1

QUANTUM MECHANICS
(PHYS4010) LECTURE NOTES

Lecture notes based on a course given by Roman Koniuk.

The course begins with a formal introduction into quantum mechanics and then
moves to solving different quantum systems and entanglement

York University, 2011

Presented by: ROMAN KONIUK

LATEXNotes by: JEFF ASAF DROR

2011
YORK UNIVERSITY
 2

CONTENTS

I. Underlying Principles of Quantum Mechanics 2

A. Postulates of Quantum Mechanics 3
B. Important Properties 3
C. The Time Evolution Operator 4

II. Particle in a Box 5

III. Dirac Notation 7

IV. Hilbert Space 9

V. Hermitian Operators 10

VI. Fourier Series 12

VII. Fourier Transform 13

A. Example: Finite Wave Train 14
B. Generalize Fourier Transforms to Three Dimensions 15

VIII. Momentum Representation 15

A. Example 1: Hydrogen in the 1s state 16
B. Example 2:Harmonic Oscillator 16

IX. Commutators 16

X. Harmonic Oscillator (Operator Method) 18

XI. Classical Harmonic Oscillator 20

XII. Time Development of Expectation Values 21

XIII. Conservation Laws 23

XIV. Tunneling 24

XV. Angular Momentum 27

XVI. Orbital Angular Momentum 29

XVII. Hydrogen 32

XVIII. Matrix Mechanics 38

XIX. Spin 45

XX. Addition of Angular Momentum 48

XXI. Entanglement 51

XXII. Variational Method 55

I. UNDERLYING PRINCIPLES OF QUANTUM MECHANICS

Lecture 1 - Sept. 7th 2011

The course website is http://www.yorku.ca/koniuk/PHYS4010/index.htm .
 3

A. Postulates of Quantum Mechanics

1. To every observable there corresponds an operator. For example to the observable A(e.g. energy,momentum,
position, etc.) there corresponds an operator Â. Every measurement of A gives a value, a, s.t. a is an eigenvalue
of the operator Â. i.e. for an eigenfunction of A, φ, Âφ = aφ.

2. Measurement of observable A yields the value a, and then leaves the state in the state φa . i.e.

Aφa = aφa

3. All possible information is contained in the wavefunction, Ψ(r, t).

4. Ψ developes in time according to the Schrodinger wave equation (SWE)

∂Ψ(r, t)
i~ = ĤΨ(r, t) (I.1)
∂t

B. Important Properties

The average value (expectation value) of an observable C at time t is given by

Z
hCi (t) = Ψ∗ (r, t)ĈΨ(r, t)dr (I.2)

Experimentally this would be done by preparing an ensemble of identical initial states Ψ(r, 0) and measure C at time
t. This will generate a set of values, c1 , c2 , c3 , ..., cN (where N is the number of measurements)

N
1 X
hCi = ci (I.3)
N i=1

We can also interpret this as

X
hCi = ci P (ci ) (I.4)
i

The uncertainty (or standard deviation) in c is given by ∆c:

q
2
∆c = hc2 i − hci (I.5)

The modulus of Ψ is given by

2
Ψ∗ (r, t)Ψ(r, t) = |Ψ(r, t)| (I.6)

probability to find
particle at point x
z }| {
|Ψ|2 dx = P (x)dx (I.7)

The normalization condition is:

Z ∞
P (x)dx = 1 (I.8)
−∞

Lecture 2 - Sept. 9th 2011

 4

C. The Time Evolution Operator

ˆ
Suppose Ĥ doesn’t depend on time. (i.e. Ĥ = H(r). Assume Ψ(r, t) = φ(r)T (t)) Plugging this result into the
SWE:
∂(φT )
i~ = ĤφT (I.9)
∂t
∂T
i~φ = T Ĥφ (I.10)
∂t
Divide both sides by ψ:
f g
z }| { z }| {
1 ∂T 1
i~ = Ĥφ (I.11)
T ∂t φ

Thus f (t) = g(r) (regardless of t and r) The only way this can be true is if f (t) = g(r) = E where E is some constant.
This generates two equations

∂T
i~ = ET (I.12)
∂t
and

Ĥ φ̂ = Eφ (I.13)

Notice that equation I.13 contains all the physics of the problem while the time equation contains no physics (does
not contain the Hamiltonian). Equation I.12 can be solved by:

T (t) = e−iEt/~ (I.14)

Thus if the problem is seperable then the time dependance is always given by the above equation. All this term does
is changes the phase of the wavefunction (shifts the magnitude of the wavefunction from the real and complex part
of the wavefunction). This time dependence is often referred to as trivial time dependence.
Equation I.13 is an eigenvalue problem. E is the eigenvalue and φ is the eigenfunction of the operator Ĥ. There
are typically infinitely many solutions:

Ĥφn = En φn (I.15)

With

Ψn (r, t) = φn (r)T (t) (I.16)

Ψn (r, t) = φn (r)e−iEn t/~ (I.17)

The probability density can depend on time if we don’t have an eigenfunction.

The initial value problem: If we know the Ψ(r, 0) then we can determine Ψ for all time (i.e. Ψ(r, t))

∂Ψ
i~ = ĤΨ (I.18)
∂t
∂Ψ −i
⇒ = ĤΨ (I.19)
∂t ~
∂Ψ Ĥ
⇒ +i ψ =0 (I.20)
∂t ~
∂  
⇒ exp(itĤf /~)Ψ(r, t) = 0 (I.21)
∂t

Note: For this to be true we have to assume that Ĥ = Ĥ(r)

Comments:
 5

1. Notice that in the factor exp( it~Ĥ ), Ĥ is an operator (could be a matrix). This can be understood by the real
definition of the exponential function (a power series expansion):

itĤ (itĤ)2
exp(itĤ/~) = 1 + + 2 + ... (I.22)
~ ~ 2!
This operator is defined as:
!
−1 itĤ
Û ≡ exp (I.23)
~

Lets carry on with our solution:

Z t0 Z t0
∂  
exp(itĤf /~)Ψ(r, t) dt = 0dt (I.24)
0 ∂t 0

By the fundamental theorem of calculus:

exp(it0 Ĥf /~)Ψ(r, t0 ) − exp(0)Ψ(r, 0) = 0 (I.25)

You are free to change the dummy variable t0 to t and multiplying by Û :

Ψ(r, t0 ) = exp(−it0 Ĥf /~)Ψ(r, 0) (I.26)

The operator Û is called the time evolution operator.

Note that this is true for any state. Now suppose the initial state is an eigenstate (also called stationary states)
of Ĥ. Hence:

Ψ(r, 0) = φn (r) (I.27)

⇒ Ψ(r, t) = Û φn (r) (I.28)
= exp(−iĤt/~)φn (r) (I.29)
= exp(−iEn t/~)φn (r) (I.30)

Hence in this special case the time ordering operator is just the trivial phase found earlier.

Lecture 3

II. PARTICLE IN A BOX

Step 1:
Write down the potential
(
V =0 0<x<a
V (x) = (II.1)
V (x) = ∞ x ≤ 0, x ≥ a

Step 2:
Write down the S.E.
~2 ∂ 2 ∂Ψ
− Ψ + V (x)Ψ = i~ (II.2)
2m ∂x2 ∂t
The time portion is given by (assuming a separable solution)

T (t) = exp(−iEt/~) (II.3)

 6

Thus the time independent problem is now given by

~2 ∂ 2
− φ(x) + V (x)φ(x) = Eφ(x) (II.4)
2m ∂x2
Step 3:
Outside the box the solution is trivial. It is zero.
Inside the box the equation is given by
∂2
φ(x) = −k 2 φ(x) (II.5)
∂x2
Where k is given by
2mE
k2 = (II.6)
~2
The solution is given by
φ(x) = A sin(kx) + B cos(kx) (II.7)
Step 4:
Check the boundary conditions
φ(0) = 0 = B (II.8)
φ(a) = 0 = Asin(ka) (II.9)
(II.10)
Hence we have the trivial solution (no particle) unless
ka = nπ ( n is an integer) (II.11)
r
2mE
a = nπ (II.12)
~2
 nπ 2 ~2
⇒ En = (II.13)
a 2m
Hence E developed an index and energy quantization came out. Note that k also has an index:
nπ
kn = The wave-number is also quantized (II.14)
a
 nπx 
φn (x) = A sin (II.15)
a
Note the energy levels rise rapidly they go as n2 . Note we were dealing with a homogeneous differential equation and
hence isn’t fixed yet. This can be done using the normalization condition:
(II.16)
Z ∞ Z a  nπx 
|φn |2 dx = A2 sin2 dx (II.17)
−∞ 0 a
a
A2 =1 (II.18)
2
Hence
r
2
A= (II.19)
a
Notice that the normalization constant is independent of the particular quantum number n.
Therefore
r
2  nπx  n2 ~2 π 2
φn (x) = sin , En = (II.20)
a a 2ma2
 7

Since the energy increases with decreasing a it means that quantum mechanics opposes this motion. This can be
thought of as a quantum mechanical pressure on the outside of the box.
Note that n can equal any integer
n = 1, 2, ... ( Infinity many bound states) (II.21)
√
Note the dimensions of φ are 1/ L and the dimensions of the P (the probability density) is 1/L. The energy levels

FIG. 1. The Wavefunctions of the Square Well and Their Corresponding Probability Densities

as well as the corresponding wavefunctions are shown in figure 1. Typically there are as many quantum numbers as
dimensions of the problem.
Def 1. In any problem as ~ → 0 we recover classical physics. Equivalently we can recover classical physics as E → ∞.
In our problem as E becomes large we should recover the classical distribution. The classical distribution is given
by
(
1/a 0<x<a
P (x) = (II.22)
0 x ≤ 0, x ≥ a

In the limit of large E, we have a highly oscillating function from 0 to 2/a. For any experiment that tries to measure
how likely it is we will get to any finite region it will be 1/a.
Assignment: Demonstrate by direct substitution that the first 5 eigenfunctions of the 1d well are indeed eigenfunctions
of the potential. Plot the first 5 eigenfunctions as well as the first 5 probability distributions. Due date: Wednesday.

Lecture 4 (September 14th, 2011)

Note: Assignment 2 is up.

III. DIRAC NOTATION

The Dirac notation is a more abstract notation then was used up till now though it is more powerful. There is a
corresponding notation in geometry referred to as coordinate free notation. The idea is a vector has a meaning before
a coordinate system is assigned to it. Similarly in quantum mechanics we have states which are independent of a
particular representation.
A state labeled by a quantum number α is denoted by a “ket”. The symbol for the ket of α
|αi (III.1)
In the spatial representation the state |αi is given by Ψα (x). For example
Z ∞
hα|βi = ψβ∗ (x)ψa (x)dx (III.2)
−∞
 8

Note that this is analogous to

A·B (III.3)

The analogy is even more evident by writing the integral as

N
X
lim ψβ∗ (xi )ψα (xi )∆x (III.4)
∆xi →0
N →∞ i=1

Thus the wavefunctions are infinite dimension vectors. In coordinate free notation equation III.2 is given by

hβ|αi (III.5)

The “thing you dot kets with” is called a bra:

hβ| (III.6)

Bras and kets obey the following rules (a is some constant)

hβ|aαi = a hβ|αi (III.7)

haβ|αi = a∗ hβ|αi (III.8)
∗
hβ|αi = hα|βi (III.9)
hα| + hβ| = hα + β| (III.10)

The state |x0 i is an eigenstate of the position operator x̂.

eigenvalue
z}|{
x̂ |x0 i = x0 |x0 i (III.11)

hx0 |xi = δ(x0 − x) The Dirac Delta Function (III.12)

This makes sense since this expression reads what is the amplitude that a state |x0 i coincides with the state |x|. These
two states are distinctly different unless x = x0 . The overlap of a position with the state α is

hx0 |αi = ψα (x0 ) (III.13)

This expression can be read as what is the amplitude that a state α is at x. Suppose you have a complete set of states
labeled by integers (an orthonormal set):
X
|ni hn| = Iˆ (III.14)
n

To see that this is truly the identity:

X
Iˆ |mi = |ni hn|mi (III.15)
n
X
= |ni δnm (III.16)
n
= |mi (III.17)

Lecture 5: September 16, 2011

Notes for the assignment:
Z
hxi = P (x, t)xdx (III.18)

