Justin Young

young.1022@osu.edu
The Ohio State University
Department of Physics

Performing Fourier Transforms in Mathematica

Mathematica is one of many numerical software packages that offers
support for Fast Fourier Transform algorithms. You can perform
manipulations with discrete data that you have collected in the laboratory, as
well as with continuous, analytical functions. This tutorial introduces some of
the common functions used to perform these calculations as well as some
basic concepts and caveats that are specific to Mathematica.
Load data file into an array
It is often the case that you have a text file that contains a number of discrete data points that you
have collected and need to analyze. This data file may be a comma-, space-, or tab- delimited
(.csv) file, text file (.txt), or any other file supported by Mathematica. (Run the "$ImportFormats"
command to see all supported data types.) In this tutorial, we will use the "Import" command to
import an existing text file that contains some electric field data into an array (Mathematica calls
arrays 'lists'). You will need to specify the path to your data file.
Important! If your data is in a non-standard format, you might need to modify the way this notebook
imports your data. The if statement below checks how the data file is structured and modifies the
array appropriately in order for the rest of this notebook to work. This code assumes you have a
data file with a separate row for each data point. It can handle data points specified as real field
values at each time, in two columns, as well as data that has been separated into real and
imaginary components at each time, in three columns. If your data file is structured differently, you
will have to modify how the data arrays are being referenced throughout the rest of the notebook.
This is not difficult--if you have an array called myArray, Mathematica references elements of the
array as myArray[[4]], for the 4th entry, for example (it does NOT use a 0-based index). For multidimensional arrays, you can reference elements using myArray[[4]][[2]], in order to get the 2nd
element of the 4th element of the array, for example.
data = Import@"sample.txt", "Table"D;
n = Length@dataD;
If@Length@data@@1DDD == 3,
data = Table@8data@@iDD@@1DD, data@@iDD@@2DD + * data@@iDD@@3DD<, 8i, 1, n<DD;

MyTutorial.nb

Plot the waveform in the time domain

It is useful to plot the data both to get an idea of what it looks like in the time domain and to make
sure it loaded properly.
Plot the real part of the electric field E(t) by providing an array of (t,Re(E(t))) data points and
passing it to the ListPlot function. The array is created using the Table function. (If you are familar
with basic programming structures, Table is essentially a 'for' loop.)
In[139]:=

ListPlot@Table@8data@@iDD@@1DD, Re@data@@iDD@@2DDD<, 8i, 1, n<D,

Joined True, PlotRange All, FrameLabel 8"t", "Re@EHtLD"<, Frame TrueD
1.0
0.8

Out[139]=

Re@EHtLD

0.6
0.4
0.2
0.0
-0.2
-10

-5

0
t

10

Plot the imaginary part of the electric field.

In[140]:=

ListPlot@Table@8data@@iDD@@1DD, Im@data@@iDD@@2DDD<, 8i, 1, n<D,

Joined True, PlotRange All, FrameLabel 8"t", "Im@EHtLD"<, Frame TrueD
0.0

Out[140]=

Im@EHtLD

-0.2

-0.4

-0.6

-10

-5

0
t

10

MyTutorial.nb

Plot the intensity of the field.

In[141]:=

ListPlotATableA9data@@iDD@@1DD, Abs@data@@iDD@@2DDD2 =, 8i, 1, n<E,

Joined True, PlotRange All, FrameLabel 8"t", "Intensity"<, Frame TrueE
1.0

Out[141]=

Intensity

0.8

0.6

0.4

0.2

0.0
-10

-5

0
t

10

Transform the data into the frequency domain

Find the temporal spacing between data points. In this example, the spacing is the same between
all the data points.
In[142]:=
Out[142]=

Dt = Hdata@@2DD@@1DD - data@@1DD@@1DDL
0.1

Create an array of the field amplitudes to pass to the Fourier function.

In[143]:=

fieldValues = Table@data@@iDD@@2DD, 8i, 1, n<D;

Compute the Fourier transform E() using the built-in function. Important! The sample data array is
ordered from negative times to positive times. However, Mathematica requires that the array
passed to the Fourier function be ordered starting with the t=0 element, ascending to positive time
elements, then negative time elements. To account for this, we run the RotateLeft command on our
data array. The array that results after taking the transform is ordered starting from zero-frequency,
increasing to the maximum frequency; then it jumps to the most negative frequency, and increases
to zero. Therefore, when we want to plot the result of the transform in frequency space, we need to
shift the array again after the transform. We use RotateRight (by the same amount) before plotting,
so that our data is ordered from negative frequencies to positive frequencies, with the zero
frequency component in the middle.
In[144]:=

FFTValues = Fourier@RotateLeft@fieldValues, n 2 - 1D, FourierParameters 81, - 1<D;

FFTValues = RotateRight@FFTValues, n 2 - 1D;

Find the spectral spacing (in Hz).

1
In[146]:=

D =
n Dt

Out[146]=

0.0390625

Create a list of frequency-axis values (in angular Hz) to use in the plot of E().

MyTutorial.nb

Create a list of frequency-axis values (in angular Hz) to use in the plot of E().
In[147]:=

frequencyValues = Table@D * Hi - n 2L, 8i, 1, n<D;

Plot the waveform in the frequency domain

Plot the real part of E().
In[148]:=

ListPlot@Table@8frequencyValues@@iDD, Re@FFTValues@@iDDD<, 8i, 1, n<D,

Joined True, PlotRange All, FrameLabel 8"", "Re@EHLD"<, Frame TrueD

Out[148]=

Re@EHLD

10

-4

-2

Plot the imaginary part.

