My job requires a fair amount of frequency identification, so I've gotten reasonably acquainted with Fourier analysis. Last week someone made the statement that a "Prony analysis" might be better to sort out two close frequencies (117 Hz and 120 Hz) in a five second record sampled at 2000 Hz.
I started to formulate an example sheet for this analysis, using as an example a known signal. When Prime 3.0 began to frustrate me (declaring both "Find" and "Minerr" as undefined variables after the solve block ha been happily computing for some time,) I switched to version 15 (old tried and true.)
Now in addition to not being able to make sense of what Prony was doing I have two sets of answers for what is (as closely as I could mimic) the same set of math.
Attached are two files; the Prime file lays out the Prony method as found in the literature. The version 15 file is a quickly generated version of the Prime file, I've tried to keep the notation the same for ease of comparison. Neither file is correct. Neither file properly finds the correct frequency (the version 15 file could find it if I told it what is was.