P (x, t) = Ψ∗ (x, t)Ψ(x, t) (III.19)

 9

1
Ψ(x, t) = √ φ1 (x)e−iω1 t + φ2 (x)e−iω2 t


(III.20)
2

~2 π 2
E1 = ~ω1 = (III.21)
2ma2
Continuing with the lecture

hx|αi = ψα (x) (III.22)

hβ|xi = ψβ (x)∗ (III.23)
|αi = ket, ”vector” (III.24)
hβ| = bra, ”dual vector” (III.25)

Dual vectors are the objects that come from the scalar product with kets.
Z
hα|βi = ψα∗ (x)ψβ (x)dx (III.26)

Aside (Cultural):The place where these are completely different objects are in differential geometry (general relativity).
The objects dual to vectors are called one-forms. Maxwell’s equals can be written in differential geometry language
as:

dF = 0 (III.27)
d∗F =J (III.28)

IV. HILBERT SPACE

Recall Cartesian 3-space. We have vectors:

v = (vx , vy , vz ) (IV.1)
v = êx vx + êy vy + êz vz (IV.2)

We say that (êx , êy , êz ) form a basis that spans the vector space. In other words we can build any vector in the vector
space by a linear combination of the basis vectors.
Inner Product:

v · u = vx ux + vy uy + vz uz (IV.3)

Length:
√
v · u = |v| (IV.4)

1. The Hilbert space(H) is analogous to this space.

The space is linear
For a constant ”a” aφ is also in the Hilbert space
If ψ and φ are in H then so is ψ + φ.
2. There is an inner product
Z
hα|βi = ψα∗ (x)ψβ (x)dx (IV.5)

3. Any element of H has a ”length”

hα|αi = ||α||2 (IV.6)

4. H is complete: Every Cauchy sequence of functions (or states) in H converges to an element in H.

i.e. H contains the limit points)
To understand this definition we need to define the Cauchy sequence:
 10

Def 2. Cauchy Sequence: A sequence that as we go on in the sequence the difference gets smaller and smaller
between neighboring number.

Mathematically, given the sequence {xi } any  for some i:

|xi+1 − xi | ≤  (IV.7)
A Cauchy sequence can converge to a number that is not rational (the limit point can be outside the set). We
say the irrationals complete the rationals.

Lecture 6: September 19, 2011

A quick summary of the different notations is given bellow
3-D Representation Bro-ket
R ∗
v·v =0 ψα (x)ψβ (x)dx = 0 hα|βi = 0
(eˆ1 , eˆ2 , eˆ3 ) (ψ1 , ψ2 , ..., ) (|1i , |2i , |3i , ...)
P P P
v = i ai eˆi φ(x) = n an φn (x) |αi = n an |ni
R ∗
eˆi · eˆj = δi,j φ (x)φm (x)dx = δn,m hn|mi = δn,m
R n∗
eˆn · v = an φn (x)ψα (x)dx = an hn|αi = an

X
|αi = am |mi (IV.8)
m
X
hn|αi = hn |am | mi (IV.9)
m
X
= am hn|mi (IV.10)
m
X
= am δn,m (IV.11)
m
= an (IV.12)
Word of warning: Sometimes we choose basis sets in H that are labeled by a continuous valued label. An example of
this is the wave number: |ki. This can be used as:
hx|ki = eikx (IV.13)
 Z  Z ∞
k| |x| hx| k dx = (1)dx = ∞ (IV.14)
x −∞

The reason the overlap is infinity is because the plane waves are non-normalizable.

V. HERMITIAN OPERATORS
D E
Consider the operator  such that  |αi is also in H and β|Â|α . If there is another operator designated by †
such that
D E D E
† β|α = β|Âα (V.1)

Then we say that † is the Hermitian adjoint of  (It does not mean that  is hermitian).
Simplest possible operator  = a (where a is some number)

†
a β|α = hβ|aαi = a hβ|αi = ha∗ β|αi

(V.2)

Hence a† = a∗ . i.e. the hermitian adjoint of a complex number is its complex conjugate.
∂
Consider the operator D̂ = ∂x .
D E Z ∞ ∂
β|D̂α = dxψβ∗ (x) ψα (x) (V.3)
−∞ ∂x
 11

By integration by parts:
∞ 0 Z ∞  
∗
 :
 ∂ ∗
−

= ψβ(x)ψ β (x)
  dx ψ (x) ψα (x) (V.4)


−∞ −∞ ∂x β

The surface terms cancel due to normalization condition. Hence

D̂† = −D̂ (V.5)

In the special case where

† =  (V.6)

we say the operators are Hermitian.

∂ ∂ † ∂


Lecture 7 - Sept 21, 20111 Recall ∂x is not Hermitian, but ∂x = − ∂x .
Consider

 |ni = an |ni (V.7)

Then
D   E
n Â n = hn |an | ni (V.8)
 

= an hn|ni (V.9)
(V.10)

But if  is Hermitian then

D E
Ân|n = han n|ni (V.11)
= a∗n hn|ni (V.12)

Hence an = a∗n (an is real). Consider now

D   E
m Â n = hm |an | ni (V.13)
 

= an hm|ni (V.14)

But
D   E D E
m Â n = † m|n (V.15)
 
D E
= Âm|n (V.16)
= ham m|ni (V.17)
= a∗m hm|ni = am hm|ni (V.18)

Therefore

(an − am ) hm|ni = 0 (V.19)

If an 6= am then

hm|ni (V.20)

Therefore eigenstate of Hermitian operators are orthogonal.

 12

VI. FOURIER SERIES

Suppose f (x) is periodic with period 2. i.e. For some a and all n:
f (x + na) = f (x) (VI.1)
Then we can write
( P∞
a0
2 + n=1 an cos(nx) for odd f
f (x) = P∞ (VI.2)
n=1 bn sin(nπx) for odd f
Where for all n:
Z 2 Z 2
an = f (t) cos(nπt)dt bn = f (t) sin(nπt)dt (VI.3)
0 0
Exercise: Prove that any function can be written as a Fourier expansion
For any interval -L to L for a general periodic function:
∞
a0 X   nπx   nπx 
f (x) = + an cos + bn sin (VI.4)
2 n=1
L L
Where
Z L 

1 nπt
an = f (t) cos dt (VI.5)
L −L L
1 L
Z  
nπt
bn = f (t) sin dt (VI.6)
L −L L
For an example consider the square wave:
(
0 −π <x<0
f (x) = (VI.7)
1 0<x<π

a0 = 1 (VI.8)
1 π
Z
an = cos (nt) dt = 0 It is an odd function shifted upwards (VI.9)
π 0
1 π
Z
bn = sin (nt) dt (VI.10)
π 0
 
1 1 cos nπ
= − (VI.11)
π n n
1
= (1 − (−1)n ) (VI.12)
nπ
Therefore
(
2
nπ odd n
bn = (VI.13)
0 even n
 
1 2 sin 3x
f (x) = + sin x + + ... (VI.14)
2 π 3
The orthogonality statement
Z 2π
sin (mx) sin (nx) = πδnm Unless m = n = 0 (VI.15)
0
Z 2π
cos (mx) cos (nx) = πδnm (VI.16)
0
Z 2π
sin (mx) cos (nx) = πδnm = 0 Unless m = n = 2π (VI.17)
0
 13

VII. FOURIER TRANSFORM

Lecture 8 - September 23rd, 2011

The Fourier series of a function f is given by
∞
a0 X  nπx   nπx 
f (x) = + an cos + bn sin (VII.1)
2 n=1
L L

Where
Z L  
1 nπt
an = f (t) cosdt (VII.2)
L
−L l
1 L
Z  
nπt
bn = f (t) sin dt (VII.3)
L −L L
Explicitly:
 πnx  Z L   ∞  nπx  Z ∞  
1 1 nπt 1X nπt
f (x) = f (t)dt + cos f (t) cos dt + sin f (t) sin dt (VII.4)
2L L L −L L L n= L −L L
Z L ∞ Z
1 1X L  nπ 
= f (t)dt + f (t) cos (t − x) dt (VII.5)
2L −L L n= −L L

Now let L → ∞ (non-periodic function). Set

nπ
=ω (VII.6)
L
π
This is equivalent to making the spacing between neighboring ω’s (∆ω = L ) goes to 0. i.e. ω is continuous. In this
limit:
∞ Z ∞
1X
f (x) = ∆ω f (t) cos (ω (t − x)) dt (VII.7)
π n=1 −∞
P∞
Notice the first integral went to zero due to the L1 dependence and no infinite summation over n. π1 n=1 ∆ω is just
the definition of an integral. Thus
1 ∞
Z Z ∞
f (x) = dω f (t) cos (ω (t − x)) dt (VII.8)
π 0 −∞

Notice that in the integrand we have an even function of ω. Therefore we can divide by 2 and change the ω integral
to:
Z ∞ Z ∞
1
f (x) = dω f (t) cos (ω (t − x)) dt (VII.9)
2π −∞ −∞

Now notice that the limits are symmetric so we can add an odd function of ω in the integrand (it will just go to zero
anyways)
Z ∞ Z ∞
1
f (x) = dω f (t) cos (ω (t − x)) + if (t) sin (ω (t − x)) dt (VII.10)
2π −∞ −∞
Z ∞ Z ∞
1
= dω f (t)e−iω(t−x) dt (VII.11)
2π −∞ −∞
F −1 (x) F (ω)
zZ }| {Z
z }| {
∞ ∞
1
= eiωx dω f (t)e−iωt dt (VII.12)
2π −∞ −∞

Each of these F are transformations on the function f (t). Whatever the first one does, the second one undoes. F is
called the Fourier transform of f (t). F −1 (x) is the inverse Fourier transform of g(ω).
F F −1
f (t) −→ g(ω) f (t) ←−−− g(ω) (VII.13)
g(ω) is analogous to an .
 14

Lecture 9 - Sept 30th, 2011

Make up lecture Tuesday 2:30 pm.
Recall:
Z ∞  Z ∞ 
1
f (x) = f (t) ei(t−x)ω dω dt (VII.14)
−∞ 2π −∞

Definitions:
Z ∞
1
g(ω) = √ f (t)e−iωt dt (VII.15)
2π −∞

We say that g(ω) is the Fourier transform of the function f (t). The function
Z ∞
1
f (t) = √ g(ω)eiωt dω (VII.16)
2π −∞

is called the inverse Fourier transform.

A. Example: Finite Wave Train

(
sinω0 t |t| < Nω0π
f (t) = (VII.17)
0 |t| Nω0π
 
Nπ
The reason it is ω0 is because sin ω0 t → sin ωNω0
π
= 0. Note that this is an odd function. Remember that

eiωt = cos ωt + i sin ωt (VII.18)

Since f (t) has a particular parity (i.e. is either odd or even) then only one cos or sin survive. The Fourier transform
can be split into two integrals:
r Z ∞
2
gs (ω) = f (t) sin ωtdt Fourier-Sin Transform (VII.19)
π 0
r Z ∞
2
gc (ω) = f (t) cos ωtdt Fourier-Cos Transform (VII.20)
π 0
Since we have an odd function we only have to do the sin integral:
r Z N π/ω
2 0

gs (ω) = sin ω0 t sin ωtdt (VII.21)

π
r Z0 N π/ω
2 0
1
= (cos (ω0 − ω) t − cos (ω + ω0 )) dt (VII.22)
π 0 2
 
1 sin ((ω0 − ω) N π/ω0 ) sin ((ω0 + ω) N π/ω0 )
=√ − (VII.23)
2π ω0 − ω ω0 + ω
Consider the case in which ω  1. Further lets focus on the region where ω ∼ ω0 . This allows us to concentrate on
the first term.
1 sin (ω0 − ω) N π/ω0
gs ω ∼ √ (VII.24)
2π ω0 − ω
Taking the limit as ω0 → ω
1 Nπ
lim gs (ω) = √ (VII.25)
x→0 2π ω0
 15

At what values does gs (ω) vanish?

