Function Differences with torch.hann_window

torch.hann_window

torch.hann_window(
    window_length,
    periodic=True,
    *,
    dtype=None,
    layout=torch.strided,
    device=None,
    requires_grad=False
) -> Tensor

For more information, see torch.hann_window.

mindspore.numpy.hanning

mindspore.numpy.hanning(M) -> Tensor

For more information, see mindspore.numpy.hanning.

Differences

PyTorch: Return the Hanning window with the same size as window_length. The periodic parameter determines whether the returned window will remove the last duplicate value of the symmetric window.

MindSpore: MindSpore API basically implements the same function as PyTorch, but it lacks the parameter periodic, which is equivalent to setting periodic to False.

Code Example 1

The periodic parameter in the PyTorch operator determines whether the return window will remove the last repeated value of the symmetric window, while the MindSpore operator lacks the parameter, equivalent to setting periodic to False.

# PyTorch
import torch

torch_output = torch.hann_window(12, periodic=False)
print(torch_output.numpy())
# [0.         0.07937324 0.29229248 0.57115734 0.82743037 0.97974646
#  0.9797465  0.8274305  0.5711575  0.29229265 0.07937327 0.        ]

# MindSpore
import mindspore

ms_output = mindspore.numpy.hanning(12)
print(ms_output)
# [0.         0.07937324 0.29229248 0.57115734 0.82743037 0.97974646
#  0.9797465  0.8274305  0.5711575  0.29229265 0.07937327 0.        ]