Fourier Transformations & Understanding Uncertainty

Uncertainty Relation: Fourier Analysis of Wave Packets

Group 1: Brian Allgeier, Chris Browder, Brian Kim

(7 December 2015)

Introduction:
In Quantum Mechanics, wave functions offer statistical information about possible results, not
the exact outcome of an experiment. Two wave functions of interest are the position wave
function and the momentum wave function; these wave functions represent the components of a
state vector in the position basis and the momentum basis, respectively. These two can be related
through the Fourier Transform and the Inverse Fourier Transform. This demonstration
application will allow the users to define their own momentum wave function and use the inverse
Fourier transformation to find the position wave function. If the Fourier transform were to be
used on the resulting wave function, the result would then be the original momentum wave
function. Both wave functions visually show the wave packets in momentum and position space.
The momentum wave packet is a Gaussian while the corresponding position wave packet is a
Gaussian envelope which contains an internal oscillatory wave. The two wave packets are related
by the uncertainty principle, which states that the more defined (i.e. more localized) the wave
function is in one basis, the less defined the corresponding wave function will be in the other
basis. The application will also create an interactive 3-dimensional model that relates the two
wave functions through the inverse Fourier transformation, or more specifically, by visually
decomposing a Gaussian wave packet (in momentum space) into a specified number of
components waves (in position space). Using this application will give the user a stronger
knowledge of the relationship between the Fourier transform, inverse Fourier transform, and the
relationship between wave packets in momentum space and position space as determined by the
Uncertainty Principle.

Concept:
Wavelength and Position
In Griffith’s Introduction to Quantum Mechanics, Griffith brings up the following scenario: one
steadily oscillates a long rope and generates a standing wave, proceeding to ask the odd question,
“Where exactly is the wave?” This question is odd since the wave is not exactly anywhere; it is
spread out over the length of the rope. A more reasonable question would be, “What is its
wavelength?” However, if one were to give the same rope a sudden jerk to create a single,
narrow bump travelling down the rope, then the situation is reversed: one could answer for the
location of the wave, but it would be difficult to interpret what a wavelength would be. This is a
very basic model of the uncertainty relationship; one is able to achieve a well-defined
wavelength (localized in momentum space) or a well-defined position (localized in position
space) but not both at the same time.
Wave Packet
The position wave function is a superposition of potentially infinite sinusoidal waves. The
amplitude of a wave with a certain wave number 𝑘𝑘0 is given by the momentum wave function
evaluated at the corresponding point ℏ𝑘𝑘0 in momentum space. The position wave function will
contain internal “ripples” whose peaks and troughs form the shape of an external Gaussian
“envelope” or “group.” This is known as the wave packet. By using the 3-dimensional model
generated from the application, this relationship can be seen visually.

Uncertainty Principle
The uncertainty of measurements is related to the variance (𝜎𝜎 2 ) of the distribution. The general
uncertainty principle is defined as
2
1
2 2 � �
𝜎𝜎𝐴𝐴 𝜎𝜎𝐵𝐵 ≥ � 〈�𝐴𝐴, 𝐵𝐵 �〉�
2𝑖𝑖
̂ �
where 𝐴𝐴 and 𝐵𝐵 are the A and B hermitian operators respectively. The term �𝐴𝐴, � 𝐵𝐵� � = 𝐴𝐴̂𝐵𝐵� − 𝐵𝐵� 𝐴𝐴̂ is
the commutator of the two hermitian operators. When two operators commute, the commutator is
zero. Mathematically, this allows the product of the two variances to equal zero. However, when
the commutator is not zero, the two operators are incompatible observables. The uncertainty
principle is the result of the statistical interpretation of quantum mechanics.

The demonstration application focuses on two vector spaces: the momentum space and the
position space. By using the momentum and position operators, one is able to reproduce the
Heisenberg uncertainty principle that states that the more defined (i.e. localized) the momentum
or position wave function is in its respective basis, the less defined (i.e. more spread out) the
corresponding wave function of the other basis. By the canonical commutation relation, [𝑥𝑥�, 𝑝𝑝̂ ] =
𝑖𝑖ℏ, where 𝑥𝑥� is the position operator and 𝑝𝑝̂ is the momentum operator and ℏ is the reduced
Planck’s constant.
2 2
1 2
1 2
ℏ 2
𝜎𝜎𝑥𝑥 𝜎𝜎𝑝𝑝 ≥ � 〈[𝑥𝑥,
� 𝑝𝑝̂ ]〉� = � 𝑖𝑖ℏ� = � �
2𝑖𝑖 2𝑖𝑖 2

ℏ Heisenberg Uncertainty Principle

𝜎𝜎𝑥𝑥 𝜎𝜎𝑝𝑝 ≥
2

where 𝜎𝜎 is the standard deviation, defined as the uncertainty in the distributions.

Fourier Transform and Inverse Fourier Transform
The inverse Fourier transform can be operated on the momentum wave function to derive the
position wave function:
∞
1
Ψ(𝑥𝑥, 𝑡𝑡) = � 𝜙𝜙(𝑝𝑝, 𝑡𝑡)𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖/ℏ 𝑑𝑑𝑑𝑑
√2𝜋𝜋ℏ −∞
where Ψ(𝑥𝑥, 𝑡𝑡) is the position wave function and the 𝜙𝜙(𝑘𝑘) is the momentum wave function.

