Has anyone done any Fourier analysis in Arev?

http://www.falstad.com/fourier/

I've done it using AREV with *.wav file data, to try for Fourier based compression (storing only the sin/cosine coefficients instead of the *.wav sound itself), but the reconstructed sounds were not that accurate (audibly noisy / distorted). As ever, there's a trade-off between precision and time to process. I did work out an RBASIC arccos() function once, to do great circle distance calculations between a zip code latitude/longitude and a "nearest" branch office lat/long, if that is any use to you:

temp=¹⁾ +

(cos(office.lat/57.2958) * (cos(zip.lat/57.2958)*cos²⁾))

negtemp=temp*-1

ArcCos=atan³⁾) + 2 * atan(1)

I've got my own arc cosine function:

FUNCTION ACOS(X)

*

* This function uses ATAN to return the arc cosine of the argument

IF X=0 THEN RETURN 90

RETURN_VALUE=ATAN⁴⁾/X)

IF RETURN_VALUE LT 0 THEN RETURN_VALUE += 180

RETURN RETURN_VALUE

You didn't actually mention what you wanted to do by way of Fourier analysis. What's the application?

I own a weather station and have accumulated about a year and a half of weather data. I want to find patterns in the barometric pressure. There are obvious patterns like the daily and twice daily cycles that according to meteorologists are related to heating. But I'm interested in other patterns, for example tidal effects and longer term periods that may relate to the passing of fronts, etc.

I expect some cycles to be unpredictable and chaotic, but I expect others to recur regularly.

There may be a simpler way to ascertain trends.

Take the data as an array of numbers.

Set your "transformation" function as "Reduce every nth array value by x" and then compress the data using pkzip. If the resulting zip file is smaller then keep that iteration/transformation and repeat the function on the modified data using new values of x and n.

View this thread on the forum...