“But why must I treat the measuring device

classically? What will happen to me if I don’t??”

--Eugene Wigner

“When I hear of Schrödinger’s cat, I reach

for my gun.” --Stephen W. Hawking

“There is obviously no such limitation – I can

measure the energy and look at my watch;
then I know both energy and time!”
--L. D. Landau, on the time-energy
uncertainty principle
 Lab 3 Comments
-Lab 3 meets this week. (So does Discussion, and
there *is* a quiz, so don’t skip...)
-If you are normally assigned to 132 Loomis, go
to 257 Loomis instead.
-You will need your “Active Directory” Login
{www.ad.uiuc.edu}
-You can save a lot of time by reading the lab
ahead of time – it’s a tutorial on how to draw
wavefunctions
“Quantum Information”

One of the most modern applications of QM

quantum computing, quantum communication – cryptography, teleportation,
quantum metrology

Prof. Kwiat will give an optional 214-level lecture on this topic

Sunday, March 1

3 pm, 151 Loomis
 Lecture 9: Superposition &
Time-Dependent Quantum States
(x,t=0)
U= U=

0 L x

(x,t0)
U= U=
x

0 L x
period T = 1/f = 2t0
with f = (E2-E1)/h

Snapshot at two times

Probability density oscillates back
and forth between left and right sides
 Overview


Superposition of states and particle motion

‘Packet States’ in a Box

Measurement in quantum physics

Schrödinger’s Cat

Time-Energy Uncertainty Principle
 Time-independent SEQ

Up to now, we have considered quantum particles in
“stationary states,” and have ignored their time dependence

Remember that these special states

were associated with a single energy
(from solution to the SEQ)
“eigenstates”
 2 d 2  ( x)
 2
 U ( x) ( x)  E ( x)
2m dx
“Functions that
fit”: ( = 2L/n) “Doesn’t fit”:
(x)
(x)
U= U=
n=1 n=3

0 L x
0 L x
n=2
 Review
Complex Numbers
The equation, ei = cos + isin, might be new to you. It is a convenient way to
represent complex numbers. It also (once you are used to it) makes trigonometry
simpler. a) Draw an Argand diagram of ei. b) Suppose that  varies with time,
 = t. How does the Argand diagram behave?

Solution:
a) Im(A) a) The Argand diagram of a complex number, A,
A = ei
puts Re(A) on the x-axis and Im(A) on the y-
axis. Notice the trig relation between the x and
 y components.  is the angle of A from the real
Re(A) axis. In an Argand diagram, ei looks like a
vector of length 1, and components (cos, sin).

Im(A) A = eit b) At t = 0,  = 0, so A = 1 (no imaginary

b)
component). As time progresses, A rotates
 = t
counterclockwise with angular frequency .
t=0
Re(A) cei (c and  both real), is a complex number of magnitude,
|c|. The magnitude of a complex number, A, is |A| = (A*A),
where A* is the complex conjugate of A: Im(A*) = -Im(A).
 Lecture 9, Act i
i  1
We know that
1. What is (-i)i?
a. –i
b. -1
c. +1

2. What is 1/i?
a. –1
b. -i
c. +i

3. What is |ei|2 ?
a. 0
b. e2i
c. 1
 Lecture 9, Act i
i  1
We know that
1. What is (-i)i?
a. –i
b. -1
c. +1

2. What is 1/i?
a. –1
b. -i
c. +i

3. What is |ei|2 ?
a. 0
b. e2i
c. 1

(-i)i = -i2 = -(-1) = +1

1  1  i  i
      i
i  i  i  1

e i 2
  ei   e i   ei i  e0  1
 Time-dependent SEQ

To explore how particle wavefunctions evolve with time,
which is useful for a number of applications as we shall see,
we need to consider the time-dependent SEQ:

 d  ( x, t )
2 2
d  ( x, t )
 2
 U ( x )  ( x, t )  i 
2m dx dt i2 = -1

This equation describes the full time- and space

dependence of a quantum particle in a potential U(x),
replacing the classical particle dynamics law, F=ma


Important feature: Superposition Principle

The time-dependent SEQ is linear in  (a constant times  is also a
solution), and so the Superposition Principle applies:
If 1 and 2 are solutions to the time-dependent SEQ, then so is any linear
combination of 1 and 2 (example: 0.6 1 + 0.8i2)
 Time-dependence of Energy Eigenstates


Example 1: Time-evolution of an “eigenstate” x
0 L

An “eigenstate”  is described by a single E, so we can write:

 2 d 2  ( x, t ) d  ( x, t )
 2
 U ( x )  ( x , t )  E  ( x, t )  i 
2m dx dt
i  t E
This equation has the solution:  ( x , t )   ( x )e with  

This wavefunction has a complex time-dependence, containing “i”.

But, we are mostly interested in what we measure, often |(x,t)|2:

2 i t i  t 2
 (x, t)   (x)e *
 (x)e   (x)
As previously stated, the probability density |(x,t)|2 associated with eigenstates of the SEQ doesn’t change with
time.

Thus the name for states with well-defined energies … Stationary States
 Time-dependence of Superpositions
with different E’s

It is possible that a particle can be in a superposition of “eigenstates” with different energies.

Such superpositions are also solutions of the time-dependent SEQ!

What does it mean that a particle is “in two states”. What is its E?

Let’s see how these superpositions evolve with time.


Consider a simple example using our trusty “particle in an infinite square well” system:

A particle is described by a wavefunction involving a superposition of the two lowest infinite square well states (n=1
and 2)

See nice animations at http://www.falstad.com/qm1d/

(x)
 i1t  i2t
 ( x, t )   1 ( x ) e   2 ( x) e U= U=

E E 
1  1 2  2
 
h2 0 L x
E1 
8 mL2
E2  4 E1 
 Particle Motion in a Box
The probability density is given by: |(x,t)|2 :
2 You can prove this
 (x, t)  12   22  21  2 cos((2  1 )t) using  on the
previous page.
( We used the identity: 2cos  ei  e -i )

Because the cos term oscillates between ±1, |(x,t)|2 oscillates between:
2 2
 ( x , t )  ( 1   2 ) 2
and  ( x ,t )  ( 1   2 )2

Probability (x,t1) (x,t2)

0 L x 0 L x

particle localized on left side of well particle localized on right side of well

The frequency of oscillation between these two extremes is = (E2-E1)/, or f = (E2-E1)/h. This is
precisely the frequency of a photon that would make a transition between the two states.
 Particle Motion in a Box: Example

Consider the numerical example: (x,t=0)
U= U=
An electron in the infinite square well
potential is initially (at t=0) confined to
the left side of the well, and is described
by the following wavefunction: 0 L x

2    2   (x,t0)
 ( x, t  0)  A  sin  x   sin  x  
L  L   L  U= U=

If the well width is L = 0.5 nm, determine

the time to it takes for the particle to
“move” to the right side of the well. 0 L x
 Particle Motion in a Box: Example

Consider the numerical example: (x,t=0)
U= U=
An electron in the infinite square well
potential is initially (at t=0) confined to
the left side of the well, and is described
by the following wavefunction: 0 L x

2    2   (x,t0)
 ( x, t  0)  A  sin  x   sin  x  
L  L   L  U= U=

If the well width is L = 0.5 nm, determine

the time to it takes for the particle to
“move” to the right side of the well. 0 L x
h2 1.505 eV  nm2 period T = 1/f = 2t0
En    n  2L/n with f = (E2-E1)/h
2me n2 n2
1.505 eV  nm 2 1.505 eV  nm 2
E1    1.505 eV En  E1n2
4 L2 4(.5nm)2

T h h 4.136 1015 eV  sec

to      4.6  1016 sec
2 2  E 2  E1  2  3E1  2  3 1.5eV 
 Interference beats

Note that all the motion comes from changing interference terms
between components with different E’s.

The frequencies of those changes are the frequencies at which the
interference terms change: the beat frequencies between the
different components

“beat frequency” = frequency difference – e.g., f2-f1


So the important energies determining how things move are energy
differences hf1-hf2, etc.

E1-E2 etc.

