[OT] Fourier analysis question using Gnumeric

[Mike]

On Wed, Sep 26, 2012 at 1:10 AM, Walter Dnes <waltdnes-SLHPyeZ9y/tg9hUCZPvPmw at public.gmane.org> wrote:
>   This is a personal project, not a homework assignment<G>.  I have a
> rather noisy series of approximately 24,000 data points.  Gnumeric's FFT
> analysis insists on a whole-number power of 2, so I'll use the last
> 16,384 data points.  I highlighted the data, and in the menu selected...
>
> ==> Statistics ==> Dependant observations ==> Fourier analysis
>
>   The final output was 2 columns ("Real" and "Imaginary") of 16,384
> rows of numbers.  Now what?  How do I interpret the output?  Anybody
> here with some stats knowledge?
>

Oi. Stats <=~=> Spectral analysis. Once I resigned myself to this (a
while ago), I breathed a sigh of relief and went about  my day :-)

First, you might want to pad your 24000 data to 32768 with
uninteresting data (zero etc.), if you care about having it all
counted.
Second, you can obtain amplitude and phase from your re,im complex output:
  amplitude = sqrt (re^2 + im*2)
  phase = arctan (im / re)
which tends to be a bit more physically meaningful than the complex data.

Cheers,
Mike
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists