On "cleaning" residual values from complex matrices

[Roberto Hernandez]

Hello everyone,

I've been doing a lot of processing with FFT. While the magnitude of the 
results I've been getting makes perfect sense, I noticed the phase was 
totally incongruent. After looking into the problem, I realized it boils 
down to residual values. What's happening is that values that should be 
zero show up as something like +/-eps +/-eps*i, so that throws off any 
phase calculations.

Is there any function that "cleans" residual values (makes them zero)? I 
looked up "precision" in the help files and found information on the 
built in variables 'realmin', 'realmax' and 'eps'. However, I found no 
"cleaning" functions.

If it's not available, I don't have a problem with writing my own. I'm 
not an expert in numerical analysis, so I figure it's very likely that 
I'll overlook something. My idea is to check for values where the 
real/imaginary part is <= 'eps' and truncating the real/imaginary part 
for those values to zero.

If there is already a function that does this, please let me know. 
Otherwise, if I do have to write my own, is there anything else the 
function should do besides what I already mentioned?

TIA,

Roberto




-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------