Sunteți pe pagina 1din 26

PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

Additional Notes
These notes are to help explain some new topics and methods. They are not meant to
replace your own notes. Links to further resources are also given. There will be 28 lecture
slots. One of these will be used for the midterm exam.

Table of Contents
8/29/17: Lecture 1 (Introduction) ....................................................................................................... 2
8/31/17: Lecture 2: Statistics, Probability and Wavefunction normalization (Griffiths sections
1.2, 1.3 and 1.4) .................................................................................................................................. 3
9/5/17: Lecture 3: Momentum, Operators and Ehrenfest's Theorem ............................................. 5
9/7/17: Lecture 4: Concept review and The Time-Independent Schrdinger Equation ................. 6
9/12/17: Lecture 5: A particle in an infinite potential well and Completeness of Stationary
States (Griffiths Section 2.2) ............................................................................................................. 7
9/14/17: Lecture 6: Combining stationary states (including Griffiths Ch. 2, Problem 7) .............. 9
9/19/17: Lecture 7: The 1D SHO: Ladder operators, the ground state, and exited states .......... 10
9/26/17: Lecture 8: 1D SHO review, Hermitian conjugate of an operator, raising and lowering
operators for the 1D SHO & free particle dispersion (Section 2.4) ............................................... 12
9/28/17: Lecture 9: Concept review; wavepacket spreading; phase and group velocity ............ 14
10/3/17: Lecture 10: Probability current; Bound and scattering states of a delta-function
potential well (Section 2.5) .............................................................................................................. 15
10/5/17: Lecture 11: Transmission and reflection coefficients; Review of HW1 .......................... 17
10/10/17: Lecture 12: The finite potential well (Section 2.6) .......................................................... 18
10/12/17: Lecture 13: A first foray into formalism (Chapter 3, Sections 3.1-3.3) ......................... 20
10/17/17: Lecture 14: Expectation values in Dirac notation; Compatible observables; Time
dependence of the expectation value of an operator .................................................................... 21
10/19/17: Lecture 15: Ehrenfests Theorem and concept review .................................................. 23
10/24/17: Lecture 16: Generalised Uncertainty Principle (G.U.P.) and End of Chapter 3 ............ 24

1
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

8/29/17: Lecture 1 (Introduction)


What is a quantum object?
We answered the question of what is a quantum object by considering the de Broglie
wavelength of a particle:
= / .
Here, is the momentum of the particle and = 6.63 1034 Js is the Planck constant. [NB:
we proved that the units of the Planck constant were correct in class.]

As an example of a quantum object, we considered the de Broglie wavelength of a 1 keV


electron and contrasted that with the de Broglie wavelength of a billiard ball (see also HW1)

We then considered the concept of electron diffraction through a double slit. 1 This is a
famous example of particle-wave duality and of uncertainty. The electrons will diffract if the
slit separation is comparable to the de Broglie wavelength. In general, quantum mechanics is
relevant, as Griffiths puts it (Problem 18): "when the de Broglie wavelength of the particle in
question is greater than the characteristic size of the system." Note: the interference fringes
associated with double-slit diffraction of electrons appear after many electrons have been
diffracted. We cannot tell "which" slit an individual electron went through.

Finally, we examined a wave-like solution to the full, time-dependent Schrdinger equation:


2 2
2
+ =
2

We looked at the case of = 0 for all space. Then, a right-going wave proportional to
() is a solution to the above equatiion. For such a wave, the relation between angular
frequency, and wavenumber, of the particle it describes is:
2
= 2 .
This is what is known as a dispersion relation between and .2 Here, the angular frequency
of the wave is = 2 where is the frequency in Hertz. Similarly, the wavenmber = 2/
where is the wavelength of the wave in meters.

We contrasted the above dispersion relation with that for light (for which = ) . We
concluded that, unlike for light, particle-waves with different wavelengths can have different
speeds in a zero potential, and thus a wavepacket made of different Fourier component
waves will change shape with time, even if the potential has no time dependence. We will
come back to this later.

N.B. It will be useful to revise your description of waves (left- and right-going wave in one-
dimension) in preparation for future work on one-dimensional potentials.

1 For a revision of double-slit diffraction using light, see: http://hyperphysics.phy-


astr.gsu.edu/hbase/phyopt/mulslid.html#c2. For a discussion of double-slit diffraction of electrons, see:
http://www.nyu.edu/classes/tuckerman/adv.chem/lectures/lecture_5/node1.html
2 For a revision of group and phase velocity, see: http://physics.gmu.edu/~dmaria/590 Web

Page/public_html/qm_topics/phase_vel/phase.html
2
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

8/31/17: Lecture 2: Statistics, Probability and Wavefunction normalization (Griffiths


sections 1.2, 1.3 and 1.4)

Expectation value
We revised discrete and continuous probability distributions and noted that the expectation
value of some quantity () in a probability distribution () is (in the continuous case) given
by the integral:
< () >= ()()

Here, the < > brackets indicate the expectation value is being taken.

Standard deviation
We proved the useful relation that:
()2 =< 2 > < >2

One of the main points to note in this derivation is the fact that the expectation value of , i.e.
< >, is just a number and can be brought outside summations, integrals etc.

Quantum expectation values


We revised the concept of quantum probability density, noting that the wavefunction, , of a
particle is a solution to the Schrdinger equation for that particle, but that the wavefunction
does not have a direct physical meaning in terms of quantum expectation. Instead, the
square of the modulus of the wavefunction, ||2 is the probability density. This means that
the probability of finding a particle between and + is:

() = |()|2

if the particle is described by wavefunction () in one-dimensional space.

