search

styfenschaer / rocket-fft

Rocket-FFT makes Numba aware of numpy.fft and scipy.fft. Rocket-FFT takes its name from the PocketFFT Fast Fourier Transformation library that powers it, and Numba's goal of making your scientific Python code blazingly fast - like a rocket. 🚀
BSD 3-Clause "New" or "Revised" License
67 stars 2 forks source link

compatibility issue with numpy version > 1.25 #4

Closed EliousMondal closed 1 year ago

EliousMondal commented 1 year ago

Thanks a lot for making this. I am incorporated this in my code and it works really good. But I have a small issue for now. I tried to install rocket-fft and realised that I need numpy version ~ 1.21 because of which I have to uninstall my numpy and install a lower version to make the rocket-fft and some lower version of other packages like scipy, mpi4py, matplotlib. Is there a newer version of rocket-fft which is compatible with higher versions of numpy? or am I installing it wrong (I just did 'pip install rocket-fft')?

styfenschaer commented 1 year ago

Hi @EliousMondal

I am glad that you find it helpful.

Rocket-FFT does not have any requirements for the Numpy version. However, Numba, which is one of the strict requirements for Rocket-FFT, does. E.g. Numba 0.57 requires at least Numpy 1.21 and the maximum supported version is (if I remember correctly) 1.24. If you need Numpy >1.25, you can check if the release candidate of Numba 0.58 works for you, which supports up to and including Numpy 1.26.

If you still think that the problems originate from Rocket-FFT, please let me know and share the corresponding error message.

EliousMondal commented 1 year ago

I no longer have a problem with Rocket-FFT. Thanks again for the package.