Sie sind auf Seite 1von 77

IMPORTANT NOTICE

Sections in grey fonts are from last year.


They can give you reasonable guidance about where this
course is headed but they may change before we use them
for the 2010 course.
Notes for Physics 115a (Quantum Mechanics)
Equation Chapter 1 Section 1© 2010 Andreas Albrecht
Course web site: http://www.physics.ucdavis.edu/Cosmology/P115a/

Table of Contents
Notes for Physics 115a (Quantum Mechanics) ................................................................... 1
1. Intro ............................................................................................................................. 3
2. Probabilities ................................................................................................................ 5
2.1. Probabilities in everyday experience.................................................................... 5
3. The Schrödinger equation ........................................................................................... 5
3.1. Properties of the wavefunction............................................................................. 6
3.2. Discrete analogy for continuous variables ........................................................... 6
3.3. The wavefunction as a vector ............................................................................... 7
3.4. Conservation of probability and Unitarity ........................................................... 8
4. Momentum ................................................................................................................ 10
4.1. The momentum operator and the uncertainty relation ....................................... 10
4.2. Momentum and position eigenstates .................................................................. 12
4.3. de Broglie wavelength........................................................................................ 13
4.4. Ehrenfest’s Theorem .......................................................................................... 13
5. Energy Eigenstates (stationary states) ...................................................................... 16
5.1. Changing Basis in Bra-Ket notation................................................................... 16
5.2. Energy Eigenstates ............................................................................................. 20
5.3. Real wavefunctions for Energy Eigenstates ....................................................... 22
6. The infinite square well ............................................................................................ 22
6.1. Stationary states.................................................................................................. 23
6.2. More about boundary conditions........................................................................ 25
6.3. Discussion of HW3 ............................................................................................ 25
6.4. Griffiths Examples 2.2 and 2.3........................................................................... 25
7. The harmonic oscillator ............................................................................................ 27
7.1. Energy lower bound (Griffiths problem 2.2) ..................................................... 27
7.2. The harmonic oscillator potential ....................................................................... 27
7.3. The “ladder operators” ....................................................................................... 28
7.4. Constructing the harmonic oscillator energy eigenstates ................................... 31
7.5. Normalization ..................................................................................................... 32
7.6. Working with ladder operators ........................................................................... 34
7.7. Hermite Polynomials .......................................................................................... 37

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 1


7.8. Discussion of ψ n ( x ) ....................................................................................... 37
7.9. Power Series Method ......................................................................................... 38
8. The Free Particle ....................................................................................................... 38
8.1. Energy Eigenstates ............................................................................................. 38
8.2. Normalization ..................................................................................................... 39
8.3. An example ........................................................................................................ 40
8.4. Group velocity .................................................................................................... 41
9. The delta function potential ...................................................................................... 42
9.1. Bound states and scattering states ...................................................................... 42
9.2. Delta Function Well ........................................................................................... 43
9.3. Matching a wavefunction across infinite changes in the potential..................... 44
9.4. The bound state .................................................................................................. 45
9.5. Scattering states .................................................................................................. 46
9.6. The delta function peak ...................................................................................... 49
10. The finite square well............................................................................................. 50
10.1. Functional form .............................................................................................. 50
10.2. Matching at the boundaries............................................................................. 51
10.3. Scattering states .............................................................................................. 52
11. Review ................................................................................................................... 54
11.1. Review of eigenstates for flat potentials......................................................... 54
12. The energy-time uncertainty principle ................................................................... 55
13. Multiple systems, coherence and measurement ..................................................... 57
13.1. Combining two systems.................................................................................. 57
13.2. Properties of the wavefunction v2 .................................................................. 60
13.3. Coherence and Decoherence .......................................................................... 62
13.3.1. Nomenclature .......................................................................................... 62
13.3.2. The “relativity” of Decoherence ............................................................. 62
13.3.3. Examples ................................................................................................. 63
13.4. Measurement .................................................................................................. 64
13.4.1. General Discussion.................................................................................. 64
13.4.2. Description of quantum measurement..................................................... 64
13.5. Quantum measurements as Hermitian operators. ........................................... 65
13.5.1. The measurement process ....................................................................... 66
13.5.2. Interpretation ........................................................................................... 67
13.5.3. Can we see the many worlds? ................................................................. 72
14. More about measurements and uncertainties ......................................................... 75
14.1. Compatibility of measurements ...................................................................... 75
14.2. Generalized uncertainty principle ................................................................... 76

Start Mar 30 Lecture


Intro
HW 1 assigned on Web
Reading Assigned on Web
Check office hours OK?:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 2


1. Intro

Why I am excited about teaching this course:


• QM 1st really turned me on to Physics
• I still think it is amazing how our predecessors just wanted to
understand how the atom works and they wound up revolutionizing
our understanding of the physical world
• New course for me (not Cosmo/Astro). I will learn from it!
• I’ve thought a lot about fundamental questions in QM
• I’m excited to share this “rite of passage” with you.

Questions for class (what year, connections x2)

What I expect from you:


• Respect this exciting topic
• Get into it / Work hard
• See me for advice if you are having trouble
• USE OFFICE HOURS (for fun and practical help)
• Don’t worry about being a whiz! Acquire a strong work ethic

The weird stuff:


• Probabilities
• Interference/Uncertainty principle
• Measurement apparatus as link between classical and quantum
worlds
• What is reality?
• Many Worlds
• Fear Factor/Sociology/Historical Baggage (Griffiths section 1.2)
• QM is less weird than it first looks (and less weird than Griffiths
claims). (My position, not universally held even among the very
best physicists.)
• Almost all awkward questions can be addressed “putting the
measurement apparatus into the Schrödinger equation”. (See my
“following a collapsing wavefunction” paper linked on the course
web page)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 3


• PUNCH LINE: Learn the technical side well. That will put you in
the best position to understand what is weird and what is not. Thus,
technical knowledge is basically the sole emphasis of this course.
Come to office hours to talk about the weird stuff (plus brief
comments during lectures and a lecture or two at the end if
possible). NB Also, the technical side is universally agreed
among all physicists.

 Review General Info handout (linked on web)

 Measurement illustration (to introduce next section)

 General comment about probability in everyday life

 Classical equation for particle of mass m in 1 dimension


dV
F = mx = − (1.1)
dx

 Schrodinger equation for a particle of mass m moving in a potential V


(1d):

∂Ψ 2 ∂ 2Ψ
i =− +V Ψ
∂t 2m ∂x 2 (1.2)
• Linear equation (only one power of Ψ )
• Ψ ( x, t ) is the “wavefunction” (complex)
• i= −1

•  1.054572 × 10-34 m 2 kg / s =J ⋅ s is “Planck’s


constant” (technically h = 2π  is “Planck’s Constant” and  is the
“Reduced Planck’s Constant”, but few quibble about the words. The
letter you write says exactly what you mean).
• The probability of finding the particle between positions x1 and x2 is

P ([ x1 , x2 ] , t ) =
∫ ( x, t ) Ψ ( x, t )dx
x2
Ψ *
(1.3)
x1
(the “Born rule”).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 4


• Ψ ( x, t ) can be negative, so adding wavefunctions can *reduce*
probability in a given region. (NB, due to linearity, adding two
solutions to Eqn. (1.2) gives another solution.) Give an example at the
board.

• NB I replace Griffiths section 1.2 with Eqn (1.3) (Plus technical


discussion of how we match quantum states with our perceptions of
the world, including what constitutes a measurement.)

• NB: Don’t ask how we can talk about QM in a classical way (as in
Griffiths section 1.2), ask how classical world emerges from a
fundamentally quantum picture.

• One real number total (CM) vs one complex number at every point in
space (QM)

• Adding wavefunctions can *reduce* probability in a given location.


(NB, due to linearity, adding two solutions to Eqn. (1.2) gives another
solution.)

Start April 1 Lecture


Comment on HW1: Tricky integral
Questions?

2. Probabilities
Equation Section (Next)

2.1. Probabilities in everyday experience


Discuss section 1.3 from text (plus additional related material not in section
1.3) using Slides
http://www.physics.ucdavis.edu/Cosmology/P115a/Notes/S1_Prob.ppt

Continuous parameters: Finish Slides, Discuss probability density at the


board. Discuss the units of a wave function when there are varying numbers
of continuous coordinates.

3. The Schrödinger equation


Equation Section (Next)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 5


3.1. Properties of the wavefunction
In class problem (to be done in small groups):
Consider the following particle wavefunctions
 A −1 < x < 1
Ψ1 ( x, 0 ) =
 (3.1)
0 x ≥1
and
 B −1 < x ≤ 0

Ψ 2 ( x, 0 ) =  − B 0 < x < 1 (3.2)
0 x ≥1

a) Find the values of A and B that normalize the wavefunctions so that
the total probability (of finding the particle anywhere) is unity in each
case.
b) Plot the probability density (vs x ) for each wavefunction.

Start April 6 Lecture


Collect HW1:
HW 2 assigned on web
New reading assigned on web
Continue discuss at board.

Now consider a new wavefunction


Ψ 3 ≡ Ψ1 + Ψ 2 (3.3)
c) Is Ψ 3 normalized?
d) Plot the probability density (vs x ) for Ψ 3

Discuss how adding wavefunctions can be very different from adding


probabilities.

3.2. Discrete analogy for continuous variables


This section draws on appendix A

Let us consider the “discrete analogy” of the wavefunction in the continuous


variable x . This was discussed at the end of the previous lecture but not
included in these written notes:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 6



∫ ψ *ψ dx ↔∑ψ * ( xi )ψ ( xi ) ∆x (3.4)
−∞
i

where the sum runs over some discrete approximation to the continuous
variable x . For all we know, real space might be discrete at some very small
length scale, so there should be no loss of generality by using Eqn. (3.4) for
sufficiently fine grids.

One can absorb the ∆x into the “wavefunction” by defining


ψˆ i ≡ ψ ( xi ) ∆x (3.5)
which allows one to write

∑ψ ( x )ψ ( x ) ∆x =∑ψˆ ψˆ
i
*
i i
i
*
i i (3.6)

3.3. The wavefunction as a vector


One can think of ψˆ as a vector in a vector space. We write such vectors as
ψ̂ or ψˆ . The vector space of interest for quantum mechanics has the inner
product (or “dot product”) written as ψˆ ψˆ . The quantities ψˆ i are the
components of the vector in the basis { i } which correspond to the discrete
positions { xi } . The inner product expressed in this basis is given by

ψˆ ψˆ ≡ ∑ψˆ i*ψˆ i ↔ ∫ ψ * ( x )ψ ( x ) dx (3.7)
−∞
i

This implies that ψˆ is the transpose and complex conjugate of ψˆ . (This


is called the “Hermitian conjugate”). The transpose comes about because
you need to multiply a row vector times a column vector to get the sum of
components given in Eqn.(3.7). By convention ψˆ is a column vector. The
components ψˆ i are given by
i ψˆ (3.8)
and the probability that the particle can be found at position xi is given by

ψˆ i*ψˆ i ↔ ψ * ( xi )ψ ( xi ) dx
2
ψˆ i i ψˆ ≡ i ψˆ = (3.9)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 7


3.4. Conservation of probability and Unitarity
We want time evolution not to change the fact that the total probability is
unity:
ψˆ ψˆ
= ∑
=ψˆ ψˆ
i
*
i i 1 (3.10)
This is sometimes called “conservation of probability. That means we want
ψˆ ( t =
ψˆ ( t ) ψˆ ( t ) = 0 ) ≡ ψˆ ( 0 ) ψˆ ( 0 ) =
0 ) ψˆ ( t = 1 (3.11)
One of the principles of quantum mechanics is that the time evolution from
an initial “state vector” ψˆ ( 0 ) to the state vector at a later time ψ ( t ) is
given by a linear operator 1:
ψˆ ( t ) = T ( t ) ψˆ ( 0 ) (3.12)
Taking the Hermitian conjugate:
ψˆ ( t ) = ψˆ ( 0 ) T† ( t ) (3.13)
where the “ † ” denotes the Hermitian conjugate (complex conjugate and
transpose) of the operator. Equation (3.11) becomes
ψˆ ( 0 )=
T†T ψˆ ( 0 ) ψ
=ˆ ( 0 ) ψˆ ( 0 ) 1 (3.14)
or
T† T = 1 (3.15)
Eqn. (3.15) is the statement that “ T is unitary”.
Compare with the scalar case: T
*
T = 1. This is true in general if
T = eiθ (3.16)
for any real value of θ . Generalizing to the matrix case, what about:

1
In general one can talk about T ( t2 , t1 ) which evolves the system from t1 to t2
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 8
 eiθ1 0 0  0
 
0 eiθ2   
T=
    0  (3.17)
 
iθ N 
 0 0 0 e 
This certainly obeys Eqn. (3.16), but one need not look at T in the basis
where it is diagonal. The full generalization of Eqn. (3.16) is

T (t ) = e
iΘ ( t )
(3.18)

Where Θ† = Θ ( Θ is Hermitian) so that when you do diagonalize T the θi ’s


are real.

