaeTRETpGp8hFddREl6iFPEQRRw/oCT1bd9aj9WK9Lkpz1rLnEP2A9fYJoBaP3w==</latexit>

<latexit sha1_base64="3Lvv/2py0TPfyypIVa1RsN1A4yk=">AAAB7nicdVDLSsNAFJ34rPVVdelmsAiuwqSNbboruHFnBWMLbSiT6SQdOpPEmYlQSn/CjQsVt36PO//G6UNQ0QMXDufcy733hBlnSiP0Ya2srq1vbBa2its7u3v7pYPDW5XmklCfpDyVnRAryllCfc00p51MUixCTtvh6GLmt++pVCxNbvQ4o4HAccIiRrA2Uqd3JWiM+3G/VEa247q1Wh0iu3KOXM8zpOqhRtWDjo3mKIMlWv3Se2+QklzQRBOOleo6KNPBBEvNCKfTYi9XNMNkhGPaNTTBgqpgMr93Ck+NMoBRKk0lGs7V7xMTLJQai9B0CqyH6rc3E//yurmOvGDCkizXNCGLRVHOoU7h7Hk4YJISzceGYCKZuRWSIZaYaBNR0YTw9Sn8n/gVu2Gja7fc9JZpFMAxOAFnwAF10ASXoAV8QAAHD+AJPFt31qP1Yr0uWles5cwR+AHr7RO+a4/0</latexit>
laser
|ei
<latexit sha1_base64="AKN4AzQz98BfgCI2j54Gg1M3MOM=">AAAB73icdVBNT8JAEN3iF+IX6tHLRmLiqWkRBG4kXjxiYgUDDdkuA2zYbpvdrQmp/AovHtR49e9489+4FEzU6EsmeXlvJjPzgpgzpR3nw8qtrK6tb+Q3C1vbO7t7xf2DGxUlkoJHIx7JTkAUcCbA00xz6MQSSBhwaAeTi7nfvgOpWCSu9TQGPyQjwYaMEm2k23voSSJGHPrFkmNX6+duw8WOXXYrVTcjTq18VsWu7WQooSVa/eJ7bxDRJAShKSdKdV0n1n5KpGaUw6zQSxTEhE7ICLqGChKC8tPs4Bk+McoADyNpSmicqd8nUhIqNQ0D0xkSPVa/vbn4l9dN9LDup0zEiQZBF4uGCcc6wvPv8YBJoJpPDSFUMnMrpmMiCdUmo4IJ4etT/D/xynbDdq4qpWZ9mUYeHaFjdIpcVENNdIlayEMUhegBPaFnS1qP1ov1umjNWcuZQ/QD1tsnyhGQkg==</latexit>

|gi
<latexit sha1_base64="U5+Kt32cfilqSNhRe9AomITbWzQ=">AAAB73icdVBNS8NAEN3Ur1q/qh69LBbBU9jU1ra3ghePFYyttKFstpt06WYTdjdCif0VXjyoePXvePPfuE0rqOiDgcd7M8zM8xPOlEbowyqsrK6tbxQ3S1vbO7t75f2DGxWnklCXxDyWPR8rypmgrmaa014iKY58Trv+5GLud++oVCwW13qaUC/CoWABI1gb6fY+HEgsQk6H5Qqy681zp+VAZFedWt3JCWpUz+rQsVGOCliiMyy/D0YxSSMqNOFYqb6DEu1lWGpGOJ2VBqmiCSYTHNK+oQJHVHlZfvAMnhhlBINYmhIa5ur3iQxHSk0j33RGWI/Vb28u/uX1Ux00vYyJJNVUkMWiIOVQx3D+PRwxSYnmU0MwkczcCskYS0y0yahkQvj6FP5P3KrdstFVrdJuLtMogiNwDE6BAxqgDS5BB7iAgAg8gCfwbEnr0XqxXhetBWs5cwh+wHr7BM0lkJQ=</latexit>

Figure 1. Sketches of the Raman driving and optomechanical cooling schemes.

 19

C. Graded exercise 7b: Optomechanical sideband cooling

In the last exercise of the course we will study one research-level problem: how to cool down a mechanical resonator
by using laser light. It is research level in the sense that the theory behind it, which is what you we will do in
this exercise, was developed only about a decade ago, and the related experiments started only around the same
time. Altogether, these developments gave birth to the now well-established field of quantum optomechanics, which
is currently a subject of active investigation by many groups all around the world.
1. Consider an optical cavity with an the end mirror coupled to a fixed wall through a spring (Fig. 1), so that
it can oscillate at some frequency ⌦ (much smaller than optical frequencies). Take the position of the mirror
around its equilibrium location as a dynamical variable ẑ, whose evolution (in the absence of coupling to other
2
1 2
systems) is provided by a harmonic oscillator Hamiltonian Ĥm = 2m p̂z + m⌦ 2
2 ẑ , where m is the mirror’s mass
and p̂z its momentum, so that [ẑ, p̂z ] = i~. Consider now one mode of the cavity with annihilation operator â
and resonance frequency !c when the cavity has length L. Using the fact that the cavity’s length is L + ẑ, and
assuming that the mirror’s deviations from equilibrium are small compared to L, show that the Hamiltonian of
the system can be approximated by

Ĥ = ~⌦b̂† b̂ + ~!c ↠â + ~g0 ↠â(b̂† + b̂), (44)

where we have introduced the annihilation and creation operators for the mechanical oscillator, b̂ and b̂† , respec-
tively, and the so-called optomechanical coupling rate is given
zzpf
g0 = !c , (45)
L
p
where zzpf = ~/2⌦m refers to the zero-point fluctuations of the mirror’s position.
Hint: You will need to use the expression that we saw at the beginning of the lectures for the resonance
frequencies inside a cavity as a function of its length.
2. Assume now that the temperature is low enough for the number of thermal photons at optical frequencies to
be negligible, but large enough for the mechanical oscillator to have a lot of thermal excitations or phonons n̄
(this is what naturally happens at room temperature, since typical optical frequencies are hundreds of THz while
mechanical oscillations are never larger than GHz). Moreover, assume as well that the cavity is driven by a
laser, so that the master equation of the whole system reads
" #
dˆ⇢ Ĥ
= + Ee i!L t ↠E ⇤ ei!L t â , ⇢ˆ + Da [ˆ
⇢] + (n̄ + 1)Db [ˆ
⇢] + n̄Db† [ˆ
⇢]. (46)
dt i~

Show that there exists a transformation operator Ûc (t) such that the state ⇢ˆ0 (t) = Ûc† (t)ˆ
⇢(t)Ûc (t) in the new
picture satisfies the time-independent master equation

⇢0
dˆ h i
= i⌦b̂† b̂ + i c â
†
â + E ↠E ⇤ â ig0 ↠â(b̂† + b̂), ⇢ˆ0 + Da [ˆ
⇢0 ] + (n̄ + 1)Db [ˆ
⇢0 ] + n̄Db† [ˆ
⇢0 ], (47)
dt
where c = !L !c .
3. Assume that the state is coherent, that is, ⇢ˆ0 (t) = | (t)ih (t)|⌦|↵(t)ih↵(t)|, and show that the coherent amplitudes
satisfy the equations of motion
⇤
↵˙ = E [ i c + ig0 ( + )] ↵, (48a)
˙ = ( + i⌦) ig0 |↵| . 2
(48b)

Hint: Find the equations of motion of hâi and hb̂i like we did in class for other problems starting from the master
equation, and use the fact that coherent states are right and left eigenstates of the annihilation and creation
operators, respectively.
4. Consider now a displacement operator with time-dependent amplitude
(t)ĉ† ⇤
(t)ĉ
D̂c [ (t)] = e , (49)
 20

where ĉ and ĉ† are annihilation and creation operators, respectively, that is, [ĉ, ĉ† ] = 1. Show that its time-
derivative can be written as
✓ ◆
dD̂c [ (t)] † ⇤ ˙ ⇤ ˙⇤
= D̂c [ (t)] ˙ ĉ ˙ ĉ + . (50)
dt 2

d
Hint: Take and ⇤
as independent variables, so that you can write the time derivative as dt = ˙ @@ + ˙ ⇤ @ @ ⇤ ,
2
| (t)| /2 (t)ĉ† ⇤
(t)ĉ
and apply it to the normally-ordered form of the displacement operator D̂c [ (t)] = e e e .

