Re: FFT and 'uneven' time data

[Martijn Brouwer]

Hi Robert,
FFT assumes on an equidistant grid. That is why you need to specify a
frequency axis. If your data are not equidistant you have three choices:
- resample your data
- calculate the Fourier transform using its definiation (slow)
- cut you data in part that have an equidistant axis and add the partial
FFT spectra.

Martijn


On Mon, 2009-04-20 at 10:00 -0700, address@hidden wrote:
> How do I easily handle obtaining an FFT when the data is NOT evenly spaced
> in time?
> 
> I wrote my own program to create a new variable with evenly spaced data
> using linear interpolation, which works fine, except it does not decimate,
> it only adds in so the smallest step becomes the final step [obviously
> creating a huge data base], plus with data that has jitter, does not work
> as well.
> 
> Is there some way to obtain an FFT directly given the time base of the
> data is not evenly spaced *and* contains jitter?
> 
> Robert
> 
> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
>