Physics 285b: Homework # 1

Due Wednesday, Sept. 25 by 11:59 pm

Questions or assistance:
Samuel Buckley-Bonanno (sbuckleybonanno@g.harvard.edu)

1. Optical transitions and angular momentum

This problem reviews some aspects of atomic transitions which will be relevant
for the course. For calculating transition matrix elements, one often makes use of the
Wigner-Eckart theorem, which can be found in most quantum mechanics textbooks,
e.g. Mertzbacher p432-439 or Sakurai p238-242.
An atom of total angular momentum F has a spontaneous radiation rate A that
consists of a sum of the rates of all transitions out of F . For examples, assume it
radiates to a lower level with angular momentum F ′ = F − 1. The problem is to
find the rates for the various allowed transitions, i.e. the fraction of the radiation
that goes into each of the possible transitions (F, m) → (F ′ , m′ ) (where m stands for
mF , the quantum number associated with Fz ). Each of the rates is proportional to
|⟨F, m|Y1,q |F ′ , m′ ⟩|2 with q = ±1, 0. The rates can be found by either direct evaluation
of matrix elements or by applying the following considerations:
(1) The sum of the rates out of each state F, m must equal A
(2) The sum of the rates into each state F ′ , m′ must equal A 2F
2F +1
′ +1 .

(3) An unpolarized mixture of radiators in level F must emit equal intensities of

light with each of the three polarization components (z, σ± ).
(4) The rate for a transition (F, m → F ′ , m′ ) must be the same as for (F, −m →
F ′ , −m′ ).
Consider the situation F = 2, F ′ = 1. Designate the transitions by letters as
follows:
a : m = 2 → m′ = 1
b : m = 1 → m′ = 1
c : m = 0 → m′ = 1
d : m = 1 → m′ = 0
e : m = 0 → m′ = 0

Note that this list of transitions is not exhaustive, and additional transitions may
be important.

(a) Find the rates for a through e in terms of A using the appropriate version of
the Wigner-Eckart theorem, and make a figure of your results. (Clebsch-Gordan
coefficients can either be worked out from first principles, taken from a table in
a quantum mechanics or spectroscopy text, or computed with Mathematica)

(b) Using the symmetry considerations and conservation of probabilities (i.e. the
total number of decaying atoms from level F should be equal to the total number
arriving to F ′ ) show rules (2) and (4) must be true.

(c) Find the rates for a through e using rules (1)–(4).

2. Landé g-factor in a weak magnetic field
Recall that in atoms, angular momenta are coupled. First, L ⃗ and S⃗ couple into
J by spin-orbit coupling L · S, giving the fine structure. Then, J and I⃗ couple into
⃗ ⃗ ⃗ ⃗
F⃗ through I⃗ · J,
⃗ which gives the hyperfine structure. This means that L ⃗ and S
⃗ are
⃗ ⃗
determined up to a common rotation around the common axis J, while J itself and
I⃗ are determined up to rotation around their common axis F⃗ .
How do these couplings affect an atom’s interaction with an external magnetic
field? When the coupling with the external field is much weaker than coupling be-
tween atomic angular momenta, the external field will not destroy the fine structure
and hyperfine structure couplings. As L ⃗ and S ⃗ rotate around their common axis J, ⃗
their total magnetic moment µ ⃗L + µ ⃗
⃗ S also rotates around J. The result is that the
component of µ ⃗L + µ⃗ S orthogonal to J⃗ effectively cancels out, and does not interact
with the external field. Therefore, the effective magnetic moment relative to the ex-
ternal field is the component parallel to J.⃗ (This is similar in the coupling of J⃗ and
⃗ only the projection of the magnetic moment in the F⃗ direction is effective.)
I;
The magnetic moments involved in the spin-orbit coupling are
⃗
L ⃗
S
⃗ L = −gL µB
µ and ⃗ S = −gS µB
µ
ℏ ℏ
where gL = 1 and gS ≈ 2. By the argument above, there is some effective magnetic
coupling µ
⃗ J that describes the interaction between the weak external magnetic field
and the overall magnetic moment from the spin-orbit coupling, given by

J⃗ ℏ µL,J + µS,J
⃗ J = −gJ µB ,
µ where gJ = −
ℏ µB ⃗
|J|

and µL,J and µS,J are the scalar projections of µ

⃗ L and µ ⃗
⃗ J onto J.

(a) Derive an expression for the Landé g-factor gJ , in terms of only J, S and L.

The magnetic moments involved in the hyperfine interaction are

J⃗ ⃗
S
⃗ J = −gJ µB
µ and ⃗ I = −gI µN
µ
ℏ ℏ

where µN is the nuclear magnetic moment. If the magnetic field is weaker than the I⃗·J⃗
coupling, the field can again be treated as a perturbation. The effective interaction
with the external field will be

HB = −gF (⃗µJ + µ ⃗ ≈ −gF µ⃗J · B

⃗I) · B ⃗

where for the approximation at the end, one neglects µ ⃗ I since the nuclear moment µN
is much smaller than the electronic moment µB . It is important to remember that
nevertheless, I⃗ and J⃗ couple into F⃗ first, and only then to B.⃗ Thus, the above logic
of the coupling between L ⃗ and S⃗ is the same for the coupling between I⃗ and J. ⃗ In
particular, gF should follow from the projection of µ ⃗
⃗ J onto F .
 |3〉
∆ω

|2〉

ω12
ν

|1〉

Figure 1: Problem 3

(b) Derive an expression for the Landé g-factor gF , taking into account nuclear spin,
in terms of only F , J, I, and gJ .

3. Two-level approximation
Consider a three-level system, shown in Figure 1, with two excited states, each
coupled to the ground state with identical selection rules and matrix elements. The
energy spacing between states |2⟩ and |3⟩ is ∆ω, which is much smaller than the
optical frequency ν and the frequency separation between 1 and 2, ω12 . An oscillating
(optical) field is tuned near resonance for the transition |1⟩ → |2⟩. It will, in principle,
also induce transitions from |1⟩ to |3⟩.

(a) Derive equations of motion for the probability amplitudes ci associated with
each level. Assume that the applied field has frequency ν = ω12 + δ and equal
Rabi frequencies for both of the optical transitions (Ω).

(b) Suppose now that ν = ω12 and ∆ω is much larger than the Rabi frequency Ω.
Evaluate the effect of state |3⟩ on Rabi oscillations to the second order in 1/∆ω.

(c) Show that the leading order correction to the Rabi dynamics due to the state |3⟩
can be compensated by slightly detuning the laser field. Determine the detuning
δ = ν − ω12 needed. Provide physical explanation for your result. Describe the
physics behind the next relevant correction (i.e. how the coupling of |1⟩ to |3⟩
leads to this effect).

(d) What conditions must the Rabi frequency Ω and total interaction time τ satisfy
in order that we can neglect the effect of state |3⟩? Consider the D1 transition
line in 87 Rb, where the hyperfine splitting of the excited state is about 0.8 GHz.
Given appropriate Ω and τ , can we treat a transition between hyperfine sublevels
as a two-level system for laser-cooled atoms? What about room temperature
atoms? The natural linewidth of the excited state of 87 Rb is about 5 MHz and
the Doppler broadening at room temperature is about 500 MHz.
 |2〉
δ(t)

|1〉

Figure 2: Problem 4

4. Dressed states and adiabatic passage

Consider a two-level system with frequency ω subject to an E/M field of fixed
intensity (Rabi frequency) but varying detuning δ(t) = ν(t) − ω (Figure 2). Assume
that initially the detuning is large (|δ(0)| ≫ |Ω|) and negative (ν < ω). The detuning
δ(t) is then slowly scanned through resonance to (|δ(T )| ≫ |Ω| and ν > ω) .

(a) Determine the instantaneous (adiabatic) eigenenergies and eigenstates as a func-

tion of δ(t).

(b) If the initial state of the system is |ψ⟩ = |1⟩, what is the final state, including
relevant phases? Consider now the situation when the detuning is adiabati-
cally returned to its initial value. What is the final state vector of the system,
including relevant phases? Provide a physical interpretation for various accu-
mulated phases, and describe how they might be measured in terms of physical
observables.

(c) For a given Ω, how slowly should the detuning change so that the resulting
evolution is adiabatic?
 |3〉
δ1 δ2

Ω2

Ω1
|2〉

|1〉

Figure 3: Problem 5

5. Raman transitions and adiabatic elimination

Consider a 3-level system interacting with two separate E/M fields, as shown in
Figure 3. Assume the fields only interact with atoms via the couplings shown (e.g.
field 1 does not couple |2⟩ to |3⟩).

