From: Andrew Veliath <dre...@eideti.com>
Subject: Help with DFT/FFT
Date: 1995/04/27
Message-ID: <3nod5j$o40@ayrton.eideti.com>#1/1
X-Deja-AN: 101781716
distribution: inet
content-type: text/plain; charset=iso-8859-1
organization: eideti.com Systems Corp.
x-url: news:sci.math.num-analysis
mime-version: 1.0
newsgroups: sci.math.num-analysis
x-mailer: Mozilla 1.1b3 (X11; I; Linux 1.2.5 i486)

I am writing a signal analyzer and right now I am using the FFT from Numerical
Recipies.  However, I still don't have too much background in Fourier analysis
and the "right thing to do"--right now the program is displaying a complex
array, alternating even and odd coefficients.  This results in a somewhat
toothed appearance, so I am wondering if someone who knows much more about this
could help me with regards about what the best way to display coefficient data?

Also, I have a problem with the FFT routine from Num. Recipies, it is supposed
to accept a real array on half the size of the buffer or a complex array the
full size of the buffer.  I am giving it a quarter size real array right now
since if I give it a half-size array of data it gives a segmentation fault.
Has anyone had a problem with the FFT given in there?  I am using it but it was
sent to me by someone so I can't check it out and am trying to avoid debugging
it by hand if I have to.   When using the quarter size the second half of the
array clearly shows the untouched waveform so I have to scale it out right now,
but again the routine segvs if I give it the half-size.  

Also does anyone have any ideas on the best way to do logarithmic scaling for
amplitude?  Right now it works, but don't think it is very efficient.  For
frequency log scale I am using frequency indexing scheme which returns a
position and a length to determine rms values on the coefficient array.

I apologize for my ignorance, I know there are very knowledgeable people on
here! 

---------------------------------------------------------------------------
Best regards,                              Government RULES!!!
Andrew Veliath <dre...@eideti.com>        
http://eideti.com/~drewvel
---------------------------------------------------------------------------

From: Andrew Veliath <dre...@eideti.com>
Subject: Need Free FFT in C
Date: 1995/05/02
Message-ID: <3o48qs$bgn@ayrton.eideti.com>#1/1
X-Deja-AN: 101905108
distribution: inet
content-type: text/plain; charset=iso-8859-1
organization: eideti.com Systems Corp.
x-url: news:sci.math.num-analysis
mime-version: 1.0
newsgroups: sci.math.num-analysis
x-mailer: Mozilla 1.1b3 (X11; I; Linux 1.2.7 i486)

Does anyone know of a FREEly useable FFT written in C on the net?  I was just
informed of the copyright of the Numerical Recipies routines and had to remove
it from my program!

---------------------------------------------------------------------------
Best regards,                              Government RULES!!!
Andrew Veliath <dre...@eideti.com>        
http://eideti.com/~drewvel
---------------------------------------------------------------------------