Solutions Quantum Mechanics Week 3

QM Instruction Team
Universiteit Utrecht
November 2024

3 Approximation methods in Quantum Physics

A: Perturbation theory in the harmonic oscillator
The harmonic oscillator is one of the best studied problems in quantum mechanics and the
exact eigenstates and eigenvalues have been known for decades. Perturbation theory is a
method to approximate the ground state energy of systems whose Hamiltonian we cannot
simply solve. In this exercise you will apply it to a problem with a known solution.
3.1 Displaced Harmonic Oscillator
The Hamiltonian of the standard Harmonic Oscillator is given by
p̂2 1
H= + mω 2 x2 . (1)
2m 2

We now consider the shifted oscillator given by

p̂2 1
H= + mω 2 (x − b)2 . (2)
2m 2

a) Show that this shifted harmonic oscillator should have exactly the same eigenenergies
as the original one, with the wavefunctions ψi (x + b). Assuming the shift b is very small,
we may treat this as a perturbation problem. We say V = V0 (x) + V 0 (x), where
1
V0 (x) = mω 2 x2 (3)
2
0 1 1
V (x) = mω 2 ((x − b)2 − x2 ) = mω 2 (b2 − 2xb) (4)
2 2
(5)
3 Approximation methods in Quantum Physics Problem class QM

b) Show that the first order correction to the energy of any state is 12 mω 2 b2 . Is this in
line with your previous answer?

Solution:

We use (7.9) from Griffiths. The term with x is odd under x → −x so it disappears when
taking the bracket. Onlt the constant term remains.

c) Find the second order correction to the energies of the ground state and of the first
excited state. (Hint: Use the methods from Example 2.5 in Griffiths)

Solution:
p
We use (in a notation that is more common than that of the book) x̂ = ~/2mω(â + ↠)
Now we can express
p p p
x̂|ji = ~/2mω( j|j − 1i + j + 1|j + 1i)

From which we find that

p p p
hi|x̂|ji = ~/2mω( jhi|j − 1i + j + 1hi|j + 1i)
p p p
= ~/2mω( jδi,j−1 + j + 1δi,j−1 )

For the ground state we now fill in (7.15),

(2)
X |hi|V 0 |ji|2
Ei =
j6=i
Ei − Ej

We only need to take into account j = 1, and simply filling in gives us

(2) 1
E0 = − mω 2 b2
2
For the first excited state (i = 1) we need to take into account two terms; they have
opposite sign and add up to the same result. Remarkably, the suprious energy correction
that we saw in the first order perturbation theory is canceled by the second order, and the
correct energy is regained. Note that this is a fortuitous result: Higher order corrections
are nonzero and hence take us away from the correct result. Taking into account a finite
number of extra orders in the perturbation series does not always improve the result.

2
Problem class QM 3 Approximation methods in Quantum Physics

3.2 Numerical particle in a box [Key !]

Insights: This exercise is not just a programming exercise. The finite-step discretization
method establishes a connection between eigenvalue equations for linear operators and
finite-size matrices. This numerical method is very useful for the poster miniproject.
In this exercise we want to look at the free particle Schrödinger equation in a box by
discretizing space and converting the problem to a matrix problem. We start with a
discretization of the Schrödinger equation in real space. In order to do so you have to
assume that xi = ia where i is an integer and a is the lattice spacing meaning we represent
space as a chain. This works well for wavelengths >> 2a, or for momenta << π~/a, but
less well for higher momenta. In continuum, the Schrödinger equation reads
~2 d2
Eψ(x) = − ψ(x) + V (x)ψ(x) (6)
2m dx2
We want to describe a particle in a box meaning

0 for 0 < x < L
V (x) = (7)
∞ else
We discretize the first derivative on a lattice by using the center of the bonds
df (xi ) f (xi+1/2 ) − f (xi−1/2 )
≈ (8)
dx a
Iterating this leads to the discrete Schrödinger equation
~2
Eψ(xi ) = − (ψ(xi+1 ) + ψ(xi−1 ) − 2ψ(xi )) . (9)
2ma2
To simplify calculations and avoid numerical overflow or underflow caused by the appear-
ance of very large or small numbers, we rewrite the equation in Hartree atomic units
(abbreviated as a.u.), in which ~ = 1 and m = 1. Now a is the lattice constant in units
of a0 .
1
Eψ(xi ) = − (ψ(xi+1 ) + ψ(xi−1 ) − 2ψ(xi )) [Hartee Units (a.u.)]. (10)
2a2

In order to implement the hard wall condition we consider a chain of length L = (N − 1)a
and label the sites from i = 1, ..., N . The wavefuncvtion is zero everywhere else. In order
to solve the differential equation it is useful to think of the wave function as a vector,
meaning ψ ~ = (ψ(x1 ), ψ(x2 ), ..., ψ(xN ))T and we can write the discretized Schrödinger
equation as
~ = Ĥ ψ
Eψ ~ (11)

