dsplit() in PyTorch

*Memos:

• My post explains tensor_split().
• My post explains split().
• My post explains vsplit().
• My post explains hsplit().
• My post explains chunk() and unbind().

dsplit() can split a 3D or more D tensor into one or more tensors depthwise as shown below:

*Memos:

• dsplit() can be used with torch or a tensor.
• The 1st argument(tensor of int, float, complex or bool) with torch or using a tensor(tensor of int, float, complex or bool) is input(Required).
• The 2nd argument(int) with torch or the 1st argument(int) with a tensor is sections(Required).
• The 2nd argument(tuple of int or list of int) with torch or the 1st argument(tuple of int or list of int) with a tensor is indices(Required).
• The total number of the zero or more elements of the one or more returned tensors changes.
• The one or more returned tensors keep the dimension of the input tensor.

import torch

my_tensor = torch.tensor([[[0, 1, 2, 3],
                           [4, 5, 6, 7],
                           [8, 9, 10, 11]]])
torch.dsplit(input=my_tensor, sections=1)
my_tensor.dsplit(sections=1)
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),)

torch.dsplit(input=my_tensor, sections=2)
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, sections=4)
# (tensor([[[0], [4], [8]]]),
#  tensor([[[1], [5], [9]]]),
#  tensor([[[2], [6], [10]]]),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(0,))
torch.dsplit(input=my_tensor, indices=(-4,))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(1,))
torch.dsplit(input=my_tensor, indices=(-3,))
# (tensor([[[0], [4], [8]]]),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(2,))
torch.dsplit(input=my_tensor, indices=(-2,))
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, indices=(3,))
torch.dsplit(input=my_tensor, indices=(-1,))
# (tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(4,))
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64))

torch.dsplit(input=my_tensor, indices=(0, 0))
torch.dsplit(input=my_tensor, indices=(0, -4))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(0, 1))
torch.dsplit(input=my_tensor, indices=(0, -3))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0], [4], [8]]]),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(0, 2))
torch.dsplit(input=my_tensor, indices=(0, -2))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, indices=(0, 3))
torch.dsplit(input=my_tensor, indices=(0, -1))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(0, 4))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64))

torch.dsplit(input=my_tensor, indices=(1, 0))
torch.dsplit(input=my_tensor, indices=(1, -4))
# (tensor([[[0], [4], [8]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(1, 1))
torch.dsplit(input=my_tensor, indices=(1, -3))
# (tensor([[[0], [4], [8]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(1, 2))
torch.dsplit(input=my_tensor, indices=(1, -2))
# (tensor([[[0], [4], [8]]]),
#  tensor([[[1], [5], [9]]]),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, indices=(1, 3))
torch.dsplit(input=my_tensor, indices=(1, -1))
# (tensor([[[0], [4], [8]]]),
#  tensor([[[1, 2], [5, 6], [9, 10]]]),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(1, 4))
# (tensor([[[0], [4], [8]]]),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64))

torch.dsplit(input=my_tensor, indices=(2, 0))
torch.dsplit(input=my_tensor, indices=(2, -4))
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(2, 1))
torch.dsplit(input=my_tensor, indices=(2, -3))
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(2, 2))
torch.dsplit(input=my_tensor, indices=(2, -2))
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, indices=(2, 3))
torch.dsplit(input=my_tensor, indices=(2, -1))
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([[[2], [6], [10]]]),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(2, 4))
# (tensor([[[0, 1], [4, 5], [8, 9]]]),
#  tensor([[[2, 3], [6, 7], [10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64))

torch.dsplit(input=my_tensor, indices=(3, 0))
torch.dsplit(input=my_tensor, indices=(3, -4))
# (tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(3, 1))
torch.dsplit(input=my_tensor, indices=(3, -3))
# (tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(3, 2))
torch.dsplit(input=my_tensor, indices=(3, -2))
# (tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, indices=(3, 3))
torch.dsplit(input=my_tensor, indices=(3, -1))
# (tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(3, 4))
# (tensor([[[0, 1, 2], [4, 5, 6], [8, 9, 10]]]),
#  tensor([[[3], [7], [11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64))

torch.dsplit(input=my_tensor, indices=(4, 0))
torch.dsplit(input=my_tensor, indices=(4, -4))
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(4, 1))
torch.dsplit(input=my_tensor, indices=(4, -3))
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[1, 2, 3], [5, 6, 7], [9, 10, 11]]]))

torch.dsplit(input=my_tensor, indices=(4, 2))
torch.dsplit(input=my_tensor, indices=(4, -2))
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[2, 3], [6, 7], [10, 11]]]))

torch.dsplit(input=my_tensor, indices=(4, 3))
torch.dsplit(input=my_tensor, indices=(4, -1))
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[3], [7], [11]]]))

torch.dsplit(input=my_tensor, indices=(4, 4))
# (tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([], size=(1, 3, 0), dtype=torch.int64))

torch.dsplit(input=my_tensor, indices=(0, 0, 0))
torch.dsplit(input=my_tensor, indices=(0, 0, -4))
torch.dsplit(input=my_tensor, indices=(0, -4, 0))
torch.dsplit(input=my_tensor, indices=(0, -4, -4))
# (tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([], size=(1, 3, 0), dtype=torch.int64),
#  tensor([[[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]]))
etc.

my_tensor = torch.tensor([[[0., 1., 2., 3.],
                           [4., 5., 6., 7.],
                           [8., 9., 10., 11.]]])
torch.dsplit(input=my_tensor, sections=1)
# (tensor([[[0., 1., 2., 3.],
#           [4., 5., 6., 7.],
#           [8., 9., 10., 11.]]]),)

my_tensor = torch.tensor([[[0.+0.j, 1.+0.j, 2.+0.j, 3.+0.j],
                           [4.+0.j, 5.+0.j, 6.+0.j, 7.+0.j],
                           [8.+0.j, 9.+0.j, 10.+0.j, 11.+0.j]]])
torch.dsplit(input=my_tensor, sections=1)
# (tensor([[[0.+0.j, 1.+0.j, 2.+0.j, 3.+0.j],
#           [4.+0.j, 5.+0.j, 6.+0.j, 7.+0.j],
#           [8.+0.j, 9.+0.j, 10.+0.j, 11.+0.j]]]),)

my_tensor = torch.tensor([[[True, False, True, False],
                           [False, True, False, True],
                           [True, False, True, False]]])
torch.dsplit(input=my_tensor, sections=1)
# (tensor([[[True, False, True, False],
#           [False, True, False, True],
#           [True, False, True, False]]]),)