Solutions Homework 9

Physics 486, Fall 2014

University of Illinois at Urbana-Champaign

October 24, 2014

Problem I: Commutators

(a)
[y, px ] f (x, y) = y(ix ) f (x, y) + ix (y f (x, y)) = iy fx (x, y) + iy fx (x, y) = 0
(b) Let us write down all the commutation relations between ra and pb , where ra = x, y, z and pb = px , py , pz .
[ra , rb ] = 0
[pa , pb ] = 0
[ra , pb ] = ihab
We know that La = amn rm pn , where repeated indices are summed over.Then
[La , Lb ] = [amn rm pn , bstrs pt ] = amn bst [rm pn , rs pt ] =
We use the following identity (which you can prove by yourself)
[AB,CD] = A[B,C]D + [A,C]BD +CA[B, D] +C[A, D]B
to obtain:
[La , Lb ] = amn bst (rm [pn , rs ]pt + [rm , rs ]pn pt + rs rm [pn , pt ] + rs [rm , pt ]pn )
Using the commutation relations between position and momentum:
[La , Lb ] = i
hamn bst (rm pt ns + rs pn mt )
[La , Lb ] = i
hamn bnt rm pt + iatn bst rs pn
amn bnt rm pt = nam nbt rm pt = (ab mt at mb )rm pt = ab rm pm + rb pa
where repeated indices are summed over. Similarly,
atn bst rs pn = tantbsrs pn = (ab ns as nb )rs pn = ab rs ps + ra pb
We finally obtain:
[La , Lb ] = i
hab rm pm i
hrb pa i
hab rs ps + i
hra pb = i
h(ra pb rb pa ) = i
habc Lc
(c)
L2 = Lm Lm

with summation over repeated indices.

[La , L2 ] = [La , Lm Lm ] = [La , Lm ]Lm + Lm [La , Lm ] = amc Lc Lm + amc Lm Lc = amc Lc Lm + acm Lc Lm = 0
(d)
[Lz , L ] = [Lz , Lx iLy ] = [Lz , Lx ] i[Lz , Ly ] = iLy Lx =
h(Lx iLy ) =
hL
where we have used the results from (c).
[L2 , L ] = [L2 , Lx iLy ] = [L2 , Lx ] i[L2 , Ly ] = 0
where we have used the results from (b).
(e) Since Ylm is a simultaneous vector of Lz , L2 , it must be the case that if Ylm is a simultaneous vector of
2
L , L+ , it must also be a simultaneous vector of Lz , L+ , which is equivalent to [Lz , L+ ]Ylm = 0. Using part (d)
[Lz , L+ ]Ylm = hL
+Ylm
so that [Lz , L+ ]Ylm = 0 translates to
L+Ylm = 0
which implies l = m. Thus Yll is an equivalent vector of L2 , L+ (and Lz ):
L2Yll = l(l + 1)Yll
L+Yll = 0Yll
LzYll = lYll

Problem 2: Matrices of angular momentum operators

Before computing, it is useful to discuss some properties of the 3D harmonic oscillator. In any energy
eigenbasis, the hamiltonian is block diagonal, with each block corresponding to a sector of the Hilbert
space of a given energy En = (n + 3/2)
h. Since Lz , L2 and H are mutually commuting operators (as they are
for any 3D hamiltonian with a radial potential), they can be simultaneously diagonalized. Thus, the Lz and
L2 matrices will also be block diagonal (same blocks as the hamiltonian) in any energy eigenbasis.
(a) Lets write the three energy eigenstates in the position basis. Since hxyz|i jk i = hxi i hy j i hzk i, where
|n i is the n 1d harmonic oscillator eigenstate, we have that
100 (x, y, z) = 1 (x)0 (y)0 (z) = C1C02 e

m(x2 +y2 +z2 )

2
h

Moving to spherical coordinates:

mr2
2h

r sin cos

mr2
2h

r sin cos

2
mr
2
h

r sin sin

100 (r, , ) = C1C02 e

Similarly,
100 (r, , ) = C1C02 e
010 (r, , ) = C1C02 e

001 (r, , ) = C1C02 e

mr2
2h

In spherical coordinates, Lz is represented by:

Lz = i
We find:
Lz |001 i = 0
2

r cos

Lz |100 i = i |010 i
Lz |010 i = i |100 i
so that:

0
L1 = i
0

i 0
0 0
0 0

L1 is hermitian as it corresponds to the representation of an observable operator. In spherical coordinates

L2 is given by:



1 2
1
2
sin
2
L =
sin

sin 2
We find:
L2 |001 i = 2 |001 i
L2 |010 i = 2 |001 i
L2 |001 i = 2 |001 i
so that

2
L2 = 0
0

0
2
0

0
0
2

(b) Since |100 i and |200 i correspond to eigenstates of different energy and Lz and L2 are block diagonal
in any energy eigenbasis, we find:
h100 |Lz |200 i = 0
h100 |L2 |200 i = 0
(c) Making use of
Lz |100 i = i |010 i
Lz |010 i = i |100 i
we can exponentiate easily to find:
eitL1 |100 i = cos(t)100 sin(t) |010 i
Thus, for t = n, |100 i rotates back to itself, and for t = 2n+1
2 , |100 i rotates to |010 i.
(d) Using Alpha, we find

cost sint 0
cost 0
eitL1 = sint
0
0
0
which is the same as the rotation matrix around the z-axis.
(e) The result does not change since Lz is block diagonal in the energy basis |i jk i.
(f) The operator eitLz rotates a wave-function by t around the z-axis. We notice the similarity of Lz = i
with the usual px = ix . Thus we can see the action of Lz as translation in the compactified direction.
(g) Since Lz |Ylm i = m |Ylm i, then
eitLz |Ylm i = eitm |Ylm i
We see that for any t, |Ylm i rotates back to itself (up to a phase).
(h) The probability density | h, |Ylm i |2 does not depend on .