Griffiths Example 1
As a classical probability example, we considered Example 1 from Griffiths. This was the
case of a classical particle being dropped from the top of a cliff. We converted a probability of
observing the particle in a unit of time :

() = / (where is the total time for the particle to fall),

into a probability of observing the particle in a range of 1-dimensional space, from to + :



() = 2

We demonstrated that normalization of the probability works in both cases (integrating time
up to or space up to the height of the drop, . And we showed that the expectation value of
the particle position is one third of the way down from the top of the cliff.

We noted that a similar techique can be used to describe the classical probability of finding a
particle undergoing simple harmonic oscillation (SHO) at a particular part of its oscillation. In
particular, we will see in later lectures that such classical probabilities are reproduced by the
3
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I
high energy limit of the quantum probability probability distribution. This principle, to be
discussed later, is the Correspondence Principle.

Quantum mechanical normalization


Normalization of a wavefunction () (in the case of a one-dimensional space) means
that |()|2 = 1 where the integral is performed over all space.

The wave function must be square integrable. This generally means that the
wavefunction will decay to zero at infinity. You can often use this property in
integration proofs (see the proof of the form of the momentum operator).

If normalization is performed, then this normalization is valid for all time. In other
words, a wavefunction can change shape with time but the integral (over all space) of
the square of its modulus is unchanged and is equal to 1. This proof is not shown in
class but can be referred to in Section 1.4 of Griffiths. It is a consequence of the
Schrdinger equation. We noted that the method within the proof, of turning a single
differential in time into a double differential in space (that is then integrated) is also
used in the momentum operator derivation in Section 1.5 (see next lecture).

It's good at this stage to try Problem 7 in Griffiths to practice your differential
math and your integration.

4
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

9/5/17: Lecture 3: Momentum, Operators and Ehrenfest's Theorem


Momentum
By looking at Schrdinger eqn, we converted between time and space variables (using
similar math to that found in Section 1.4) to calculate (see Eq. [1.33]):
<>
< >= =

from which we extracted the momentum operator:



= =

Please note that I have used a little "hat" above the in order to show that this is the
momentum operator and not some other meaning of "p".

Similarly, the expectation value of an operator , associated with an observable


quantity , is
< >=
We used the momentum operator to examine a "commutator". This concept is covered
a little later in Griffiths, but we first examined it here in Lecture 3. The commutator of
two operators, say and is written as:
[, ] =


We saw that 3 [ , ] = This result will be linked to the Generalized Uncertainty
Principle later in the course.

We also examined the kinetic energy operator, = 2 /2 and observed that the
Hamiltonian , is the operator for + where is the potential energy operator. We
will return to this in the next chapter's discussion.
<>
Lastly, we considered the time derivative (Griffiths problem 1.7).4 We found that:

=

which resembles Newton's second law if we associate the right hand side with a force.

The latter is an example of Ehrenfest's Theorem: that expectation values of operators


obey classical relations. We'll see some further examples of this theorem in later
lectures.

3 Remember to take care when evaluating commutators since the order of the operators is very important and
they act on a wavefunction. So any differential in the operator is acting on everything (including a wavefunction)
to the right of it.
4 We referred to this source of solutions to Griffiths problems during the lecture:
http://www.physicspages.com/index-physics-quantum-mechanics/griffiths-introduction-to-quantum-mechanics-
problems/ Another solution source is https://www.scribd.com/doc/242288766/Introduction-to-Quantum-
Mechanics-SolutionManual-D-Griffiths-1995
5
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

9/7/17: Lecture 4: Concept review and The Time-Independent Schrdinger Equation


Concept review
We started with a review of concepts in probability, the Copenhagen interpretation,
wavefunction normalization and the physics of (classical and quantum) waves. This
was done using a conceptual presentation.5

Time-independent Schrdinger equation (Griffiths Chapter 2)


If the potential in the time-dependent Schrdinger equation is time-independent6 such
that V(, ) = (), then we can write the solution, (, ) as the product of functions
that only depend on space and on time: (, ) = ()().

Instead, we find that () satisfies a time-independent Schrdinger equation:


2 2
+ =
2 2

The overall wavefunction, (, ) has to be normalized. However, the temporal part,


(), is proportional to / and so the normalisation relies on () (the spatial part).
Therefore the probability density, |(, )|2 = |()|2 has no time dependence! For
this reason, solutions of the time-independent Schrdinger equation are called
stationary states.

The "Hamiltonian" operator, has eigenvalues that are total energy of system in a
given state, (, ). Each state has a defined energy, E .

Lastly, we proved that if (, ) is an solution of the Schrdinger equation for the


system, then a wavefunction c (, ) that is a linear combination of several such
states is also a solution of the same Schrdinger equation (see HW1).

Furthermore, because the time dependence of the wavefunction of stationary states


with different associated energies is different 7 then the linear combination of such
stationary states is not stationary. We explored an example of the latter in class, by
making the wavefunction (x, t) = 1 (, ) + 2 (, ) and examining | (x, t)|2 .
This is effectively Example 1 of Griffiths, Chapter 2. It's a good idea to review this
example now.

5 The presentation used in class was adapted from here:


http://www.colorado.edu/physics/EducationIssues/Quantum/source_documents/materials/Concept-
Tests/ConceptTests_part1_all.ppt (where the answer key is in the "notes" field of the Powerpoint field)
6 If there is time-dependence in the potential, then energy might not be conserved - a situation that we do not

consider for now.