 
Nπ
sin (ω0 − ω) =0 (VII.26)
ω0
Nπ
⇒ (ω0 − ω) = nπ (VII.27)
ω0
nω0
⇒ω0 − ω = = ∆ω (VII.28)
N
Thus the distance between nodes (node spacing) is ω0 /N . The width of the main peak is given by 2ω0 /N . The

Lecture 10 - Oct 3rd, 2011

Test 1: Nov 2nd
Test 2: Nov. 21st

B. Generalize Fourier Transforms to Three Dimensions

Recall the 1d definition

Z ∞
1
g(k) = √ f (x)e−ikx dx (VII.29)
2π −∞

The three dimensional analog is

 3 Z ∞ Z ∞ Z ∞
1
g(k) = √ f (x, y, z)e−ik·x dxdydz (VII.30)
2π −∞ −∞ −∞

VIII. MOMENTUM REPRESENTATION

hx|αi = ψα (x) (VIII.1)

This is equal to probability amplitude to find a particle at position x for the state |αi. Analogously,

hp|αi = φα (p) (VIII.2)

is the probability amplitude to find a particle with momentum p for the state |αi. The functions ψα (x) and φα (p)
are Fourier transforms of one another
FT
ψα (x) −−→ φα (p) (VIII.3)

The probability density in momentum (p) space is given by

Pα (p) = φ∗α φα (p) (VIII.4)

The dimensionless quantity that represents the probability of finding the particle with momentum between p and
p + dp is:

Pα (p)dp (VIII.5)

In three dimensions:
Z Z Z
1
φα (p) = 3/2
ψα (r)ei(r·p)/~ dr (VIII.6)
(2π~)
 16

A. Example 1: Hydrogen in the 1s state

 1/2
1
ψ1s (r) = e−r/a0 (VIII.7)
πa30
2
~
Where a0 = me 2 is the Bohr radius. Notice that the wave function is spherically symmetrical. Note if we want to

find the probability of finding the electron between radius r and r + dr is

∗
ψ1s (r)ψ1s (r)r2 dr (VIII.8)
The probability density is the greatest at the origin. However the radius of a spherical shell around the origin is so
small that the probability to find an electron at that radius is zero. The most likely radius to find the electron is a0 .
However the most likely place to find the electron is the origin since that is where there are the highest density.
The Fourier transform of this functions turns out to be (not proven)
3/2
23/2 a0 ~5/2
φ1s (p) = (VIII.9)
π (a2o p2 + ~2 )2
Notice the Fourier transforms are really different functions. The expectation value of position is given by
Z ∞
hxi = ψ ∗ (x)xψ(x)dx (VIII.10)
−∞

There are two ways to find the expectation value of momentum:

Z ∞
~ ∂
hpi = ψ ∗ (x) ψ(x)dx (VIII.11)
−∞ i ∂x
Z ∞
= φ∗ (p)pφ(p)dp (VIII.12)
−∞

B. Example 2:Harmonic Oscillator

The representations of the harmonic oscillator (ground?) state in position and momentum space are given by
√
mk/(2~))x2
ψ(x) ∝ e−( (VIII.13)
−1
√ p2
φ(p) ∝ e 2~ mk (VIII.14)
Notice the functional forms are the same but one gets narrower as the other gets wider with α.

Lecture 11 - October 5th, 2011

Makelectures delayed to October 25, and November 1st

IX. COMMUTATORS

h i
Â, B̂ = ÂB̂ − B̂ Â (IX.1)
h i
If Â, B̂ = 0 we say the operators commute.
 
~ ∂
[x̂, p̂] = x, (IX.2)
i ∂x
  
~ ∂ ∂ ~ ∂
= x − + x = i~ (IX.3)
i ∂x ∂x i ∂x
 17
h i
If Â, B̂ = 0 then one can find simultaneous eigenfunctions of  and B̂. I.e.

Âψ = λψ (IX.4)
0
B̂ψ = λ ψ (IX.5)
h i
If Â, B̂ 6= 0 then one cannot find simultaneous eigenfunctions.
The expectation value of the distance from the average value is given by
 D E2 
∆2A ≡ Â − Â (IX.6)

This value is referring the uncertainty. To avoid confusion lets expand this definition:
 D E2  D D E E
 −  = Â2 − 2  + Â2 (IX.7)
D E D E2 D E
= Â2 − 2 Â + Â2 (IX.8)
D E D E2
= Â2 − Â (IX.9)

Remember its implied that there is some state α at either end. i.e.
D  D E  D E E
∆2A = α  Â − Â Â − Â  α (IX.10)
 
D D E  D E E
= Â − Â α  Â − Â  α (IX.11)
 

Call this term hf |f i where

 D E
|f i ≡ Â − Â |αi (IX.12)

likewise

∆2B = hg|gi (IX.13)

 D E
Where |gi = B̂ − B̂ |αi. But recall from mathematics the Schwarz inequality says:

2
hf |f i hg|gi ≥ |hf |gi| (IX.14)

Therefore
2
∆2A ∆2β = hf |f i hg|gi ≥ |hf |gi| (IX.15)

remember that in general hf |gi is in general a complex number. Lets call this number z. Then
2
z − z∗

2 2 2 2
|z| = (<z) + (=z) ≥ (=z) = (IX.16)
2i
Thus
 2
1
∆2A ∆2B ≥ [hf |gi − hg|f i] (IX.17)
2i
Consider
D D E  D EE
hf |gi = Â − Â B̂ − B̂ (IX.18)
D D E D E D E D EE
= ÂB̂ − Â B̂ − Â B̂ + Â B̂ (IX.19)
D E D ED E
= ÂB̂ − Â B̂ (IX.20)
 18
D E D ED E
hg|f i = B̂ Â − B̂ Â (IX.21)

D E D E
hf |gi − hg|f i = ÂB̂ − B̂ Â (IX.22)
D E
= ÂB̂ − B̂ Â (IX.23)
Dh iE
= Â, B̂ (IX.24)

Hence
 2
1 Dh iE
∆2A ∆2β ≥ Â, B̂ (IX.25)
2i

Example:

[x̂, p̂] = i~ (IX.26)

For momentum and position then:

 2
i~
∆2x ∆2p ≥ (IX.27)
2i

Equivalently:

~2
∆2x ∆2p ≥ (IX.28)
4
~
∆ x ∆p ≥ (IX.29)
2
When none commuting operators correspond to observables, we can them incompatible observables.

Lecture 12 - Oct. 7th, 2011

X. HARMONIC OSCILLATOR (OPERATOR METHOD)

Recall the classical harmonic oscillator

F = −kx (X.1)
2
d x
m = −kx (X.2)
dt2
⇒ x = A sin ωt + B cos ωt (X.3)
2
kx
V = (X.4)
2
The use of the harmonic oscillator model is that almost any potential can be approximated as a harmonic oscillator.
Now lets work with the quantum analogue. The Hamiltonian is given by

~2 ∂ 2 ψ 1
H=− + mω 2 x2 ψ (X.5)
2m ∂x2 2
The S.E. is given by
1
p2 + (mωx)2 ψ = Eψ


(X.6)
2m
 19

The key idea is to factor the operator p2 + (mωx)2 . Consider if we were factoring two numbers u and v:

u2 + v 2 = (iu + v) (−iu + v) (X.7)

Lets define
1 1
a− ≡ √ (ip + mωx) ; a+ ≡ √ (−ip + mωx) (X.8)
2~mω 2~mω
Now if everything commuted then we would have H = a− a+ . However this is not the case as shown below:

1
a− a+ = (ip + mωx) (−ip + mωx) (X.9)
2~mω
1
p2 + m2 ω 2 x2 + imω (px − xp)


= (X.10)
2~mω
1
p2 + m2 ω 2 x2 − imωi~


= (X.11)
2~mω
1 1
= H+ (X.12)
~ω 2
In the future we require a+ a− so we may as well calculate it here:

1
a+ a− = (−ip + mωx) (ip + mωx) (X.13)
2~mω
1
p2 + m2 ω 2 x2 − imω (px − xp)


= (X.14)
2~mω 
1 1
= H − iω (i~) (X.15)
~ω 2
 
1 1
= H− (X.16)
~ω 2

Rearranging gives
 
1
H = ~ω a+ a− + (X.17)
2

The commutator of a+ , a− is
   
1 1 1 1
[a+ , a− ] = H− − H+ (X.18)
~ω 2 ~ω 2
= −1 (X.19)

Hence

Hψ = Eψ (X.20)
 
1
~ω a+ a− + ψ = Eψ (X.21)
2

Now suppose that we take

 
1
H (a+ ψn ) = a+ a− + a+ ψn (X.22)
2
 
1
= ~ωa+ H + 1 ψn (X.23)
~ω
 
En
= ~ωa + + 1 ψn (X.24)
~ω
= (E + ~ω) a+ ψn (X.25)
 20

Hence the energy of the state acted on or “raised” by a+ is increased by a factor of ~ω.
Now if we apply a− continuously to some state ψ then at one point you will get the state with no particle there at all
(zero)

a − ψo = 0 (X.26)

Hence

1
√ (ip + mωx) ψ0 = 0 (X.27)
2~mω
∂ψ
~ + mωxψ = 0 (X.28)
∂x
∂ψ ω
= −m xψ (X.29)
∂x ~
−mω 2
ψ = e 2~ x (X.30)

XI. CLASSICAL HARMONIC OSCILLATOR

F = ma (XI.1)
−kx = ma (XI.2)

d2 x
m + kx = 0 (XI.3)
dt2
d2 x
+ ωo2 x = 0 (XI.4)
dt2

x(t) = A cos ωo t (XI.5)

1 1
E(t) = mẋ2 (t) + kx2 (t) (XI.6)
2 2

Plugging the solution into the energy gives

1 1
E(t) = mA2 ωo2 sin2 ωo t + kA2 cos2 ωo t (XI.7)
2 2
1 2
= kA (XI.8)
2

Alternatively we can write

1 1 1
mẋ2 + kx2 = kA2 (XI.9)
2 2 2 p
ẋ = ωo A2 − x2 (XI.10)

The probability to be between x and dx is

dt
P (x)dx = (XI.11)
T /2
 21

2π
Where T is the period of oscillation. The period is given by T = ωo . Thus:

dt
P (x)dx = (XI.12)
T /2
2ωo dt
= (XI.13)
2π
2ωo dx
= (XI.14)
2π ẋ
2ωo dx
= √ (XI.15)
2π ωo A2 − x2
dx
= √ (XI.16)
π A2 − x 2
The probability of finding the particle everywhere should be 1. Hence:
Z A
= P (x)dx (XI.17)
−A
Z A
1 dx
= √ =1 (XI.18)
−A π A2 − x2
Let x = Ay, dx = Ady.
Z A
A dy
= p (XI.19)
−A π A − A2 y 2
2
Z A
1 dy
= p (XI.20)
−A π 12 − y 2
=1 (XI.21)

Note that the probability density is infinite at x = A, however its not a problem since the probability is finite (as it
must be since our probability density is normalized)

XII. TIME DEVELOPMENT OF EXPECTATION VALUES

Consider
D E
 (XII.1)

In the x-representation for example we integrate this over all positions x. Hence the expectation value isn’t a function
of x. In other words:
D E D E
d  ∂ Â
= (XII.2)
dt ∂t
D E
∂ Â ∂
Z ∞
= ψ ∗ (x)Âψα (x)dx (XII.3)
∂t ∂t −∞ α
Z ∞
∂  ∗ 
= ψα (x)Âψα dx (XII.4)
−∞ ∂t
Z ∞
∂ψα∗ (x) ∂ Â ∂ψα (x)
= Âψα (x) + ψα∗ (x) ψα (x) + ψα∗ (x)Â dx (XII.5)
−∞ ∂t ∂t ∂t
Now the Schrodinger equation says that:
∂ψ ∂ψ ∗
i~ = Ĥψ; −~ = Hψ ∗ (XII.6)
∂t ∂t
 22

Thus
!
∞ ∞
∂ψ ∗ Âψ
Z Z
i ∗ ∗ h ∗ ∂ Â
dx = Ĥψ Âψ − ψ ÂĤψ + ψ ψ dx (XII.7)
−∞ ∂t −∞ ~ i ∂t

Using representation free notation:

D E *   +!
d  i D D E E h  ∂  
= Ĥα|Âα − α|ÂĤ|α + α α (XII.8)
 
dt ~ i  ∂t 

But the Hamiltonian is Hermitian. Hence

* +
d hAi i Dh iE ∂ Â
= Ĥ, Â + (XII.9)
dt ~ ∂t
∂ Â
In most cases ∂t = 0. In other words we’re mostly interested in operators that don’t have explicit time dependence.
In this case
D E
d  i Dh iE
= Ĥ, Â (XII.10)
dt ~
h i D E
If Ĥ,  = 0, then ddt = 0 . Therefore if  commutes with the Hamiltonian then  is a constant of motion. As
an example consider the free case of V = 0. Then:
h i p̂2
p̂, Ĥ = 0; Ĥ = (XII.11)
2m
Hence momentum is conserved.
Now consider the non free case:
p̂2
Ĥ =
+V (XII.12)
2m
 2 
d hxi i i p̂
= h[H, x]i = + V, x (XII.13)
dt ~ ~ 2m
But V will certainly commute with x since V contains no derivatives but is just a function of x. Hence:
 2 
d hxi i p̂
= ,x (XII.14)
dt ~ 2m
i
 2 
= p̂ , x (XII.15)
2m~
* 0 +
i z }| {
= p̂p̂x − xp̂p̂ −p̂xp̂ + p̂xp̂ (XII.16)
2m~
i
hp̂ [p̂, x] + [p̂, x] p̂i
= (XII.17)
2m~
i
= h−i~p̂ − i~p̂i (XII.18)
2m~
i
= h−i~p̂ − i~p̂i (XII.19)
2m~
hpi
= (XII.20)
m
This is an example of Ehrnfest’s principle which says that classical physics emerges as an average of quantum proba-
bilities. Next consider
h i
Ĥ, p̂ = [V, p] (XII.21)
∂V
= i~ (XII.22)
∂x
 23

but
d hp̂i i Dh iE
= Ĥ, p̂ (XII.23)
dt ~ 
d hp̂i ∂V
=− (XII.24)
dt ∂x
This is Newton’s second law!