3
3 Approximation methods in Quantum Physics Problem class QM

where
 
2t −t 0 .. 0

 −t 2t −t 0 0 

Ĥ = 
 .. .. .. .. .. 
 (12)
 0 0 −t 2t −t 
0 .. 0 −t 2t

is a band-diagonal N × N matrix and t = 1/2a2 . We have now converted the Schrödinger

equation to a linear algebra eigenvalue problem.
a) For a = 0.1 a.u. and L = 20 a.u. (So N = 201, the endpoints are lattice points),
numerically calculate the eigenvalues and eigenvectors (which we will call wavefunctions
for simplicity). Sort the eigenvectors and eigenvalues from low to high energy.
b) Plot and examine the wavefunctions and compare them to the known solutions for the
continuum Infinite Square Well.
c) List the first 5 eigenvalues and compare them to the known solutions.
d) List the 20th, 40th, 60th ... 200th eigenvalue and compare them to the known solu-
tions.
e) Make a plot of eigenenergy versus level number. As a reference, show the exact result
from Griffiths in the same plot.
f) For which states, and in what sense, is the lattice chain (also called tight-binding)
approximation a reasonable approximation of the continuum physics?
g) Discussion question: In many problems such as this, the symbol i is used for an index,
while it also represents the imaginary unit. Similarly, sometimes an index label m is used,
the same symbol we use for mass. Is that a problem? Why (not)? Given that this type
of ambiguity is widespread in theoretical, numerical and experimental physics papers and
even textbooks, should we take care to avoid such ambiguities in all course materials?
h) Advanced question: Is it the best approximation to have the endpoints of the interval
(where the boundary condition is enforced) as lattice points? See if you can improve and
approximate the exact result better.
Note: Sample code for this problem, and for a similar calculation, including the harmonic
oscillator potential, is available on Blackboard.

B: Perturbation Theory
3.3 First order perturbation theory in the infinite square well [Key!]

4
Problem class QM 3 Approximation methods in Quantum Physics

Study Example 7.1 (Griffiths 3rd Ed.). Consider whether you think the result is plausible.
The result of perturbation theory is a good approximation for small V0 . What happens
to the wavefunctions and the energies when V0 is very large?
Insights: This should give you an idea of perturbation theory and its limitations.

3.4 Perturbation theory in an exactly solvable problem [Furher practice]

Insights: Applying perturbation theory to a problem that can be solved exactly is an

important consistency check.
Solve problem 6.2 (Griffiths 2nd Ed.) / 7.2 (Griffiths 3rd Ed.).

Solution:
p p
(a)
p We have ω → (1 + )ω, so E n → (1 + )En . Using the Taylor expansion
(1 + ) = 1+ 2 − 8  +..., we rewrite this to second order as En → En + 12 En − 18 2 En +....
1 1 2

(b) We have H 0 = 21 mω 2 x2 . We can use the representation of x in terms of ladder

operators, see Example 5 in chapter 2, to find
~
x2 = [(a+)2 + a+ a− + a− a+ + (a− )2 ],
2mω
which leads to
~ ~
hψn |x2 |ψn i = (n + 1 + n) = (2n + 1) .
2mω 2mω
Now therefore
1 ~ 1 
hψn |H 0 |ψn i = mω 2 (2n + 1) = (n + 1/2)~ω = En .
2 2mω 2 2
We see this is the same first-order term as from the Taylor expansion.

3.5 An approximation of an approximation [ Key !]

Insights: A relatively simple problem where the results of perturbation theory can be
visualized.
Revisit example 7.1 (Griffiths 3rd Ed.) and especially the second case (Figure 7.3). We
now want to calculate the wavefunction correction to the ground state to first order,
and the energy corrections to the ground state in second order perturbation theory. The
results involve infinite sums, which we will approximate by taking into account only the
first term.

5
3 Approximation methods in Quantum Physics Problem class QM

a) Calculate the matrix elements V1m ≡ hψ10 |V 0 |ψm

0
i, for m = 2, 3, 4, ... (Hint: You should
find that it is nonzero only for even m).

b) The series 7.13 (Griffiths 3rd Ed.) gives the perturbation to the wavefunction. List
the prefactors for the leading terms m = 2, 4, 6.

c) We will now continue using only the m = 2 term. Discuss whether you think this
could be a good approximation.

d) Calculate the m = 2 term explicitly. What will be its effect on the wavefunction? Is
it what you expect?

e) Now write the series expression for the second-order correction to the energy.

f) (Extra! Difficult!) can you sum the infinite series? If not, can you prove convergence?

g) Calculate the first three terms. How accurate do you expect the result based on the
first three terms to be?

h) Modify the code in the exercise “Numerical particle in a box” to include the same
(0)
perturbation. Use V0 = 0.1E1 . Compare the results of perturbation theory with the
finite-difference approximation.

