auto-corelation with unevenly time samples

[jecampagne]

Hi,
Is there a way to use nifty-ls to compute auto-correlation of for instance unevenly samples of astronomical light curves (as well as cross correlation of lightcurves with different time sampling)?
Best

[lgarrison]

I don't think one would use nifty-ls directly for this, as it just computes Lomb-Scargle periodograms, rather than power spectra or correlation functions. But if you wanted to compute correlation functions in Fourier space, you could certainly use finufft, which underpins nifty-ls. I think you would use a type-1 NUFFT to transform the non-uniform time samples to a uniform Fourier grid, take the conjugate product in the usual way, then either use a uniform FFT to transform back to a uniform time sampling, or a type-2 NUFFT to transform from the uniform Fourier grid to the desired arbitrary lags.

[jecampagne]

oh, thanks for the nice adwise. I will try and let you know. May be it is more appropriate in an other github. Let me know.