XIII. CONSERVATION LAWS

• In classical physics one can show that the invariance of physics under time translations leads to energy conser-
vation.
• In a previous lecture we have derived the time evolution operator:
!
−iĤt
U = exp ; U † = U −1 (XIII.1)
~

Consider the expectation value of a time evolved state α:

hU α |H| U αi = α U † HU  α

  
(XIII.2)

Note that
 2
Ht −iHt 1
U = exp (−iHt/~) = 1 − i + + ... (XIII.3)
~ ~ 2
Hence U commutes with H. Thus

hU α |H| U αi = hα |H| αi (XIII.4)

So the expectation values of Ĥ is constant. In other words

d hEi
=0 (XIII.5)
dt
This conservation of the energy. This can also be shown by the formula derived in the previous section for the
expectation value of an operator.

Consider any old function:

1
f (x + ξ) = f (x) + f 0 (x) (ξ) + f 00 (x) ξ 2 + ...


(XIII.6)
  2!
∂
= exp ξ f (x) (XIII.7)
∂x
Consider the operator we just derived:
   
∂ i ~ ∂
T ≡ exp ξ = exp ξ (XIII.8)
∂x ~ i ∂x
 
i
= exp ξ p̂ (XIII.9)
~
Comparing with the time evolution operator we infer that this is the spatial translation operator. Alternatively by
looking at the definition you can see that when you let the operator act on some function f (x) and it brings you to
the function f (x + ξ).
p̂2
For the free Hamiltonian Ĥ = 2m ,
h i
T, Ĥ = 0 (XIII.10)
 24

Therefore
 
dp̂
=0 (XIII.11)
dt

Thus invariance under spatial transformation gives us momentum conservation.

Now consider rotations:

f (φ) = f (φ + ∆φ) (XIII.12)

 
i ~ ∂
= exp ∆φ f (φ) (XIII.13)
~ i ∂φ
 
i
= exp ∆φLˆz (XIII.14)
~

Define this operator as

 
i
R(∆φ) = exp ∆φLˆz (XIII.15)
~

This operator rotates the function by an amount ∆φ. If the rotation operator commutes with the Hamiltonian then
angular momentum will be conserved. i.e.

d hLz i
[Lz , H] = 0 ⇒ =0 (XIII.16)
dt
Next consider the parity operator (P).

Pf (x) = f (−x) (XIII.17)

If f (x) = f (−x) then we say f (x) even. If f (x) = −f (−x) then we say that f (x) is odd. For even f the eigenvalue of
P is 1 if f is odd then the eigenvalue of P is −1. If

d hPi
[P, H] = 0 ⇒ =0 (XIII.18)
dt
Weak interactions violate parity conservation.

Lecture 15: October 21st, 2011

XIV. TUNNELING

Test 1: November 1
Make up lecture is in room HNE on Nov. 1st, Nov 8th
Consider the potential

x < −a
 V =0
V = −a < x < a V = Vo (XIV.1)

x>a V =0


Consider particles coming from the left

−~2 d2 ψ
+ V ψ = Eψ (XIV.2)
2m dx2
For x < a:
−~2 d2 ψ
= Eψ (XIV.3)
2m dx2
 25

Particles from left Particles from right

~2 k12
z }| { z }| {
ψ= Aeik1 x + Be−ik1 x ; E= (XIV.4)
2m
We keep both terms because particles may reflect back off the barrier.
In the region −a < x < a:
−~2 d2 ψ
+ Vo ψ = Eψ (XIV.5)
2m dx2
Bring the potential to the other side we have the same equation we have before with a new k:
Particles from left Particles from right
z }| { z }| {
ψ= Ceik2 x + De−ik2 x (XIV.6)
2m
Where k2 = ~2 (E − Vo ).
In x > a:
−~2 d2 ψ
ψ = Eψ (XIV.7)
2m dx2

Particles from left Particles from right

~2 k12
z }| { z }| {
ψ= Geik1 x + Ge−ik1 x ; E= (XIV.8)
2m
But here we can’t have any left movers, thus
Particles from left
~2 k12
z }| {
ψ= F eik1 x ; E= (XIV.9)
2m
2 2
In this situation we can choose the energy E and |A| . E corresponds to the energy of the particles and |A| corresponds
to how many particles per second are incoming. B,C,D, and F are output that we get from the boundary conditions.
We require continuity of the wavefunction as well as first derivative in order to have continuity of the second derivative
(for S.E.).
The boundary conditions are (where the points where the S.E. was solved now correspond to the labels, I,II, and III):
ψI (−a) = ψII (−a) (XIV.10)
ψI0 (−a) = ψII0
(−a) (XIV.11)
ψII (a) = ψIII (a) (XIV.12)
0 0
ψII (a) = ψIII (a) (XIV.13)
All we really care to know is the transmission and reflection coefficients so we may as well set A = 1. If A = 1 then
we know that
2 2
1 = |B| + |F | (XIV.14)
The boundary conditions tell us that
e−ik1 a + Beika = Ce−k2 a + Deik2 a (XIV.15)
ik1 e−k1 a − Beik1 a = ik2 Ce−k2 a − Deik2 a



(XIV.16)
ik2 a −ik2 a ik1 a
Ce + De = Fe (XIV.17)
ik2 a −ik2 a ik1 a


ik2 Ce + De = ik1 F e (XIV.18)
These equations can be solved using Mathematica. Switching to dimensionless variables: Define y = E/Vo and
2
η = 8ma
~2
Vo
the transmission coefficient comes out to
1
T = √ (XIV.19)
sinh2 ( 1−y )
4(1−y)y +1

The points where the transmission is one are called transmission resonances. The transmission function in different
regimes is shown in figure 2.
 26

FIG. 2. The transmission function in different regimes T (η, y). The greater the value of η the more classical the transmission
function

Lecture 16 - Oct. 24, 2011

The probability density for different types of transmission are shown in figure 3 Note that on a test we may be required
to sketch these probability densities for different amounts of reflection or transmission.
 27

XV. ANGULAR MOMENTUM

Classically:

L=r×p (XV.1)
 
x̂ ŷ ẑ
= det  x y z  (XV.2)
 
px py pz
= x̂ (ypz − ypy ) + ŷ (zpx − xpz ) + ẑ (xpy − ypx ) (XV.3)
(XV.4)

Quantumly:

h ∂ ~ ∂
Lx = y −z (XV.5)
i ∂z i ∂y

The 3d operator angular momentum operator can be written

~
L= r×∇ (XV.6)
i
In order to find the following commutator makes use the the commutator:

[px , f (x)] = (px f (x)) + f px − f px (XV.7)

= px f (x) (XV.8)

[Lx , Ly ] = Lx Ly − Ly Lx (XV.9)
= (ypz − zpy ) (zpx − xpz ) − (zpx − xpz ) (ypz − zpy ) (XV.10)
~ ~
= ypx − xpy (XV.11)
i i
~
= (−Lz ) (XV.12)
i
= i~Lz (XV.13)

Exercise: Do this . The commutators can be summarized by the following

[Li , Lj ] = i~ijk Lk (XV.14)

Where the symbol ijk is called the Levi-Cevita and is defined as


1
 ijk cyclic
ijk
 ≡ −1 ijk anti-cyclic (XV.15)

0 otherwise


Lecture 17 - October 26th, 2011

L2 = L2x + L2y + L2z (XV.16)

Using this it is straightforward to show that

Lx , L2 = 0
 
(XV.17)
Ly , L2 = 0
 
(XV.18)
Lz , L2 = 0
 
(XV.19)
 28

We can pick a component (typically we pick Lz ) and find simultaneous eigenfunctions of Lz , L2 .

Note on notation: L is used for orbital angular momentum while S is used for spin angular momentum (intrinsic). J
is used for total angular momentum e.g. L + S, L1 + L2 . No matter what the type of angular momentum we are
dealing with is, L, J, or S the commutation relations are identical. e.g.
[Jx , Jy ] = ~Jz (XV.20)
Define the ladder operators as:

J+ = Jx + iJy (XV.21)
†
J− = Jx − iJy = J+ (XV.22)

List of commutation relations:

[Jz , J+ ] = ~J+ (XV.23)
[Jz , J− ] = −~J− (XV.24)
 2 
J , J+ = 0 (XV.25)
 2 
J , J− = 0 (XV.26)
Further we can show that
J 2 = J∓ J± + Jz2 ± ~Jz (XV.27)

[J+ , J− ] = 2~Jz (XV.28)

Let
Jz |mi = m~ |mi (XV.29)
Consider
Jz (J+ φm ) = (~J+ + J+ Jz ) φm (XV.30)
= ~J+ + J+ m~φm (XV.31)
= J+ (m + 1) ~J+ φm (XV.32)
Thus the state J+ φm is an eigenfunction of Jz with the eigenvalue ~ (m + 1). Thus J+ is called a raising or ladder
operator.
J+ φm ∝ φm+1 (XV.33)
J− φm ∝ φm−1 (XV.34)
Now
 2 
J , Jz = 0 (XV.35)
Therefore
J 2 φ m = ~2 K 2 φ m (XV.36)
Where K is some value (not yet known). Consider
J+ J 2 φm = J+ ~2 K 2 φm (XV.37)
2 2
= ~ K J+ φm (XV.38)
2 2
∝ ~ K φm+1 (XV.39)

J+ φm has the same eigenvalue as J 2 .

J 2 = ~K 2 = Jx2 + Jy2 + Jz2





(XV.40)
= Jx2 + Jy2 + ~2 m2



(XV.41)
(XV.42)
 29



Now Jx2 and Jy2 are positive definite quantities. The proof of this is shown below. Here |αi is some state in the
Jz basis and |αi is some state in the i basis, Ji0 is the operator is the diagonalized basis and U is the matrix that
diagonalizes Jx .

2
Ji = hα| Ji2 |αi (XV.43)
= hα| U † U Ji U † U Ji U † U |αi (XV.44)
0
= hα | Ji02 0
|α i (XV.45)
= ji2 (XV.46)

2

2

Where here ji is the eigenvalue of Ji . Thus Jx and Jy are positive definite which means that

~2 K 2 ≥ ~2 m2 (XV.47)
and hence
|K| ≥ |m| (XV.48)
This means that a given sequence of m’s must lie between |K| and − |K|. Therefore
J+ |mmax i = 0 (XV.49)
J− |mmin i = 0 (XV.50)

J 2 |mmax i = Jx2 + Jy2 + Jz2 |mmax i

(XV.51)
 2  2 !
J+ + J− J+ − J− 2
= + + Jz |mmax i (XV.52)
2 2i
1
= (J+ J− + J− J+ + J+ J− + J− J+ ) |mmax i (XV.53)
4
1
J+ J− + J− J+ + 2Jz2 |mmax i



= (XV.54)
2
~2 K 2 |mmax i = ~Jz + Jz2 |mmax i


(XV.55)
Similarly
~2 K 2 = ~2 mmin (mmin − 1) (XV.56)
Solving these equations says that
mmax = −mmin (XV.57)
or
mmin = mmax + 1 (XV.58)
We can throw out the second solution on physical grounds (we can’t have a maximum m be smaller then the minimum
m).
Therefore if mmax = j (define j this way) then mmin = −j and therefore
m = (−j, −j + 1, ..., j − 1, j) (XV.59)

J 2 |mi = ~2 j (j + 1) |mi (XV.60)

XVI. ORBITAL ANGULAR MOMENTUM

Now lets specialize to eigenfunctions of orbital angular momentum.

