[net.math] Discrete Fourier Xform Help Wanted

dont@tekig1.UUCP (Don Taylor) (11/29/84)

X
     If you are transforming a real valued time domain function, it is
possible to take advantage of the lack of imaginary values to either do
two transforms at the same time, or do one transform in ~half the time.
Or, at least so say the texts.  With the close correspondence between
the transform and its' inverse, can anyone point out a way to do a
transform and an inverse transform at the same time, with the same
number of operations, assuming real valued time domain data?  I might
even consider sacrificing phase information in the frequency domain,
if it resulted in a significant improvement.  The situation is that a
time domain signal is transformed, modified, and inverse transformed
back to the time domain.  It is done over and over and I am looking for
a way to overlay one transform with the inverse transform of the previous
data set.

Thanks in advance

Don Taylor
tektronix!tekig1!dont
(machine going down for a few days around 12/1)