Advanced Quantum Mechanics, Notes based

on online course given by Leonard Susskind

- Lecture 2

Symmetry Groups and Degeneracy

Symmetries are operations on systems that dont change the description, phenomena, the energies

eg translating an atom, interchanging electrons

Crystal lattice has a translation symmetry if you consider it to be infinite

Translation symmetry, moving it a ‘little bit’, Additional symmetry moving it a ‘unit’ of the lattice

Moving one unit of the lattice is an additional symmetry

Translation symmetry eg moving left a 'little bit'

Homogeneity of space -> translation symmetry

Isotrophy of space -> rotational symmetry

0:05:00 Degeneracy
Degeneracy is related to symmetry

Is degenerate if more than one state has the same energy

For generic Hamiltonian with discrete spectrum of energy levels it would be an odd accident to have
two levels with exactly the same energy

But doesn’t take a coincidence, takes a symmetry

Exact degeneraces require symmetries

Symmetries sometimes give degeneracies, but not always

content from lecture by Leonard Susskind

2 | AQMLecture2.nb

Example of when symmetries imply degeneracy

Rotational symmetry.

Consider a particle constrained to moving on a circular wire, ‘like a bead’

θ
x

If circle in x, y plane can describe particle position with x and y co-ordinates

But since constrained to wire only really need θ

ψ* (θ)ψ(θ) defines probability to find particle on circle

Rotation operation moves particle counter clockwise so ψ(θ) -> ψ(θ - ϵ)

∂ψ ∂ψ
∂ψ = - ϵ = -i ϵ (-i ) Put i’s in so item in bracket will be Hermitian
∂θ ∂θ

∂
Operator called angular momentum is usually denoted L = -i
∂θ

For small changes ϵ

∂ψ = -i ϵ L ψ

This identifies L as the generator of rotations.

Putting the ℏ’s back:

∂ ∂
L = -i ℏ similar to P = -i ℏ
∂θ ∂x
iϵ
∂ψ = - L ψ
ℏ

What are the eigenvectors/values of angular momentum?

L |Ψ 〉 = m |Ψ 〉 (1)
m is standard for eigenvalue of angular momentum, stands for magnetic quantum number

content from lecture by Leonard Susskind

 AQMLecture2.nb | 3

In example is rotation about z so it really Lz

imθ
∂ψ(θ)
(1) is -i ℏ = m ψ(θ) -> eigenvector for m of ψ(θ) = e ℏ
∂θ

Unlike linear momentum have the further constraint, if θ=2π the wave function should be the same

ψ(2π)=ψ(θ)

imθ im0
For θ=0 e ℏ =e ℏ =1

im2π
⟹ e ℏ =1

m
⟹ is an integer
ℏ

Usual notation is to work in units of ℏ and say m is an integer

We expect energy to depend on angular momentum

Is Energy the same for plus and minus m? ie E(m) = E(-m) ?

The difference between plus and minus is just the direction the particle is going round the ring.
The sign does not tell you anything about the physical direction it just depends on which way you
chose θ to be positive

If the particle is in a field eg magnetic then the energies are different otherwise they are the same.

With no field the energies are degenerate - ie pairs of E(m)=E(-m), except for zero.

The field breaks the degeneracy.

Note: if we think of the magnetic field as into the paper, it does not invalidate the rotational invari-
ance

x x x x x x

x x x x x x
θ
x x x x x x

x x x x x x

Rotational symmetry on its own is not enough to say the energy levels are degenerate

content from lecture by Leonard Susskind

4 | AQMLecture2.nb

Need an additional symmetry

Here it is reflection about one of the axes. Reflection is a discrete symmetry.

We’ll call ‘reflection symmetry’ ‘mirror symmetry’ and use the operator M to distinguish it from the
rotational symmetry.

Imaging mirror in x-z plane

If no mirror symmetry then energy levels of these 2 dont have to be the same

But if the descriptions of the 2 items are indistinguishable they must have the same energy.
0:28:30 Adding Mirror Symmetry to Rotational Symmetry
Note: term Mirror is used for something else in physics; dont confuse with our use here

Necessary to have 2 symmetries that dont commute to get degenerate energy levels

We’ll look for Energy levels where E(m)=E(-m)

x x x x x x x

x x x x x x x

x x x x x x x Mirror image reverses direction eg

of magnetic field which is an axial
vector. Remember magnetic field
created by currents

Show rotation and reflection dont commute

Reflection Mψ(θ)=Mψ(-θ)

∂
Operator for small rotation L=-i ℏ
∂θ

content from lecture by Leonard Susskind

 AQMLecture2.nb | 5

imθ
Consider the commutator acting on eigenvector e ℏ found above

imθ imθ imθ imθ -imθ

∂
MLe ℏ = M( -i ℏ e ℏ ) = M -i ℏ i m e ℏ = ℏmMe ℏ =ℏm e ℏ
∂θ
imθ -imθ -imθ -imθ -imθ
∂
LMe ℏ = L( e ℏ ) = -i ℏ ( e ℏ ) = -i ℏ -i m e ℏ =-ℏm e ℏ
∂θ

Not the same. So dont commute

Observations

Consider A and B as operators for 2 symmetries (either generators or the full thing):

Since each of A and B is a symmetry they commute with the Hamiltonian [A,H]=0 [B,H]=0

This doesn’t mean [A,B]=0 result could be:

- A or B or a linear combination of A and B
- C which is something new

Consider [A,B]=iC (put the i so that C will be Hermitian and conserved)

[C,H]= [ [A,B], H] = (AB-BA)H - H(AB-BA)

=ABH - BAH - HAB + BAH
=ABH - HAB
=AHB - HAB since H commutes with B
=HAB - HAB since H commutes with A
=0

Result: if A and B commute with H then the commutator of A and B commutes with H

Above also shows product of AB commutes with H but we use the commutator of A and B as it is
Hermitian whereas the product generally isn’t

Generating a new symmetry

Given 2 symmetries that dont commute you generate a symmetry which might or might not be
independent of A and B

If it is independent you can commute the new symmetry with A and B and see if you get any other
independent symmetry

Keep on until you dont get something new

Might never end, infinite number of symmetries but this is rare

Or it will close giving you a finite number of independent symmetries. ie commutation of any 2 of the
symmetries gives you one of the others (or a linear combination of the others)

You now have a commutator algebra for these symmetries and H, called a symmetry group

The Generators form a commutator algebra, a Lie algebra.

content from lecture by Leonard Susskind

6 | AQMLecture2.nb

The symmetries themselves that are built up from the generators by applying small changes transfor-
mations repeatedly, those are the group elements.

We’ll work mostly with the generators.

Abelian - where the symmetries commute with each other. Dont get anything particularly interesting

Non-Abelian - where there are non-commuting symmetries

0:50:30 Definition of a Generator

Take the unitary symmetry operation and consider a case close to the identy

U = U - iGϵ

G is called the generator

Build up continuous symmetry from applying the generator

The generators contain nearly all the interesting matrerial about the group

The algebra is a collection of generators you can add them, subtract them, multiply them by ordi-
nary numbers and commute them. Commuting is like a product in the algebra.

content from lecture by Leonard Susskind

 AQMLecture2.nb | 7

Example: angular momentum and rotation about different axes.

If rotate about x and the about y gives a rotation about z

Rotating about x then y is not the same as rotating about y then x

Can show this from commutation relations of angular momentum

Rotation about any axes form a group

The 3 components of angular momentum, the three generators, form a Lie algebra / commutator
algebra

First consider Classical Angular Momentum eg planetary orbit

Lz Y

Angular momentum Lz perpendicular to the orbit

Particle in orbit has energy E. At classical level is the energy degnerate? ie is there another orbit
with the same energy?

Yes there are many orbits with the same energy - some of these you can easily prove from rota-
tional invariance eg above shows rotation about y axis giving new configuration, angular momentum
has changed, shift in x angular momentum

Classically, combination of different rotational invariances about different axes means there must be
degeneracy ie orbits with the same energy.

0:59:00 Quantum Mechanically - need to understand the angular momentum generators

Assume particle moving in 2D not restricted to a circle

content from lecture by Leonard Susskind

8 | AQMLecture2.nb

Rotating the configuration by ϵ (small) shifts particle by ∂ x and ∂y

For small ϵ
∂x=-ϵy
∂y=ϵx

Consider change in wave function

∂ψ ∂ψ ∂ψ ∂ψ
δψ(x,y) = ∂x + ∂y = -ϵ y +ϵx
∂x ∂y ∂x ∂y

Like -ϵ y px + ϵ x py

∂ψ ∂ψ
By definition -ϵ y +ϵx = i ϵ Lz
∂x ∂y

Lz = x py - y px (part of L = r x p)
Lx = y pz - z py
Ly = z px - x pz

These are symmetries if the system is rotationally invariant eg a particle moving in a centrail force
field where rotating the system doesn’t change the Hamiltonian

[Lx ,H]=0 [Ly ,H]=0 [Lz,H]=0

1:07:50 Do they commute with each other?

If we only knew Lx and Ly could we discover Lz?

Remember: [x, y]=0 etc, px , py  = 0 etc [x, py  = 0 where the position co-ordinate differs from
momentum suffix but [x, px ] = y, py  = [z, pz] = iℏ

Using these find Lx , Ly  = [(ypz - zpy ),( z px - x pz)]

= [ypz, z px ] - [ypz,x pz] - [zpy ,z px ] + [zpy ,x pz] keeping out only non-commuting combinations
= [ypz, z px ] + [zpy ,x pz]
= ypzz px - z px ypz + zpy x pz - x pzzpy moving thro the items that commute (here all but z and pz)
= ypx pzz - px yz pz + py x zpz - x py pzz and a bit more ( x and py , and y and px commute)
= ypx [pz, z] + py x [z, pz]
= ypx [pz, z] - py x [pz, z]

content from lecture by Leonard Susskind

 AQMLecture2.nb | 9

= - Lz [pz, z]
= - Lz-i ℏ = i ℏ Lz

So we have discovered the third generator.

