MathGroup Archive 1996

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Fourier? (Q)

  • To: mathgroup at
  • Subject: [mg3614] Re: Fourier? (Q)
  • From: ianc (Ian Collier)
  • Date: Thu, 28 Mar 1996 00:10:56 -0500
  • Organization: Wolfram Research, Inc.
  • Sender: owner-wri-mathgroup at

In article <4j7hq8$af4 at>, mrsma at wrote:

> greetings:
> when i tried to go through the Fourier examples as per pgs 679-681 in
the Mma bible;
> Mma kindly returned the following:
> In[65]:=
> ?Fourier
> Fourier[list] finds the discrete Fourier transform of a list
>    of complex numbers.
> In[66]:=
> data={1,1,1,1,-1,-1,-1,-1}
> Out[66]=
> {1, 1, 1, 1, -1, -1, -1, -1}
> In[67]:=
> Fourier[data]
> Out[67]=
> Fourier[{1, 1, 1, 1, -1, -1, -1, -1}]
> i then tried the example on a PowerMac, NeXT, RS6000 and got the same reply.
> what am i missing here?
> sincerely,
> m. r.

The command Fourier in Version 2.2 has slightly different behavior than it
did in Version 2.1. If all the elements in a list are exact numbers, then
Fourier will return the input unchanged. However, if even one element is
inexact, then the transform will be calculated. It will also allow numeric
quantities that are exact numbers if N will turn them into inexact numbers;
however, it will still need at least one element to be inexact in order to
recognize the input as inexact. This is different than in Versions 2.0 and
2.1, and hence some of the examples on page 680 of the Mathematica book
(that shipped with Version 2.2) will not work as given.

In[1]:= Fourier[{-1, -1, 1, 1, 1}]

Out[1]= Fourier[{-1, -1, 1, 1, 1}]

In[2]:= Fourier[{-1, -1, 1, 1, 1.}]

Out[2]= {0.447214 + 0. I, -1.17082 - 0.850651 I, -0.17082 - 0.525731 I,

>    -0.17082 + 0.525731 I, -1.17082 + 0.850651 I}

In[3]:= Fourier[{Cos[Pi], Cos[2 Pi / 5], 1, 1}]

                         2 Pi
Out[3]= Fourier[{-1, Cos[----], 1, 1}]

In[4]:= Fourier[{Cos[Pi], Cos[2 Pi / 5], 1, 1.}]

Out[4]= {0.654508 + 0. I, -1. - 0.345492 I, -0.654508 + 0. I,

>    -1. + 0.345492 I}

This and many other frequently asked questions about Mathematica,
is addressed in the Technical Support FAQ area of the Wolfram
Research Web pages <>.

Here is the entry for Fourier:


Why does Fourier return what I typed in?

This change is mentioned in the most recent documentation, but was
inadvertently omitted from the documentation that was included with the
earliest releases of Version 2.2.

In the design of Mathematica, Fourier is currently considered a symbolic
function rather than a numerical function, and as such it is inconsistent
for Fourier to artificially lower the precision of exact arguments. The same
is true of functions like Sin and Cos, which are also considered symbolic
functions. Sin[3/2], for example, does not convert the exact number 3/2 to
the inexact number 1.5 and proceed to return an inexact result. The exact
expression Sin[3/2] is useful by itself, and it is considered undesirable to
convert it to something inexact. The same design principle has been used for

The behavior of Fourier is simply a matter of choosing whether the design
should be based on the abstract mathematical fact that Fourier is meaningful
(if perhaps a bit impractical) for exact input, or on the practical fact
that the algorithm and the most common applications are strictly numerical.

You can restore Version 2.1 behavior by adding the following rule for

In[5]:= Unprotect[Fourier]

In[6]:= Fourier[e_] := With[{ne = N[e]},
        Fourier[ne] /; Precision[ne] =!= Infinity] /;
                                     Precision[e] === Infinity

In[7]:= Protect[Fourier]

The change in Fourier was debated in the technical group, and the decision
was made to make the change. We have, however, been keeping track of
comments from users, and it is entirely possible that this change will be
reversed or otherwise modified for some future release of Mathematica.


I hope this helps.


Ian Collier
Wolfram Research, Inc.
tel:(217) 398-0700   fax:(217) 398-0747    ianc at
Wolfram Research Home Page:


  • Prev by Date: Re: addressing matrix elements
  • Next by Date: Re: Plot problem
  • Previous by thread: Fourier? (Q)
  • Next by thread: Re: Fourier? (Q)