Now the Schrödinger equation is about differential amounts of time


evolution. To talk about those we can perform a Taylor expansion for small
values of t given by ∆t :
T (=
∆Θt ) e
iΘ ∆t 
 
≈ 1+ i 1∆t + O ( ∆t )
2 (3.19)
where Θ1 must also be Hermitian. This implies that
∂T
= iΘ1
∂t (3.20)
Start April 8 Lecture

or

ψˆ = iΘ1 ψˆ (3.21)
∂t
Note that Θ1 has units of inverse time. By convention we take the i to the
other side giving

−i ψˆ =
Θ1 ψˆ (3.22)
∂t

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 9


It turns out that for quantum mechanical systems with classical analogues
Θ1 is nothing other than the quantum equivalent of the classical energy or
Hamiltonian. Of course one has to get the units right and this is
accomplished by constructing
HΘ≡ − 1 (3.23)
(which also establishes the sign convention). This leads to the standard form
of the Schrodinger equation

i ψˆ = H ψˆ (3.24)
∂t
As long as H is Hermitian time evolution according to Eqn. (3.24) will
preserve the normalization of ψˆ . For the standard 1 dimensional
Schrödinger equation given by Eqn. (1.2) the Hamiltonian is
2 ∂ 2
H=
− +V (3.25)
2m ∂x 2
This operator can be shown to be Hermitian once we understand how to

 ∂2 
interpret  2  , so this is a special case of Eqn. (3.24) (and thus it is not
 ∂x 
surprising that Griffiths verifies conservation of probability in section 1.4).
We have derived a more general result in this section, namely that whenever
H is Hermitian time evolution according to Eqn. (3.24) will preserve
normalization (or “conserve probability).

In the (very common) case where H is time independent (required for


conservation of energy) one can integrate Eqn. (3.24) to find
T ( t2 , t1 ) = e − iH(t2 −t1 ) /  (3.26)
(This uses math facts such as Eqn A.91 in Griffiths.)

4. Momentum
Equation Section (Next)

4.1. The momentum operator and the uncertainty relation


The momentum operator is give by

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 10


 ∂
p≡ (4.1)
i ∂x
Some facts about the momentum operator:
• Why? Because it works (fits data)
• xp ≠ px This is usually discussed in terms of the “commutator”:
[ x, p ] ≡ xp − px =
i (4.2)
Proof (operate on a wavefunction)
 ∂ ∂ 
[ ]
x, p ψ =
−i   x ψ − xψ 
 ∂x ∂x 
 ∂ ∂ 
= −i  x ψ −ψ − x ψ 
 ∂x ∂x  (4.3)

= iψ
so

[ x, p ] = i (4.4)
• uncertainty principle. One can show (see eqn 3.62 in text) that for any
operators A and B
2
1 
σ A2σ B2 ≥  [ A, B ]  (4.5)
 2i 
for momentum and position this leads to

σ xσ p ≥   (4.6)
2
•  can’t know both x and p with precision.
 ∂ 

• What is   ? If you know p has real eigenvalues must have


 ∂x 
p† = p (4.7)
so
 ∂   ∂

  = (4.8)
 i ∂x  i ∂x
thus

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 11


 ∂   ∂

−   = (4.9)
i  ∂x  i ∂x
so
 ∂  ∂

−  = (4.10)
 ∂x  ∂x


Can relate Eqn. (4.10) to discrete version of the operator.
∂x
CAREFUL:
∞  ∂ 
p = ∫ ψ* ψ dx
−∞
 i ∂x 
(4.11)
∞  ∂ 
≠ ∫−∞  i ∂x ψ ψ dx
*

Start April 13 Lecture


Comment on HW1: Integrals of even and odd functions
Questions?

4.2. Momentum and position eigenstates


To work on in class: Using
ψ = Aeikx (4.12)
evaluate
pψ (4.13)
p and p 2 . Note, you should just leave
1
A2 ≡ (4.14)
∫ (e ) e

ikx * ikx
dx
−∞
or
1
A2 ≡ (4.15)
∫ (e ) e
L
ikx * ikx
dx
−L
without worrying about how to do the integral.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 12


• What do these results tell us about ψ ? Discuss how ψ is an
eigenstate of p with eigenvalue k .
• How well determined is position in a momentum eigenstate?
• What is a position eigenstate? Need:
xψ a = aψ a (4.16)
satisfied by
ψ
= a δ ( x − a) (4.17)
• How well determined is momentum in a position eigenstate? (note
that the Fourier transform of a delta function is a constant  a delta
function “contains” all possible momenta).

 Comment on HW problem 2.2: Note how much easier it is to do the


algebraic version, vs Griffiths 1.16.

4.3. de Broglie wavelength


Definition:
h 2π 
λ≡ = (4.18)
p p
“the wavelength of the quantum momentum eigenstate corresponding to
classical momentum p”.
2π  6.63 ×10-34 m
=λ = (4.19)
p p
kg m/s
In groups: find an example of an object (with a mass and a speed) which
has a de Broglie wavelength of 1m. (Related to this week’s HW problem
2.4)

NOTE: The non-commutation of x and p is one of the “weird” aspects of


quantum mechanics (in addition working with probabilities and
wavefunctions that add to give non-additive behavior of the probabilities).

4.4. Ehrenfest’s Theorem


Worked out at the start of section 1.5 in Griffiths:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 13


d x
p =m (4.20)
dt
Also here is problem 1.7 from Griffiths worked out algebraically: Prove:
d ∂
p = − V (4.21)
dt ∂x
Proof:
d ∂  ∂ 
= ψ p ψ  ψ p ψ + ψ p ψ  (4.22)
dt  ∂t   ∂t 
(NB use partial derivative when the expression depends on more than one
variable, to indicate differentiation with respect to only one variable) now
use the Schrödinger equation:
∂ H
ψ = ψ (4.23)
∂t i
and its Hermitian conjugate
∂ H
ψ = −ψ (4.24)
∂t i
in Eqn. (4.22) to get
d
dt
ψ pψ =
1
i
{− ψ Hp ψ + ψ pH ψ }
(4.25)
−1
= [ H , p]
i
plugging in the Hamiltonian gives
−1 −1  p 2  
= [ ]
H , p  + V  ,p
i i   2m  
(4.26)
−1  p 2
 −1
= + [V , p ]
i  2m 
, p
i
Since p commutes with itself (commutative property of derivatives) the first
term is zero and we get

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 14


d −1
p = [V , p ]
dt i
−1   ∂ 
= V,
i i  ∂x 
∂ ∂
= V − V (4.27)
∂x ∂x
∂  ∂  ∂
=V −  V  −V
∂x  ∂x  ∂x
 ∂ 
= −  V
 ∂x 
which is the desired result. Note I have used the chain rule to get to the 4th
line of Eqn. (4.27) (if you find this confusing, re-write this with
wavefunctions and integrals over x).

Start April 15 Lecture


Review MT1 study guide
Plan xtra office hours (Wed April21 3pm?, Thur April 22
10am?
Questions?

Basically the upshot is that for quantum systems that correspond to known
classical systems, the equations for the quantum operators typically
correspond to the classical equations for the variables. Then you can just
take the expectation value of the operator equations and “prove” Ehrenfest’s
theorem (namely that the expectation values obey the classical equations).
But note a couple of caveats:

• Discus how a symmetrical wavefunction starting “at the top” of a


symmetrical upside-down potential obeys the trivial Ehrenfest
d d
equations: =
p =
V ' 0 and x = 0 . Thus Ehrenfest’s
dt dt

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 15


Theorem does not necessarily help you fully understand the
emergence of classical behavior in quantum systems.

• There is not always a unique meaning to “the equations for the


quantum operators typically correspond to the classical equations for
the variables”. For example, if the classical equation has a term of the
2 2 2 2
form x p , should the corresponding quantum expression be x p or
xpxp etc. These are different expressions when expressed as
quantum operators. Note that this feature does not represent a
fundamental problem with quantum mechanics (as some people tried
to teach me many years ago). There simply are different quantum
theories that have the same classical limit. We are not able to
differentiate right from wrong among these options when only
observing classical phenomena, but so what?

5. Energy Eigenstates (stationary states)


Equation Section (Next)

5.1. Changing Basis in Bra-Ket notation


Here is are a few technical points about linear algebra in the bra-ket
notation. Just as when we discuss vectors in our 3-d space, one can express
the vectors in different bases or “coordinates”

3-d Example: A vector a in a 3-d coordinate system.

One can consider a basis or coordinate system given by the orthonormal


  
(mutually orthogonal and normalized) basis vectors {e1 , e2 , e3 } .
Orthonormal means that
 
δ ij
ei ⋅ e j = (5.1)

The coordinates of the vector a in the coordinate system given by the basis

{ei } are given by:
 
ai= ei ⋅ a (5.2)

We often give a as an “ordered triple” of its coordinates:

a ↔ ( a1 , a2 , a3 ) (5.3)
The dot product can be written in abstract
 form:

a ⋅b (5.4)
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 16
Or in terms of coordinates
 
a ⋅ b= a1b1 + a2b2 + a3b3 (5.5)
 
One can also express a in a different basis/coordinate system given by {ei′} .
For example, if one rotates the x and y axes counterclockwise by an angle θ
the new basis vectors can be written in the original coordinates as
      
e1′ ↔ ( cosθ ,sin θ ,0 ) =( e1′ ⋅ e1 , e1′ ⋅ e2 , e1′ ⋅ e3 )
      
e2′ ↔ ( − sin θ ,cosθ ,0 )= ( e2′ ⋅ e1 , e2′ ⋅ e2 , e2′ ⋅ e3 ) (5.6)
      
e2′ ↔ ( 0,0,1) =( e3′ ⋅ e1 , e3′ ⋅ e2 , e3′ ⋅ e3 )

Bases in Bra-Ket notation: Vectors

A vector space can be spanned by a basis { i } which obeys the


orthonormality relation
i j = δ ij (5.7)
A vector can be ψ expressed in terms of its components
ψ ↔ ( 1 ψ , 2 ψ , 3 ψ ,) (5.8)

Similar to Eqn. (5.5), the inner product of two vectors can be express in
terms of components:
ψ 2 ψ1 = ∑ ψ 2 i i ψ1 (5.9)
i

One basis commonly used for particles are { x } the eigenstates of the
i
position operator x :
x xi = xi xi (5.10)
The standard particle wavefunction ψ ( x ) is the continuum limit of the
components of ψ in the { xi } as discussed in section 3.2.

Bases in Bra-Ket notation: Operators

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 17


A linear operator A is an object which maps a ket ψ into a new ket given
by
ψ′ = Aψ (5.11)
One can work in a specific basis { i } in which case A is represented by a
matrix Aij ≡ i A j and the operation of A on vector ψ is represented
by matrix multiplication on the stated expressed as the components of ψ .
this is written as
=iψ′ ∑=
j
A jψ ∑ i A
ij
j
j jψ (5.12)

Often one writes


=A ∑=
k k A j j ∑k
kj kj
Akj j (5.13)

and one can see that Eqn. (5.12) is achieved by dotting Eqn. (5.13) by ψ
from the right and i from the left (and using Eqns. (5.11) and (5.7) ). Also,
one can write
∑ψ k k A j
A ≡ ψ Aψ =
kj

(5.14)
= ∑ ψ k Akj j ψ
kj

(The second equality involved inserting Eqn (5.13)) For any operator A one
can define a complete orthonormal space of eigenstates Ai defined by { }
A Ai = Ai Ai (5.15)
and the orthonormality condition
Ai Aj = δ ij (5.16)
One can use the { A } as a basis in which case A takes a special form
i ij

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 18


A = ∑ Ak Ak A Aj Aj
kj

= ∑ Ak Ak Aj Aj Aj
kj

= ∑ Ak Aj Ak Aj Aj (5.17)
kj

= ∑ Ak Ajδ kj Aj
kj

= ∑ Aj Aj Aj
j
In this case
Aij = Aiδ ij (5.18)
Where Ai are the eigenvalues and the matrix Aij is diagonal in the
eigenvector basis. In the diagonal basis (using the last line of Eqn (5.17) for
the second equality)
A ≡ ψ Aψ = ∑
ψ j Aj j ψ
j
(5.19)

Note that in this section A has different meaning depending on the number
of subscripts. Ai are the eigenvalues of A and Aij are the matrix elements of
A . In the special case where the matrix elements are expressed in the
eigenbasis, the two are related by Eqn. (5.18).