Solution:

The matrix element (we leave out the factor V0 for now)
Z a/2 Z a/2
(0)∗ (0) 2 πx mπx
V1m = dxψ1 (x)ψm (x) = dx sin( ) sin( )
0 a 0 a a
cos(θ−ϕ)−cos(θ+ϕ)
Using sin θ sin ϕ = 2
, this becomes

a/2
(m − 1)πx
Z
1 (m + 1)πx
dx cos( ) − cos( )
a 0 a a

where the primitive is easily found (assuming m > 1),

1 (m − 1)πx 1 (m + 1)πx a/2

sin( )− sin( )|0
(m − 1)π a (m + 1)π a

1 (m − 1)π 1 (m + 1)π
= sin( )− sin( )
(m − 1)π 2 (m + 1)π 2

6
Problem class QM 3 Approximation methods in Quantum Physics

Now, if m is odd, both m + 1 and m − 1 are even, so the sine functions evaluate to 0. If m
is even, the sine functions are sin( (m−1)π
2
) = 1 for m = 2, 6, 10, ..., and sin( (m−1)π
2
) = −1
for m = 4, 8, 12, ... The other term always has the opposite sign, so we find
(m − 1) + (m + 1) 2m
V1m = 2
(−1)m/2−1 = (−1)m/2
(m − 1)π (1 − m2 )π
m Exact Value Numerical Approximation
4
2 3π
0.4244
8
4 − 15π −0.1696
12
6 35π
0.1088
16
8 − 63π −0.0807
The terms after 2 decrease in size and are alternating in sign. We now turn to the
perturbation of the wavefunction,
(0) (0)
X hψm |H 0 |ψn i
ψn(1) = (0) (0)
(0)
ψm .
m6=n En − Em
(0)
Here we can substitute the matrix elements in the numerator, and the energies En =
n2 π 2 ~2
2ma2
to find (re-introducing the prefactor V0 ),

(1)
X Vm1 (0)
ψ1 = V0 (0) (0)
ψm .
m6=1 E1 − Em
r
(1) 2 2ma2 X 2m mπx
ψ1 = V0 (−1)m/2 sin( ).
a π 3 ~2 m=2,4,...
2
(1 − m )2 a

We’re only looking at the first term now,

r
(1) 2 8ma2 2πx
ψ1 ≈ −V0 3 2
sin( ).
a 9π ~ a
We see that this term is negative where the half-barrier is, i.e. the probability shifts away
from the barrier as expected.
Let’s now look at the second order correction,
(0) (0)
X |hψm |H 0 |ψn i|2
En(2) = (0) (0)
.
m6=n En − Em
and filling in,
(0) (0)
(2)
X |hψm |H 0 |ψ1 i|2
E1 = (0) (0)
.
m=2,4,6,... E1 − Em

7
3 Approximation methods in Quantum Physics Problem class QM

(2) 8ma2 V02 X m2

E1 = .
π 4 ~2 m=2,4,6,...
(1 − m2 )3

(2) 4V02 X m2
E1 = (0)
.
π 2 E1 m=2,4,6,...
(1 − m2 )3
The series can be summed or −π 2 /64, and convergence can be shown easily as for large m
the terms become of order 1/m4 . Let’s look at the first three nonzero terms of the series:
m Exact Fraction Numerical Approximation
4
2 − 27 −0.1481
16
4 − 3375 −0.0047
36
6 − 42875 −0.0008
4 16 36
Sum for m = 2, 4, 6 − 27 − 3375 − 42875 −0.1537
1 2
Infinite Series Sum − 64 π −0.1542
This leads to the result
(2) V02
E1 = − (0)
,
16E1
summing the full series, or a result that is only a few percent smaller if only the first three
terms are taken into account.
The numerical result is available as a separate python file. It agrees very well with the
(0)
perturbation result untill the perturbation becomes larger than about 0.25E1 . We leave
it to you to also visualize the wave functions.