5. Move to another picture defined by the transformation operator Û (t) = D̂b [ (t)]D̂a [↵(t)], where the amplitudes ↵
and satisfy the classical equations of motion (48). Show that the transformed state ⇢˜(t) = Û † (t)ˆ ⇢0 Û (t) satisfies
the master equation

⇢ h
d˜ i
= i⌦b̂† b̂ + i (t)↠â i[g(t)↠+ g ⇤ (t)â + g0 ↠â](b̂† + b̂), ⇢˜ + Da [˜
⇢] + (n̄ + 1)Db [˜
⇢] + n̄Db† [˜
⇢], (51)
dt
where we have defined the dressed detuning (t) = c g0 [ (t) + ⇤ (t)] and the dressed optomechanical coupling
g(t) = g0 ↵(t).
Hint: Make the time derivative of ⇢˜(t) and use (50), (47), and the displacement formulas Û † (t)âÛ (t) = â + ↵(t)
and Û † (t)b̂Û (t) = b̂ + (t), to rewrite it as (51). You should obtain extra terms in the Hamiltonian linear in the
annihilation and creation operators, but this terms vanish identically thanks to the classical equations (48).

6. Next we are going to find an e↵ective master equation for the mechanical oscillator. Hence, we make the choices
(we assume from now on that the classical solution is time-independent)
h i
LS [˜
⇢] = i⌦b̂† b̂, ⇢˜ + (n̄ + 1)Db [˜ ⇢] + n̄Db† [˜
⇢], (52a)
⇥ ⇤
LE [˜⇢] = i ↠â, ⇢˜ + Da [˜⇢], (52b)

L1 [˜
⇢] = i (g↠+ g ⇤ â + g0 ↠â)X̂ , ⇢˜ , (52c)
| {z }
Ĥ1 /~

where we have defined the mechanical position quadrature X̂ = b̂† + b̂. Show that the stationary state of the free
cavity mode is vacuum, ⇢¯E = |0ih0|, and then prove that

PL1 P[Ŷ ] = 0, (53)

for any operator Ŷ , where the projector superoperator acts as P[Ŷ ] = trE {Ŷ } ⌦ ⇢¯E .
Hint: Just apply the definition of the projector to (53), and you should be able to write it as something
proportional to h0|â|0i and h0|↠â|0i.
P3
7. Note that the Hamiltonian can be written as Ĥ1 /~ = m=1 gm Ŝm ⌦ Êm , with the choices g2 = g = g1⇤ , g3 = g0 ,
Ŝ1 = Ŝ2 = Ŝ3 = X̂, Ê1 = â = Ê2† , and Ê3 = ↠â. Show that the environmental operators are all closed on their
own, and use the quantum regression formula to show that (⌧ > 0)
( i )⌧
lim hÂ(t)â(t + ⌧ )Ĉ(t)iE = h0|ÂâĈ|0iE e , (54a)
t!1

lim hÂ(t)â(t + ⌧ )Ĉ(t)iE = h0|Â↠Ĉ|0iE e (+i )⌧

(54b)
t!1

lim hÂ(t)â (t + ⌧ )â(t + ⌧ )Ĉ(t)iE = h0|Â↠âĈ|0iE e

† 2⌧
, (54c)
t!1

for any environmental operators  and Ĉ. Then, show that the cavity correlators required for the mechanical
e↵ective master equation read as
(+i )⌧ ( i )⌧
Cmn (⌧ ) = e m1 n2 , Knm (⌧ ) = e n1 m2 . (55)

Hint: Follow the steps followed in the example of the lectures.