Note also that the continuum notation:



A = ∫ ψ * ( x ) A ( x )ψ ( x ) dx (5.20)
−∞

assumes operator A can be expressed in diagonal from A ( x ) in the x basis


(compare with Eqns (5.14) and (5.19) . More explicitly, one might write:

A = ∫ ψ * ( x ) A ( x, x ')ψ ( x ') dxdx '
−∞

= ∫ ψ * ( x ) A ( x ) δ ( x − x ')ψ ( x ') dxdx '
−∞ (5.21)

=∫ ψ * ( x ) A ( x )ψ ( x ) dx
−∞

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 19


where the first line does not assume Α is diagonal in the x basis (and thus is
similar to Eqn. (5.14)), the second line does assume Α is diagonal in the x by
inserting an expression for A ( x, x ') that is the continuum equivalent to Eqn.
(5.18). The third line is achieved by integrating over the delta function.

Start April 20 Lecture


Return HW2

5.2. Energy Eigenstates

In algebraic form the Schrödinger equation is given by


∂ H
ψˆ = −i ψˆ (5.22)
∂t 
The hermitian operator H can be diagonalized producing eigenvalues
{ }
{Ei } and the corresponding eigenstates Ei . By definition, these obey:
H Ei = Ei Ei (5.23)
(since H is hermitian, the Ei are real) For bound systems (namely particles
trapped in wells in the potential V ( x ) the spectrum of energy eigenstates is
discrete (but often still countably infinite) so we can use the discrete index i
on the energy eigenstates and eigenvalues without apologies.

Assuming (as we probably will do for the entire course) that H is time
independent, the Schrödinger equation is trivial to integrate for the energy
eigenstates giving:
t ) exp {−iE j t / } E j ( 0 )
E j (= (5.24)
(note that here the “ t ” refers to the time evolution of the vector, not the
eigenvlaue) One can express the energy eigenstates as wavefunctions (that is
write them in the basis of x eigenstates) to get
Ψi ( x ) = x Ei (5.25)
So
Ψ j ( x, t ) =exp {−iE j t / } Ψ j ( x,0 ) (5.26)
Griffiths uses the convention that lower case ψ is
ψ j ( x ) ≡ Ψ j ( x,0 ) (5.27)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 20


The energy eigenstates are often called stationary states because the
probability density does not change with time:
{ }
Ψ *j ( x, t ) Ψ j ( x, t ) = exp −i ( E j − E j ) t /  Ψ *j ( x,0 ) Ψ j ( x,0 )
(5.28)
Ψ ( x,0 ) Ψ j ( x,0 )
= *
j
The energy eigenstates form a complete basis, so any wavefunction (or state)
can be expanded in terms of energy eigenstates:
Ψ (t = 0) = ∑ψj
j Ψ (t = 0) ψ j ≡ ∑ c j ψ j
j
(5.29)

and its time evolution can be easily determined:


(t )
Ψ= ∑c j exp {−iE j t / } ψ j (5.30)
j
So even though the energy eigenstates are “stationary”, the really tell you all
about the time evolution of any state in the space. Note that what is
stationary is the probability density, but the sum in Eqn. (5.30) is a sum over
many wavefunctions (before squaring), and as we have already seen, adding
wavefunction does not necessarily mean adding probability densities.

Also note the difference between ψ j and Ei is that


ψi
= =
Ei (t 0) (5.31)
which allows us to express the time evolution explicitly in Eqn, (5.30), or
equivalently, allows us to not write Eqn. (5.30) using the right hand side of
Eqn. (5.31) which would be more clumsy.

In Class Problem (similar to Griffiths Example 2.1)

A quantum system starts out in the following combination of energy


eigenstates:
Ψ ( t = 0 ) = a1 ψ 1 + a2 ψ 2 (5.32)
For simplicity we take both ai ’s real and we normalize with a1 + a2 =
2 2
1.
What are the probabilities of finding the system in states s as a function of
time? (without loss of generality we can assume s ψ i are both real, see
section 5.3 for specifics) Answer
s Ψ ( t )= a1 exp {−iE1t / } s ψ 1 + a2 exp {−iE2t / } s ψ 2 (5.33)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 21


So

s Ψ (=
t) + 2a1a2 cos ([ E2 − E1 ] t /  ) s ψ 1 s ψ 2
2
a1a1 s ψ 1 + a2 a2 s ψ 2
2 2

(5.34)
Note how (assuming E1 ≠ E2 ) a significant time dependence of the
probability has emerged due to the fact that different stationary states were
combined.

5.3. Real wavefunctions for Energy Eigenstates


Consider the expansion of an energy eigenstate in a particular basis:
=Ej ∑
k
k E j k ≡ ∑ ak , j k
k
(5.35)

One can write each expansion coefficient in terms of real and imaginary
parts
a=
k. j akR, j + iakI , j (5.36)
One can plug Eqn (5.35) into Eqn (5.23) and note that (as with any equation)
the real and imaginary parts must each separately be equal. That means that
≡ ∑ akR, j k
R
Ej (5.37)
k
And
≡ ∑ akI , j k
I
Ej (5.38)
k
are separately energy eigenstates (the key here is that the eigenvalues are
real, so Eqn. (5.23) does not mix the real and imaginary parts of the ak , j ) .
In particular when the basis is the position eigenstates one can chose the
wavefunctions ψ j ( x ) to be real 2. Of course, once one chooses that
convention, one must stick to it, and not re-adjust to real wavefunctions at a
later time. After all, the complex phase contains all the information about
the time evolution.

6. The infinite square well


Equation Section (Next)
2
There still is an ambiguity under ψ → −ψ . One just has to make a choice and stick to it. ψ and −ψ
represent exactly the same state.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 22


6.1. Stationary states
We now consider the case (treated in Griffiths section 2.2)
0, 0 ≤ x ≤ a
V ( x) =  (6.1)
∞ otherwise
Classically this corresponds to a frictionless particle that bounces elastically
from each boundary.

Boundary conditions (Discussed further in section 6.2 ):


• ψ continuous
∂ψ
• can be discontinuous at the barrier edge
∂x
• ψ = 0 under the barrier

Stationary states: find solutions to


 2 d 2ψ
− Eψ
= (6.2)
2m dx 2
Or
d 2ψ
2
= −k 2ψ (6.3)
dx
2mE
with k ≡ , and assuming (can be proven) E > 0

Note that (not surprisingly) we are looking for momentum eigenstates inside
p2
the well ( H = there). We’ve already discussed that momentum
2m
eigenstates take the form
ψ k A exp {−ikx}
= (6.4)
Here we will follow the convention that requires the stationary states to be
real, which gives eigenstates of the momentum operator:
= ψ k A= k cos kx ψ k Bk sin kx (6.5)
The subset of these that obeys the boundary conditions (specifically
ψ= ( 0 ) ψ=( a ) 0 ) is

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 23



=ψ k B=
k sin kx; k (6.6)
a
Normalizing appropriately gives the following complete set of stationary
states
 2  nπ 
 sin  x , 0≤ x≤a
ψ n ( x) =  a  a  (6.7)

 0 otherwise
With corresponding eigenvalues:
kn2 n 2π 2  2
=
En = (6.8)
2m 2ma 2
Discussion of Eqns (6.7) and (6.8):

• The functions are alternatively even or odd with respect to the center
of the well (corresponding to even or odd n )
• Number of nodes (places where ψ ( x ) = 0 ) = n + 1.
• Orthonormal:
ψ j ψ k = δ jk (6.9)
• Complete:

f ( x ) = ∑ cnψ n ( x ) (6.10)
n =1
• with

=cn ψ
= n f ∫ ψ n* ( x ) f ( x )dx (6.11)
−∞
(NB this is only true for f ( x ) obeying the same boundary conditions
• Thus, the general solution to the Schrödinger equation is given by
(t )
Ψ= ∑c j exp {−iE j t / } ψ j (6.12)
j
(just repeating Eqn. (5.30) ).
Start April 27 Lecture
Collect HW3
HW 4 assigned on Web
New reading assigned on Web

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 24


6.2. More about boundary conditions
How general is the completeness (Eqn (6.10) )? Consider a wavefunction
which is only non-zero under the barrier. All cn = 0 , so such
wavefunctions are not treated by this “complete set”. Still, the stationary
states we have found *are* complete for wavefunctions inside the well.

Further discussion of boundary conditions using slides

6.3. Discussion of HW3


Prob 3.1: Supposing a finite universe introduces discreteness in quantities
that we are used to think of as continuous, but the discreteness appears on
such a fine scale that we would never have noticed it so far.

Prob 3.2: Here we see that putting in the barrier removes have of the old
eigenstates, but these get “replaced” with new states. See here
http://www.chem.uci.edu/undergrad/applets/dwell/dwell.htm for an
illustration using a finite barrier. As the barrier is raised you can see the
states continuously deform in the direction of the states you found for the
infinite central barrier.

Used these slides


http://www.physics.ucdavis.edu/Cosmology/P115a/Notes/S2_SquareWellB
Cs.ppt

6.4. Griffiths Examples 2.2 and 2.3

Discussion based on text

Key points:
• Expansion in energy eigenstates
(t )
Ψ= ∑c j exp {−iE j t / } ψ j (6.13)
j
(Eqn (5.30)) with

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 25


2 a  nπ  2 a  nπ  30
x ψ ( x ) dx x  5 x ( a − x )dx
a ∫0 a ∫0
cj = sin  sin 
 a   a  a
(6.14)
 0 n even
=
8 15 / ( nπ )
3
n odd

• Energy expectation value



= ∑c Ei ≥E1
2
E i (6.15)
i =1

which gives
• Dominance of lowest energy eigenstate.
2
 8 15 
= =
2
3 
c1 0.998555... (6.16)
 π 
• Periodic behavior
Ψ * ( x, t ) Ψ=
( x, t ) ∑c cjk
*
j k { }
exp −i ( E j − Ek ) t /  ψ *j ( x )ψ k ( x ) (6.17)

Each term has a period


2π 
τ jk ≡ (6.18)
E j − Ek
Maximised for minimum E j − Ek . The energy difference is minimized for
large k and j which differ by unity approaches E1 as k and j get large

• Periodic behavior generalizes to recurrences in any finite system due


to finite number of energy eigenvalues and thus finite number of
frequencies in Eqn.(6.18)) (optional, not on exams)
• Periodic behavior of universe (optional, not on exams)?
( RU / lP )2
=T t P=
e tP e SU

(10 m /10 m ) 10−45 e10 s 25 −35 2

= t P=
120
(6.19)
e

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 26


Discussion based on Matlab program parabola.m

• Plot time evolution of probability density

 Discuss: What is the minimum possible expectation value for the energy
of all possible states for a possible particle in an infinite square well:

= ∑c Ei ≥E1
2
E i (6.20)
i =1

(using the convention where E1 = Emin )

7. The harmonic oscillator


Equation Section (Next)

7.1. Energy lower bound (Griffiths problem 2.2)


For all energy eigenstates for a potential V ,
Ei > min (V ) (7.1)
Rough proof: Rewrite the eigenvalue equation as
d 2ψ 2m
= V ( x ) − E ψ (7.2)
dx 2  2
If E < min (V ) then V ( x ) − E  > 0 everywhere. Discuss pictorially how a
i
wavefunction which obeys Eqn. (7.2) cannot be normalized.

7.2. The harmonic oscillator potential


Classically, a particle moving in a potential
1
V ( x ) = kx 2 (7.3)
2
is knows as the “harmonic oscillator” (i.e. a pendulum for small amplitudes
or a mass on an ideal spring). The oscillation frequency is
k
ω= (7.4)
m
so one can write

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 27


1
V ( x ) = mω 2 x 2 (7.5)
2
The eigenvalue equation (time independent Schrödinger equation) for a
particle moving in the potential given in Eqn. (7.5) is
 2 d 2ψ 1
− 2
+ mω 2 x 2ψ =
Eψ (7.6)
2m dx 2
Or
Hψ = Eψ (7.7)
with
p2 1
=
H + mω 2 x 2 (7.8)
2m 2

7.3. The “ladder operators”

Define p and x by 3
1 1 
=H ω  p 2 + x 2  (7.9)
2 2 

(NB I use p and x as an intermediate step to defining a± , Griffiths does


not, but he gets to the same definitions of a± .)

Start April 29 Lecture


• Return MT1

To work in class: What are the dimensions of p and x , and what is


[ x , p ]( = i ) ? (Compare with [ x, p ] = i ). Note, the classical versions of
p and x are also useful for classical studies of the SHO.

3
The tilde parameters are defined by separately equating the momentum and potential terms. My x is the
same as Griffith’s ξ used in his section 2.3.2.
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 28
Now, try and “factor” H . If p and x were numbers (not operators) one
could use the expression
u 2 + v 2 = ( iu + v )( −iu + v ) (7.10)
To factor Eqn. (7.9). Lets try this anyway and see where it takes us. Define 4
1
a± ≡ ( ip + x ) (7.11)
2
(the reason for the peculiar looking sign convention will become clear
below).

