index_select in PyTorch

Buy Me a Coffee☕

*Memos:

• My post explains select().
• My post explains masked_select().

index_select() can get the 0D or more D tensor of the zero or more elements selected with zero or more indices, not removing one dimension from the 0D or more D tensor of zero or more elements as shown below:

*Memos:regularization

• index_select() can be used with torch or a tensor.
• The 1st argument(input) with torch or using a tensor(Required-Type:tensor of int, float, complex or bool).
• The 2nd argument with torch or the 1st argument with a tensor is dim(Required-Type:int).
• The 3rd argument with torch or the 2nd argument with a tensor is index(Required-Type:tensor of int). *It must be the 0D or 1D tensor of zero or more integers.
• There is out argument with torch(Optional-Default:None-Type:tensor): *Memos:
  • out= must be used.
  • My post explains out argument.

import torch

my_tensor = torch.tensor([8, -3, 0, 1, 5, -2, -1, 4])

torch.index_select(input=my_tensor, dim=0, index=torch.tensor(4))
my_tensor.index_select(dim=0, index=torch.tensor(4))
torch.index_select(input=my_tensor, dim=-1, index=torch.tensor(4))
# tensor([5])

torch.index_select(input=my_tensor, dim=0, index=torch.tensor([5, 2, 0, 7]))
torch.index_select(input=my_tensor, dim=-1, index=torch.tensor([5, 2, 0, 7]))
# tensor([-2, 0, 8, 4])

my_tensor = torch.tensor([[8, -3, 0, 1],
                          [5, -2, -1, 4]])
torch.index_select(input=my_tensor, dim=0, index=torch.tensor(1))
torch.index_select(input=my_tensor, dim=0, index=torch.tensor([1]))
torch.index_select(input=my_tensor, dim=-2, index=torch.tensor(1))
torch.index_select(input=my_tensor, dim=-2, index=torch.tensor([1]))
# tensor([[5, -2, -1, 4]])

torch.index_select(input=my_tensor, dim=0, index=torch.tensor([1, 0, 0, 1]))
torch.index_select(input=my_tensor, dim=-2, index=torch.tensor([1, 0, 0, 1]))
# tensor([[5, -2, -1, 4],
#         [8, -3, 0, 1],
#         [8, -3, 0, 1],
#         [5, -2, -1, 4]])

torch.index_select(input=my_tensor, dim=1, index=torch.tensor([3, 1, 2]))
torch.index_select(input=my_tensor, dim=-1, index=torch.tensor([3, 1, 2]))
# tensor([[1, -3, 0],
#         [4, -2, -1]])

my_tensor = torch.tensor([[[8, -3], [0, 1]],
                          [[5, -2], [-1, 4]]])
torch.index_select(input=my_tensor, dim=2, index=torch.tensor(1))
torch.index_select(input=my_tensor, dim=2, index=torch.tensor([1]))
torch.index_select(input=my_tensor, dim=-1, index=torch.tensor(1))
torch.index_select(input=my_tensor, dim=-1, index=torch.tensor([1]))
# tensor([[[-3], [1]],
#         [[-2], [4]]])

my_tensor = torch.tensor([[[8., -3.], [0., 1.]],
                          [[5., -2.], [-1., 4.]]])
torch.index_select(input=my_tensor, dim=2, index=torch.tensor(1))
# tensor([[[-3.], [1.]],
#         [[-2.], [4.]]])

my_tensor = torch.tensor([[[8.+0.j, -3.+0.j], [0.+0.j, 1.+0.j]],
                          [[5.+0.j, -2.+0.j], [-1.+0.j, 4.+0.j]]])
torch.index_select(input=my_tensor, dim=2, index=torch.tensor(1))
# tensor([[[-3.+0.j], [1.+0.j]],
#         [[-2.+0.j], [4.+0.j]]])

my_tensor = torch.tensor([[[True, False], [True, False]],
                          [[False, True], [False, True]]])
torch.index_select(input=my_tensor, dim=2, index=torch.tensor(1))
# tensor([[[False], [False]],
#         [[True], [True]]])