 21

8. Take ĤS = ~⌦b̂† b̂ as the only part of the mechanical evolution which is relevant within the decay of the cavity
correlators. Show that
i⌦⌧ †
X̃(⌧ ) = eĤS ⌧ /i~ X̂e ĤS ⌧ /i~
= ei⌦⌧ b̂ + e b̂ , (56)

and then prove that the e↵ective mechanical master equation

Z t h i
d˜
⇢S
⇢S ] + |g|2
= LS [˜ d⌧ C12 (⌧ )X̂ ⇢˜S X̂(⌧ ) K12 (⌧ )X̂ X̂(⌧ )˜
⇢S + H.c. , (57)
dt 0

1
can be written in the t  limit as
d˜
⇢S
= LS [˜
⇢S ] + L [˜
⇢S ] + L+ [˜
⇢S ] + LNRW [˜
⇢S ], (58)
dt
with
opt opt
L [˜
⇢S ] = ⇢S ] i[ ⌦ b̂† b̂, ⇢˜S ], L+ [˜
Db [˜ ⇢S ] = + Db† [˜
⇢S ] i[ ⌦+ b̂† b̂, ⇢˜S ], (59a)
" #
b̂˜
⇢S b̂ b̂† ⇢˜S b̂† b̂2 ⇢˜S ⇢˜S b̂†2
⇢S ] = |g|2
LNRW [˜ + + H.c. , (59b)
 + i( ⌦)  + i( + ⌦)  i( + ⌦)  i( ⌦)

and

opt |g|2 / |g|2 ( ± ⌦)

⌥ = 2, ⌦opt
⌥ = 2.
±⌦ 2 +( ± ⌦)
1+ 

Hint: Just perform the time integrals (taking the limit t/ 1), and reorder the terms in the proper way.

9. Argue that LNRW can be neglected within a rotating-wave approximation provided that

|g|2
p ⌧ 2⌦. (60)
2 + ( ± ⌦)2

Within this approximation, show that the e↵ective mechanical master equation can be written as

d˜
⇢S h i
= i(⌦ + ⌦ + ⌦+ )b̂† b̂, ⇢˜S + Db [˜
⇢S ] + + Db† [˜
⇢S ], (61)
dt
with
opt opt
= (n̄ + 1) + , + = n̄ + + , (62)

which can be rewritten in the standard form

d˜
⇢S h i
= i⌦e↵ b̂† b̂, ⇢˜S + e↵ (n̄e↵ + 1)Db [˜
⇢S ] + e↵ n̄e↵ Db† [˜
⇢S ], (63)
dt
with

⌦e↵ = ⌦ + ⌦ + ⌦+ , (64a)
" #
C C
e↵ = + = 1+ , (64b)
+⌦ 2 ⌦ 2
1+  1+ 
C
n̄ + ⌦ 2
+ 1+(  )
n̄e↵ = = C C
, (64c)
+ 1+ +⌦ 2 ⌦ 2
1+(  ) 1+(  )

where we have defined the cooperativity C = |g|2 / .

 22

10. Inspect n̄e↵ , and argue that the best cooling is obtained when = ⌦ (red sideband driving) and ⌦  (resolved
sideband regime). Under these conditions, show that when C 1, then we have

n̄ 2
e↵ ⇡ C, n̄e↵ ⇡ + , (65)
C 4⌦2
2
so that when C n̄, we obtain the minimum e↵ective thermal occupation n̄e↵ ⇡ 4⌦2 .

11. Self-consistency check: show that all the approximations required to derive the e↵ective master equation are
compatible with the cooling conditions.
12. Finally, argue that the asymptotic state of this equation in the original Schrödinger picture corresponds to the
displaced thermal state

⇢th (n̄e↵ )D̂b† ( ).

lim ⇢ˆS (t) = D̂b ( )¯ (66)
t!1