To work in class:
i) What is [ a− , a+ ] ? ( [ a− , a+ ] = 1)

ii) What is a± ?

Now let us see what the analog of Eqn. (7.10) gives:


1
a− a=
+ ( ip + x )( −ip + x )
2
=
2
(
1 2
p + x 2 − i ( xp
  − px   ))
(7.12)
= (
1 2
2
p + x 2 − i [ x , p ])

= (
1 2
2
p + x 2 + 1)
So
 1
=H ω  a− a+ −  (7.13)
 2
In general (by definition of the “commutator”)
= AB − [ A, B ]
BA (7.14)
So
a− a+ − [ a− , a+ ] =
a+ a− = a− a+ − 1 (7.15)
Which means we can also write

4
This is the same Eqn.2.47 of Griffiths, but I have chosen to define the dimensionless p and x variables in
order to get there in a way I think is more convenient (and commonly used in physics).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 29


 1  1
= H ω  a+ a− += 1 −  ω  a+ a− +  (7.16)
 2   2 
Thus the time independent Schrödinger (or H eigenvalue equation) reads:
 1  1
H=ψ ω  a− a+ − = ψ ω  a+ a− + = ψ Eψ (7.17)
 2   2 
Now to the punch line. It turns out that if ψ E is an eigenstate with
eigenvalue E , then a+ψ E is an eigenstate with eigenvalue E + ω  and
a−ψ E is an eigenstate with eigenvalue E − ω  .

Proof of the + case:


 1  1 
Ha+ψ = ω  a+ a− +  a+ψ = ω  a+ a− a+ + a + ψ
 2  2 
 1
= a+ ω  a− a+ + ψ
 2
 1 
= a+ ω  a− a+ − + 1 ψ (7.18)
 2 
= a+ ( H + ω )ψ
= a+ ( E + ω )ψ
= ( E + ω ) a+ψ

In class: Prove the “ − ” case.

The a± are called the “raising” and “lowering” operators.

BTW: This is why the positive and negative signs were assigned to
a± the way they were in Eqn. (7.11).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 30


7.4. Constructing the harmonic oscillator energy
eigenstates
First off, it seems we have proved that there can be energy eigenstates of
arbitrary low energy (just keep operating with a− ). This is not in conflict
with Eqn. (7.1) because it turns out (as discussed in Section 7.1) these
eigenstates will not be normalizable.

Note: if we have a non normalizable wavefunciton, it is impossible to make


it normalizable by operating with
1 1  ∂ 
=
a+ ( −ip + x ) 
x basis
→  − + 
x  (7.19)
2 2  ∂x 
Multiplying by x or differentiating will not make a non-normalizable
function normalizable. There is one way out: There is one wavefunction
that has the property that
a−ψ 0 = 0 (7.20)
That terminates the chain so that subsequent operations with a− don’t
produce different solutions. Find ψ 0 by solving
1  ∂ 
a−ψ 0=   + x ψ 0= 0 (7.21)
2  ∂x 
Or

ψ 0 = − xψ 0 (7.22)
∂x
Integrating gives
dψ 0
∫ ψ0
= − ∫ xdx
  (7.23)

Or
x 2
ln (ψ 0 ) =
− + const (7.24)
2
Which gives
= ψ 0 A exp {− x 2 / 2} (7.25)
Normalizing, and converting to x gives
 mω   mω 2 
1/ 4

= ψ 0 ( x)   exp − x  (7.26)
 π   2 

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 31


In class problem: What is E0 ? (answer:
 1 ω
Hψ=
0 ω  + −
a a + ψ=
 0 ψ0 (7.27)
 2 2
(where we have used Eqn. (7.21)) So
1
E0 = ω (7.28)
2
NB
 1   1 
Ha−ψ 0 =
 −  ω ψ =
 − 0 
a −  ω  ( 0) (7.29)
 2   2 
So the “zero” comes from the wavefunction being zero, not the energy being
zero.

Have evaluated E0 , we can use Eqn (7.18) to construct the higher energy
eigenvalue which gives
 1
En ω  n + 
= (7.30)
 2

7.5. Normalization
If n is the (normalized) energy eigenstate constructed by
n = An a+n ψ 0 (7.31)
(where An is chosen to get the normalization right). We can check the
normalization of the next state 5:

5
In the first step of Eqn. (7.32) we use a+† = a− and solve Eqn. (7.13) for H / ω
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 32
2
a+ n = n a+† a+ n
 H 1
= n + n
 ω 2 (7.32)
 1 1
= n n + +  n
 2 2
= n +1
So
a+ n = n + 1 n + 1 (7.33)
and
An
An+1 = (7.34)
n +1
Giving
1
An = (7.35)
n!
So

1 n
n = a+ ψ 0 (7.36)
n!
Also 6
2
a− n = n a−† a− n
 H 1
= n − n
  ω 2 (7.37)
 1 1
= n n + −  n
 2 2
=n
So
a=
− n n n −1 (7.38)
Start May 4 Lecture

6
In the first step of Eqn. (7.37)we use a−† = a+ and solve Eqn. (7.17) for H / ω
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 33
• Return HW3
• HW 4 assigned on web
• New reading assigned on web

7.6. Working with ladder operators

Ladder operators are useful for evaluating all kinds of expectation values
between energy eigenstates. Here are some expressions that will be useful:
a− + a+
x =
2
(7.39)
a − a+
p = −
i 2
And
H 1
= a+ a− +
ω 2
H 1
= a− a+ − (7.40)
ω 2
H a− a+ + a+ a−
=
ω 2
Now we will do Griffiths example 2.5:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 34


ω
nV n = n x 2 n
2
ω
n ( a− + a+ )( a− + a+ ) n
4
ω
= n a− a− + a+ a− + a− a+ + a+ a+ n
4
ω  H 
=  n2 n + n a− a− n + n a+ a+ n 
4  ω 
ω  H 
 n 2 n + ( n )( n − 1) n n − 2 + ( n + 1)( n + 2 ) n n + 2 
4  ω 
ω  1
=  n + 
2  2 (7.41)

Also

1
n x n
= n ( a− + a+ ) n
2
=
1
( n a− n + n a+ n ) (7.42)
2
=
1
2
(
n n n −1 + n +1 n n +1 )
=0
Now consider
1
n x m
= n ( a− + a+ ) m
2
(7.43)
=
1
( n a− m + n a+ m )
2

Giving

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 35


=
1
2
( m n m −1 + m +1 n m +1 )
(7.44)
=
1
2
( mδ n ,m−1 + m + 1δ n ,m+1 )
Note: The first line of Eqn. (7.44) is not defined for m = 0 . However, the
last line of Eqn. (7.43) is well defined since we know that a− 0 = 0 . Also,
the last line of Eqn. (7.44) is correct for all allowable m and n .

In general, to avoid concerns about treating the special case of a− 0 = 0


one can equally well think of a− operating to the left.

To understand the meaning a− operating to the left, just take the hermitian
conjugate of Eqn. (7.33)
n a− =n + 1 n + 1 (7.45)
Thus one can replace Eqn. (7.44) with

=
1
2
( n +1 n +1 m + m +1 n m +1 )
(7.46)
=
1
2
( n + 1δ n+1,m1 + m + 1δ n ,m+1 )
which does not involve the special case.

Note that one also can use


p = ω mp (7.47)
and

x= x (7.48)

to translate between expectation values of the tilde-ed and un-tilde-ed
quantities.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 36


7.7. Hermite Polynomials
The energy eigenstate wavefunctions of the harmonic oscillator are related
to some well know functions called the Hermite polynomials. Griffiths
discusses these in the context of an alternative “analytic” calculation 7 of the
energy eigenstates in his section 2.3.2 (note that Griffiths’s variable ξ is the
same as my variable x ). One can write the harmonic oscillator energy
eigenstates as
 mω 
1/ 2

H n ( x ) exp {− x 2 / 2}
1
=ψ n ( x)   (7.49)
 π  n
2 n!
Where
H0 = 1
H1 = 2 x
=
H 2 4 x 2 − 2
(7.50)
=
H 3 8 x 3 − 12 x
H 4 = 16 x 4 − 48 x 2 + 12

Note that the even and oddness of the H n ’s is the same as the even or
oddness of the index. The Hermite polynomials have been well studies by
mathematicians and there are many know theorems about them (which must
be equivalent to things you can derive about the ψ n using the algebraic
techniques used in these lectures).

7.8. Discussion of ψ n ( x )
Discuss the properties of the harmonic oscillator eigenstates using these
slides: SHO Probabilities and SHO momentum
http://www.physics.ucdavis.edu/Cosmology/P115a/Notes/SHOp.ppt
http://www.physics.ucdavis.edu/Cosmology/P115a/Notes/SHOProb.ppt

Points to discuss:
• More nearly classical probability distribution as one goes to larger
energies

7
We will not cover the analytic approach now, but it will be applied to other systems later in the 115abc
course series.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 37


• Rough notion of classical momentum as a function of position applies
for large energies (but note, these are stationary solutions with zero
expectation value for the momentum, something like combining
classical solutions with different phases)
• Note that the above discussion is not sufficient to explain the
emergence of classical behavior in quantum mechanics, but it is part
of the story.
• vs infinite square well: Similar decrease of wavelength with
increasing E
• vs infinite square well: Leakage into classically forbidden zone
(different from infinite square well).
• Increasing E results in more broad wavefunction (well gets more
broad at high E, unlike infinite square well).

7.9. Power Series Method


Read section 2.3.2. of Griffiths. Not covered in this class, but you should be
aware of these methods. You will learn and use them in other courses.

8. The Free Particle


Equation Section (Next)

8.1. Energy Eigenstates


Hamiltonian
p2 2 ∂ 2
H= = − (8.1)
2m 2m ∂x 2
So the eigenvalue equation can be written
Hψ = Eψ
↓ (8.2)
d 2ψ
2
= −k 2ψ
dx
Where
2mE
k ≡± (8.3)

We’ve already discussed (in section 4.2) that the eigenstates of p are

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 38


ψ k = A exp {ikx} (8.4)
Where the momentum eigenvalues (which can be positive or negative) are
given by
p = k (8.5)
Since the Hamiltonian depends only on p , The p eigenstates given by Eqn.
(8.4) are also eigenstates of the free particle H with eigenvalues
p 2 ( k ) 2 k 2
E (k ) =
= (8.6)
2m 2m
This means the time dependence is given by
 k 2  
Ψk ( =
x, t ) A exp i  kx − t  (8.7)
 2 m 
Which describes waves traveling to the right/left for positive/negative k .

Interestingly the propagation speed of the waves is


k p 1
=
vquantum = = vclassical (8.8)
2m 2m 2
We will see in section 8.4 that vquantum is not the speed at which a particle
actually travels.

8.2. Normalization
The energy and momentum eigenstates for a free particle (given by Eqn.
(8.4) are not normalizable:

∫ −∞
ψ k*ψ k dx= A2 ∞ (8.9)
Thus the ψ k are not realistic states for a real particle. However, thanks to
theorems from Fourier transform theory, we know that we still can use the
ψ k as a “basis”.

Realistic normalizable wavefunctions can be expressed in terms of energy


eigenstates by expressions like
1  k 2  
=Ψ ( x, t ) ∫ φ ( k ) exp i  kx − t   dk (8.10)
2π  2 m 

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 39


which is the equivalent of Eqn. (5.30) except here there are a continuum of
energy eigenstates so we have in integral instead of a sum. A normalizable
wavefunction for a free particle is called a “wave packet”.

Note that according to Fourier transform theory, one can determine φ ( k ) in


Eqn. (8.10) by performing the integral
1 ∞
φ (k ) = ∫ Ψ ( x,0 ) exp {−ikx} dx (8.11)
2π −∞

Note also that the standard “orthonormality” expression from Fourier


transform theory:

∫ exp {−ik1 x} exp {ik2 x}dx =
δ ( k1 − k2 ) (8.12)
−∞
Is equivalent to Eqn. (8.9) except δ ( 0 ) is just a more palatable way of
writing ∞ .

Also, as I have discussed before, we don’t really know if the universe is in a


some massively large box or not:

• Once can equally well expand the “free particle” in the discrete
Fourier series corresponding to the energy eigenstates of a square
well.
• The differences between the time evolution under the two different
descriptions will only be noticeable when the wave packet approaches
the edge of the well. A true free particle will just propagate on
through, but a particle that is really in the well (and thus expressed in
terms of the finite Fourier series) will bounce off the wall and come
back the other way.

