I am trying to discover what the error of the FFT implementation within Maple is.
Here is the problem:
f=some array of values
f2=iFFT( FFT( f))
Error = Absolute value (f-f2) point by point.
For example, within the IDL programming environment, Error tends to lie between 10^-14 and 10^-16 for double precision calculations, independent of array size.
Normally I would just calculate the array, then plot the results. I am having trouble getting Maple to plot an array of values, since data analysis seems not to be one of the functions that this program was designed for.
I have also been unable to find any sort of general discussion on the errors inherent with the standard algorithms, though it appears that these errors should be within factors of 10 of the minimum number on the machine.
Any suggestions, or ideas of where I can look to find some answers?