vishwa91 / pyhadamard Goto Github PK

The scaling option is not consistent. Example:

>>> a = np.ones([4, 1])
>>> ah = fast_wh.fast_wh_trans(a, True, True)

returns (4, 0, 0, 0), whereas it must return (2, 0, 0, 0).

Thanks for the code @vishwa91! It would be great to have an implementation of FWHT compatible with tensors in PyTorch/TensorFlow.

Also, an example script for using the repo might be helpful.