(a) Derive equations of motion for the slowly varying probability amplitudes in the
rotating frame (c1 (t), c2 (t), c3 (t)).

(b) Assuming that the detunings δ1 = ν1 −ω13 and δ2 = ν2 −ω23 are large but similar
(e.g. δ1 ≃ δ2 and δ1 , δ2 ≫ Ω1 , Ω2 ), show that the dynamics of the system can
be described by an effective Hamiltonian,

Ĥef f = ℏω̃1 |1⟩⟨1| + ℏω̃2 |2⟩⟨2| − ℏΩef f |1⟩⟨2| − ℏΩ∗ef f |2⟩⟨1|

and derive ω̃1,2 and Ωef f .

[Hint: In the appropriate rotating frame you can“adiabatically eliminate” the
excited state amplitudes, i.e. set ċ3 (t) = 0 and express c3 in terms of c1,2 . The
remaining evolution equations can then be described in terms of the above
effective Hamiltonian. ]
6. Coherent control – This problem is counted as part of HW 2, but it fits in here
and so is posted already. Feel free to submit now or as part of the next HW.
Consider the transition from a bound state (|b⟩ with energy ϵ0 = 0) into a 1-D
continuum induced by a quasi-monochromatic field with a time-dependent, slowly
varying amplitude f (t). The generic Hamiltonian for this system is given by

∞
X ∞
X
Ĥ = ℏ (ωk − ν)|k⟩⟨k| − ℏ (f (t)gk |k⟩⟨b| + h.c.) .
k=−∞ k=−∞

The continuum states |k⟩ represent 1-d momentum states. Assume that gk = g
is constant for all relevant k, that the dispersion relation is ωk = v0 |k|, and that the
number of |k⟩ states with k ∈ [k, k + dk] is independent of k and is given by ρ dk.
Assume that the bound state corresponds to a localized state at x = 0 and that both
positive and negative x are allowed.
This model represents a 1-dimensional “toy model” for laser-induced photoion-
ization or photodissociation. With some adjustments that confine the problem to
postive values of x, this model also describes the coupling of a single-mode cavity to
free-space modes with one semi-transparent mirror (with the time-dependent coupling
representing modulation of the mirror reflectivity). x = 0 corresponds to the positions
of the “ion”, center-of-mass of the “molecule” or the mirror surface, respectively.

(a) Assuming that the system is in the bound state |b⟩ at t = 0 derive an expression
for probability amplitude cb (t) as a function of f (t).
R∞
(b) Derive an expression for the real-space wavefunction cout (x, t) = −∞ dk ρck (t)eikx
of the outgoing (continuum-state) wavepacket.
(c) Extra Credit Consider now an inverse problem (such as photoassociation). Sup-
pose that Rthe bound state is empty at t = 0 and the incoming wavepacket
∞
cin (0, t) = −∞ dk ρck (0)ei(ν−ωk )t is given. [Equivalently, one could consider the
shape
R ∞ of the incoming wavepacket in position space at time t = 0: cin (x, 0) =
−∞
dk ρck (0)eikx .] Derive an expression for the bound state amplitude cb (t) in
terms of cin (0, t) and f (t).
(d) Extra Credit Given that cb (0) = 0 and given the shape of the incoming
wavepacket cin (0, t) that vanishes for t < 0, it is sometimes possible to find
a function f (t) such that |cb (∞)| = 1, i.e. the system is in the bound state
with unity probability. Solve for the f (t) (in terms of cin (0, t)) that maximizes
|cb (∞)| for a given cin (0, t). What conditions should the incoming wavepacket
(equivalently, ck (0)) satisfy so that this maximum possible value of |cb (∞)| is
1? Provide a physical interpretation for your results.
[R Hint: One way to approach the problem is to write cb (∞) in the form cb (∞) =
∞
0
dt h∗ (t)cin (0,R t) for some function h(t) that is determined by f (t). One can
∞
then compute 0 dt |h(t)|2 , confirm that it is bounded, think what shape of
h(t) maximizes |cb (∞)|, and solve the resulting equation for f (t).]
Note: This is the basic idea behind the field now known as “coherent control.”
For example, in principle it can be applied to the problems of efficient and
coherent photo-association of BECs, or coherent conversion between photonic
and atomic qubits. These could be potential topics for the term project.