8.3. An example
Example 2.6 from Griffiths

Consider
 A −a < x < a
Ψ ( x,0 ) =
 (8.13)
 0 otherwise

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 40


1
What is A ? (Answer )
2a
As worked out in Griffiths,
1 sin ( ka )
φ (k ) = (8.14)
πa k
Leading to
1 sin ( ka )  k 2  
=Ψ ( x, t ) ∫ exp i  kx − t   dk (8.15)
π 2a k  2 m 
The time evolution of this wavefunction is exhibited in these slides
http://www.physics.ucdavis.edu/Cosmology/P115a/Notes/G2p104.ppt

Problem 2.21 from Griffiths also discussed in class May 6 2010.

Start May 11 Lecture


• Collect HW5
• HW 4 to be returned on Thursday May 13
• MT2 Thursday May 20
• Study Guide and discussion Thursday May 13

8.4. Group velocity


The above example showed a wave packet with no average velocity. Now
let as assume that φ ( k ) is peaked around some nonzero value k0 . Let us
write Eqn. (8.10) in the form

∫ φ ( k ) exp{i ( kx − ωt )}dk
1 ∞
=Ψ ( x, t ) (8.16)
2π −∞

k 2
where here ω = . The equivalent of Eqn. (8.16) can appear in other
2m
circumstances with different functions ω ( k ) .

Now let us expand ω ( k ) as a Taylor series:


ω (k ) =ω0 + ω0′ ( k − k0 ) (8.17)
Where
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 41
ω0 ≡ ω ( k0 )
dω (8.18)
ω0′ ≡
dk k0
Keeping only the first two terms the Taylor series and changing variables to
s ≡ k − k0 (measuring relative to the central value at k0 ) one can rewrite
Eqn. (8.16) as

exp {i ( −ω0t + k0ω0′ )} ∫ φ ( k0 + s ) exp {i ( k0 + s )( x − ω0′t )} ds


1 ∞
Ψ ( x, t ) ≅
2π −∞

(8.19)

Since the integral is the same at all times if one simply makes the
substitution ( x → ( x − ω0′t ) ) one finds:
Ψ ( x, t ) ≅ exp {i ( −ω0t + k0ω0′ )} Ψ ( x − ω0′t ,0 ) (8.20)
Which is just the initial packet moving along at speed ω0′ . Thus one has
dω k0 p0
=
vgroup = = = vclassical (8.21)
dk k0 m m
So (comparing with Eqn. (8.8))
= v=
vclassical group 2v phase (8.22)
Where v phase is vquantum from Eqn. (8.8).

Note that there are corrections to Eqn. (8.20) which involve higher terms in
the Taylor expansion. They typically are related to the spreading of the
wavepacket and other phenomena that reflect the full quantum behavior of
the system.

9. The delta function potential


Equation Section (Next)My discussion closely follows Griffiths section 2.5

9.1. Bound states and scattering states


Discuss bound states and scattering states in classical physics with drawings
on the board.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 42


We have seen both types of situations in quantum mechanics (square well
and harmonic oscillator were bound, free particle was “scattering” in a
trivial way).

The following is generally true of energy eigenstates for quantum systems:

• If E < V ( x = +∞ ) and E < V ( x = −∞ ) the energy eigenstates are


bound states with a discrete index.
• If E > V ( x = +∞ ) or E > V ( x = −∞ ) the energy eigenstates are
scattering states with a continuous index

In many situations we can take V ( ±∞ ) = 0 in which case the sign of the


energy eigenvalue dictates whether the state is scattering or bound.

 For discussion in class: What do eigenstates look like in semi-infinite


potential (draw on board)? (Answer: Draw images like the ones on these
slides)

http://www.physics.ucdavis.edu/Cosmology/P115a/Notes/SHOp.ppt \

9.2. Delta Function Well


Consider the case where
V ( x ) = −αδ ( x ) (9.1)
Except at x = 0 the Hamiltonian is
p2 2 ∂
2
H= = − (9.2)
2m ∂x 2
In most of space you are solving
Hψ = Eψ
↓ (9.3)
d 2ψ
= 2
κ 2ψ E < 0
dx
d 2ψ
2
− k 2ψ E > 0
=
dx
Where
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 43
2mE
k ≡± (9.4)

(as before) and
−2mE
κ≡ (9.5)

Note that Eqns. (9.4) and (9.5) are just chosen to reflect the overall sign of E
in Eqn. (9.3)

For E < 0 the general solution is of the form


ψ ( x=
) A exp{−κ x} + B exp {κ x} (9.6)
And for E > 0 the general solution is of the form
ψ (=
x ) A exp {ikx} + B exp {−ikx} (9.7)
The solutions will take these general forms to either side of x = 0 (in
general, with different coefficients on either side). The remaining task is to
match these solutions together at x = 0 . The results from the next section
will help:

9.3. Matching a wavefunction across infinite changes in


the potential.
Let us consider the integral of the eigenvalue equation over a small region
around x = 0 :
 2 ε d 2ψ ε ε
( ) ( ) ∫−ε ψ ( x ) dx (9.8)
2m ∫−ε dx 2 ∫−ε
− dx + V x ψ x dx =
E
We are specifically interested in the limit as ε → 0 . Since ψ is always
required to be continuous, the right hand side of Eqn. (9.8) goes to zero in
this limit giving
 2 ε d 2ψ ε

2m ∫−ε dx 2
dx =
− ∫−ε V ( x )ψ ( x ) dx (9.9)
The left side can be trivially integrated to give
 2  dψ dψ  ε
 −  ∫−ε V ( x )ψ ( x ) dx
= (9.10)
2m  dx +ε dx −ε 

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 44


If V is continuous in this region the right side of Eqn (9.10) will go to zero
giving the result that ψ ′ must be continuous in that case 8.

However, for the case of the delta function potential


ε 2mα
∫ε V ( x )ψ ( x ) dx 
ε →0
→− ψ (0) (9.11)
− 2
So for this case we must have
dψ dψ  dψ  2mα
− ≡ ∆  = − 2 ψ (0) (9.12)
dx +ε dx −ε  dx  

• We will now use this result to find energy eigenstates for the delta
function potential

Start May 13 Lecture


• Return HW4
• MT2 Thursday May 20
• Study Guide and discussion of MT2

9.4. The bound state


In the case where E < 0 one has
ψ ( x=
) A exp {−κ x} + B exp {κ x} x < 0
(9.13)
ψ ( x=
) F exp{−κ x} + G exp {κ x} x > 0
To achieve normalizability one must set
A= G= 0 (9.14)
Also, continuity requires
B exp {0} = F exp {0}
(9.15)
B=F
So our solution is
=ψ ( x ) B exp {κ x} x < 0
(9.16)
ψ ( x ) =B exp {−κ x} x > 0
Now applying Eqn. (9.12) Gives

8
Once can also contemplate equation (9.10) in the case of the infinite square well and learn that the case
we considered involves a special limit where the right side remains finite but non-zero.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 45


2mα
−2 Bκ =− 2 B
 (9.17)

κ= 2

Note there is only one bound state! Every delta function potential has
exactly one bound state with wavefunction

ψ ( x)
= exp {−mα x /  2 } (9.18)

And energy
mα 2
E= − 2 (9.19)
2

9.5. Scattering states


Now consider E > 0 . For x < 0 the eigenvalue equation reads
d 2ψ 2mE
=
− ψ=
−k 2ψ (9.20)
dx 2
 2

with
2mE
k≡ (9.21)

And the general solution is
ψ (=
x ) A exp {ikx} + B exp {−ikx} (9.22)
Unlike the bound case, at this point we have no reason to exclude either
term. Similarly for x > 0 we can write
ψ (=x ) F exp {ikx} + G exp {−ikx} (9.23)
The continuity of ψ at x = 0 requires
F +G = A+ B (9.24)
The derivatives are

= ik ( F − G )
dx +
(9.25)

= ik ( A − B )
dx −
Earlier we derived Eqn. (9.12)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 46


dψ dψ  dψ  2mα
− ≡ ∆  = − ψ ( 0) (9.26)
dx +ε dx −ε  dx   2

Using Eqn. (9.25) in Eqn. (9.26) gives


2mα
ik ( F − G − A + B ) =− ( A + B) (9.27)
2
Or
F − G = A (1 + 2i β ) − B (1 − 2i β ) (9.28)
Where

β= (9.29)
2k
For each k there are two equations (Eqn. (9.28) and Eqn (9.24) ) and four
unknowns. The remaining freedom corresponds to a large variety of
possible solutions. Here is a convenient way to organize these:
First, recall that exp {ikx} is a plane wave moving to the right and
exp {−ikx} is a plane wave moving to the left. Why:
( x, ∆t ) A exp {i ( kx − Ek ∆t /  )}
Ψ k=
(9.30)
{
≡ A exp i ( k ( x − ∆x ) ) }
So the value of Ψ now has the value it used to have a bit to the left.
Similarly,
Ψ − k ( x, ∆
=t ) A exp {−i ( kx + Ek ∆t /  )}
(9.31)
{
≡ A exp −i ( k ( x + ∆x ) ) }
So Ψ now has the value it used to have a bit to the right.

Draw Picture similar to Griffiths figure 2.15 on board.

If we set G = 0 we have removed the incoming wave from the right, the
solution we get will correspond to a particle “incident from the left” with
amplitude A . Solving for B and F gives

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 47



B= A
1 − ιβ
(9.32)
1
F= A
1 − iβ
In this picture B the amplitude of the “reflected wave” and F is the
amplitude of the “transmitted wave”.

One can consider the wave incoming from the right by setting A = 0 .

In class project: Use symmetry only to determine B and F for a particle


incident from the right.
1
B= G
1 − ιβ
(9.33)

F= G
1 − iβ

Going back to the incident from the right case, the probability per unit length
2
in the incident wave is A , the probability per unit length in the reflected
2
wave is B and the probability per unit length of the transmitted wave is
2
F . From these we can construct the relative probabilities that the
incoming particle will be reflected:
2
B β2
=
R = (9.34)
A
2
1+ β 2

And that the incoming particle will be transmitted


2
F 1
=
T = (9.35)
A
2
1+ β 2
These are called the “reflection coefficient” and “transmission coefficient”
respectively. Note that we must have
R +T = 1 (9.36)
Which is indeed the case. Writing β in terms of the original parameters
gives
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 48
1
R=
1 + ( 2 2 E / mα 2 )
(9.37)

and
1
T=
1 + ( mα 2 / 2 2 E )
(9.38)

Note that higher E produces stronger transmission, a reasonable result.

Start May 18 Lecture


• Questions re Midterm?
• discuss role of α in delta function potential

Discuss relationship between these time independent states and actual time
dependent scattering (the discussion on pp 75-76 of Griffiths is good).

9.6. The delta function peak


One can use similar tools to consider the same potential with the sign
flipped:
V ( x ) = +αδ ( x ) (9.39)
You lose the bound state, but the logic of the scattering states remains the
same and you can just take α → −α in the above discussion. Since

β= Eqns. (9.32) and (9.33) will give different answers for the factors in
2k
the wavefunction, but since the transmission and refection coefficients are
ratios of probabilities (wave function squared) they wind up depending only
on α .
2

Thus the transmission and reflection coefficients are unchanged when you
flip the sign of the potential.

Discuss how this is very different from the classical case. Discuss quantum
tunneling. (Similar to the discussion at the end of section 2.5 of Griffiths).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 49


10. The finite square well
Equation Section (Next)My discussion closely follows Griffiths section 2.6

10.1. Functional form


For convenience, we place the well slightly differently on the x-axis:
−V0 −a ≤ x ≤ a
V ( x) =  (10.1)
 0 x >a
This system has both bound and scattering states. We’ll start with the bound
states.

In the region x > a the situation is similar to the exterior of the delta
function well:
Hψ = Eψ
↓ (10.2)
d 2ψ
= 2
κ 2ψ E<0
dx

With
−2mE
κ≡ (10.3)

Giving
ψ=( x ) A exp {−κ x} + B exp {κ x} x < −a
(10.4)
ψ ( x=
) F exp{−κ x} + G exp {κ x} x > a
To achieve normalizability we set A= G= 0 to get
ψ ( x ) = + B exp {κ x} x < −a
(10.5)
ψ ( x ) =F exp {−κ x} x > a
In the interior of the Schrödinger equation reads
Hψ = Eψ
↓ (10.6)
d 2ψ
2
= −l 2ψ
dx
Where
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 50
2m ( E + V0 )
l≡ (10.7)

(Remember, from section 7.1 we know that we must have E > −V0 so l is
real and positive). The general solution is
= ψ ( x ) C sin ( lx ) + D cos ( lx ) ; x < a (10.8)
(The general plane wave solutions would also work, but this form gets us to
our answer faster.)