7 It is possible for different stationary states to have the same energy and therefore the same time-dependent

part of the overall wavefunction; this is called degeneracy and is a topic that we will return to througout the
course. In my derivation of Example 1 in class, I assumed that the two stationary states that were combined to
make a new solution of the Schrdinger equation had different energies.
6
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

9/12/17: Lecture 5: A particle in an infinite potential well and Completeness of


Stationary States (Griffiths Section 2.2)

Concept review
We started with a review of concepts in linear differential equations and boundary
conditions.8

A particle in an infinite potential well


The solution of the time-independent Schrdinger equation will use boundary
conditions. For all types of potential, the wavefunction must be continuous. We also
expect that it decays at in order to be "square integrable" (normalizable).

In most cases, the derivative of the wavefunction, / must also be continuous. We


derived this by reference to the same proof that features in Section 2.5 of Griffiths.

Since the momentum operator probes the gradient of the wavefunction, the only case
where a discontinuity in / is allowed is where the potential goes to infinity. This is
the case at the edges of an infinite potential well, and so we do not need to make
/ continuous at these points.

Inside the potential well, the wavefunction satisfies


2 2
=
2 2

Outside of the potential well, the wavefunction is zero. We can simplify the
normalization by examining the probability density inside the well only. 9

Solutions to the above equation are sine or cosine functions; they have quantized
energies, E as a result of = 0 at the edges of the well.

Note that the form of the solution depends on the choice of origin. For the choice given
in Griffiths (origin at one end of the well), only sine functions are valid solutions (since
cos(0) is finite and therefore this term can have no weight).

The solutions to the infinite potential well are therefore of the form
2 2 2 ()2
= sin ( ) and have energy E = 2 ( ) =
22
2
(The normalization factor was derived in class.)

For examples of the wavefunctions, see http://falstad.com/qm1d/ (and other online


applet links given in the relevant course content folder on Blackboard).

8 The presentation used in class was adapted from here:


http://www.colorado.edu/physics/EducationIssues/Quantum/source_documents/materials/Concept-
Tests/ConceptTests_part2_all.ppt (where the answer key is in the "notes" field of the Powerpoint field)
9 In the case of a finite potential well, the normalization would have to take account of the finite probability

density outside the potential well, too.


7
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

We defined odd wavefunctions as those that were antisymmetric about a chosen point
and even ones as those that were symmetric. All of the solutions above are odd or
even about the mid-point of the well, and result in probability densities that are even
(and odd x odd function is an even function).

Hence, the average value of the position of a particle in an infinite potential well is the
mid-point of the well. However, in the states with even values of the quantum number
n = 2,4,6. .., the probability of finding the particle at the mid-point of the well is actually
zero ( = 0 at that point)!

We noted that the Correspondence Principle holds as expected; for high energy states,
the probability density is relatively flat across the well (whereas it is strongly peaked in
the middle of the well in the case of the ground state).

Completeness of (stationary) states


The set of stationary states that are solutions of the Schrdinger equation is infinite,
form a complete set and are orthogonal. The last condition means that
=
where is a Kronecker delta.

A general (wave)function, () can therefore be described by a linear superposition of


the stationary states: () = c (). This is analogous to the Fourier series and
transform examples mentioned at the start of lecture 3. We used the set of solutions of
the infinite potential well as an example of orthogonality.

We used the principle of orthogonality multiple times during the lecture, to prove that:
o the coefficients c are given by c = () ;
o normalization of () yields |c |2 = 1
o the average energy of a particle in the superposition state, () is given by
< >= |c |2 E

where E is the energy of each stationary state, .

From the last statement, we know that |c |2 give the probability of finding the system in
stationary state , if it starts in state (). This probability would be the result of
making many measurements on identical copies of the system (since the system
wavefunction would collapse, if its energy were measured, onto a single state and
so could not be re-measured in the same starting state).

8
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

9/14/17: Lecture 6: Combining stationary states (including Griffiths Ch. 2, Problem 7)

Review
We reviewed the time dependence of the overall wavefunction if the spatial state is a
linear superposition, (). This relates to Example 1 in Chapter 2. We saw that if
() = c (), then the complete solution of the Schrdinger equation is given by
(, ) = c () / (see HW1)

When we examine the probability density, |(, )|2 resulting from such a
superposition, we find that there are terms that are time-independent, such as
|c |2 |()|2 and there are terms that are time-dependent and involve / + /
and / + / terms that result in beat-like phenomena in |(, )|2 .

A linear combination of stationary states is therefore not a stationary state, since the
probability density varies with time. Note, however, that the integrated probability
density is still time-independent, meaning that wavefunction normalization is still
preserved for all time.

Time-dependent probabilities density functions can be visualized by combining


different stationary states of (for example) the infinite potential well using the applet
available at http://falstad.com/qm1d/ . Note that, if adding even functions only or odd
functions only, symmetry points and nodes are preserved.

We revised the following concepts from the previous lecture:


o the coefficients c in the above expansion are given by c = () ;
o normalization of () yields |c |2 = 1
o the average energy of a particle in the superposition state, () is given by
< >= |c |2 E

where E is the energy of each stationary state, .

We tackled some more conceptual questions (a Powerpoint presentation).10

We applied the key properties of mixtures of stationary states (see previous lecture) to
Problem 7 from Chapter 2 of Griffiths.11 In this problem, a symmetric state is given as
the starting state, (, 0), of a particle in an infinite potential well.

