[LuckyTran]

I'm familiar with how to compute correlation functions of two variables for a uniform grid via brute force or via the fft & ifft route. But I need help with doing the same correlations on a non-uniform spacing. This can be caused by either non-uniform sampling in time (because of an adaptive time-stepper) or when trying to compute spatial correlations on a modern non-uniform grid. Yes you can always resample your data via some smoother/interpolation scheme onto a uniform grid but that's not the question.

I don't know how the brute force definition needs to be changed to accommodate this but theoretically, in spectral spaces, we just need to correctly compute the Fourier transform. However,I don't have any experience doing this computation of a Fourier transform with non-uniform spacing.

edit* So I've learned that there exists many implementations of a non-uniform fft (NFFT or NUFFT) in the IEEE community (rightfully so). But the implementations I've run into so far are very cumbersome and hard to setup. If anyone has an implementation that is simple, please let me know. Meanwhile I'll be trying out some. My idea for a working implementation would be something along the lines of NFFT(data, coordinates).