Why does power_spectral_density not call cross_spectral density?
I'm looking over the code to cross-check with the paper draft and I noticed that the logic is different between PSD estimation and CSD estimation (I think introduced in !27 (merged)), this is strange to me as they are fundamentally the same operations, we just often choose to treat them differently for historical reasons.
Can we change the code to make sure power_spectral_density
is just a wrapper to cross_spectral_density
? That will enable users to coarse-grain the PSDs or welch average the CSDs if they desire.
I'm happy to put together an MR on this if it is welcome. I remember there were some edge effects for zero-frequency and Nyquist bins that caused some issues.