We wrote the starting state as the sum of all possible stationary states of the well:
o (, 0) = c () where the stationary states (sinusoids for this particular
well) are the () functions.
o The coefficients c are given by c = () ;

10 The presentation used in class was again adapted from here:


http://www.colorado.edu/physics/EducationIssues/Quantum/source_documents/materials/Concept-
Tests/ConceptTests_part2_all.ppt (where the answer key is in the "notes" field of the Powerpoint field).
11 A solution is available: http://www.physicspages.com/2012/07/11/infinite-square-well-triangular-initial-state/ or

at http://www.scribd.com/doc/242288766/Introduction-to-Quantum-Mechanics-SolutionManual-D-Griffiths-1995
9
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

o However, we know that the cofficients for = 2,4,6. .. are zero since these
stationary states are odd while the starting wavefunction is even, with respect to
the center of the well. The integral of an odd function multiplied by an even one
is zero. The coefficients of the odd terms are c = (46( 2 2 )) sin(/2)
o Indeed, in performing the integral c1 = 1 () , we found that the starting
state was about 99% in the ground state (1 ) of the well since |c1 |2 ~0.99.
o The average energy of the particle can be found from:
< >= |c |2 E

which is an infinite sum of odd- terms that vary as 1/2 since |c |2 ~1/4 while
E ~2 .

9/19/17: Lecture 7: The 1D SHO: Ladder operators, the ground state, and exited states
We started with a brief discussions of the contents of the HW1 question sheet before
turning to the main topic of the lecture: the one-dimensional (1D) simple harmonic
oscillator (SHO)

The one-dimensional (1D) simple harmonic oscillator (SHO) potential for a mass, is
1 1
() = 2 2 = 2 02 2 where = 02 is the spring constant of the oscillator. The
middle of the well is at = 0 in this definition of the potential.

This is a time-independent potential so we can apply our recent machinery, provided


that we can solve the time-independent Schrdinger equation and find the stationary
states of the system.

We re-wrote the Hamiltonian for the 1-D SHO and found that if we defined operators:
1 1
= (0 + i ) and = (0 i ).
20 2
then the Hamiltonian can be written as:

1
= 2
0
or

1
= + 2
0

Notice that I have written whereas the textbook has + . This is just a notational
difference; however, using the symbol emphasises that the raising operator and the
lowering operator are Hermitian conjugates of each other (see next lecture).

The above results will prove useful in deriving the stationary states of the simple
harmonic oscillator potential.

We used the previous relations for the "raising" and "lowering" operators, and ,
= then
respectively, to find that if a solution to the 1D SHO exists that satisfies

the state is also an eigenstate, such that:
( ) = ( + 0 )( ) .

10
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

This means that the action of the "raising" operator is to change an eigenstate of the
1D SHO into one that is also an eigenstate, but with an energy that is greater by 0 .

Similarly, the action of the "lowering" operator is to change an eigenstate of the 1D


SHO into one that is also an eigenstate, but with an energy that is lower by 0 .

Note that the proof of both of the last statements involved the use of the commutator
[, ] = 1. The use of such commutators is typical when trying to reduce the total
number of operators written in a sequence.

This means that the quantise eigenstates of a 1D SHO are separated by 0 . They
are evenly spaced in energy (unlike the eigenstates of the infinite potential well).

We saw that, in a symmetric potential, the eigenstates of the Hamiltonian are either
even or odd (prove this for yourself, and see Griffiths Problem 1(c) in Chapter 2).

Therefore it is no surprise that the states we find as eigenstates of the 1D SHO will be
even and odd functions too - see Figure 7 of Chapter 2 of Griffiths.

To find the groundstate, we asserted that the operation should result in zero, since
the ground state cannot be lowered. We therefore solved the equation = 0, using
the functional form of the lowering operator as given on the previous page.

0 2
We thus found that the ground state of the 1D SHO is proportional to 2 . From
this state we can find the other, higher energy states by applying the raising operator.
1
We also found that the ground state energy is exactly 2 0 which means that the
1
energy of the complete "ladder" of eigenstates can be written as = ( + 2)0
where = 0 corresponds to the ground state. (Contrast the fact that we used the label
= 1 for the ground state of the infinite well.)

We will explore the normalisation of states resulting from applying ladder operations
next time.

The analytic solution of the 1D SHO is contained in Section 2.3.2 of Griffiths but is not
examinable in this course.

Now try Problem 9 of Chapter 1 of Griffiths. You'll find that it refers to the 1D SHO!
You might not have realised it if you tried the problem during our study of Chapter 1.

11
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

9/26/17: Lecture 8: 1D SHO review, Hermitian conjugate of an operator, raising and


lowering operators for the 1D SHO & free particle dispersion (Section 2.4)
We started by reviewing the properties of 1D-SHO eigenstates. Note that the average
position of the particle in a 1D-SHO potential is the middle of the well ( = 0) in the
previous definition of the well.

This is made obvious by the fact that all of the eigenstates are even or odd, meaning
that the probability distribution is even. Then, the integral < >= |()|2 has an
integrand that is composed of an even function multiplied by an odd function. This
integrand is therefore odd and so < >= 0.

The result that < >= 0 makes sense because the left and right of the well are
equivalent. However, the maximum probability of finding the particle is not so obvious!

The quantum probability density |()|2 is maximum in the middle of the well, for the
ground state of the 1D SHO. This is not classically expected, since in the middle of the
well, the particle has its lowest potential energy (zero) and highest kinetic energy.

However, in the higher energy states, the classical probability distribution is essentially
recovered in the shape of the overall |()|2 function. Examine the example of
|100 ()|2 given by Griffiths in Figure 2.7. We can calculate the classical probability
distribution as being proportional to 1/(02 2 ) (where 0 is the classical limit of the
oscillator, defined by the total energy). To do this, we may a constant probability per
unit time into a probability per unit one-dimensional space: try it!.12 This classical
probability rises sharply towards the edges of the classically allowed rgime.

The equivalence of high energy quantum probabilities with those found in classical
mechanics is referred to as the correspondence principle.

