These are chat archives for SHTOOLS/SHTOOLS

4th
Aug 2016
MMesch
@MMesch
Aug 04 2016 09:54
Hey guys, in the long run I think it would be nice to have an optional implementation that doesn't depend on fftw but only on the scipy fftpack library
it would be a bit slower but always available in python
maybe we can even directly use their fortran routines without much trouble
Ilya Oshchepkov
@ioshchepkov
Aug 04 2016 11:47
It seems like FFTPACK in scipy.fft is from 2 to 3 times slower than fftw (from fftw site). Can we allow it?
MMesch
@MMesch
Aug 04 2016 12:11
For many applications definitely.
For Gravity maybe not !?
if fftpack provides the same or similar functions than fftw, it should be simple to allow for both
MMesch
@MMesch
Aug 04 2016 14:50
@MarkWieczorek I am working on the Window class. Are you as well?
Mark Wieczorek
@MarkWieczorek
Aug 04 2016 14:50
Yes, I am 99% done, so please don't commit anything.
I should be done later today.
MMesch
@MMesch
Aug 04 2016 14:55
ok. I have made some very minor changes
I'll merge the two versions later without pushing. I also wrote a new notebook
Mark Wieczorek
@MarkWieczorek
Aug 04 2016 14:57
You can see what I am up to here: It is almost done, I am just working on the multitaper spectral analysis stuff now: https://github.com/MarkWieczorek/SHTOOLS/tree/shwindow
MMesch
@MMesch
Aug 04 2016 15:10
ok, just quickly checked it. I just added a get_powerperdegree() function
I wanted to do window.get_spectral_estimate(shcoeffs) as interface to get the multitaper measurement. How did you do it now?
anyway, we just need to make a good merge later
MMesch
@MMesch
Aug 04 2016 15:18
I could use your additions very soon Mark.
Mark Wieczorek
@MarkWieczorek
Aug 04 2016 16:07
@MMesch I'm going to have to merge this tomorrow morning. Everything is here, but I need to go: https://github.com/MarkWieczorek/SHTOOLS/tree/shwindow