Hi,
We make heavy use of the FFTSpa function in our CR1000 program, and are therefore highly interested in any method of speeding up that operation. We are only interested in the real (amplitude) components of the resulting FFT. I've noted that one frequently used algorithm cleverly combines FFTs for two time series into one FFT operation to extract only real components, as detailed here: http://www.mathcs.org/java/programs/FFT/FFTInfo/c12-3.pdf - but this requires an FFT function which allows complex input. Is there some way this method, or one similar to it, could be implemented in CRBasic? We obviously do currently use the 1 option which outputs only the real amplitude values, but our testing shows that this is actually slightly slower than using the 0 option which outputs complex FFT points.
Speaking of the 0 option for outputting complex FFT points -- what is the correlation between those values and the amplitudes produced by the 1 option? The help file is very vague.
Thank you,
Benjamin