Problem 3: Three dimensional harmonic oscillator of different

We see that the hamiltonian is the sum of three distinct harmonic oscillators in the x, y, z directions with
frequency 1 , 1 , 2 . Let nx , ny , nz label the quantum numbers in the three sectors. The spectrum is given by:
1
Enx +ny ,nz = h
1 + h
2 + (nx + ny )
h1 + nzh
2
2
1
Let r =
2 . I will label a state as (m, n) where m = nx + ny and n = nz . These states are not unique because
2
of the degeneracy in m and possible accidental degeneracies in n (for special values of
1 ). Provided no
accidental degeneracies, the degeneracy in the (m, n) sector is the degeneracy of a 2-dimensional harmonic
oscillator at level m. In the figure below I plot the spectral flow obtained using Matlab, in units of h
1 = 1.

At r = 0, the system is a 2d harmonic oscillator of frequency 1 . When r  1 (that is for 0 < r < 0.2), the
first 6 unique eigenstates are the eigenstates of a 1d harmonic oscillator of frequency 2 in the z direction.
As r increases it becomes more and more energetically favorable to have 1 modes. At r > 5, the first 6
unique eigenstates are the first 6 unique eigenstates of a 2d harmonic oscillator of frequency 1 . In the
figure, cusps and jumps in the flow of each unique eigenspectrum value are due to level repulsion. The
level coming from below finds it more favorable to accept 1 modes and reduce 2 modes. As the slope of
the flow is nz (number of 2 modes) the slope decreases after such a level-level interaction and the offset
increases (the offset is nx + ny ). The level coming from above experiences a jump as the degeneracy of the
level coming from below increases, after which the slope increases, but the offset decreases. The two levels
exchange quantum numbers as a result of such interaction.
In the table below, the quantum numbers (nx + ny , nz ) of each level are given for 0 < r 1. I have listed
the accidental degeneracies, ignoring degeneracies in nx + ny . Thus to compute the degeneracy of a level we
have to multiply each entry in the table (m, n), at some value of r for a given level, by the degeneracy of a 2d
harmonic oscillator at level m and then add all of these. For example, the degeneracy of level 2 at r = 0.5 is
4

1 + 2 = 3, as the degeneracy of a 2d harmonic oscillator at level 1 is 2. The quantum numbers (m, n) for the
levels for 1 < r < m are the quantum numbers (n, m) (switching (m, n)) for 1/m < 1/r < 1. The degeneracy is
to be computed as above.

Problem 4: Superposition of Ylm

We know that
hYlm |Yl 0 m0 i = ll 0 mm0
0 s are orthogonal:
(a) Let us compute h|i. Since the Ylm


4 1 1
5
h|i = N 2
+ +
= N2
5 4 5
4

For to be normalized we thus require:

r
N=

2
4
=
5
5

. (b) We know that LzYlm = mYlm (h = 1). Thus the possible results of Lz are 1, 0, 1 with the following
probabilities
1
1
P(Lz = 1) = N 2 =
4
5
4 2 16
P(Lz = 0) = N =
5
25
1 2
4
P(Lz = 1) = N =
5
25
(c)We know that
p
L+Ylm = l(l + 1) m(m + 1)Yl(m+1)
p
LYlm = l(l + 1) m(m 1)Yl(m1)
We see that this confirms L+Yll = 0 and LYll = 0. We can now compute



2
2
1
4 2
2
2Y11 +
L+ =
2Y10 =
Y11 + Y10
2
5
5
5
5



2
2
1
4 2
2 2
2Y11 2Y10 =
L =
Y11
Y10
5
5
5
5
5

Problem 5: Squares of hermitian operators are positive

Let A be hermitian. We know A is diagonalizable. Let {|i} be the basis formed by orthogonal eigenvectors
A |i = |i, with R.. Any wave-function can be expanded in this basis as:
of A,
|i = c |i

Then:
h|A2 |i =

0 c0 c h0 |A2 |i

. We have that

A2 |i = 2 |i

and

thath0 |i = 0 , .

We then have:

h|A2 |i =

0 c0 c 2 0 ,

h|A2 |i = |c |2 2 0

Problem 6: Units of real and imaginary time path integrals

Let [] denote the units of . Then [E], [T ] are the units of energy and time.
[t]
[T ]
1
it
=
=
[ ]=
h
[h] [E][T ] [E]
We know that [kT ] = [E]. Then
[
Thus:

1
1
]=
kT
[E]

it
1
[ ]=[ ]
h
kT

Problem 7: Simultaneous eigenstates in 1d

(a) Since H = p2x +V (x),

[px , H] = [px , p2x ] + [px ,V (x)] = [px ,V (x)]

(b) Using (a) [px , H] = 0 requires [px ,V (x)] = 0. Since px is the generator of translations in the x direction, we
see that the above implies V (x) must be translationally invariant, so a constant function. We can reach this
conclusion by doing the computation. In the x-basis px acts as a derivative, so to make sure we are correctly
calculating the commutator, lets apply it to a test function f (x).
[px ,V (x)] f (x) = ix (V (x) f (x)) + iV (x)x f (x) = i f (x)V 0 (x)
. Imposing [px ,V (x)] = 0, we get f (x)V 0 (x) = 0 for arbitrary f (x). This means V 0 (x) = 0, which implies V must
be constant.