L2 φlm = ~2 l (l + 1) φlm (XVI.1)
Lz φlm = m~φlm (XVI.2)
 30

We want explicit constructions of φlm . Recall:

 
~ ∂ ∂
Lz = x −y (XVI.3)
i ∂y ∂x

We could proceed in this way but its more natural to move to spherical coordinates. The coordinate system is shown
in figure 3 Exercise: Transform Lz into spherical coordinates. This can be found by finding the transformation from

theta

phi

FIG. 3. The spherical coordinate system

   
x r
y→θ  (XVI.4)
   
z φ

and then taking the inverse of the transformation. This gives the following
   
∂ ∂ sin θ sin φ ∂ cos θ ∂
= (sin θ sin φ) + + (XVI.5)
∂y ∂r r ∂θ r sin θ ∂φ
   
∂ ∂ cos θ cos φ ∂ sin φ ∂
= (sin θ cos φ) + − (XVI.6)
∂x ∂r r ∂θ r sin θ ∂φ

By substitution (ideally blended with astute observations)

 
~ ∂ ∂ r sin θ cos φωφ r sin θ sin φ sin φ ∂
Lz = 0 +0 + + (XVI.7)
i ∂r ∂θ r sin θ r sin θ ∂θ
~ ∂
= (XVI.8)
i ∂θ
The other operators are not as compact
 
∂ ∂
Ly = i~ − cos φ + cot φ sin φ (XVI.9)
∂θ ∂φ
 
∂ ∂
Lx = i~ sin φ + cot θ cos φ (XVI.10)
∂θ ∂φ

L2 = L2x + Jy2 + Jz2 (XVI.11)

1 ∂2
   
2 1 ∂ ∂
= −~ sin θ + (XVI.12)
sin θ ∂θ ∂θ sin2 θ ∂φ2
 31

L2 φlm = ~2 l (l + 1) φlm (XVI.13)

Lz φlm = m~φlm (XVI.14)

The solutions to the differential equation is

φlm (θ, φ) = Ylm (θ, φ) (XVI.15)

The spherical harmonics are properly normalized:

Z
2
|Ylm (θ, φ)| dΩ = 1 (XVI.16)

Lz Ylm (θ, φ) = m~Ylm (θ, φ) (XVI.17)

imφ
⇒ Ylm (θ, φ) ∝ e (XVI.18)

Singlevaluedness of these solutions forces

eimφ = eimφ+im2π (XVI.19)

Hence

e2imπ = 1 (XVI.20)

In other words m ∈ Z. In order to keep the φ component normalized we require that the φ component is
1
√ eimφ (XVI.21)
2π
Now subbing into the spherical harmonics says that

eimφ Θ(θ)
Ylm (θ, φ) = √ (XVI.22)
2π
Plugging this into

L2 Ylm = ~2 l (l + 1) Ylm (XVI.23)

m2
    
1 d dΘ(θ)
⇒ sin θ + l(l + 1) − Θ(θ) = 0 (XVI.24)
sin θ dθ dθ sin2 θ
This is an ugly differential equation. Define

µ = cos θ (XVI.25)

Θ(θ)lm = Plm (µ) (XVI.26)

Where these are the associated Legendre Polynomials.

m/2 dm
Plm (µ) = (−1)m 1 − µ2 Pl (µ) (XVI.27)
dµm
Where Pl (µ) are the Legendre Polynomials defined as

1 dl
l
Pl (µ) = l l
µ2 − 1 (XVI.28)
2 l! dµ
The first few Legendre Polynomials are

Po (µ) = 1 (XVI.29)
P1 (µ) = µ (XVI.30)
1 d2 4 2

3µ2 − 1
P2 (µ) = µ − 2µ + 1 = (XVI.31)
8 dµ2 2
 32

P0

P1

-1 1
P2

The first few Associated Legendre Polynomials are

P00 (µ) = 1 = 1 (XVI.32)
P10 (µ) = µ = cos θ (XVI.33)
p
P11 (µ) = − 1 − µ2 = − sin θ (XVI.34)
Putting this all together we can write the spherical harmonics.
Ylm (θ, φ) (XVI.35)
with
L2 Ylm = ~2 l (l + 1) Ylm ; Lz Ylm = m~ (XVI.36)

 1/2
2l + 1 (l − m)!
Ylm (θ, φ) = Plm (cos θ) eimφ (XVI.37)
4π (l + m)!

The spherical harmonics are orthonormal:

Z
∗
Ylm (θ, φ) Yl0 ,m0 (θ, φ) dΩ = δm,m0 δl,l0 (XVI.38)

Due to the phase we have

∗ m
Yl,−m = (−1) Ylm (XVI.39)
The first few are tabulated below
Y0,0 = √1

4π
3 1/2
Y1,1 = − 21 2π sin θeiφ
1/2
Y1,0 = 12 π3


cos θ
1/2
1 3
sin θe−iφ


Y1,−1 = 2 π
In order to plot these we can plot |Ylm | and set |Ylm | = r. In other words the distance from the origin begin
large implies that the function is large (note this is only for plotting purposes). Its important to note there is no φ
dependence. So the plot is independent of the angle in the x,y plane.

XVII. HYDROGEN

~2 2 1 q2 Z
H=− ∇ − (XVII.1)
2m 4πo r
 33

FIG. 4. Spherical Harmonics

In this class we will use cgs units:

~2 2 1 q2 Z
H=− ∇ − (XVII.2)
2m 4πo r

Note that V = V (r) so its natural to go into spherical coordinates. Thus

FIG. 5. The Hydrogen potential energy

L2
z }| !{
2
1 ∂2
 
2 1 ∂ 2 ∂ 1 1 ∂ ∂
∇ = 2 r + 2 sin θ + (XVII.3)
r ∂r ∂r r sin θ ∂θ ∂θ sin2 θ ∂θ
~2 2 ~2 1 ∂ 2 ∂ L2
− ∇ =− r + (XVII.4)
2m 2m r2 ∂r ∂r 2mr2
p2r L2
= + (XVII.5)
2m 2mr2
 34

Theorem: The ground state of a system has the full symmetry of the problem.
This lovely coincidence that L2 showed up was of course not actually a coincidence.
2
L2 = (r × p) (XVII.6)
2 2 2
= r p − (r · p) (XVII.7)
2 2 2 2
r p = (r · p) + L (XVII.8)
2 2
p (r · p)
= + L2 r2 (XVII.9)
2m r2
2 L
= (r̂ · p) + (XVII.10)
2mr2
2 2
p L
= r + (XVII.11)
2m 2mr2
Thus
p2r L2 Ze2
 
+ − ψ (r, θ, φ) = Eψ (r, θ, φ) (XVII.12)
2m 2mr2 r
 
Using the relation L2 , H = 0 (easy to see by looking at the Hamiltonian term by term), we know that they share
eigenfunctions. Hence
ψ (r, θ, φ) = R(r)Ylm (θ, φ) (XVII.13)
Note it is always the case that if V = V (r) then ψ = R(r)Ylm (θ, φ).

Lecture 18 - November 1st, 2011

Recall (switched to MKS)
~2 1 ∂ L2 Ze2
   
2 ∂
− r + − ψ (r, θ, φ) = Eψ (r, θ, φ) (XVII.14)
2m r2 ∂r ∂r 2mr2 4πo r
Since we have a central potential (i.e. V = V (r)) we can write ψ(r) = R(r)Ylm (θ, φ)
(
Hψ = Eψ
H, L2 = 0 ⇒
 
(XVII.15)
L2 ψ = ~2 l (l + 1) ψ

Plugging in ψ(r) = R(r)Ylm (θ, φ) into our equation.

~2 1 ∂ Ze2
   
2 ∂ l (l + 1)
− r + − R(r) = ER(r) (XVII.16)
2m r2 ∂r ∂r 2mr2 4πo r
Note we have reduced a 3D problem to 1D. Introduce a new function U (r) ≡ rR(r) to simplify the differential
equation. This gives
~2 d2 ~2 l (l + 1) e2 1
 
− + − U (r) = EU (r) (XVII.17)
2m dr2 2mr2 4πo r
The nice thing about this equation is that this looks just like 1D quantum mechanics in the variable r, but the
potential term is an effective potential given by
~2 l (l + 1) e2
Vef f (r) = − (XVII.18)
2mr2 4πo r
Note that when l = 0 The potential is just the Coloumb potential. The effective potential is shown in figure 6 What
is the natural scale for this problem? We can build the scale from the constants in the problem:
 2 γ
α β e
[m] [~] =L (XVII.19)
4πo
2 β
γ
M L3
  
α ML
[M ] =L (XVII.20)
T T2
 35

Angular momentum

Coloumb

Centrifugal Barrier

FIG. 6. The Effective potential of Hydrogen

So we know that
α+β+γ =0 (XVII.21)
−β − 2γ = 0 (XVII.22)
2β + 2γ = 1 (XVII.23)
The solution of this set of equations is
α = −1 (XVII.24)
β=2 (XVII.25)
γ = −1 (XVII.26)
Thus
4πo ~2
[L] = = 0.53 × 10−10 m ≡ ao (XVII.27)
me2
Note that with little work we already know the natural scale for atomic systems (Å).
Its also convenient to introduce
r
−2mE 2
κ≡ ; ρ ≡ κr; ρo = (XVII.28)
~ ao κ
With these changes we get
d2 v
 
ρo l (l + 1)
= 1− + v (XVII.29)
dρ2 ρ ρ2
Consider ρ → ∞.
d2 v
=v (XVII.30)
dρ2
⇒ u(ρ) = Ae−ρ +  ρ
Be (XVII.31)
 36

B must be zero due to normalization requirements. Now consider ρ → 0:

d2 v l (l + 1)
2
= v (XVII.32)
dρ ρ2
−l
u(ρ) = Cρl+1 + 
Dρ (XVII.33)

Where for normalization we require D = 0. Substitute u(ρ) = ρl+1 e−l v(ρ) into the radial equation

d2 v dv
ρ + 2 (l + 1 − ρ) + (lo − 2 (l + 1)) v = 0 (XVII.34)
dρ2 dρ
Frobenius method gives that the result is the associated Laguerre polynomials.
p d
p
Lpq−p (x) ≡ (−1) dx Lq (x)
d q
(e−x xq )


Lq ≡ ex dx
L0 = 1
L1 = 1 − x
L2 = x2 − 4x + 2
L00 = 1
L01 = −x + 1
L02 = x2 − 4x + 2

Reassemble our radial wavefunctions:

1 l+1 −ρ 2l+1
Rnl = ρ e Ln−l−1 (2π) (XVII.35)
r
1 r
Where κ = ao n , ρ = ao n . The energies are

E1
En = (XVII.36)
n2
Where
1 e2 1
E1 = − mc2 α2 ; α≡ ≈ (XVII.37)
2 4πo ~c 137
α is called the fine structure constant. It is a measure of the electromagnetic interaction.

November 7th, 2011

The radial equation is

~2 1 ∂ e2 1
   
2 ∂ 2 l (l
+ 1)
− r +~ − R(r) = ER(r) (XVII.38)
2m r2 ∂r ∂r 2mr2 4πo r

Notice that the only place were we have l dependence is in the middle term. Also notice that there is no m dependence
in the Schrodinger equation. This reflects the fact that our problem is rotationally invariant. m measures the projection
of l into the z axis and our choice of z is arbitrary. We can break this rotational symmetry (by for example adding a
magnetic field) in this case the energy would depend on m. Note there has been a change in notation (at least with
his online notes):
√
4πo ~2 −2mE
ρ ≡ 2κr; ao = = 0.53 Å; κ = (XVII.39)
me2 ~
The new differential equation is

d2 u
 
1 n l(l + 1)
= − + (XVII.40)
dρ2 4 ρ ρ2
 37

1
Where n = ao κ . The result is

u(ρ) = ρl+1 e−ρ/2 u(ρ) (XVII.41)

Using the Frobenius method we find that we don’t have well behaved solutions unless the series Frobenius series
terminates. The ones that are finite are the Laguerre polynomials. They occur when n is an integer. Now we can
find the radial term:
Rnl (ρ) = ρl e−ρ/2 L2l+1
n+l (ρ) (XVII.42)
Notice that the Rnl (ρ) depends on n and l and we call n the principle quantum number.
1 mc2 α2 e2 1
En = − ; α≡ ≈ (XVII.43)
2 n2 4πo ~c 137
Notice the eigenvalues don’t have any angular dependence! Thus two very different eigenfunctions
 3/2  
1 r
R20 = 2− e−r/2ao (XVII.44)
2ao ao
 3/2  
1 1 r
R21 = √ e−r/2ao (XVII.45)
2ao 3 a o

have the same eigenvalues! This mysterious degeneracy in the eigenvalues is due to a hidden symmetry in the problem.

R10

R21

R20

FIG. 7. The Radial terms of the Hydrogen wavefunction

ψnlm (r) = Rnl (r)Ylm (φ, θ) (XVII.46)

Z
ψn∗ 0 l0 m0 ψnlm r2 sin θdrdθdφ = δn0 n δl0 l δm0 m (XVII.47)
 38

The radial probability distribution is

P (r) = r2 Rnl
2
(r) (XVII.48)
For the state 1s state:
P1,0 (r) = r2 R10
2
(r) (XVII.49)
To find the maximum we take the derivative and set it to zero:
dP10
=0 (XVII.50)
dr
2r 2
2re ao − e−2r/ao = 0 (XVII.51)
ao
r2
 
r− e−2r/ao = 0 (XVII.52)
ao
This is zero at r = 0, r = ∞, r = ao . It is a minima at ∞ and 0. It is a maximum at ao .
Z
hri = R10 (r)rR10 (r)r2 dr (XVII.53)
Z ∞
= P10 rdr (XVII.54)
0

Where P10 (r) is the radial probability distribution.

Z ∞
4 2 −2r/ao
hri = r e rdr (XVII.55)
o a3o
Define x = 2r/ao → dx = 2dr/ao
Z ∞
ao 3 −x
hri = x e dx (XVII.56)
o 4
ao
= 3! (XVII.57)
4
3ao
= (XVII.58)
2
Another interesting quantity is
hri = hxi x̂ + hyi ŷ + hzi ŷ (XVII.59)
=0 (XVII.60)
This must be zero since we have reflection symmetry. This is true for all Hydrogen eigenstates (but not for the linear
combinations)

Lecture 19 - November 19th, 2011

XVIII. MATRIX MECHANICS

Consider some basis set

B1 = (φ1 , φ2 , ..., .φn ) (XVIII.1)
We have already encountered some basis sets
r !
2  nπy 
sin (XVIII.2)
a a
 2

An Hn (ξ)e−ξ /2 (XVIII.3)
(Rnl (r)Ylm (φ, θ)) (XVIII.4)
 39

Recall in Dirac notation

X
|αi = |ni hn|αi (XVIII.5)
n

hn|αi = an (XVIII.6)
X
|αi = an |ni (XVIII.7)
n

Knowing what basis B and all the an is equivalent to knowing |αi.

Consider an operator F (arbitrary). Let
F |α0 i = |αi (XVIII.8)
Now lets construct hq|αi. Where |qi is some basis state.
hq|αi = |q |F | α0 i (XVIII.9)
X
= hq |F | ni hn|α0 i (XVIII.10)
n
X
aq = Fqn a0n (XVIII.11)
n

Where we have defined

Fqn ≡ hq |F | ni (XVIII.12)
Z
= φ∗q F φn dr (XVIII.13)

We call Fqn a matrix element.

X
aq = Fqn a0n (XVIII.14)
n

a01
    
a1 F11 F12 ...
a  F F
 2  =  21 22 ... 

 a02 
 (XVIII.15)
.. .. .. .. ..
. . . . .
We also call Fqn the matrix representation of the operator F.
Consider
G |ni = gn |ni (XVIII.16)
This is an eigenvalue equation.
 
g1 0 0
G =  0 g2 0  (XVIII.17)
 
.
0 0 ..
The eigenfunctions are just

0
 . 
 .. 
 
0
 
|ni =  1  (XVIII.18)
 
 
0
 .. 
 
 . 
0
Recall some matrix properties
 40

1. Matrix multiplication:
X
(AB)nq = Anp Bpq (XVIII.19)
p

2. Inverse:

A−1 A = I (XVIII.20)

1 i+j
A−1 = Aadj ; Aadj = ATcof actor ; Acof = (−1) Mij (XVIII.21)
det A

3. Determinant:
X n
Y
det(A) = sgn(σ) Ai,σ(n) (XVIII.22)
σ∈Sn i=1

Where σ is the particular permutation. Sn is the set of permutations. sgn(σ) = +1, −1 depending on if we have
an even or odd permutation.
4. Symmetry: If

AT = A (XVIII.23)

Then A is symmetric. If

AT = −A (XVIII.24)

Then A is antisymmetric.
5. Trace:
X
Tr (A) = Ann (XVIII.25)
n

6. Hermitian Adjoint:

A† = A∗T (XVIII.26)

A†qn = A∗nq (XVIII.27)

For Hermitian operators:

A† = A (XVIII.28)

The proof of Hermitian operators corresponding to real numbers in terms of matrices is straightforward. It also
shows that Hermitian operators correspond to Hermitian matrices. Exercise try this!
7. Unitary:

U † = U −1 (XVIII.29)

We have seen one unitary operator already. The time evolution

e−iHt/~ (XVIII.30)

The importance of unitary operators is that they maintain inner products.

hβ 0 |α0 i = hU β|U αi (XVIII.31)

= β U † U  α

  
(XVIII.32)
= hβαi (XVIII.33)
 41

How operators transform?

F |αi = |βi (XVIII.34)

F U −1 |α0 i = U −1 |β 0 i (XVIII.35)
−1 0 −1 0
UFU |α i = U U |β i (XVIII.36)
F |α0 i = |β 0 i
0
(XVIII.37)

Hence

F 0 = U F U −1 (XVIII.38)

Lecture 22 - November 11, 2011

Recall the harmonic oscillator
~2 ∂ 2 1
H=− 2
+ mωo c2 (XVIII.39)
2m ∂x 2

2
B = e−ξ /2
(Ao Ho (ξ), A1 H1 (ξ), ...) (XVIII.40)
mωo
Where ξ 2 = β 2 x2 , β 2 = ~ . Another way to denote the basis is

B = (|0i , |1i , |2i , ...) (XVIII.41)

The action of the Hamiltonian

 
1
H |ni = n+ ~ωo |ni (XVIII.42)
2

In this basis the Hamiltonian:

1
 
2 ~ωo 0 ...
3
H=
 0 2 ~ωo 0 

(XVIII.43)
.. ..
. 0 .

Since we are in the eigenbasis We see that H is diagonal. Now consider

xmn = hm| x |n| (XVIII.44)

We can evaluate these matrix elements:

Z
xmn = Am Hm (ξ)xAn Hn (ξ)dx (XVIII.45)

However alternatively we can be clever and use ladder operators.

1
a+ = √ (−ip + mωx) (XVIII.46)
2~mω
1
a− = √ (ip + mωx) (XVIII.47)
2~mω

1/2
a+ |ni = (n + 1) |n + 1i (XVIII.48)
 
a− |ni = n1/2 |n − 1i (XVIII.49)
 42

We can write the x operator as

1 1
x= √ (a− + a+ ) (XVIII.50)
2β

What is the a+ matrix?

(a+ )mn = hm| a+ |ni (XVIII.51)

1/2
= hm| (n + 1) |ni (XVIII.52)
1/2
= (n + 1) hm|n + 1i (XVIII.53)
1/2
= (n + 1) δm,n+1 (XVIII.54)

(a− )mn = n1/2 δm,n−1 (XVIII.55)

Now we can construct the x

 
0 1 0 0 ...
√
1 0 2 0 0
 
√ √

1 10 2 0

3 0

x= √ (XVIII.56)

2β  √ ..
 
.. 
0 0 3 . . 
.. .. .. . .
 
..
. . . . .

This is a tridiagonal matrix.

For bonus marks: Find the eigenvectors of this matrix numerically using Mathematica.
What is a+ a−
 
0 0 0 ...
0 1 0 ... 
 
a+ a− =  ..  (XVIII.57)
0
 0 2 . 

.. .. .. ..
. . . .

This is simply the number operator. Thus we can write

 
1
H = ~ωo N + (XVIII.58)
2

Angular momentum:
 2 
L , Lz = 0 (XVIII.59)

L2 |lmi = ~2 l (l + 1) |lmi (XVIII.60)

Lz |lmi = m~ |lmi (XVIII.61)

 
0 0 00 0 0
0 2 00 0 0 
 
0 0 20 0 0
 
2 2

L =~   (XVIII.62)
0 0 00 2 0 

0 0 00 0 6 
 
..
0 0 0 0 0 .
 43
 
0 0 0 00 0
0 1 0 00 0 
 
0 0 0 00 0
 
2

Lz = ~   (XVIII.63)
0 0 0 −1
0 0 

0 0 0 00 2 
 
..
0 0 0 0 0 .

L+ = Lx + iLy (XVIII.64)
L− = Lx − iLy (XVIII.65)

1/2
L± |l, mi = ((l ∓ m) (l ± m + 1)) |l, m ± 1i (XVIII.66)

1/2
hl, m| L± |l0 m0 i = (l0 ∓ m0 ) (l0 ± m0 + 1) ~δll0 δn,m0 ±1 (XVIII.67)

Let’s look at the l = 1 subspace:

 
2 0 0
L2 = ~2  0 2 0  (XVIII.68)
 
0 0 2

 
0 1 0
~ 
Lx = √  1 0 1  (XVIII.69)

2
0 1 0

 
−i 0
~ 
Lx = √  i 0 −i  (XVIII.70)

2
0 i 0

 
1 0 0
Lz = ~  0 0 0  (XVIII.71)
 
0 0 −1

The eigenvectors of Lz are

  
  
1 0 0
|l = 1, m = 1i =  0  ; |l = 1, m = 0i =  1  ; |l = 1, m = −1i =  0  (XVIII.72)
     
0 0 1

Lecture 23 - November 14th, 2011

On the test:

• Tunneling

• Angular momentum

• Hydrogen
 44

The eigenvalues of L2 are 2~2 (the same for all m). To diagonalize Lx is to

Lx v = λv (XVIII.73)
(Lx − λ) v = 0 (XVIII.74)

One solution is the trivial solution v = 0, but this isn’t interesting. To get interesting solutions:

det (Lx − λ) (XVIII.75)

 
−λ ~
√
2
0
 = 0 − λ λ2 − ~2 /2 + λ~2 /2
 

det 

~
√
2
−λ ~
√
2  =0 (XVIII.76)
0 ~
√
2
−λ

This gives eigenvalues are 0, ±~. To get eigenvectors plug λ’s back in. Plugging in λ = ~:
  
−~ ~
√ 0
 ~ 2  v1
 √ −~ √~    v2 =0

(XVIII.77)
 2 2 
0 √~
−~ v3
2
 
−1 √12 0

 1   v1
√ −1 √1   v =0 (XVIII.78)

~
 2 2  2
1 v
0 √
2
−1 3

Solving the systems gives the following eigenvalues:

 
1
1 √ 
λ=~ → 2 (XVIII.79)
2
1
 
1
1 
λ=0→ √  0  (XVIII.80)

2
−1
 
1
1 √ 
λ = −~ →  − 2  (XVIII.81)
2
1

The transformation is a Unitary Matrix (as expected):

√  
1 2 1
1√ √ 
U=  2 0 − 2 (XVIII.82)
2 √
1 − 2 1

In order to diagonalize Lx :
 √     √ 
1 2 1 0 1 0 1 2 1
1√ √  ~  1√ √ 
U Lx U −1 =  2 0 − 2√ 1 0 1  2 0 − 2 (XVIII.83)
2 √ 2 2 √
1 − 2 1 0 1 0 1 − 2 1
 
1 0 0
= ~ 0 0 0  (XVIII.84)
 
0 0 −1
 45

In terms of spherical harmonics what are the eigenvectors? Thus the eigenvector for λ = ~:
 
1 √
1√  1  
 2  → = Y1,1 (φ, θ) + 2Y1,0 (φ, θ) + Y1,−1 (φ, θ) (XVIII.85)
2 2
1
r
1 3
sin θe−iφ + 2 cos θ + sin θeiφ


= (XVIII.86)
4 2π
r
1 3
= (cos θ + −i sin θ sin φ) (XVIII.87)
2 2π

1 3
cos2 θ + sin2 θ sin2 φ sin θ


P (φ, θ) = (XVIII.88)
4 2π

Z
P (φ, θ)dθdφ = 1 (XVIII.89)

Notice that since these aren’t eigenfunctions of z we know longer have rotational
 symmetry of φ! Now we can check
∂ ∂
whether this truly is an eigenfunction of Lx = i~ sin φ ∂θ + cos θ cos φ ∂φ . Notice that finding the eigenfunction of
this operator is very difficult without the matrix formulation.

