search

serazing / xscale

Xscale a library of multi-dimensional signal processing tools using parallel computing
Apache License 2.0
25 stars 12 forks source link

Adding documentation in xscale.fft to explain the use of zero-padding #11

Open stephanieleroux opened 6 years ago

stephanieleroux commented 6 years ago

Hi Guillaume, I am using your fft routines. Very convenient to process gridded data! Just wondering if you have implemented yet some option to pad with zeros both ends of the data before performing the fft?

serazing commented 6 years ago

This should work if you precise the nfft argument and you use a number larger than your dataset dimensions, as done in dask.array.fft

I 've just noticed that nfft is missing in the current docstring of xscale.fft.fft. This should be corrected and mentioned in the documentation. If you're willing to contribute to the documentation, it will be very welcome !

I have also implemented fitting methods xscale.signal.fitting but they require more work as You should be able to remove a linear trend on multi-dimensional dataset using xscale.signal.fitting.detrend. These functions require however additional testing.

stephanieleroux commented 6 years ago

ok, great. I know the nfft argument from the scipy.signal package but i had not seen it in your doc indeed. I will test this and try to contribute to the doc asap.

I haven't tried any detrending methods yet as my current dataset is already detrended, but yeah, i'll let you know when i have used/tested it.