Just like in Newton’s physics: absolute energies aren’t important (you
can pick where to call U=0 by convenience), it’s energy differences
that are important.

http://www.falstad.com/qm1d/
 Normalizing Superpositions

It’s a mathematical fact that any two eigenstates with different
eigenvalues (of any measurable, including energy) are ORTHOGONAL

Meaning:

 ( x) 2 ( x)dx  0
*
1

To normalize a superposition of normalized energy eigenstates,

make the sum of the absolute squares of their coefficients equal 1.

If   a 1  b 2
where  1and  2 have different eigenvalues then

 ( x) dx   (a 1  b 2 ) dx  a  b
2 2 2 2

|a|2 is the probability that the particle would be found in state “1”
|b|2 is the probability that the particle would be found in state “2”
|a|2 + |b|2 = 1
|a|2 and |b|2 don’t change in time because 1 and 2 are energy eigenstates!
 Lecture 9, Act 2
Consider a particle in an infinite potential
well, which at t= 0 is in the state:
 ( x, t )  0.5 2 ( x)  A 2 4 ( x)
0 L x
with 2(x) and 4(x) both normalized.
1. What is A2?
a. 0.5 b. 0.707 c. 0.866

2. At some later time t, what is the probability density at the exact

center of the well?
a. 0 b. 1 c. It depends on the precise time t.
 Lecture 9, Act 2
Consider a particle in an infinite potential
well, which at t= 0 is in the state:
 ( x, t )  0.5 2 ( x)  A 2 4 ( x)
0 L x
with 2(x) and 4(x) both normalized.
1. What is A2?
a. 0.5 b. 0.707 c. 0.866
As stated, the question is not well posed, since A 2 could be complex.
However, let’s assume that A2 is real (or that we were asked for |A2|).
We are told that 2(x) and 4(x) are both normalized. Therefore:
0.52 + |A2|2 = 1, or |A2| = sqrt(1 – 0.25) = sqrt(0.75) = 0.866
2. At some later time t, what is the probability density at the exact
center of the well?
a. 0 b. 1 c. It depends on the precise time t.
 Lecture 9, Act 2
Consider a particle in an infinite potential
well, which at t= 0 is in the state:
 ( x, t )  0.5 2 ( x)  A 2 4 ( x)
0 L x
with 2(x) and 4(x) both normalized.
1. What is A2?
a. 0.5 b. 0.707 c. 0.866
As stated, the question is not well posed, since A 2 could be complex.
However, let’s assume that A2 is real (or that we were asked for |A2|).
We are told that 2(x) and 4(x) are both normalized. Therefore:
0.52 + |A2|2 = 1, or |A2| = sqrt(1 – 0.25) = sqrt(0.75) = 0.866
2. At some later time t, what is the probability density at the exact
center of the well?
a. 0 b. 1 c. It depends on the precise time t.
In general, the probability distribution of a superposition of energy eigenstates
does depend on time. However, both of these solutions always have a node at
L/2. Therefore, every possible superposition of them also has a node at L/2.
 Measurements of E or x

The important new result concerning superpositions of
energy eigenstates is that these superpositions represent
quantum particles that are moving. Consider:
 ( x, t )  A1 1 ( x)e  i1t  A 2 2 ( x)e  i 2t

But what happens if we try to measure E on a wavefunction
which involves more than one energy?

We can still only measure one of the allowed energies,
i.e., one of the eigenstate energies (e.g., only E1 or E2 in (x,t) above)!

If (x,t) is normalized, |A1|2 and |A2|2give us the probabilities that

energies E1 and E2, respectively, will be measured in an experiment!
(x,t)

What about measurements of |(x,t)| ? 2

If we make a large # of measurements at

time t, the result should look like the
probability function |(x,t)|2 at that time.
0 L x
 Lecture 9, Act 3
Consider a particle in an infinite potential well, which at t = 0 is in the
state:

 ( x, t )  0.5 ( x)  0.866 4 ( x)
with 2(x) and 4(x) both2normalized. 0 L x
1. We now measure the energy of the particle. What value is observed?
a. E2 b. E4 c. 0.25 E2 + 0.75 E4
d. It depends on when we measure the energy.

2. If E2 is observed, what is the state of the particle after the measurement?

 Lecture 9, Act 3
Consider a particle in an infinite potential
well, which at t = 0 is in the state:
 ( x, t )  0.5 2 ( x)  0.866 4 ( x)
0 L x
with 2(x) and 4(x) both normalized.
1. We now measure the energy of the particle. What value is observed?
a. E2 b. E4 c. 0.25 E2 + 0.75 E4
d. It depends on when we measure the energy.
Since we are measuring energy, we can only get one of the eigen-energies,
E2 or E4. The probability of measuring E2 is 25%. The probability of measuring
E4 is 75%.
The average energy (if we were to measure a large ensemble of similar
particles) is the weighted sum of the energies: 0.25 E2 + 0.75 E4
2. If E2 is observed, what is the state of the particle after the
measurement?
 Lecture 9, Act 3
Consider a particle in an infinite potential
well, which at t = 0 is in the state:
 ( x, t )  0.5 2 ( x)  0.866 4 ( x)
0 L x
with 2(x) and 4(x) both normalized.
1. We now measure the energy of the particle. What value is observed?
a. E2 b. E4 c. 0.25 E2 + 0.75 E4
d. It depends on when we measure the energy.
Since we are measuring energy, we can only get one of the eigen-energies,
E2 or E4. The probability of measuring E2 is 25%. The probability of measuring
E4 is 75%.
The average energy (if we were to measure a large ensemble of similar
particles) is the weighted sum of the energies: 0.25 E2 + 0.75 E4
2. If E2 is observed, what is the state of the particle after the
measurement?
Our measurement has “collapsed” the original wavefunction. It is now simply
in the eigenstate associated with the measurement result E 2: 2(x)
 Schrödinger’s Cat:
How far do we take superpositions?

We now know that we can put a quantum object into a superposition of states.

We also know that we can “measure” it. But if quantum mechanics is completely
correct, shouldn’t our measurement apparatus end up in a superposition state too?

This result is best exemplified by the famous “Schrödinger’s cat” paradox


A radioactive nucleus can decay, emitting an alpha
particle.

The alpha particle is detected with a geiger
counter, whose firing releases a hammer, which
breaks a bottle, which releases cyanide, which kills
a cat.
 Schrödinger’s Cat

Strictly according to QM, because we
can’t know without measuring when the
decay happened, until we look inside the
box, the cat is in a superposition* of
being both alive and dead!

1
 cat    (alive)   (dead )
2 http://www.physics.uiuc.edu/Research/QI/Photonics/movies/cat.swf

And in fact, strictly according to QM, we then are put into a quantum superposition* of having seen a live and a dead cat!!

Where does it end?!?

it doesn’t end (“wavefunction of the universe”)

there is some change in physics (quantum  classical)

”many-worlds” interpretation…

In any event, the correlations to the rest of the system cause decoherence and the appearance of “collapse”.

* More correctly, the atom and the cat and us become “entangled”.
 Motion of a Free Particle
Wavefunction of a free particle with single K.E.

A free particle moves without applied forces; so we set U(x) = 0.

 2 d 2  ( x, t ) d  ( x, t )
  i  i  1
2m dx 2 dt

Traveling wave solution  ( x ,t )  Aei( kx t ) Wavefunction of free particle

p2
Check it. Take the derivatives: classicall y, E
 2k 2 2m
d   with p  momentum
 ik Aei ( kx t ) 2m
dx and E  kinetic energy
d 2
 (ik ) 2 Aei ( kx t )   k 2 Aei ( kx t ) From DeBroglie, p = ħk = h/.
dx 2
Now we see that E =
d
 (i )Aei ( kx t ) ħ= hf
dt So this is consistent.
 Wavepackets

The plane-wave wavefunction for a free particle is problematic

 ( x ,t )  Aei( kx t )

It describes a particle with well defined momentum, p = ħk, but completely uncertain position. It’s
equally likely to be anywhere!