Lx |1, 1ix = ~ |1, 1ix (XVIII.90)

Note that these are m = 1 eigenfunctions in the x basis.

XIX. SPIN

Spin is the intrinsic angular momentum of a particle. There is no coordinate representation for spin since it doesn’t
exist in real space. We will construct the matrix representation. We still expect the angular momentum commutation
relations to hold. For example

[Sx , Sy ] = i~Sz ; (XIX.1)

or more generally we expect

[Si , Sj ] = i~ijk Sk (XIX.2)

We can define

S+ = Sx + iSy ; S− = Sx − iSy (XIX.3)

We expect S 2 = Sx2 + Sy2 + Sz2 to obey

S 2 |s, ms i = s (s + 1) ~2 |s, ms i (XIX.4)

Sz |s, ms i = ms |s, ms i (XIX.5)
(XIX.6)

Where ms = (−s, −s + 1, ..., s − 1, s). Pions, mesons, etc all carry s = 0. Electrons, protons, neutrons, neutrinos, etc.
all carry s = 12 . The force carriers such as photons, W and Z bosons, gluons, carry s = 1. The graviton (if it exists)
carries s = 2. We will focus on spin half objects since all stable matter is mostly made up of spin half particles.
1 1 13 2 1 1
S2 | , i = ~ | , i (XIX.7)
2 2 22 2 2
1 1 1 1 1
Sz | , i = ~| , i (XIX.8)
2 2 2 2 2
 46

Recall
1/2
L± |l, mi = ~ (l (l + 1) − m (m ± 1)) |l, m ± 1i (XIX.9)

    1/2
1 1 1 1 1 1
S− | , i = +1 − −1 (XIX.10)
2 2 2 2 2 2
1 1
= ~| ,− i (XIX.11)
2 2

1 1
S+ | , i = 0 (XIX.12)
2 2

1 1 3 1 1
S 2 | , − i = ~2 | , − i (XIX.13)
2 2 4 2 2
1 1 1 1 1
Sz | , − i = − ~ | , − i (XIX.14)
2 2 2 2 2
1 1 1 1
S+ | , − i = ~ | , i (XIX.15)
2 2 2 2
1 1
S− | , − i = 0 (XIX.16)
2 2
Lets denote |↑i → |αi, |↓i → |βi.
S+ |αi = 0 (XIX.17)
S− |αi = ~ |βi (XIX.18)
S+ |βi = ~ |αi (XIX.19)
S− |βi = 0 (XIX.20)

S± = Sx ± iSy (XIX.21)

~
Sx |αi = |βi (XIX.22)
2
~
Sy |αi = i |βi (XIX.23)
2
~
Sz |αi = |αi (XIX.24)
2
~
Sx |βi = |αi (XIX.25)
2
~
Sy |βi = −i |αi (XIX.26)
2
−~
Sz |βi = |βi (XIX.27)
2
To construct the Sx matrix:
!
hα|Sx i α hα|Sx i β
Sx = (XIX.28)
hβ|Sx i α hβ|Sx i β
!
~ 0 1
(XIX.29)
2 1 0

!
~ 0 −i
Sy = (XIX.30)
2 i 0
!
~ 1 0
Sz = (XIX.31)
2 0 −1
 47
! !
1 0
|αi → ; |βi → (XIX.32)
0 1

We can write S = ~2 σ. Where

σ = (σx , σy , σz ) (XIX.33)

S 2 = Sx2 + Sy2 + Sz2 (XIX.34)

  !
2 1 1 1 0
=~ +1 (XIX.35)
2 2 0 1
!
23 1 0
=~ (XIX.36)
4 0 1

Hence any two component vector is an eigenstate of S 2 . Note that these spinors are not a vector in the sence of R3
or R2 . Of course its a vector in Hilbert space. The geometrical meaning is not a 2D or 3D geometrical vector but its
a new geometric object called a spinor. Geometrically there are scalars, vectors, and tensors. Spinors fit in between
scalars and vectors. Under rotations spinors do not behave trasnform like vectors but like spinors.
! ! ! !!
~2 0 1 0 −i 0 −i 0 1
Sx Sy + Sy Sx = + (XIX.37)
4 1 0 i 0 i 0 1 0
=0 (XIX.38)

Doing this for all directions:

{Sx , Sy } = 0 (XIX.39)

Where { , } is the anticommutator. This relation implies the Pauli-Exclusion principle.

Spin has many important consequences. The electron is charge. Can think of electron as a spinning ball of charge
(though this is a very incorrect picture). Thus we have a current loop and hence a magnet. The electron has a
magnetic moment. The potential associated with this magnetic moment is

V = −µ · B (XIX.40)

Where
e
µ= S (XIX.41)
mc

Lecture 25- November 16, 2011

There was a small change in the notes:
  √
2 1 1
1 √ √
U= − 2 0 2 (XIX.42)

2 √
1 − 2 1

If you compare this with Ly you can see that

U = I + iLy − L2y (XIX.43)

But

eiθLy = I + i sin θLy + cos θL2y − L2y (XIX.44)

is the rotation operator for rotations. If θ = π2 then this is just the U !. This makes sence since the matrix that
diagonalizes Lx can be interpreted as a rotation about the y axis.
 48

XX. ADDITION OF ANGULAR MOMENTUM

Examples
An electron with both orbital and spin angular momentum:

J=L+S (XX.1)

Two electron system (e.g. Helium):

J = S1 + S2 (XX.2)

We will now consider two electrons:

2
J 2 = (S1 + S2 ) (XX.3)

Where the subscript denotes which electron we are talking about.

[S1 , S2 ] = 0 (XX.4)

Since the operators act on completely different spaces.

J 2 = S12 + S22 + 2S1 · S2 (XX.5)

but
2
(S1 + S2 ) = S12 + S22 + 2S1 · S2 (XX.6)

Which implies that

2S1 · S2 = S+,1 S−,2 + S−,1 S+,2 + 2Sz,1 Sz,2 (XX.7)

Hence

J 2 = S12 + S22 + S+,1 S−,2 + S−,1 S+,2 + 2Sz,1 Sz,2 (XX.8)

! !
0 1 0 0
S+ = ~ ; S− = ~ (XX.9)
0 0 1 0

Hence

S+ |↑i = 0; S− |↑i = ~ |↓i (XX.10)

S+ |↓i = ~ |↑i ; S− |↓i = 0 (XX.11)

J 2 |↑↑i = S12 + S22 + S+,1 S−,2 + S−,1 S+,2 + 2Sz,1 Sz,2 |↑↑i (XX.12)
2
3 2 3 ~
= ~ |↑↑i + ~2 |↑↑i + |↑↑i (XX.13)
4 4 2
= 2~2 |↑↑i (XX.14)
= J (J + J) ; Where J = 1 (XX.15)

J 2 |↑↓i = S12 + S22 + S+,1 S−,2 + S−,1 S+,2 + 2Sz,1 Sz,2 |↑↓i (XX.16)
~2
 
3 2 3 2
= ~ + ~ |↑↓i + 0 + ~2 |↓↑i − |↑↓i (XX.17)
4 4 2
= ~2 (|↑↓i + |↓↑i) (XX.18)

This is not an eigenfunction of J 2 . By symmetry of electrons 1 and 2 we know that

J 2 |↑↓i = ~2 (|↑↓i + |↓↑i) (XX.19)

 49

but adding the two results above:

J 2 (|↑↓i + |↓↑i) = 2~2 (|↑↓i + |↓↑i) (XX.20)

We can also subtract them

J 2 (|↑↓i − |↓↑i) = 0 (XX.21)

J 2 |↓↓i = 2~2 |↓↓i (XX.22)

This gives us the spin singlet and the spin triplet states. Thus electrons in Helium can be in both spin singlet and
spin singlet.
Exercise: take L = 1 and S = 21 and find the eigenfunctions of J = L + S. Also try S1 + S2 + S3 .

In general if we do

J = L1 + L2 (XX.23)

l1
X l2
X
|l1 l2 , jmi = |l1 m1 i |l2 m2 i hl1 m1 , l2 m2 |jmi (XX.24)
m1 =−l1 m2 =−l2

Jz |↑↑i = Sz1 + Sz2

(XX.25)
 
~ ~
= + |↑↑i (XX.26)
2 2
(XX.27)

J 2 |↑↑i = 1(2)~2 |↑↑i (XX.28)

The spin triplet is formed by J = 1:

|↑↑i (XX.29)
1
√ (|↑↓i + |↓↑i) (XX.30)
2
|↓↓i (XX.31)

The spin singlet is formed by J = 0:

1
√ (|↑↓i − |↓↑i) (XX.32)
2

 
1
1
Jz √ (|↑↓i + |↓↑i) = Sz1 + Sz2 √ (|↑↓i + |↓↑i) (XX.33)
2 2
 
1 ~ ~ ~ ~
= √ |↑↓i − |↓↑i − |↑↓i + |↓↑i (XX.34)
2 2 2 2 2
=0 (XX.35)

The spin states are orthonormal:

h↑ | ↑i = 1 (XX.36)
h↓ | ↓i = 1 (XX.37)
h↑ | ↓i = 0 (XX.38)
h↓ | ↑i (XX.39)
 50
!
~ 0 1
Lets find the matrix U that diagonalizes the Sx = 2 matrix.
1 0
!
−λ ~ ~
det 2 =0⇒λ=± (XX.40)
~
2 −λ 2

! !
~ −1 1 1
=0 (XX.41)
2 1 −1 1
Hence the eigenvector is (after normalization)
!
1 1
√ (XX.42)
2 1
The other eigenvector (easy to check) is
!
1 1
√ (XX.43)
2 −1
Thus the transformation matrix is
!
1 1 −1
U† = √ = U −1 (XX.44)
2 1 1
Here we are changing the quantization axis from z to x. So the rotation matrix is (rotation about the y axis)
2 3
iθσy (iθ) σy2 (iθ) σy3
eiθSy /~ = 1 + + + 3 ... (XX.45)
2 4 2 2 6
But the squares the Pauli spin matrices are 1!. Thus
iθσy θ2 1 θ3 σy 1
eiθSy /~ = 1 + − − 3 + ... (XX.46)
2 4 2 2 6
= cos (θ/2) + i sin (θ/2) σy (XX.47)
!
cos (θ/2) sin (θ/2)
= (XX.48)
− sin (θ/2) cos (θ/2)
For θ = π/2:
!
iπSy /2~ 1 1 1
e =√ (XX.49)
2 −1 1
Notice that for θ = 2π
!
−1 0
U (θ = 2π) = (XX.50)
0 −1
Thus when we rotate by 2π we don’t get the identity!
We can split the eigenstates of Sx into eigenstates of Sz :
!
1 1 1
√ = √ (|↑i + |↓i) (XX.51)
2 1 2
One notation for this vector is
1
|→i = √ (|↑i + |↓i) (XX.52)
2

Lecture 26 - November 23rd, 2011

 51

XXI. ENTANGLEMENT

The idea of entanglement first came out in the paper called the “EPR paper” by Einstein, Podolsky, and Rosen
(1935). This paper discussed the philosophy of entanglement but never thought that the different interpretations of
quantum mechanics could really be proved right or wrong. However a paper by Bell showed that they could and
Bell produced Bell’s inequalities. Einstein’s problem with quantum mechanics ws that he thought that “Quantum
mehanics is not complete, it doesn’t describe objective reality”.
The EPR experiment is as follows. Consider a π 0 particle which has spin J = 0. It decays through