The Fourier transform can be used on the momentum wave function to derive the momentum
wave function:
∞
1
𝜙𝜙(𝑝𝑝, 𝑡𝑡) = � Ψ(𝑥𝑥, 𝑡𝑡)𝑒𝑒 −𝑖𝑖𝑖𝑖𝑖𝑖/ℏ 𝑑𝑑𝑑𝑑
√2𝜋𝜋ℏ −∞

Plancherel’s Theorem states that the inverse Fourier transform of a Fourier transform is the
starting function. Therefore, if either the momentum or position wave function is known, then
the other can be determined using the inverse Fourier transform and the Fourier transform,
respectively. Then, by using Fourier transform and the inverse Fourier transform, respectively,
the original wave function can be recovered. Through the use of these transformations, one can
freely change from momentum space to position space, or vice versa.
Demonstration Application of the Fourier Analysis of the Wave Packet and the Uncertainty
Relation

The demonstration application helps the user understand the concept of the Fourier transform
and inverse Fourier transform. After the momentum wave function is defined, the application
will produce the position wave function but also create a 3-dimensional model that combines the
momentum and the individual sinusoidal waves that are superimposed to create the wave
function through the relationship of the Fourier transform.

By defining a Gaussian momentum wave function (see Figure 1), one can use the demonstration
application to find the position wave function (see Figure 2) that is enveloped by a Gaussian
wave packet. This is performed by taking the defined momentum wave function and performing
the inverse Fourier transformation.

Figure 1: The defined momentum wave function

Figure 2: Position wave packet resulting from the inverse Fourier

transform of the defined momentum wave function. It can be seen that
the wave function is "grouped" together to make the wave packet.
The uncertainty of the momentum wave function is defined by the user and the uncertainty of the
position wave function will be calculated by the application. It is then shown that the product of
the uncertainty of the momentum and position wave function is greater than or equal to ℏ⁄2 (i.e.
Heisenberg uncertainty principle will be conserved).

In fact, for any Gaussian wave packet (i.e. any wave functions produced with this application),
the Uncertainty Principle holds to equality: 𝜎𝜎𝑥𝑥 𝜎𝜎𝑝𝑝 = ℏ⁄2. However, in the images above, the
product of the uncertainties shown is much larger than ℏ⁄2; this occurs because of the use of a
small finite number of component waves in the application simulation (this in turn allows for the
reappearance of identical wave packets at regular intervals in x-space, not seen in the image). In
order to increase the accuracy of the uncertainty product, larger numbers of component wave are
used (see images below); in reality, the Fourier transformation uses an infinite number of
infinitely-densely-spaced component waves, which can only be approximated in the simulation.

The application will produce a 3-D model that relate the momentum and position space through
the Fourier transform. The model will show the relationship of the momentum wave function and
the individual sinusoidal waves that forms the position wave function when superimposed
(Figure 3, 4, 5).
Figure 3: The 3-D model that relates the momentum
space and position space through the Fourier transform
Figure 4: The momentum wave function is made up of the
amplitudes of the individual waves that are superimposed to create
the wave function

Figure 5: The individual waves that are superimposed to create

the position wave function.
Examples of the Demonstration Application
Three examples will be shown with the application. Each example will minimize the
uncertainties while still conserving the uncertainty principle.

1. Momentum uncertainty and position uncertainty are the same:

Figure 6: A Gaussian momentum wave function is defined

and the application will perform the inverse Fourier

Figure 7: Many individual sinusoidal position waves

were used in this 3-D model. The amount of waves can
be changed in this application.
2. Momentum uncertainty is large and position uncertainty is small:

By giving the momentum wave function a large uncertainty (i.e. make the wave spread
out), the position wave packet is able to have a small uncertainty (i.e. localized).

Figure 8: The momentum wave was given a large uncertainty which

resulted in a position wave function with a small uncertainty. The
Heisenberg uncertainty principle is still conserved.
3. Momentum uncertainty is small and position uncertainty is large:

By giving the momentum wave function a small uncertainty (i.e. localized), the position
wave packet must have a large uncertainty (i.e. spread out).

Figure 9:The momentum wave was given a small uncertainty (to simulate
localization) which resulted in a position wave function with a large
uncertainty. The Heisenberg uncertainty principle is still conserved.
How to use the application

When one starts the program, it will give the user an introduction, the current distribution, and a
list of its functions (Figure 10).

Hello there! Welcome to

Fourier Transformations &
Understanding Uncertainty

Your current transformation is:

Gaussian k-distribution centered at 10 with sigma 1

showing 11 component waves, 5 < k < 15 & -5 < x < 5

To view your current transformation, just type "go".

To adjust the Gaussian center, type "cen=#" for center at
#.
To adjust the Gaussian sigma, type "sig=#" for a sigma of
#.
To adjust the number of wave components shown, type
"num=#".
To adjust k scale from a to b, type "klo=a" and "khi=b".
To adjust x scale from c to d, type "xlo=c" and "xhi=d".
To view probability distributions (|f|^2), type "sqd=1".
Separate multiple commands with a comma.
If you wanna quit, just say "bye".
>>>

Figure 10: The program’s introductory instructions

The momentum wave function can be determined by the functions “cen=#” and “sig=#”. To
determine range of k values to be viewed, use the functions “klo=#”, and “khi=#”.

For the position wave function, the range of x values to be viewed is determined with the
functions “xlo=#” and “xhi=#”. To control the number of individual sinusoidal waves that are
superimposed to form the wave function, use the function “num=#”.

To view the wave functions and the 3-D model of the Fourier Transform, one just needs to type
“go”.

Note: One can view the actual probability distributions (instead of just the amplitudes) with the
"sqd=1" command, as shown.
To exit the image window, just close the window. Then a new transformation can be defined.
When the user wants to quick the program, use the “bye” command.

Example
A Gaussian momentum wave function with a peak centered at the k value 15, a k value range
from 0 to 30, with 100 decomposed sinusoidal waves use the command shown on Figure 11.

>>> num=100, cen=15, klo=0, khi=30, go

Figure 11: The example command