The Lie algebra for the 3 components of the angular momentum

[Lx , Ly  = i Lz
Ly , Lz] = i Lx
[Lz, Lx ] = i Ly

Question: should we prove that Lz is not a linear combination of Lx and Ly ?

Response: Yes, could prove by showing that Lz does not cummute with any linear combination of
Lx and Ly

A Lie algebra has ‘commutation’ as the ‘multiplication’. It contains a group of components where
commutation doesn’t give anything that isn’t in the group in some form already.

Also have [Lx , H]=0 Ly , H]=0 [Lz, H]=0 so each component of angular momentum is conserved.

1:19:00 Creation and Annihilation

Remember for Harmonic Oscillator we had closed commutation relations for creation and annihila-
tion operators.

Here we look at creation/annihilation operators that change the z component of angular momentum
(arbitrary choice)

Is usual in QM to chose a representation eg we have a position representation and a momentum

representation and in spin we have the σz representation.

Doesn’t matter which representation you choose - just pick one and stick with it

Here we pick the z component of angular momentum and its eigenvectors (we could have picked x
or y)

Lz|m> = m |m>

|m> is eigenvector of Lz with eigenvalue m

What are the possible values of m?

In Harmonic Oscillator we did this by inventing operators that raised and lowered the enrgy, here
will use operators that raise and lower m

Note: audience pointed out Harmonic operator not covered in lectures. But is covered in book.

Let’s define:

Lx + i Ly = L+

content from lecture by Leonard Susskind

10 | AQMLecture2.nb

Lx - i Ly = L-

from earlier
[Lx , Lz] = - iLy
[iLy , Lz] = - Lx

[L+ , Lz] = - L+ from adding the above two equations

[L- , Lz] = L- from subtracting

Is another way of writing the Lie algebra

So we have
Lz|m> = m |m> and [L+ , Lz] = - L+ which combine to give

[L+ , Lz] m > = - L+ m>

L + Lz m > - L z L + m > = - L + m >
L + m m > - Lz L + m > = - L + m > and swapping sided for two of the terms
L+ m m > + L + m > = Lz L+ m >
(m + 1) L+ m > = Lz L+ m > or Lz L+ m > = (m + 1) L+ m >

so L+ |m> is an eigenvector of Lz with eigenvalue m + 1

Doing the same thing with L- gives us an eigenvector with an eigenvalue of one lower

If we plot on a vertical axis the values of m we only know they have integer spacing but not exact
position

3
2
m values
1
at integer
spacing but 0
dont know -1
exact postion -2
-3

There is another possibility in the up and down sequence, getting zero ie the sequence of m might
terminate at some point going up and going
down.

content from lecture by Leonard Susskind

 AQMLecture2.nb | 11

L+ might kill the vector (give null state) at some point going up and L- might kill the vector going down.

If it terminates then there is a relationship between where it terminates above and below because
we have rotational invariance

So imagine rotating the above diagram by 180° so Lz becomes -Lz

If really rotationally invariant the eigenvalues for Lz must be the same as for -Lz, so red dots must
be symmetrically placed

Two ways of doing this:

1. red dot on the origin and others at integer spacing above and below it
1 1
2. a red dot an and - and others at integer spacing above and below them
2 2

For orbital angular momentum we already proved it was integer spacing using ψ(0)=ψ(2π)

For now we’ll forget the half integer one that corresponds to spin angular momentum, will do this
later

The sequence may or may not terminate, let’s assume it does, we have a multiplet of states that are
somehow related to each other and which have to have the same energy

Why the same energy?

If H |m> = Em |m> then lets go to |m+1> which is L+ |m> and finds its energy

H L+ |m> = Em L+ |m> = Em+1 |m + 1>

But H commutes with all components of the angular momentum so

L+ H |m> = Em+1 |m + 1>

Since H |m> = Em |m> this gives
L+Em |m> = Em+1 |m + 1>
EmL+ |m> = Em+1 |m + 1> as Em is just a number
Em |m + 1> = Em+1 |m + 1>

So Em=Em+1 they have the same energy

Can do the same with minus 1

So if we find an eigenvector of Lz which happens also to be an eigenvector of energy then we can

apply the raising and lower operators and, until we reach any end points, create states with the
same energy.

This is the idea of degeneracy following from symmetries that dont commute with each other

When we have non-commuting symmetries it tells us there will be degeneracies.

This is the quantum equivalent of the rotation of the classical orbit that we looked at earlier

content from lecture by Leonard Susskind

12 | AQMLecture2.nb

Raising and lowering operators are typical of this sort of closed algebra

Can make this concrete by looking at specific wave functions

content from lecture by Leonard Susskind