I'm following the official LIGO tutorial and I want to do a whitening process now. I computed FFT using hf=Fourier[data] then PSD using psd=PeriodogramArray[data, Round[1/dt], 1, HannWindow]. In next step I interpolated PSD inter_psd=Interpolation[psd]. In tutorial they divide hf/(np.sqrt(interp_psd(freqs) /dt/2.)) where freqs=np.fft.rfftfreq(Nt, dt), dt=1/4096, Nt=131072. So, as I understand, they divide only by the ASD values for the discrete Fourier transform sample frequencies. Is this possible to obtain those freqs in some of Mathematica functions? Or is there another possibilty to whiten the data?
Asked
Active
Viewed 437 times
3
astrotom94
- 41
- 2
-
1I have slides here that have an example of subtracting an estimated dominant sinusoid from data, in effect whitening, in order to get the next frequency component estimated. Not positive this will help but it looks like you are wanting to do something similar. – Daniel Lichtblau Aug 01 '16 at 21:40