You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The upcomping SciPy 1.4.0 release will include a new subpackage scipy.fft that supercedes scipy.fftpack. This new interface matches the behaviour of numpy.fft closely including real to complex rffts but is different in a few ways (e.g. dtype casting).
This interface also includes a new backend mechanism (scipy/scipy#10383) that would allow mkl_fft to implement the scipy.fft interface without any monkey patching. e.g. from pyFFTW/pyFFTW#269:
importscipy.fftscipy.fft.fft([1]) # Calls scipy's own implementationfrompyfftw.interfacesimportscipy_fftscipy.fft.set_global_backend(scipy_fft)
scipy.fft.fft([1]) # Calls into pyfftw
See scipy_fft.py for what needs to be done on the mkl_fft side to support this.
It's also unfortunate that the name mkl_fft._scipy_fft is already taken. Any ideas on how to move forward with this?
The text was updated successfully, but these errors were encountered:
The upcomping SciPy 1.4.0 release will include a new subpackage
scipy.fft
that supercedesscipy.fftpack
. This new interface matches the behaviour ofnumpy.fft
closely including real to complexrfft
s but is different in a few ways (e.g. dtype casting).This interface also includes a new backend mechanism (scipy/scipy#10383) that would allow
mkl_fft
to implement thescipy.fft
interface without any monkey patching. e.g. from pyFFTW/pyFFTW#269:See
scipy_fft.py
for what needs to be done on themkl_fft
side to support this.It's also unfortunate that the name
mkl_fft._scipy_fft
is already taken. Any ideas on how to move forward with this?The text was updated successfully, but these errors were encountered: