tensorplay.functional.as_strided_scatter

tensorplay.functional.as_strided_scatter(input, src, size, stride, storage_offset=None) → Tensor[source]

Embeds the values of the src tensor into input along the elements corresponding to the result of calling input.as_strided(size, stride, storage_offset).

This function returns a tensor with fresh storage; it does not return a view.

Parameters:

• input (Tensor) – the input tensor.

• size (tuple or ints) – the shape of the output tensor

• stride (tuple or ints) – the stride of the output tensor

• storage_offset (int, optional) – the offset in the underlying storage of the output tensor

Note

src must be of the proper size in order to be embedded into input. Specifically, it should have the same shape as tensorplay.as_strided(input, size, stride, storage_offset)

Example:

>>> a = tensorplay.arange(4).reshape(2, 2) + 1
>>> a
tensor([[1, 2],
        [3, 4]])
>>> b = tensorplay.zeros(3, 3)
>>> b
tensor([[0., 0., 0.],
        [0., 0., 0.],
        [0., 0., 0.]])
>>> tensorplay.as_strided_scatter(b, a, (2, 2), (1, 2))
tensor([[1., 3., 2.],
        [4., 0., 0.],
        [0., 0., 0.]])