10.2. Matching at the boundaries


Now we must match up the solutions in the different regions. We’ll use
continuity of ψ , and since the potential is finite we get to use continuity of
ψ ′ as well.

One can prove (see Griffiths problem 2.1c) that for a symmetric potential
well the solutions are either even or odd. We can use that symmetry
ψ (a) = ±ψ ( −a ) to relate the matching at a to that at −a , so lets focus on
−a .

Now we’ll consider the even functions (so C = 0 ).

Continuity of ψ at a gives
F exp {−κ a} =
D cos ( la ) (10.9)
and the continuity of ψ ′ gives
−κ F exp {−κ a} =
−lD sin ( la ) (10.10)
Dividing the above two equations gives
κ = l tan ( la ) (10.11)
Since both κ and l depend on E , we can solve Eqn. (10.11) for E to get
the allowed energies. It helps to use the variables
z ≡ la (10.12)
And
a
z0 ≡ 2mV0 (10.13)

From the definitions of κ and l one can get
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 51
κa
= z02 − z 2 (10.14)
And Eqn. (10.11) becomes
tan ( z )
= ( z0 / z ) −1
2
(10.15)

Plot figure 2.18 from Griffiths and discuss:

Start May 25 Lecture


• Collect HW6
• HW 7 assigned on web
• discuss office hours

1) Wide “and” deep

When z0  1

tan z ≈ z0 / z
So
n 2π 2  2
En + V0 ≅ (10.16)
2m ( 2a )
2

Which corresponds to half the energy eigenvalues of the infinite square well.
(Remember we are only doing the even functions right now.) Note that for
finite V0 there are only a finite number of bound states.

2) Shallow “and” narrow

As z0 takes on smaller values (show how one curve moves to the left) the
total bound state number falls, but there always is one remaining. (Note
π
that the lowest odd state drops out for z0 <
2

10.3. Scattering states


Starting on the left:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 52


ψ ( x ) A exp {ikx} + B exp {−ikx} x < −a
= (10.17)
2mE
Where k ≡ .

Inside the well
ψ ( x ) C sin ( lx ) + D cos ( lx ) ;
= x <a (10.18)
With
2m ( E + V0 )
l≡ (10.19)

Let’s consider a scattering state incoming from the left, so we remove the
wave incoming from the right leaving
= ψ ( x ) F exp {ikx} x > a (10.20)

Continuity of ψ and ψ ′ at − a gives


A exp {−ika} + B exp {ika} =
−C sin ( la ) + D cos ( la ) (10.21)
And
ik  A exp {−ika} − B exp {= ika} l C cos ( la ) + D sin ( la )  (10.22)
While both continuities at a give
C sin ( la ) + D cos ( la ) =
F exp {ika} (10.23)
and
l C cos ( la ) − D sin ( la )  =
ikF exp {ika} (10.24)
One can eliminate C and D and solve for B and F to give
sin ( 2la ) 2
= B i
2kl
( l − k2 )F (10.25)
and
exp {−2ika} A
F=
(k ) sin ( 2la )
(10.26)
2
+l 2

cos ( 2la ) − i
2lk
Using T ≡ F / A and going back to the original variables gives
2 2

−1
 V02 2  2a 
T=
 1 + sin  2 m ( E + V )
0  (10.27)
 4 E ( E + V0 )   

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 53


Note that there is perfect transmission ( T = 1) at the zeros of the sin which
occur at
2a
2m ( E + V0 ) =
nπ (10.28)

Which gives
n 2π 2  2
En + V0 = (10.29)
2m ( 2a )
2

Sketch Griffiths figure 2.19 on the board and discuss. (Think destructive
interference on reflected wave)

11. Review
Equation Section (Next)

11.1. Review of eigenstates for flat potentials


The time independent Schrödinger equation can be written as
d 2ψ 2m
= V ( x ) − E ψ (11.1)
dx 2  2
If V ( x ) = const one has

d 2ψ
= Cψ (11.2)
dx 2
If E > V then C < 0 and Eqn. (11.2) is solved by
ψ A exp ±i −C x
= { } (11.3)
Which gives oscillating behavior.

If E < V then C > 0 and Eqn. (11.2) is solved by


=ψ A exp ± C x { } (11.4)
which gives decaying or growing behavior.

If V ( x ) is piecewise constant then the solution is produced by combining


the above solutions in their relevant regions using appropriate boundary
conditions:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 54


BOUNDARY CONDITONS:
1) ψ ( x ) is continuous everywhere

d
2) ψ ( x ) is continuous everywhere if V does not have an infinite
dx
discontiunity. Otherwise there may be discontinuities in ∂ xψ ( x ) . We have
seen specific examples of these in the infinite square well (see section 6.1)
and for the delta function potential (see section 9.3).

3) ψ ( x ) is normalizable (for bound states).

Comment: Note that even when V ( x ) is not constant the rough


characteristics described above apply. For example the harmonic oscillator
energy eigenstates have oscillating behavior in regions where E > V and
decaying behavior in regions where E < V .

12. The energy-time uncertainty principle


Equation Section (Next)
From section 3.5.3 of Griffiths (eqn 3.72, eqn 3.72, eqn 3.73)

One can differentiate the expectation value of hermitian operator Q to get

d d ∂  ∂  ∂ 
Q= Ψ Q Ψ=  Ψ Q Ψ + Ψ  Q  Ψ + Ψ Q Ψ 
dt dt  ∂t   ∂t   ∂t 
(12.1)
Using
∂ H
Ψ= Ψ (12.2)
dt i

∂ H
Ψ =− Ψ (12.3)
dt i
gives

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 55


d i ∂Q
= Q [ H, Q ] + (12.4)
dt  ∂t
Note that in the case where there is no explicit time dependence of the
∂Q
operator Q (that is = 0 then Eqn. (12.4) makes sense in a
∂t
straightforward way: If [ H, Q ] = 0 then the eigenstates of Q should also
be eigenstates of H , the stationary states, so Q should be time dependent.
It seems intuitive that the degree of time dependence of Q should depend
on the degree to which [ H, Q ] ≠ 0 .

One can now set A = H and B = Q in Eqn. (14.4) and combine the result
with Eqn. (12.4) to get
2 2
  1  d Q   d Q 
2 2
1
σ H2 σ Q2 ≥  [ H, Q ]  =  =    (12.5)
 2i   2i i dt    
2 dt 
or
 d Q
σ Hσ Q ≥ (12.6)
2 dt
now if we define ∆E ≡ σ H and
σQ
∆t = (12.7)
d Q / dt
Then Eqn. (12.6) can be written

∆E ∆t ≥ (12.8)
2
Start May 27 Lecture
• MT2?

This is an “uncertainty principle for E and t ”, but note that due to the
somewhat imprecise definitions (especially the definition of ∆t by Eqn.
(12.7) which depends on the choice of Q ) this is not as concrete a result as
Eqn.(4.6). However, it can be very helpful in building intuition.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 56


For example an interesting perspective can be gained by rewriting the above
as
d Q / dt 2
≤ ∆E (12.9)
σQ 
This gives a bound on the rate of change of the expectation value of any
hermitian operator Q (in units of its standard deviation) in terms of the
standard deviation of the energy distribution ( ∆E ). The more narrowly the
energies are spread, the more slowly the expectation values of all
observables (in fact all hermitian operators) can evolve. Of course, in the
limit of totally sharply peaked energy ( ∆E = 0 ) nothing evolves at all. The
system is in a stationary state.

13. Multiple systems, coherence and measurement


Equation Section (Next) We have so far only discussed the behavior of
“single systems”. The real world is made of many physical objects
(including a great many harmonic oscillators for example). This section
introduces some of the tools that we use to handle this multiplicity of
systems in quantum mechanics. Along the way, we will introduce the
important concept of quantum coherence, and introduce tools that are key to
understanding quantum measurements.

This discussion is quite different from what is found in Griffiths, so I will try
and make these notes self-contained.

13.1. Combining two systems


We have discussed the behavior of simple systems whose quantum states
live in a simple vector space. For example, we can consider a vector space
{ }
spanned by the orthonormal basis i and write any state in that space in
that basis as

ψ = ∑ ci i (13.1)
i
Now suppose we want to consider a large quantum systems made up of two
subsystems designated S A and S B . Suppose S A is spanned by the

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 57


orthonormal basis { i } and S
A B is spanned by the orthonormal basis

{ i } . One construct a larger vector space out of S


B A and S B (formally
denoted S A × S B ). One can construct a basis that spans S A × S B (formally
written { i } × { j } ) which contains elements
A B

ij A×B
= i A
j B
(13.2)
for all the i ’s and j ’s that label the respective bases for S A and S B . The
inner product in S A × S B is given by
=
A×B ij kl A×B A
i k A B j l B δ ik δ jl
= (13.3)

Very important: It only takes a difference in one of the subsystems to cause


orthognality. For example
A×B 3,5 3,9 A×B = A 3 3 A B 5 9 B =δ 3,3δ 5,9 =1 × 0 =0 (13.4)
The fact that subsystem S A is in the same state in both basis states in Eqn.
(13.4) does not prevent the two states from having an inner product (or
“overlap”) of zero.

The general state in the combined space can be expanded in the basis to give

ψ A×B
= ∑ c=
ij
ij ∑c ij A×B
ij
ij i A
× j B
(13.5)
nd
(the 2 equality gives an alternate form that will be useful below).

It is tempting to think of states in the combined space such that each system
is in a specific state: For example one could have system S A in state φ A
and system S B in state ξ B
.

Let us define the expansion coefficients for each subsystems state in the
subsystem basis as

φ A
= ∑ ai i A
(13.6)
i
and

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 58


ξ B
= ∑ bi i B
(13.7)
i
Then one could construct the following state for the combined system:
ψ A×B
=φ A
ξ B

   
=  ∑ ai i A  ∑ j
× b j B
(13.8)
 i   j 

= ∑ ai b j ij A×B
ij
Note that it does *not* make sense to write
ψ =
A×B
φ A
+ξ B
(13.9)
Each term in Eqn. (13.9)is defined in a *different* vector space, so Eqn.
(13.9) does not make sense.

Note that the state given in Eqn. (13.8) is not nearly as general as the general
state given in Eqn (13.5). For example if the sizes 9 of the subspaces are N A
and N B respectively, then Eqn. (13.8) contains only 2 ( N A + N B ) − 1
independent real parameters whereas the general state (Eqn. (13.5)) contains
( 2 N A × 2 N B ) − 1 independent real parameters. Note that the “ −1” comes
from the normalization constraint and each the “2’s” come from the complex
numbers ai , bi and cij each have two real paramters . For subsystems larger
than dimension 2, there are many possible states that cannot be written in the
product form of Eqn.(13.8).

If the state of the combined system cannot be written in the product form,
the two subsystem are said to be in an “entangled state”.

Discuss examples from everyday experience (such as a macroscopic


pendulum interacting with the air and photons in the room) where
entanglement is a natural consequence of interactions with the environment.
Start June 1 Lecture
• Return HW 6
9
Here size means the number of orthogonal basis states required to span the space, aka the “dimension” of
the space.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 59


• Turn in HW 7
• For Thursday: Final Review and Course Eval’s
• Write Eqns. (13.8) and (13.5) to start discussion.

13.2. Properties of the wavefunction v2


Loss of coherence through correlation: An illustration

In Section 3.1 we considered a simple exercise. I’ve reproduced in here with


the correct normalizations put in:
In class problem (to be done in small groups):
Consider the following particle wavefunctions
 1
 −1 < x < 1
Ψ1 ( x, 0 ) =
 2 (13.10)
 0 x ≥1

and
 1
 2 −1 < x ≤ 0

 1
Ψ 2 ( x, 0 ) =  − 0 < x <1 (13.11)
 2
 0 x ≥1


a) Find the values of A and B that normalize the wavefunctions so that the
total probability (of finding the particle anywhere) is unity in each case.
(These normalizations are give this time around)
b) Plot the probability density (vs x ) for each wavefunction. (Both have
constant probability densities for −1 < x < 1 )

Now consider a new wavefunction


Ψ1 + Ψ 2
Ψ3 ≡ (13.12)
2
Is Ψ 3 normalized? (Yes, once the 2 is included, as it is done in Eqn.(13.12), but not
in section 3.1)
e) Plot the probability density (vs x ) for Ψ 3

We learned that even though both states had equal probably of finding the
particle for positive or negative x , when the two wavefunctions were added

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 60


we found that there was zero probability of finding the particle for all
positive values of x .

Now let us consider a variation on this problem.

Suppose the particle is just one subsystem, and is combined with another
(dimension 2) subsystem. The second subsystem is spanned by the
orthonormal basis 1 B , 2 { B }

Now consider the state for the whole system given by:

( t 0 ) A=
Ψ= ×B
1
2
( ψ1 A
1 B + ψ2 A
2 B ) (13.13)

Now we can evaluation the probability of finding the particle in position


eigenstate x A by evaluating
2 2
=
xΨ x ψ1 A
1 B + x ψ2 A
2 B

( B 1 ψ1 x +
= B
2 ψ2 x )( x ψ1 A
1 B + x ψ2 A
2 B )
= B
11 B
ψ1 x x ψ1 + B
12 B
ψ1 x x ψ 2 (13.14)
+B 21 B
ψ 2 x x ψ1 + B
22 B
ψ2 x x ψ2
= ψ1 ( x) + ψ 2 ( x)
2 2

Notes:

1) The first line of Eqn. (13.14)uses facts about how to define probabilities
of subsystems which may not (yet) be familiar to you.

2) The probability of finding the particle at x > 0 is nonzero in this case!


(Compare this with the original problem).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 61


13.3. Coherence and Decoherence
13.3.1. Nomenclature
Here are some of the words that go with the situation discussed in Section
13.2. Note that you can already understand this simple example just using
what we have discussed in section 13.2 The additional words (called out in
bullets below) just introduce you to terminology that is often used to discuss
this type of situation.

• We say that the particle and system B are “entangled” or “correlated”


because Eqn. (13.13) cannot be written in the form of a single product
of states (in the two subsystems).
• We also say the particle is in an “incoherent state” or is “decohered”
due to its entanglement with system B.
• The state
Ψ 3 ≡ Ψ1 + Ψ 2 (13.15)
• is considered a “coherent superposition”.
• We attribute the cancellation of the wavefunction (or “probability
amplitude”) for 0 < x < 1 in the case where the particle was not
entangled with system B to “quantum coherence”.
• A common example of such quantum coherence is often discussed in
the “double slit experiment” where the coherent addition of the
wavefunctions emerging from each slit leads to interesting patterns of
cancellations and enhancements.
• The entanglement described by Eqn. (13.13) prevents the cancellation
in the region 0 < x < 1 and thus we say the entanglement “destroys the
coherence of the particle” or “decoheres” the particle.
13.3.2. The “relativity” of Decoherence

Instead of Eqn. (13.13) we could have chosen a different state


1  ψ1 + ψ2 ψ1 − ψ2 
( t 0 ) A=
Ψ= ×B

A A
1B+ A A
2 B  (13.16)
2 2 2 
Eqn. (13.16) also has an “entangled” form. But note that the states entangled
with 1 B and 2 B are quite different, and seem to involve some of the
quantum coherence we discussed for the non-entangled case.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 62


For example the probability amplitude for 0 < x < 1 cancels to zero in the state
correlated with 1 B (similarly the region −1 < x < 0 has zero probability in
the state correlated with 2 B
). These cancellations are the result of
“quantum coherence” much like that discussed for the coherence case in
Section 3.1, even though there is entanglement. Note, however that the
probability for finding the particle at point x is the same for the
wavefunctions given in Eqn. (13.13) and in Eqn.
(13.16) The upshot: There are many interesting effects that come about
when quantum systems become entangled. These often are associated with
the “destruction of quantum coherence” or “decoherence”. However, be
careful: The presence of decoherence or entanglement does not mean that
there are no effects associated with quantum coherence (as illustrated in the
above example). One just has to carefully crank through the calculation and
see how things turn out.

This section is called the “relativity” of decoherence because typically the


decoherence is best understood in terms of which states are correlated with
orthogonal states in the other system. That is, decoherence can be different
depending on which states are entangled (for example, one can imagine
entangling either position or momentum eigenstates of a particle).

13.3.3. Examples

Discuss how interacting systems are the norm, and often the interaction will
change the state of at least one of the systems dramatically. That will
rapidly lead to quantum coherence even if you imagine starting out in a
coherence non-entangled state.

Discuss:
• Large pendulum interacting with air and light.
• Double slit interacting with air/light (or “measuring which slit”)
• Bose-Einstein condensation, neutrino mixing,
superfluidity/conductivity (examples of maintaining coherence despite
the “environment”).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 63


13.4. Measurement
13.4.1. General Discussion

Ask class: What are important properties of a measurement (quantum or


otherwise). Look for these points:

• Correlate apparatus with state of system being measured


• Do not change the state of the system being measured
• Be able to read the apparatus (compare with air/photon bouncing off
macroscopic pendulum).
• Able to keep reliable record of measurement.
Start June 3 Lecture
Discuss office hours next week
Review final exam study guide
Discuss quantum measurement
Section 13.5.1
Discussion
Course Evaluations

13.4.2. Description of quantum measurement

A good quantum measurement apparatus will:


1. Correlate apparatus with state of system being measured
2. Not change the state of the system being measured (as best as possible…
see below)
3. Be readable and enable a “permanent” record to be made

Here is a simple model of a quantum measurement.

“S” indicates the system being measured and “A” indicates the apparatus.

Before the measurement let us assume the system is in “pure” product state,
with the apparatus in the “zero” state:
Ψ ( t0 ) ψ S 0A
= S× A
(13.17)

At some time after the measurement the system and apparatus become
entangled:
Ψ ( t1 ) S×=
A
c1 ψ 1 S
1 A + c2 ψ 2 S
2 A
+ c3 ψ 3 S
3 A
+ ... (13.18)

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 64


Note that the system is now in an entangled state, which is not generally the
same as the initial state. A key part of a quantum measurement apparatus is
that it is identified with a “measurement basis” or “pointer basis” in the
system space. This basis (the ψ i S ’s in Eqn. (13.18) define the quantity
being measured (i.e. position or momentum).

The only way a measurement cannot change the state of the system is if the
system starts out in a single pointer basis state 10. The next best thing one
can hope for in terms of the measurement not impacting the system is that
the ci ’s given in Eqn. (13.18) are given by
ci = S
ψi ψ S
(13.19)
where ψ S
is the state which appears in Eqn. (13.18). In other words, the
probabilities of finding the system in the state ψ i S
are unchanged by the
measurement (although the probabilities of finding the system in linear
combinations, or “coherent superpositions” of the ψ i S will in general be
changed). More specifically, one could have written Eqn. (13.17) as
Ψ ( t0 ) S× A
= c1 ψ 1 S
+ c2 ψ 2 S
+ c3 ψ 3 SA
+ ... 0 A
(13.20)

Which would then evolve under a good measurement into Eqn. (13.18) with
the same values for the ci ’s. For this process, the probability for finding the
system in state ψ i ( ≡ ci ) is the same before and after the measurement.
2
S

With a measurement apparatus constructed in this way, one can see that
2
ci really is “the probability that the system will be measured to be in state
ψ i S ” (despite the fact that Griffiths tells you not to talk that way)

13.5. Quantum measurements as Hermitian operators.

From the System’s point of view, a measurement apparatus is represented by


a set of (pointer basis) states (in the system space) which get correlated with
the apparatus during the measurement process (for an illustration of the
10
This is related to the “famous” theorem: “ A quantum state cannot be cloned” (not covered in this class).

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 65


properties of the Hamiltonian that are required to do this see my “Following
a collapsing wavefunction” paper which is linked at the bottom of the course
web page.).

For a good measurement, the correlations (or entanglement) must proceed


without changing the probabilities of finding the system in the pointer basis
states.

Without loss of generality, on can think of the pointer basis states as


eigenstates of a hermitian operator. Furthermore one can assign each
eigenstate the eigenvalue ai given by the reading on the apparatus (such as
position, charge, volts etc) corresponding to the state of the apparatus i A

that gets correlated with that state. Thus, once given the ψ i S
’s and the
apparatus readings ai one has all one needs to construct the hermitian
operator:

A ≡ ψ i S ai S ψ i
i
(13.21)

This is the operator the corresponds to the statement “Observables are


represented by hermitian operators” in Griffiths section 3.2. But from the
(more modern) point of view I present here, the relationship between
measurements and Hermitian operators is a matter of convenience rather
than a defining principle.

One immediate benefit to defining A as in Eqn. (13.21) can be seen from:


Ψ A Ψ= ∑ i
Ψ ψi S
ai S
ψi Ψ
(13.22)
= ∑ ci ai
2

i
2
≡ ψi Ψ
2
Since ci is the probability of the apparatus achieving a
result ai , then Ψ A Ψ is the average (or “expectation value”) of the
outcome of the experiment.

13.5.1. The measurement process

Here is what happens when you measure a quantum system:

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 66


Let the system to be measured be initially in state φ in space S . The
measurement or “observable” is represented by operator A which also is
defined in space S . A has eigenvalues ai and eigenstates { i } .
0) Prior to the measurement, the probability of a result ai is given by
2
ci ≡ i φ
2
, where i is the eigenstate corresponding to eigenvalue ai
and φ is the state of the system before the measurement. Contrary to the
comments at the bottom of page 36 of Griffiths, I do not think it is “bad
2
language” to say “ ci is the probability of finding the system in state i ”
1) The apparatus interacts with the system
2) The apparatus registers a single reading corresponding to eigenvalue ai
for one particular value of i .
3) From then on, all subsequent calculations may assume that the system is
in i with the same value of i given in 2) with unit probability. (this is the
“collapse of the wavefunction”).

Note that this formalism implies that the quantum expectation value of A
corresponds to the normal classical expression for a probabilistic expectation
2
value where the probabilities are given by ci :
A ≡ φ Aφ
= ∑ φ i ai i φ (13.23)
i

= ∑ ci ai
2

Where we used Eqn.(5.17) in the second line of Eqn. (13.23) (the notation is
slightly different, but basically we are exploiting the fact that i are the
eigenstates of A .
13.5.2. Interpretation

Physicists have different views of what additional words they like to


associate with 3). These views are often passionately held, but for the most
part do not lead to practical differences in how they work with real data.
I’ve outlined my impressions of these interpretational perspectives in the
next paragraphs. No doubt others will have their own versions. (I’ve even

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 67


had arguments with colleagues about which one of these to call the
“Copenhagen” interpretation.)

 “Many worlds” or “Everett”. The wavefunction only ever really


changes in the sense of Eqn. (13.17)vs Eqn.(13.20) . All possible outcomes
are still represented. This type of process can cascade through many
subsystems (such as the experimentalist noticing the outcome and recording
it in her notebook). But all these processes never change the fact that the
full wavefunction continues to contain different “branches” or terms
expressing different correlations (such as the terms in (13.18) ) which
continue to reflect all possible outcomes. The reassessment of the
probabilities given in step 3) above only affects the realization of a given
observer in a particular correlated state that she is on a particular branch.
Other states of the same observer are represented on other branches and they
are busily reassessing their probabilities according to their experimental
outcomes.

The “collapse of the wavefunction” only ever reflects an approximation that


is convenient for a particular observer. The lack of interference between
branches which allows such an approximation to be a good one is a key
aspect of “classical behavior” in a quantum system. Such behavior is
realized with abundance in our world, and is strongly connected to the
“stability of the record” mentioned in the section entitled “Description of a
quantum measurement”.

A key aspect of this perspective is the notion that even though the state of
the universe includes different copies of an observer (or an apparatus) which
experience different outcomes of a measurement, each of these copies can
safely assume that their outcome is “the only one” and can count on their
future interactions with the system to be consistent with that. To understand
how this works, think about what practical actions an observer or apparatus
might take to bolster (or undermine) his/hers/its confidence in the outcome
of a measurement (such as repeated re-measurement). If one thinks about
how these actions are reflected in the full quantum state one can see that re-
measurement just results in additional entanglement with new records that
are highly correlated on each branch. It is this correlation of subsequent
measurements that corresponds to the confidence of each observer/apparatus
that they may simply take a single outcome for the measurement and ignore
the others. Note that this confidence (i.e. tight correlation of subsequent

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 68


measurements) is shared across all branches of the wavefunction, even
though each branch describes a different outcome.

If the observer wants to reexamine the situation in a more refined way, it


might be possible (especially in specially designed situations) to pick up
“quantum fluctuations” which represent interference from the other branches
(due to imperfections in the apparatus, especially in quality of the record of
the result). In the many worlds picture, the other branches are “always
there”. They can enable you to do a better calculation which takes into
account imperfections in the apparatus (vs assuming absolute wavefunction
collapse). However, it typically the goal of designers of measurement
apparatuses to avoid this effect (basically, “errors due to quantum
fluctuations”) and indeed it is usually straightforward to reduce such errors
to a completely negligible level.

 “Copenhagen”: There is a definite dividing line between the quantum


and classical worlds. Objects in the classical world (generally, macroscopic
objects) “roll the dice” upon the measurement of a quantum system and
wind up with only one classical outcome. If someone does a refined
experiment of the sort described in the previous paragraph for which a many
worlds advocate would need the other branches to explain the quantum
fluctuations, the Copenhagen advocate would simply redraw the lines
between quantum and classical to allow the subtle fluctuations to be located
on the quantum side. In practice there are so many extremely classical
systems that are very far from the boundary between quantum and classical
behavior that the Copenhagen advocate need not worry about seeing the
classical world completely consumed by these boundary shifts.

