Closed dizcza closed 4 years ago
Hi @dizcza,
Thanks for the pull request! I will take a look at these as soon as I can.
A couple of quick comments:
There are a lot of different changes here. Some stylistic and some more functional. It would be helpful to maybe break this up into a couple of PRs with some justification for why you think they should be changes. For example, f-strings, while nice, are a stylistic change and also requiring a whole new minor version of python. I do like f-strings an I'm inclined to be okay with it, but it's sort of fundamentally different than the other stuff.
While I agree that computing the entire cross-spectral matrix with negative frequencies isn't ideal, one major thing is pairwise_spectral_granger_prediction
, and possibly a couple others depends on a cross-spectral matrix with negative frequencies. So using the rfft everywhere will break these. It is probably a better move to compute the full cross-spectral matrix only when needed and otherwise compute only the one with non-negative frequencies.
Hi @dizcza --
Could we cherry pick these commits you made:
I could do them myself, but since they are your work, and I would like them in the library....
Ok, I'll do it later
Thank you!
I'm thinking of using your package in https://github.com/NeuralEnsemble/elephant. To learn your package, I started with boiling down the complexity of decorators and properties. Here is a list of updates for you to consider to squash and merge.
_protected
duplicated fields in Multitaper.keys()
function that triggers error when you dodir(multitaper)
ordir(connectivity)
.power()
to.power
)