By adding together (“superposing”) waves with a range of wave vectors k, we can produce a localized
wave packet. We can imagine such a packet in space:

x

We saw in Lecture 6 that the required spread in k-vectors (and by
p = ħk, momentum states) is determined by the Heisenberg
Uncertainty Principle: p·x ≈ ħ
We’ll now talk about states with some “k”, but really we mean packets with
a narrow range of k’s. They do have an (approximate) position so it means
something to say where “are”.
 Free particle motion

Consider a free particle that is moving in space

At any time it is located in some region of space ~ x

The particle must have a spread in momenta as required by the
uncertainty principle - p > h/x

The different components (different p’s, different E’s) of the wave
packet move at different velocities – they change phase at different
rates (f=E/h).

At a point where the components had been in phase, they won’t be any
more.

At a point where they had been out of phase, they will become in phase.

x x x

Position of maximum probability moves

and the width of probability distribution spreads out

NOTE: This only happens for massive particles, not photons 

 FYI: Time-Energy Uncertainty Principle

Now that we are considering time-dependent problems, it is a good time to introduce another application of the Heisenberg Uncertainty Principle, based on measurements of energy and time. We start
from our previous result:


Sometimes this is further transformed as follows:


The last line is a standard result from Fourier wave analysis; this should not surprise us – the Uncertainty Principle arises simply because particles behave as waves!

 x 
p x     cp       E t  
 c 

E t       t  
 t  1  f t  1/ 2
 FYI:E t Uncertainty Principle Example*
A particular optical fiber transmits light over the range 1300-1600 nm (corresponding to a frequency range
of 2.3x1014 Hz to 1.9x1014 Hz). How long (approximately) is the shortest pulse that can propagate down this
fiber?

*This problem obviously does not require “quantum mechanics” per se. However,
due to the Correspondence Principle, the quantum constraints on single photons
also apply at the classical-pulse level.
 FYI:E t Uncertainty Principle Example*
A particular optical fiber transmits light over the range 1300-1600 nm (corresponding to a frequency range
of 2.3x1014 Hz to 1.9x1014 Hz). How long (approximately) is the shortest pulse that can propagate down this
fiber?

 t  1  2 f t  1
t  1/ 2 f
 1/(2 0.4 10 Hz) 14

 4  1015 s  4 fs
Note: This means the upper limit to data transmission is
~1/(4fs) = 2.5x1014 bits/second = 250 Gb/s
*This problem obviously does not require “quantum mechanics” per se. However,
due to the Correspondence Principle, the quantum constraints on single photons
also apply at the classical-pulse level.
 Supplement: Quantum Information References

Quantum computing

employs superpositions of quantum states - astoundingly good for certain
parallel computation, may use entangled states for error checking

www.newscientist.com/nsplus/insight/quantum/48.html

http://www.cs.caltech.edu/~westside/quantum-intro.html


Quantum cryptography

employs single photons or entangled pairs of photons to generate a secret
key. Can determine if there has been eavesdropping in information transfer

cam.qubit.org/articles/crypto/quantum.php

library.lanl.gov/cgi-bin/getfile?00783355.pdf

Quantum teleportation

employs an entangled state to produce an exact replica of a third quantum state at a
different point in space

www.research.ibm.com/quantuminfo/teleportation

www.quantum.univie.ac.at/research/photonentangle/teleport/index.html

Scientific American, April 2000
 Supplement: Free particle motion

It turns out (next slide) that the constructive interference region
for a matter wavepacket moves at the “group velocity”
v = h/m = p/m


So there’s a simple correspondence between the quantum picture and
our classical picture of particles moving around with momentum
p = mv.


But the quantum packet will spread out in the long run, since it has a
range of p, so the correspondence is never perfect.

x x x

Position of maximum probability moves

and the width of probability distribution spreads out
 Supplement: Group velocity

Say a wave-packet starts out at x=0 at t=0.

meaning each harmonic component has the same phase there.

After time t

the harmonic component at 1 will have changed phase by t1

the harmonic component at 2 will have changed phase by t2

The phase difference between these components at x=0
will now be t(2- 1)
To find the point x where they’re in phase, we need to find where
the phase difference from moving downstream by x cancels that:

t(2- 1)=x(k2-k1) Or for small differences in ,k: td=xdk

Result
vg=x/t=d/dk

In this case vg=p/m