Assessment of the two views:

In practice there is no difference between the two pictures when it comes to


interpreting most laboratory results. There are many systems which are so
solidly in the classical domain that we will never observe their quantum
fluctuations. So while the many worlds person might say “I am only
collapsing the wavefunction as an approximation that makes my calculation
tractable” while a Copenhagen person might say “I am collapsing the
wavefunction because this is the only correct way to describe the physical

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 69


world”. There is in most cases no practical way of teasing out observable
differences which could help us determine which is right.

Copenhagen advocates have to deal with the fact that the boundary between
classical and quantum appears to be established by fiat and is not derivable
from first principles. Some hope that eventually someone will figure out
how to make such a derivation while others insist that everything is fine just
as it is and physicists should get used to using different rules in different
domains.

Many worlds people need deal with the fact that when they describe the full
fundamental calculation (without imposing wavefunction collapse as a
practical approximation) they are describing a calculation which is
preoccupied almost entirely with describing all the things that could have
happened but did not. These are all the branches that are not realized in our
experience but which must be kept in the calculation, at least in principle, to
fully describe the many worlds picture. To many this seems very strange.
My personal view is that we should wish to understand this issue better. My
guess is that further insights into this issue will involve better understanding
how small subsystems of the Universe (which is what we are) should view
probabilities, rather than any major change in the formulation of quantum
mechanics.

A few physicists that I know express skepticism that we understand


consciousness well enough to be sure that the picture of having one’s brain
(and body) split across different branches of the wavefunction (and having
different experiences on each branch) is consistent with what we experience.
They not convinced that simple operations (such as the repeated
measurements described above) that describe “confidence building” on the
part of the apparatus or observer are the whole story in describing our
conscious experience and are thus unconvinced that the many worlds are
consistent with our “consciousness that there is only one outcome”. Even
though I suspect there are plenty of exciting new discoveries (probably
involving novel collective phenomena of neurons) on the path to better
understanding consciousness my personal view is that these discoveries are
highly unlikely to undermine the Everett interpretation).

There are two areas where I feel the many worlds is at a distinct advantage.
One is the field of “cosmology”, the study of the entire universe (past,
present and future). Many aspects of the cosmology appear to be well
Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 70
understood and have very successful theoretical calculations that agree well
with data. Many cosmology calculations involve novel boundaries between
classical and quantum. For example in the most successful model of the
origins of cosmic structure, all structure in the universe (galaxies, stars,
planets, us) originated as quantum fluctuations of a particular field (the so-
called “inflaton field”). The many worlds picture does fine in this context,
but I am much less impressed with how the Copenhagen picture manages (I
just don’t think it is that easy to draw an absolute boundary between the
classical and quantum domains in current cosmology theory). However,
there is much we still do not understand in cosmology, and it is certainly
conceivable that changes to quantum physics will be needed before we get a
more complete picture.

The other area of advantage is in quantum computation.

The basic point behind quantum computation is that the real world, by being
quantum mechanical, is actually doing a much harder “calculation” to move
forward in time as compared with than a classical one. One can see this by
imagining programming a Schrödinger equation on a computer (one
wavefunction = infinity real degrees of freedom) vs a classical particle (just
two real degrees of freedom in 1-D, x and p).

If one designs a computer in terms of classical degrees of freedom, one is


missing out on the opportunity to more fully harness nature, which is doing
the harder (quantum) “calculation”. The idea behind quantum computation
is to harness the full quantum nature of the physical world to make more
powerful computers. The theory has already been done to show that
revolutionary increases in computing power are possible.

To do this, one must design a computer that can preserve quantum coherence
of bits. Classically a bit can have two possible states: [1] and [0] . A bit in a
quantum computer must be able to be in any state of the form:
= bit c0 [ 0] + c1 [1] (13.24)
and the quantum coherence must be preserved as the computer evolves.
That demand is technically challenging, but the potential rewards are great,
and this is a very interesting area of ongoing research. Such “Q-bits” can get
in interesting entangled states with each other and each “Everett world” is
effectively doing one piece of a parallel computation. The trick is to limit
interactions with the environment so that the different Everett branches do

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 71


not get entangled with non-measurable microscopic environment degrees of
freedom (such as microscopic motions in the air or in states of emitted
radiation). If such entanglement is strictly controlled, the many branches of
the computation can be measured coherently in the end. Dramatic
computational benefits can in principle be derived from this type of quantum
parallelism.

Technically, one can think about quantum computers from the Copenhagen
point of view (by simply moving the boundary between the classical and
quantum worlds outside of the quantum computer), but there is no doubt that
the freedom from worries about these issues enabled by the many worlds
picture has been an essential part of some recent advances. I know for a fact
that the person who started the current revolution in quantum computation
(David Deutsch of Oxford) is a serious adherent to the many worlds view
and I suspect that way of thinking made it much more natural for him to
explore that territory

Probably the most general advantage to the many worlds picture is that it
allows the practitioner to not be intimidated by novel boundaries between
quantum and classical behavior. The many worlds picture offers a complete
set of tools for forging ahead in any situation. Adherents to the Copenhagen
picture are more likely to avoid situations where the boundaries between
classical and quantum are not already established from experience (such as
in cosmology or quantum computation).

13.5.3. Can we see the many worlds?

We build up our picture of the physical world by making a huge number of


measurements and attempting to assemble a consistent story from the results.

To illustrate this process, consider an observer who repeatedly observes a


two-state system “S” and records the result of the measurement in other
“record” subsystems of the physical world “R1”, “R2”, “R3”… . These
record subsystems might be ink in a lab notebook, bits on a computer disk or
human memories in our brains. All these are very complicated systems, but
we will consider them in a highly idealized and simplified form below.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 72


Let { 1 S , 2 } represent the basis in which “S” will be measured, and let us
S

describe the “RN” subspace by { 0 , 1 , 2 } where the 0


RN RN RN
state RN

represents the record subsystem before any record has been made. To further
simplify things, consider a universe “U” in which only three measurements
are made so U =S × R1 × R 2 × R3 .

Suppose time t0 the system is unmeasured, and is in the state


=ψ S α10 1 S + α 20 2 S (13.25)
giving a state for “U” given by
ψ=( t0 ) U (α10 1 S + α 20 2 S ) 0 R1 0 R2
0 R3
(13.26)
after a measurement is made in “R1” is made at time t1 , the state for “U”
will in general be given by
ψ ( t1 ) U = (α111 1 S 1 R1 + α121 1 S 2 R1 + α 21
1
2 S 1 R1 + α 22
1
2 S 2 R1 ) 0 R 2 0 R3
(13.27)
(look carefully at the positions of the brackets in Eqn. (13.27)and also in
Eqn. (13.29)below, they are important, but have showed up a bit small in
this document).

If the measurement apparatus is a good one, the coefficients in Eqn. (13.27)


will obey:
α ii1 = α i0
(13.28)
α=1
12 α= 1
21 0
giving
= ψ ( t1 ) U (α10 1 S 1 R1 + α 20 2 S 2 R1 ) 0 R 2 0 R 3 (13.29)

In general after all three measurements are made the state of “U” will be

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 73


ψ ( t3 ) U
= α1111
3
1 S 1 R1 1 R 2 1 R 3
+α1112
3
1S1 R1
1 R2
2 R3
+α 3
1121 1 S 1 R1 2 R 2 1 R 3
+α 3
1122 1 S 1 R1 2 R 2 2 R 3
+α 3
1211 1 S 2 R1 1 R 2 1 R 3
+α 3
1212 1 S 2 R1 1 R 2 2 R 3
+α 3
1221 1 S 2 R1 2 R 2 1 R 3
+α 3
1222 1 S 2 R1 2 R 2 2 R 3
+α 3
2111 2 S 1 R1 1 R 2 1 R 3
+α 3
2112 2 S 1 R1 1 R 2 2 R 3
+α 3
2121 2 S 1 R1 2 R 2 1 R 3
+α 3
2122 2 S 1 R1 2 R 2 2 R 3
+α 3
2211 2 S 2 R1 1 R 2 1 R 3
+α 3
2212 2 S 2 R1 1 R 2 2 R 3
(13.30)
+α 3
2221 2 S 2 R1 2 R 2 1 R 3
+α 3
2222 2 S 2 R1 2 R 2 2 R 3
What one expects for the different values of α ijkl
3
depends on the details.
Suppose “S” has no inclination to evolve on its own (so H S = 1 ). If all three
records represent new measurements of “S” then one expects
α10 i = j = k = l = 1
 0
α ijkl
3
= α 2 i = j = k = l = 2 (13.31)
0 otherwise

Alternatively R3 might record a different sort of measurement: A check
whether R1 and R2 recorded the same values. If 1 R 3 means “yes” and
2 R 3 means “no”, then (assuming the records are good ones) one again
expects Eqn. (13.28) to hold leading (again) to
= ψ ( t3 ) U α10 1 S 1 R1 1 R 2 1 R 3 + α 20 2 S 2 R1 2 R 2 1 R 3 (13.32)
(a great simplification over the general form given by Eqn.(13.30)). One
could say Eqn. (13.32) represents two “Everett worlds”, each of which
contains observers who are completely confident in their own state for “S”
(but each of which have different ideas of what that state is). In a state like

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 74


that given by Eqn.(13.32), every possible cross-check bolsters the
confidence in each world that they know the state of “S”. (This idea can be
extended to any number of records.) Of course no measurements and
records are perfect, and in principle some of the cross-terms in Eqn. (13.30)
will not be precisely zero in a realistic situation. Then one will need the full
form of Eqn. (13.30) to get correct answers, and thus need the “many
worlds” to explain one’s results. Whenever the role of the cross-terms is
measurable we do see them, but those are typically situations that are very
specially constructed in the lab. In most situations we encounter in the
macroscopic “classical” world, the role of the cross-terms is vastly below
any measurable level and they can be neglected.

A believer in the many worlds picture will be pleased to note that one can
quantitatively predict when the cross-terms would be measurable in this
picture, and thus express confidence that the quantum-classical boundary is
well understood.

A skeptic of the many worlds picture will object to the fact that two worlds
are represented, when we only experience one, and may not be happy with
the above argument that we have no way of telling that the other worlds are
there.

14. More about measurements and uncertainties


Equation Section (Next)

14.1. Compatibility of measurements


One can follow one measurement with another measurement, possibly with a
different apparatus. Let us consider the case where the 2nd measurement
occurs rapidly enough that the time evolution of the state between the two
measurements can be neglected.

In general the new measurement could have a pointer basis that is different
from that of the first measurement. In that case, the state of the system
could be different, possibly very different, after the 2nd measurement takes
place. This situation is said to describe “incompatible measurements”, and
the consequence of this incompatibility is that it is meaningless to talk about
the system having simultaneous values of the observables corresponding to
both measurements.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 75


Mathematically, the statement that two measurement devices have the same
pointer bases is equivalent to that statement that the two hermitian operators
that represent those measurements (say A and B ) commute:
[ A, B ] = 0 (14.1)
Eqn. (14.1) is equivalent to saying that the measurements corresponding to
A and B are compatible, or equivalently that the observables represented
by A and B are simultaneously measurable. Following one measurement
by another will not change the state of the system.

Example: We have encountered this situation in a partial sense with the free
particle. Since [ H, p ] = 0 , the energy and momentum can be determined
simultaneously. If you measure E , you can predict with certainty that a
subsequent measurement of p would yield. p = ± 2mE . Aside from the
sign of the momentum, the state is already well determined by the energy
measurement. If we measured p first, then the result of subsequent energy
p2
measurement would be completely determined to be E = .
2m
Similarly, if
[ A, B ] ≠ 0 (14.2)
The corresponding measurements are incompatible. Following one
measurement by the other will change the state of the system. (See
Homework 8)

Example: Position and moment are the classic example of incompatible


observables. Since [ x, p ] = i , measuring the position of a particle after
measuring its momentum will dramatically change the state of the particle.

14.2. Generalized uncertainty principle


As we discussed earlier in this course, and as proved in detail in section
3.5.1 of Griffiths, the commutator is related to an inequality for the variances
of any pair of observables. Using the definition
σ A2 ≡ A 2 − A
2
(14.3)
one can derive

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 76


2
1 
σ σ ≥  [ A, B ] 
2
A
2
B (14.4)
 2i 
This is a generalization consequence of observables being incompatible, and
gives some intuition which applies to the case where the wavefunction is not
an eigenstate of either observable.

Phys 115 Lecture notes © 2010 Andreas Albrecht 6/9/2010 2:32:31 PM 77

Das könnte Ihnen auch gefallen