In[149]:=

ListPlot@Table@8frequencyValues@@iDD, Im@FFTValues@@iDDD<, 8i, 1, n<D,

Joined True, PlotRange All, FrameLabel 8"", "Im@EHLD"<, Frame TrueD

Out[149]=

Im@EHLD

-5

-10
-4

-2

MyTutorial.nb

Plot the frequency power intensity spectrum.

In[150]:=

ListPlotATableA9frequencyValues@@iDD, Abs@FFTValues@@iDDD2 =, 8i, 1, n<E,

PlotRange All, Joined True, FrameLabel 8"", "Intensity"<, Frame TrueE
200

Out[150]=

Intensity

150

100

50

0
-4

-2

Plot the spectral phase of the waveform versus frequency

Since complex values have an associated phase in the complex plane, we want to plot the spectral
phase as a function of frequency. However, the Arg(z) function is periodic in multiples of 2.
Therefore, we need to 'unwrap' the angles. Unfortunately, Mathematica does not provide a built-in
function to do this. Here is some short code that provides this functionality, returns the phase
values in an array, and plots the values.
In[151]:=

phase = Arg@FFTValuesD;
phase = FoldList@Round@H1 - 2L H2 PiLD 2 Pi + 2 &, First@phaseD, Rest@phaseDD;
ListPlot@Table@8frequencyValues@@iDD, phase@@iDD<, 8i, 1, n<D,
Joined True, PlotRange All, FrameLabel 8"", ""<, Frame TrueD

-5

-10

Out[153]=

-15

-20

-25
-4

-2

MyTutorial.nb

Inverse transform back into the time domain

Perform a reverse Fourier transform (remembering to shift the array before the transform, since we
plotted it above ordered from negative to positive values!).
In[154]:=

originalFFTValues =
InverseFourier@RotateLeft@FFTValues, n 2 - 1D, FourierParameters 81, - 1<D;

Plot the intensity versus time (remembering to shift the array before plotting...you should be good
at this by now)
In[155]:=

originalPulse = RotateRight@originalFFTValues, n 2 - 1D;

ListPlotATableA9data@@iDD@@1DD, Abs@originalPulse@@iDDD2 =, 8i, 1, n<E,
Joined True, PlotRange All, FrameLabel 8"t", "Intensity"<, Frame TrueE
1.0

0.8

Intensity

Out[156]=

0.6

0.4

0.2

0.0
-10

0
t

-5

10

As expected, we get the same pulse back as before, since we did not apply any transfer function to
it in the frequency domain. This is a good test to see if your software is working.

Working with analytic functions

You can take the Fourier transform of an analytic function as follows ...
In[157]:=

FourierTransformAExpA- a t2 E, t, , FourierParameters 81, - 1<E

-

4a

Out[157]=

... as well as its inverse

-

In[158]:=

4a

InverseFourierTransformB

, , t, FourierParameters 81, - 1<F

a

Out[158]=

-a t

You can use these methods to create discrete data sets from a given analytic expression, for
example, by sampling at specified times and frequencies.

MyTutorial.nb

You can use these methods to create discrete data sets from a given analytic expression, for
example, by sampling at specified times and frequencies.

Possible Issues
Here are some common issues you might run into when using Mathematica, as well as general
things to keep in mind when performing Fourier transforms.

When working with analytic functions, be aware of the sign convention used for plane waves traveling to
the left and those traveling to the right.
You are free to choose the sign convention you prefer when distinguishing between forward and reverse
transforms, as long as you are consistent. Mathematica' s convention is different from that used in class,
but you can force Mathematica to use a specified convention by setting the FourierParameters property in
the Fourier and InverseFourier functions. I have set my transforms to use the sign convention used in
class. See the documentation for more information.
When generating sample data to work with, make sure your sample window is large enough in order to
accomodate the entire signal of interest. This may involve using more or less data points, or setting
different temporal and spectral spacings.
It is easy to get the array rotations mixed up when preparing your array for forward and reverse
transforms, and in plotting. However, it is easy to tell if your array ordering is incorrect; your data will
appear to be discontinuous. Perform another rotation and try plotting it again.
In order to avoid array rotation problems when you make a change in your notebook, such as in a transfer
function, it is useful to re-evaluate the entire notebook from start to end in order to properly reset your
arrays.
Make sure you are using frequency values or angular frequency values, as appropriate.
It is often convenient to work only with the pulse envelope. Therefore, be sure to distinguish between
instantaneous and carrier frequencies in your calculations.
Check the scale of your units to be sure everything matches up! (e.g. Hz vs. THz)

Summary
With these tools, you should now be able to perform forward and reverse Fourier transforms of
discrete and continuous data sets in Mathematica. You have seen that the Table function is one of
the most useful functions in Mathematica to create and manipulate arrays. You have also seen
some of the ways that you can visualize your data, by way of examining the magnitudes and
phases of complex-valued fields, as well as their frequency spectra. The best way to learn about
Mathematica's capabilities is to simply start using the software and try as many examples as
possible. The documentation is quite helpful when you need a description of what certain functions
do, what their assumptions are, and what parameters they can take. Most manual pages give you
examples to work with that you can modify for your own purposes. The web is also a great
resource for finding examples. Have fun transfoming, please email me if you find any errors in this
document, and remember, transforms are just robots in disguise!