3.6 Two levels [Practice]

When a system such as an atom has two levels that are relatively close together in energy
while all other levels have very different energies, the denominators in the expressions
(7.13) and (7.15) suppress the terms corresponding to those “far-detuned” levels. As a
result, we can make the two-level approximation, where we only take into account the
two relevant levels a and b. The wavefunction may now be written as a vector,
 
a
|ψi = ≡ a|ψa i + b|ψb i.
b
The potential then can be written as a (2x2) matrix,
 
Vaa Vab
V = .
Vba Vbb
0
with the matrix elements Vmn ≡ hψm |V |ψn0 i. The unperturbed Hamiltonian is
 0 
0 Ea 0
H = (13)
0 Eb0

8
Problem class QM 3 Approximation methods in Quantum Physics

from which we see that

 theunperturbed
 eigenenergies are Ea , Eb , and the corresponding
1 0
eigenvectors are and . The perturbation is
0 1
 0
Vaa Vab0

0
H =λ (14)
Vba0 Vbb0

where both H 0 and H 0 are hermitian (λ is a real constant).

(a) Find the exact energies for this two-level system.
(b) Expand your result from (a) to second order in λ (Taylor series) and set λ = 1,
subsequently. Verify that the terms in the series agree with the result from perturbation
theory. Assume Eb > Ea .
0
(c) Setting Vaa = Vbb0 = 0, show that the series in (b) only converges if

Vab0 1
0 0
< . (15)
Eb − Ea 2

Solution:

9
 Solution to problem 7.3 (2021)

too it i is
Ca
H H H I II
Ea tE5 d Vaa Usb I V EdiE5eNVaa Vbb
2
4EaE5tNEaVbbeE5Vaa
EI 2

a2 UaaUssVabVbT

b 9
lb Ho EI EI eigenstates

correction
fcistadu
Ea Ei incl this 3 b our

Ea dVaa told
Y Y f 01N
Es Esta lo 1

Eso Ngbtd R2
secondorder
ko.INT18 s7lb l2 azlVbaI
sea 22 E Ebo Ed Eso
lli0 lYaaVasbb l9 azlVasI
s Es.az Ef Eso EE
Eso

Ea EatenVaa talked
3
01
EI Eso
Olds
Es Esoadvised
Problem class QM 3 Approximation methods in Quantum Physics

C: The variational principle in the context of the har-

monic oscillator
. The variational method allows to approximate the ground state energy of systems whose
Hamiltonian we cannot simply solve. In this exercise you will apply it to a problem with
a known solution.

3.7 Ground state and first excited state [Repetition, related to Example 8.1.
Try to solve without looking at the Example.]
The Hamiltonian we consider is given by
p̂2 1
H= + mω 2 x2 . (16)
2m 2
2
a) Make an ansatz for the wavefunction according to ψa = e−ax . Take care of normaliza-
tion. Show that a minimization with respect to the variational parameter a leads to the
exact eigenstate and eigenenergy of the groundstate. Make sure when taking derivatives,
that you include the derivative of the normalization factor and apply the product rule.

Solution:

https://chat.openai.com/share/16302166-7814-4eef-bcef-3c8a1f5ed794
2
b) Now make an ansatz according to ψa = xe−ax (with the proper normalization factor
left for you to evaluate). What is the energy you find and which exact eigenstate does it
correspond to. Give an argument why this could have been expected to not return the
ground state. How would you make an ansatz for higher excited states?

Solution:

Following the exact same procedure we find the same value for a, but now the energy
is that of the first excited state E = 23 ~ω. The ground state of a quantum harmonic
2
oscillator is symmetric and does not have nodes. The ansatz ψa = xe−ax is asymmetric
(odd function) and has a node at x = 0, which is characteristic of excited states. In
quantum mechanics, the ground state of a system is always nodeless and has the lowest
possible energy. Since this wavefunction has a node, it cannot represent the ground state.
It has the correct symmetry for ths first excited state. For the higher excited states we
could look at wavefunction that have more nodes, but unless we can get the positions of
the nodes right a priori, it will be necessary to introduce more parameters.

11
3 Approximation methods in Quantum Physics Problem class QM

3.8 Part II: Shifted harmonic oscillator [Key!]

Insights: Learn a simple case where the variational principle can be used with multiple
variational parameters.
The Hamiltonian we consider is now given by

p̂2 1
H= + mω 2 (x − b)2 , (17)
2m 2
a shifted harmonic oscillator.
2
a) Use the ansatz for the wavefunction according to ψa = e−a(x−x0 ) (take care of normal-
ization!). Show that a minimization with respect to the variational parameters a and x0
leads to the exact eigenstate and eigenenergy of the groundstate

Solution:
2
We find for the kinetic energy term a~2m
exactly as above, and for the potential energy
2
mω (4a(x0−b) +1)
2
term 8a
. We can now set the derivative with respect to x0 to zero, which
leads directly to the expected result x0 = b, and then the problem reduces to the case
above.

b) What ansatz would you use for the first excited state?

Solution:

There are a number of possibilities. We could chose one that has a zero fixed at x = b or
introduce additional parameters. Important is that there must be a node, which is either
fixed at the correct point or can be moved there by optimizing the parameters.

12