Hermitian conjugate of an operator


We used the example of the raising and lowering operator for the states of a 1D SHO
to set out the definition of the Hermitian conjugate, of an operator, :


= ( )

As we'll see in a few lectures from now, an operator is Hermitian or self-adjoint (the
same meaning, just different words to describe this property), then = . We
checked that the momentum operator is indeed Hermitian. Note that the raising
operator and the lowering operator for the 1D SHO are Hermitian conjugates of each
other, i.e. () = and ( ) = .

Effects of the raising and lowering operators for the 1D SHO

12 We saw a similar example of calculating classical probability in Lecture 2: the case of a rock falling down a
cliff.
12
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

We used the above integral relation in our proof of the effect of the raising and lowering
operators for the 1D SHO:
= + 1+1
= 1

We noted that we need to use this relation when applying the raising and lowering
operators to stationary states. (In practice you will be given these relations in exams
and HWs.)

The free particle


We next reviewed the dispersion relation of a free particle (discussed in lecture 1),
finding that its energy is given by = 2 2 /2 and which is equivalent to 2 /2
where the symbols have their usual meanings.

We saw that the plane wave solutions of the Schrdinger equation for a free particle
are not normalisable. Instead, the flow of particles is conserved (see later in the
course). However, a wavepacket that is a sum of plane waves is a solution. Each
different plane wave, will have its own time dependence associated (
where = 2 /2).

Therefore the total wavefunction of a wavepacket can be written as



1
(, ) = ()
2
where the function () is the amplitude of the plane wave, . We see that this is
an extension of the Fourier transform in (, ) space, with the addition of a time-
dependence in the overall function.

From the above expression, we may find () using the inverse Fourier transform:

1
() = (, = 0)
2
1
The prefactors in the last two equations are a matter of convention - the choice
2
merely alters any normalisation prefactors in the wavefunction.

Phase and group velocity


The phase velocity is the velocity of an individual frequency, or wavenumber. The
group velocity is the velocity of a wavepacket - the velocity at which the modulation
itself moves. 13 From the previous relation = 2 /2 we may find that the phase
velocity is given by / = /2 while the group velocity is given by the differential:
/ = /. The free particle is "dispersive". In this case, the group velocity is
twice the phase velocity.

13 For a nice exposition of group and phase velocity, see: http://www.feynmanlectures.caltech.edu/I_48.html


13
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

Contrast this with the case of a electromagnetic wave in a vacuum, for which the phase
velocity and group velocity are the same (the speed of light). That is because, for light,
/ = .

In practice this means that a wavepacket will spread as time passes. Each frequency
component of the wavepacket has its own phase velocity.

Now try Example 6 of Chapter 2 of Griffiths. This is a case of a wavepacket changing


shape with time. It does so because each of its component waves travel at different
phase velocities! You will need to remember that the Fourier transform of a "top hat"
(box-shaped) function of width 2 is a "sinc" function (sin()/).14

9/28/17: Lecture 9: Concept review; wavepacket spreading; phase and group velocity

Concept review
We started with a review of concepts related to 1D SHO ladder operators and to free
particle wave packets.15

Wavepacket spreading
We went through Example 6 of Chapter 2 of Griffiths. This is a case of a wavepacket
changing shape with time. We saw that the Fourier transform of a "top hat" (box-
shaped) function of width 2 is a "sinc" function (sin()/). We examined the limits
of small and large top hat width (in space) and saw a manifestation of the uncertainty
principle, in the inverse relation between the wavepacket's spread in space and the
spread of k-components that it is comprised of.

At this stage, it's very useful to revise Fourier transforms (e.g. to prove that the Fourier
transform of a Gaussian function is also a Gaussian function) and to revise the
properties of delta functions (before our work on a delta function potential next time).

Phase and group velocity


We finished with the proof of group velocity as being / for any dispersion function
() (see Eq. [105-106] of Chapter 2 of Griffiths). For the particular case of a free
particle, we see that the group velocity is twice the size of the phase velocity.16

14 For some worked examples of Fourier transforms, see: http://www.thefouriertransform.com/pairs/fourier.php


15 The presentation used in class was adapted from here:
http://www.colorado.edu/physics/EducationIssues/Quantum/source_documents/materials/Concept-
Tests/ConceptTests_part2_all.ppt (where the answer key is in the "notes" field of the Powerpoint field)
16 An example from classical physics of the opposite case, namely where group velocity is smaller than the

phase velocity is that of surface waves on deep water, see: http://www.feynmanlectures.caltech.edu/I_51.html


14
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

10/3/17: Lecture 10: Probability current; Bound and scattering states of a delta-function
potential well (Section 2.5)
After a brief discussion of the possible solutions of the time-independent Schrdinger
equation for various potential wells and steps, we examined the concept of probability
current density.

Probability current (see e.g. Gasiorowicz Chapter 3 or Problems 1.14 and 4.41 of Griffiths)
We saw that the probability current for a one-dimensional state can be expressed as


(, ) = ( )
2
where
(, ) (, )
=

This tells us that a finite change, with time, in the probability of finding a particle in a
particular region of space is associated with a net gradient of probability current (in
space). Its an example of a conservation law here, in terms of probability.

The integral of the above relation over all space yields the expected result of no
change with time in the case of the total probability density (which equals 1 for all
time):

(, )
(, ) = = () () = 0

In three dimensions, the current density continuity equation can be expressed as


(, )
= . (, )

which yields (using Stokes Theorem):

(, ) 3 = . (, ) 3 = (, ) .
volume volume area

Delta-function potential well (Section 2.5)


We then solved our first example of potential well that is not infinite at the edges. This
was a delta-function well potential, = ()

We first looked at the case of the total state energy being within the well (a bound
state).

