[Numpy-discussion] Re: Numpy with eigen c++ binding

2023-06-11 Thread Matthieu Brucher
What would be the point? You would need to add support in Eigen for most of what makes Numpy Numpy. Eigen is really about 2D, Tensors are not even close to be something working properly (not even talking about the dynamic number of dimensions that breaks), and I'm not talking about broadcasting rul

Re: [Numpy-discussion] IRFFT of sound

2021-01-24 Thread Matthieu Brucher
Hi, You need the angle to restore the sound. Otherwise you don't get the phase information. Just store the angle information and apply it? Cheers, Matthieu Le dim. 24 janv. 2021 à 14:27, a écrit : > > Hello friends! > Maybe that's not directly a Numpy question but I think it can be solved by

Re: [Numpy-discussion] Direct GPU support on NumPy

2018-01-02 Thread Matthieu Brucher
Hi, Let's say that Numpy provides a GPU version on GPU. How would that work with all the packages that expect the memory to be allocated on CPU? It's not that Numpy refuses a GPU implementation, it's that it wouldn't solve the problem of GPU/CPU having different memory. When/if nVidia decides (fin

Re: [Numpy-discussion] Reshape 2D array into 3D

2017-07-10 Thread Matthieu Brucher
Hi, This works, but reshape doesn't move data around. What happens is that the data is flattened and then reshaped. If your 5 is not supposed to move, you should create a 2,5,4 array and then copy the two slices by hand, or use transpose (make it 5,4,2 and then transpose to 2,5,4=. Matthieu Le 1