Memory problem with scipy and fft

[sujan1746]

Hi, when I use MSNoise, I can see some errors related to memory occurred by sf.fftn or sf.ifft etc and stopped.
Calculation of fft related things can be calculated with numpy and scipy, and when I did quick test, using numpy is faster (almost 2~3 times).
When I see the code, you used numpy for other parts of the fft calculation except fft itself.
So I wonder why you used scipy for calculation of fft.
Is there a reason that I did not recognize? If there is, I have to know to solve memory problem.

And cc computation is proceeded with 0.6~0.8 %Memory, but it sometimes goes up almost 6.8%.
I have no idea to solve it.

The error message is as below even though memory is enough. (related to sf.fftn, sf.ifft, sf.fftfreq in s03compute_no_rotation.py)
numpy.core._exceptions._ArrayMemoryError: Unable to allocate 84.6 MiB for an array with shape (140, 79233) and data type float64

p.s. even numpy.fft also shows memory problem with the same code. Do you have any idea for this problem ?

[ThomasLecocq]

dupplicate of #301 , closing;