We used a boundary condition (as in an earlier lecture on the infinite square potential
well) that, at the well edges, the wavefunction is continuous. At any kind of infinite
well, step, or barrier we do not use the continuity of the wavefunction derivative.
Instead we can only employ continuity of the wavefunction.

To examine the Schrdinger equation solutions, we divided the region up so that


solutions were identified with different regions of space. This is a general tactic that we

15
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I
employ with any potential step, barrier or well (see following lecture on the finite
potential well).

We found that there is indeed a discontinuity in the derivative of the wavefunction at


the edge of the well, and the size of the discontinuity is proportional to the amplitude of
the delta function well, (see Equation 2.125 and Lecture 5)

We can see that this kind of well has exactly one bound state (Equation 2.129).17

We then looked at scattering states in the same delta-function potential well. These
are states that have energy greater than the potential landscape, for all space.

These states are travelling waves, not bound states, and their wavevector will change
with the potential landscape. For example, when the potential decreases, since the
total energy of the state is fixed, the kinetic energy of the scattering state increases.
This kinetic energy is 2 2 /2 (see previous lectures). Therefore when the potential
decreases, should increase, meaning that the wavelength of the scattering state will
decrease ( = 2/).

We initially allowed only right- travelling waves in the region to the left of the delta-
function potential well. This simplified the problem by considering the situation that
there is only an incoming wave from the left. We thus reduced the number of solutions
we considered to:
o an incoming wave on the left of the potential
o a reflected wave on the left of the potential
o a transmitted wave on the right of the potential18

The resulting solution allows us to compare the amplitude of the reflected wave and
the transmitted wave to that of the incident wave. The modulus squared of these ratios
give us the so-called Reflection coefficient and Transmission coefficient for the
potential well.

We note that, because of the conservation of particle flux, the presence of the potential
well should not result in any creation or loss of particles from the incident wave. This
means that the reflection coefficient and the transmission coefficient sum to 1 (since all
particles are either reflected or transmitted).

The textbook goes on to consider the case of a delta function barrier and you should
read and understand this, too.

Problems to try: Griffiths Problem 2.24, 2.27

17We will see that the finite potential well, under certain conditions can have just one bound state, too.
18 Notice that the left-travelling wave on the right side of the potential is no longer written since this has no
relevance in the case of an incoming wave from the left. In the nomenclature of Figure 15 of Chapter 2, if we
consider just an incoming wave from the left, = 0.
16
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

10/5/17: Lecture 11: Transmission and reflection coefficients; Review of HW1

Transmission and reflection coefficients


The first 30 minutes of the lecture were devoted to a review of the previous lecture's
material, and in particular and examination of the transmission and reflection
coefficients ( and ) derived at the end of Lecture 10.

We examined the large-wavevector (large kinetic energy) and small wavevector (low
kinetic energy) limits of and and observed that these correspond to expected
results from classical wave mechanics.19 We noted that the limit of large-wavevector
scattering corresponds to full transmission and scattering of a low-wavevector wave
corresponds to reflection with a phase shift.

Review of HW1
The rest of the lecture was devoted to a discussion of HW1. Numerical solutions will
be posted on Blackboard.

19 In the lecture, I displayed a video of the reflection and transmission of waves on a string, from this website:
http://www.animations.physics.unsw.edu.au/jw/waves_superposition_reflection.htm - densities
17
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

10/10/17: Lecture 12: The finite potential well (Section 2.6)


This is a classic problem in one-dimensional quantum mechanics. Unlike the delta-
function potential well considered in the previous lectures, the solutions have to found
numerically.20

The finite width of the potential well in this case means that we have 3 regions in which
we need to solve the Schrdinger equation.

Because the potential well is finite, we have two boundary conditions whenever the
potential changes: the wavefunction is continuous and its derivative is continuous.

Again, in the case of bound states (those with total energy less than the top of the well,
but greater than the bottom of the well), the state will evanesce (decay) at the well
edges and will have a sinusoidal appearance inside the well.

In the case of scattering states (those with total energy greater than the potential
landscape, for all space) the kinetic energy of the state will change at the well
boundary, with the wavevector of the state increasing in the well region (for the same
reason as in the delta-function potential well).

As the well is symmetric, the probability distribution associated with bound states must
also be symmetric. This means that there are two families of bound state solutions:
odd and even functions (with respect to the middle of the well). These states are
quantised, and have energies that depend on the depth of the well.

We showed that the ground state wavefunction belongs to the even series and has no
nodes. The ground state is always present, no matter what the depth of the potential
well is (see the 1D online well applet mentioned earlier).

As the potential well depth is increased, more bound states can be accommodated.
We discussed the condition at which the first excited state (the state above the ground
state) is just accommodated. In this case, it would have the longest possible
wavelength that accommodates one node, with zero gradient of the wavefunction at
the well edges (so that the wavefunction outside the well can decay, rather than grow,
outside the well as is required).

Just as with the delta-function well, we can examine scattering states and calculate
reflection and transmission coefficients for a wave that is incident from the left. Please
examine Equations 2.158 to 2.171 for this aspect of the well's properties.

20Except when we ask very particular questions such as towards the end of the lecture, asking "at what well
depth does the first excited state just no longer exist?" Then, as shown in the lecture, we can appeal to
symmetry arguments to find the particular answer. In general, however, the solution of the Schrdinger equation
for the finite potential well is made numerically.
18
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

We also motivated the exploration


of the finite quantum well with a
discussion of the use of quantum
wells in lasing (a LASER is a
device for Light Amplification by
Stimulated Emission of Radiation).
A well in the conduction band of a
semiconductor device can trap
electrons which can then be used
to provide the "overpopulation" of
energy levels that is required to Figure L11: A schematic band structure of a device
produce large amounts of light consisting of a layer of semiconducting (In,Ga)As
upon decay to a lower energy state. between two layers of InP. There is a finite well for
The latter decay is triggered by electrons in the conduction band and a finite well
incident radiation, hence the term for holes in the valence band, within the (In,Ga)As
"stimulated emission". layer.

This is a use of the bound state physics. Materials with different electronic band gaps
are grown in layers (in the z direction in Figure L11 above). If, for example, a
semiconducting material is sandwiched between two others with larger band gaps, the
resulting electron bands have a finite well in the (In,Ga)As material:

The concepts of bands will be introduced in PHYS 4350 when we start to discuss solid
state physics in the context of electroncs as identical (fermionic) particles. Alferov and
Kroemer were awarded the Nobel Prize in Physics for their research on
semiconducting heterojunctions.21 An advantage of the use of a solid material with a
known bandgap is that the light emitted when electrons fall from the high energy
conduction band to the valence band will be monochromatic.

Problems to try: Griffiths Problem 2.34, 2.38, 2.47* (harder)

21
The Nobel Prize website is an excellent resource, with transcriptions of the speeches and lay
summaries of the prize-winning research. For the topic of heterostructure devices, see:
http://www.nobelprize.org/nobel_prizes/physics/laureates/2000/
19
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

10/12/17: Lecture 13: A first foray into formalism (Chapter 3, Sections 3.1-3.3)

We discussed the idea of Hilbert Space and how Dirac bra-ket notation works:
o | denotes the state
o | denotes the state (the complex conjugate of )

o | denotes the overlap integral which can be thought of as akin
to a scalar product of vectors expressed in an orthonormal basis.

Hermitian conjugates and Hermitian operators


We defined the Hermitian conjugate, of an operator via an integral:


= ( )

If an operator is Hermitian or self-adjoint (the same meaning, just different words to


describe this property), then = . We checked that the momentum operator is
indeed Hermitian.

Observables correspond to Hermitian operators

We saw that the expectation value of a Hermitian operator is always real.

In fact, eigenvalues of Hermitian operators are real (the proof is given in Section 3.3.1
and was discussed in class).

Orthogonality of eigenstates of Hermitian operators


Eigenfunctions of Hermitian operators with distinct eigenvalues (i.e. states that are not
degenerate) are orthogonal.

Let two states, |1 and |2 be eigenstates of a Hermitian operator, :


|1 = 1 |1 and |2 = 2 |2

Then we have that 1 ||2 = 2 1 |2

But since is Hermitian, we may also write that


1 ||2 = 2 | |1 = 2 ||1 = 1 2 |1

So we see that 2 1 |2 = 1 2 |1 , meaning that (1 2 )2 |1 = 0

This tells us that, if there is no degeneracy among the eigenstates (different states
with the same eigenvalue) then 2 |1 = 0 , i.e. the eigenstates of a Hermitian
operator are orthogonal. 22

22If there is some degeneracy among the eigenstates of the operators, a linear combination of states can be
constructed by the "Gram-Schmidt" method, such that the revised states are orthogonal (if eigenstates share
eigenvalues, they are not guaranteed to be orthogonal, as we have just seen).
20
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

If we look at the situation in which we replace |2 by |1 then we see that the above
expressions show that (1 1 )1 |1 = 0 which demonstrates that the eigenvalues
of a Hermitian operator are real.

The matrix element, || may also be defined: . Here, the operator
acts on state | and the result is overlapped with | .

We may check that if a wavefunction is written in terms of a basis of normalised,


orthogonal states (an orthonormal basis), the squares of the modulus of the
coefficients of each term add to 1. This means that if the set of functions { } is
orthonormal, and we write:

| = |

for a normalised state | then:
= |
and
| |2 = | | = | = 1

We will return to this in the generalised statistical interpretation of quantum mechanics


(Section 3.4).

10/17/17: Lecture 14: Expectation values in Dirac notation; Compatible observables;


Time dependence of the expectation value of an operator

The lecture began with a discussion of the questions on HW2. This led to a discussion
of expectation values using a complete basis.

Expectation values
The expectation value of an operator in some state | can be written as
= ||

If we express the state | in terms of the eigenstates {| } of operator :


| = | where | = |
then we find that
= | |2

This result makes sense - we see that the expectation value of the operator is the
weighted sum of all possible values, using weightings given by the "overlap" of the
state with each eigenfunction of . See also Eqs. [3.49] and [3.51] in Griffiths.

21
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

We then turned to some further aspects of operator algebra, compatible observables,


and the time dependence of operator expectation values.

Operator algebra
The addition of operators can be done in any order.

The product of two operators is not necessarily reversible: the product of and is
but the product is not necessarily the same because the commutator: [, ] =
is not necessarily zero.

The Hermitian conjugate of + is + .

The Hermitian conjugate of is . Therefore the Hermitian conjugate of is


only also equal to if both operators and are Hermitian and their commutator is
zero.

If and are Hermitian but do not commute, the anticommutator + does


commute (try it!)

We tackled an example of a small problem of reducing an operator combination. The


case we solved was [ 2 , 2 ] = . We solved this by adding and subtracting
terms such as so that terms such as could be simplified to

[,
] = .

Note that this kind of calculation reappeared at the end of the lecture when we
examined the time dependence of the expectation value of an operator.

Compatible observables
In this lecture, we first examined the consequences of two operators commuting.

We showed that, if [, ] = 0 for two Hermitian operators, and , then the two
operators are said to share eigenstates (the eigenstates of the one operator are also
eigenstates of the other). We also say that such operators are compatible.

A derivation is as follows. Consider two Hermitian operators, and , which have


non-degenerate eigenstates (there is a one-to-one mapping between their eigenstates
and eigenvalues). Then:
| = | so (| ) = (( | )

But if [, ] = 0 then = and so we have that (| ) = | . However, we


already saw that the left hand side was equal to (( | ) and so | is an
eigenstate of with eigenvalue . This means that | is proportional to | , so
| is an eigenstate of as well!

22
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

Time dependence of
The next concept examines the time dependence of the expectation value of a
Hermitian operator in the case where the operator itself has no time dependence. In

other words, we are going to find in cases where = 0.

To start, we considered the time-dependent Schrdinger equation and how the


= .
wavefunction of a system evolves with time:


We have that = || and so = || = || + ||




= || + ||


= ( || || ) 23


= (| | | |)

and so

, ]|
= |[


The problem set for next time was to show that = for a particle moving in a
time-independent potential. Note that this involves finding [ 2 , ] in similar fashion to
the [ 2 , 2 ] calculation earlier in the lecture.

10/19/17: Lecture 15: Ehrenfests Theorem and concept review

Ehrenfest's Theorem
This lecture began by reviewing some of the material discussed in the previous
session, and in particular the time dependence of the expectation value of an operator.


We then looked at the example of the time evolution of average position, of a
particle in a time-independent one-dimensional potential. In that case, the general
relation gives:

= |[ , ]|

The operator commutes with the potential energy operator in this instance, but not
with the kinetic energy operator.


So , ]| = |[, ]| = |[ 2 , ]|
= |[
2

23 where we are using the properties of the complex conjugation.


23
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

The commutator can be written as


= + = [,
] + [ ,]
2
=


which will lead to Ehrenfest's Theorm: =

Ehrenfest's Theorem is a particular case for the expectation value of the position and
the momentum operators. For a particle of mass moving in a time-independent
potential we have just showed that

=

and now in class we further showed that

These two results both have classical analogues: the first looks like the classical
relation between position and momentum and the second resembles Newton's second
law if we associate the right hand side with a force. Let's be careful: we notice that the


right hand side of the second equation involves = () for some force, ().
So does the average of () actually look the same as () near , the classical
value of position?

We can answer this question by making an expansion of the force around = :


( )2
() = () + ( ) () + ()+. . .
2

We may see that the third term is the variance in and so if the quantum variation in


can be neglected,24 we have that () () and the classical result =
() is recovered. Ehrenfest's Theorem is therefore an example of the
Correspondence Principle (Lecture 8).

Concept review
The rest of the lecture reviewed concepts in formalism and statistical interpretation by
means of a series of multiple choice examples.25

10/24/17: Lecture 16: Generalised Uncertainty Principle (G.U.P.) and End of Chapter 3
We discussed some of the questions in HW2 at the start of the lecture.

24 See, for example: https://books.google.com/books?id=kvPxCAAAQBAJ&pg=PA30


25 The presentation used in class was adapted from here:
https://www.colorado.edu/physics/EducationIssues/Quantum/source_documents/materials/Concept-
Tests/ConceptTests_part3_all.ppt (where the answer key is in the "notes" field of the Powerpoint field).
24
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

The main result of this lecture was to derive the G.U.P.. An example derivation is
given in Griffiths, Section 3.5. We found that:
1
|[, ]|
2

We saw that the expected uncertainty relation for is recovered from this
expression.

We may also see that a Gaussian wavepacket has minimum uncertainty, since the

product for it is exactly 2. For this reason, the ground state of a 1D-SHO (which
is a Gaussian wavepacket) satisfies minimum uncertainty.

For a formal proof of the latter, see Section 3.5.2 of Griffiths.

For the energy-time uncertainty relation, see Section 3.5.3 of Griffiths. Note that we
use a relation that is superficially similar to that of the position-momentum uncertainty
relation:


2

Here we have to remark that the definition of is particular: it is the time that some
property of the system takes to change by one standard deviation. This time is a
characteristic timescale for change within the system under consideration.

10/26/17: Lecture 17: Concept review


We reviewed concepts from Chapters 2 and 3 including further questions on formalism
and observables.25

10/31/17: Lecture 18: Concept review


We looked over the practice midterm exam (see Blackboard)

25
PHYS 4300 Atomic Phys and QM I PHYS 7410X Quantum Mechanics I

11/2/17: Lecture 19: Quantum Mechanics in three dimensions


We started with a further practice question on wavefunctions that are a mixture of
stationary states.

Then, this lecture examined quantum mechanics in three dimension, starting off in a
similar way to Chapter 4 of Griffiths, noting the extension of operator formalism (such
as for position and linear momentum) from one dimension to three dimensions.

We then moved directly to a description of quantum mechanical angular momentum


(section 4.3 of Griffiths) in three dimensions, forming operators for the angular
momentum, , , about Cartesian axes by using appropriate combinations of
position and linear momentum operators. From the commutation properties of the
latter, we obtained a set of commutation relations for , , : [ , ] = and so
on.

From this we saw that we can specify one component of the angular momentum but no
more (because of the Generalized Uncertainty Principle), unless all components of the
angular momentum are zero(!)

We also showed that [2 , ] = 0 for any angular momentum component, . This


means that the total angular momentum and one component can be known
simultaneously. By the same token, it means that eigenstates of one angular
momentum component are eigenstates of the total angular momentum.

26

S-ar putea să vă placă și