π 0 → e+ + e− (XXI.1)

Angular momentum is conserved and therefore the e+ , e− pair has total J = 0. Hence the spin configuration is
1
|0, 0i = √ (↑↓ − ↓↑) (XXI.2)
2
Entanglement is displayed in figure 8

e- e+
z z
50%
x x
z
x
x

FIG. 8. Entanglement of a positron and elecron pair

Lecture 27 -November 25th, 2011

Interesting paper: “Is the moon there when nobody looks? Reality and quantum theory”, Physics Today, April 1985,
by David Mermin.
Consider two operators O1 and O2 . Define the correlator as
hO O i
p 1 2 (XXI.3)
hO12 i hO22 i

If O1 = O2 , then the correlator is equal to 1. The most common use of correlators is

hO(x1 , t1 )O(x1 , t2 )i (XXI.4)

Def 3. Entanglement: We say a state is entangled if it cannot be written a as a simple product of one particle
states.
For example: If we could write the Helium atom as

ψ1 (x)ψ2 (x) (XXI.5)

then we don’t have an entangled state. However

1
√ (|↑↓i − |↓↑i) (XXI.6)
2
is entangled.
Consider a box with a button on it There a few key points
1. Brick stops light from flashing
2. Everytime both random number generators has pointers pointing in the same direction we get perfect anticor-
relation. In order words we have GR or RG if we have 11,22, or 33.
 52

L R
G R G R

1 1

2 3 2 3

3 Switch Positions
2 Lights

FIG. 9. Entanglement figure

3. There are four possible outcomes

4. If one pays attention to the point directions then half the time we have GG or RR and half the time we have
GR and RG
Classical point of view says that particles have genes. In order for the paricles to have perfect anticorrelation we
require the particles 1 and 2 have the following gene table.
Particle one genes Particle two genes
123 123
GGG RRR
GGR RRG
GRG RGR
RGG GRR
RRG GGR
RGR GRG
GRR RGG
RRR GGG
Consider the GGR -RRG row. There are 9 possible pointer settings:
11, 12, 21, 22, 33
13, 23, 31, 32
The top row gives anti-correlation while the bottom row gives the same colours. For this gene the probably is 5/9 for
anti-correlation. The row that we analyzed is the same form as all the rows but the first and law ones. On row 1 and
row 8 we get perfect anti-correlation. Therefore identical results for 6 of the 8 rows (genes) and perfect anti-correlation
for the other two rows.
The classical conclusion is that if we pay no attention to pointer positions we will get anti-correlation with a
probabilty greater then 5/9. This is the simplest way to state Bell’s inequality. The experiment violates Bell’s
inequality. This refutes the EPR claim of the existence of an element of physical reality.
The EPR reality criteria is:“If without in any way disturbing the system we can predict with certainty the value of
a physical quantity then there exists an element of physical reality corresponding to this physical quantity.”
Recall
1
π 0 → χ = √ (|↑↓i − |↓↑i) (XXI.7)
2
 53

Recall that
J 2 χ = 0χ (XXI.8)
We want to know that correlator:
hχ| S1z S2z |χi (XXI.9)
First lets do the simpler correlator
11
S1z S2z |↑↓i = −~2 |↑↓i (XXI.10)
22

~2
S1z S2z |↓↑i = − |↓↑i (XXI.11)
4
Now
~2
S1z S2z χ = − (XXI.12)
4
We can finally compute the normalized correlator:
hχ| S1z S2z |χi
q = −1 (XXI.13)
2 2
hχ| (S1z ) |χi hχ| (S2z ) |χi

This tells you that the spins in this state are perfectly anti-correlated. Now consider the operators in the x direction
hχ| S1x S2x |χi
q (XXI.14)
2 2
hχ| (S1x ) |χi hχ| (S2x ) |χi

To get this correlator we need

~2
S1x S2x |↑↓i = |↓↑i (XXI.15)
4
~2
S1x S2x |↑↓i = |↓↑i (XXI.16)
4
~2
S1x S2x |↓↑i = |↑↓i (XXI.17)
4

~2
S1x S2x χ = − χ (XXI.18)
4
Thus
hχ| S1x S2x |χi
q = −1 (XXI.19)
2 2
hχ| (S1x ) |χi hχ| (S2x ) |χi

The same is true for the S1y S2y we have perfect anti-correlation.
C (S1x S2z ) |↑↓i = (S1x ↑) (S2z ) (XXI.20)
! ! ! !
~ 0 1 1 ~ 1 0 0
= (XXI.21)
2 1 0 0 2 0 −1 1
~2
=− |↓↓i (XXI.22)
4
Similarly
~2
S1x S2z |↓↑i = |↑↑i (XXI.23)
4
 54

~2 1
S1x S2z χ = − √ (↓↓ + ↑↑) (XXI.24)
4 2

|χ| S1x S2z |χi = 0 (XXI.25)

Hence there is no correlation.

Lecture 28 - November 30th, 2011

Recall the EPR experiment:

π 0 → e+ + e− (XXI.26)

The results were

1. If the spin measurements are on the same axis then perfect anti-correlation
2. If you don’t pay attention to which axis then no-corelation
Classical thinking:
If you construct gene tables to satisfy the first condition then the second condition will produce correlation.
Quantum thinking :

1
π 0 → √ (↑↓ − ↓↑) ≡ χ; J = 0, Jz = 0 (XXI.27)
2
The three pointers in the experiment
√
are actually
√
the directions in which we are actually meauring the spin: The
vectors are (nˆ1 = eˆz , nˆ2 = 23 eˆx − 12 eˆz , nˆ3 = − 23 eˆx − 21 eˆz ). Recall the definition of S
~
S= σ (XXI.28)
2
Where σ 1 is the spin matrix corresponding to particle 1. To find the correlator we first find the relations:
−~2
S1z S2z ↑↓= ↑↓ (XXI.29)
4
The correlator of S1 · n̂1 S2 · nˆ1 (without normalization) is

hχ| S1 · n̂1 S2 · nˆ1 |χi = hχ| S1z S2z |χi (XXI.30)

= −1 (XXI.31)

Now lets calculate

√ ! √ !
3 x 1 z 3 x 1 z
hχ| (S1 · n̂2 ) (S2 · n̂2 ) |χi = hχ| S − S1 S − S2 |χi (XXI.32)
2 1 2 2 2 2
√ √
3 1 3 z x 3 x z
= hχ| S1x S2x + S1z S2z − S1 S2 − S S |χi (XXI.33)
4 4 4 4 1 2
Define this operator in the bracket as O.
!
~ 0 1
Sx = (XXI.34)
2 1 0

! !
~ 0 1 1
Sx ↑ = (XXI.35)
2 1 0 0
!
~ 0
(XXI.36)
2 1
 55

Sx ↓ = − ↑ (XXI.37)

√ √ !
~2 3 1 3 3
O ↑↓ = ↓↑ − ↑↓ − ↑↑ + ↓↓ (XXI.38)
4 4 4 4 4

√ √ !
~2 3 1 3 3
O ↓↑ = ↑↓ − ↓↑ + ↓↓ − ↑↑ (XXI.39)
4 4 4 4 4

Finally
1
Oχ = O √ (↑↓ − ↓↑) (XXI.40)
2
~2
=− χ (XXI.41)
4
Thus these are perfectly anti-correlated.
If we don’t pay any attention to the spin then we should get zero for the following: (i 6= j)
X X X
hχ| S1 · n̂i S2 · n̂j |χi = hχ| S1 · n̂i S2 · n̂j |χi (XXI.42)
i,j i j
!  
X X
= hχ| S1 · n̂i S2 ·  n̂j  |χi (XXI.43)
i j

=0 (XXI.44)

Since the vector sum over all the three vectors is trivially zero. To summarize: Bell’s inequality follows from classical
thinking. Bell’s inequality is violated by experiment and quantum mechanics. Bell’s inequality is a statistical state-
ment. In other words you have to do many measurements and sum the results. This is the way things stood until
GHZ came along (Greenburger, Horne, and Zolnger) and they proposed an experiment and carried it out which is
not a statistical experiment (one-shot experiment). In this experiment 100% of the time quantum mechanics predicts
one thing and classical mechanics predicts another. You are responsible for this experiment for the exam.
We have solved most of the exactly solvable problems in quantum mechanics but we have approximation schemes for
the others.

• Time-dependant perturbation theory

• Time-independant perturbation thoery

• Degenerate perturbation theory

• WKB approximation

• Variational method

Lecture 29 - December 1st, 2011

XXII. VARIATIONAL METHOD

This is an approximation technique. If |αi is some trial state to the equation:

Hψ = Eψ (XXII.1)
 56

Then we consider
hα |H| αi (XXII.2)
As an apprimation to the energy. To get the trial state we make an educated guess. There is no way we’ll get the
exact solution. If we minimize
hα| H |αi (XXII.3)
this quantity with respect to a parameter then we assured of getting as closed as possilbe the exact solution. Mathe-
matically:
d hα| H |αi
=0 (XXII.4)
da
For some parameter a is the best possible solution. We call a a variational parameter and we use it to parametrize
our wavefunction. e.g. ψ(x, a). Why does it work? This works because
X
|αi = an |ni (XXII.5)
n

X
hα| H |αi = an a∗n0 hn0 | H |ni (XXII.6)
n,n0

but since this is an exact solution H |ni = En |ni

X X 2
hα |H| αi = a∗n an0 En δn,n0 = |an | En (XXII.7)
n,n0 n
X 2
≥ Eo |an | (XXII.8)
n
≥ Eo (XXII.9)
Mimizing the energy is maximizing ao . As an example consider Hydrogen
~2 1 d2 r 1 L2 Ze2
   
− + − ψ(r, φ, θ) = Eψ(r, φ, θ) (XXII.10)
2m r dr2 2m r2 r
Because the problem has rotational symmetry we know that ψ(r, φ, θ) = R(r)Ylm (φ, θ). Subsituting in we will get
~2 1 d2 r ~2 Ze2
 
H=− + l (l + 1) − (XXII.11)
2m r dr2 2mr2 r
We will discuss l = 0. Thus
~2 1 d2 r Ze2
 
H=− − (XXII.12)
2m r dr2 r

3/2 −r/a
Our trial solution is R(r) = 2 a1 e , where a is the variational parameter (not ao ). This is silly since we know
that R(r) above contains within it the exact solution. Note once we decide we want to try an exponential solution we
are essentially forced to this form by dimensions and normalization (even the 2!). First lets find this (define γ in this
way).
~2 1 d2 r Ze2
   
HR(r) = − − γe−r/a (XXII.13)
2m 4 dr2 r
~2 e2
   
2 1
=γ − − + 2 −Z e−r/a (XXII.14)
2m ar a r

∞
~2 e2
Z    
2 1
ha| H |ai = γ 2 r2 dr − − + 2 −Z e−r/a (XXII.15)
o 2m ar a r
Z ∞  2
e2 ~2 2 −r/a
 
4 ~ −r/a
= 3 −Z re − r e dr (XXII.16)
a 0 ma r 2ma2
 57

but
∞
a3
Z
r2 e−r/a dr = (XXII.17)
0 4

and
∞
a2
Z
re−r/a dr = (XXII.18)
0 4

 2  2
~2 a3

4 ~ 2 a
ha| H |ai = 3 − Ze − (XXII.19)
a ma 4 2ma2 4
 2 2 2

4 ~ a e a
= 3 −Z (XXII.20)
a 8m 1 4
 2
e2

~
= − Z (XXII.21)
2ma2 a
(XXII.22)

We have calculated the energy as a function of our variational parameter a.

~2 e2
E(a) = − Z (XXII.23)
2ma2 a
The Coloumb term is pushing the electron inwards while the Quantum mechanics forms the repulsive term pushing

Quantum Mechanics

Coloumb Term

FIG. 10. The energy contributions from the varitional method

the electron outwards.

~2 Ze2 ~2
E 0 (a) = − 3
+ 2 = 0a = (XXII.24)
ma a 2e2 m
To no suprise E 0 (a) is minimized at the Bohr radius. Plugging this back in you get the exact energy, E(a) = − 12 mc2 α2 .