Name: is_floating_point(), is_complex() and is_nonzero() in PyTorch
Rating: 1.2 (8446 reviews)
Author: hyperkai

*Memos:

My post explains isnan(), isinf(), isposinf() and isneginf().
My post explains isfinite(), isreal() and isin().

is_floating_point() can check the 0D or more D tensor of zero or more elements is float type, getting the scalar of a boolean value as shown below:

*Memos:

is_floating_point() can be used with torch or a tensor.
The 1st argument with torch or using a tensor is input(Required-Type:tensor of int, float, complex or bool).

import torch

my_tensor = torch.tensor([])
my_tensor = torch.tensor(8.)
my_tensor = torch.tensor(float('nan'))
my_tensor = torch.tensor(float('inf'))
my_tensor = torch.tensor(float('-inf'))

torch.is_floating_point(input=my_tensor)
my_tensor.is_floating_point()
# True

my_tensor = torch.tensor(-5)
my_tensor = torch.tensor(3.+0.j)
my_tensor = torch.tensor(3.+7.j)
my_tensor = torch.tensor(True)

torch.is_floating_point(input=my_tensor)
# False

my_tensor = torch.tensor([8.,
                          float('nan'),
                          float('inf'),
                          float('-inf')])
torch.is_floating_point(input=my_tensor)
# True

my_tensor = torch.tensor([[8., float('nan')],
                          [float('inf'), float('-inf')]])
torch.is_floating_point(input=my_tensor)
# True

my_tensor = torch.tensor([[[8., float('nan')],
                           [float('inf'), float('-inf')]],
                          [[-5, 3.+0.j],
                           [3.+7.j, True]]]) # complex64
torch.is_floating_point(input=my_tensor)
# False

is_complex() can check the 0D or more D tensor of zero or more elements is complex type, getting the scalar of a boolean value as shown below:

*Memos:

is_complex() can be used with torch or a tensor.
The 1st argument with torch or using a tensor is input(Required-Type:tensor of int, float, complex or bool).

import torch

my_tensor = torch.tensor(3.+0.j)
my_tensor = torch.tensor(3.+7.j)

torch.is_complex(input=my_tensor)
my_tensor.is_complex()
# True

my_tensor = torch.tensor([])
my_tensor = torch.tensor(-5)
my_tensor = torch.tensor(8.)
my_tensor = torch.tensor(float('nan'))
my_tensor = torch.tensor(float('inf'))
my_tensor = torch.tensor(float('-inf'))
my_tensor = torch.tensor(True)

torch.is_complex(input=my_tensor)
# False

my_tensor = torch.tensor([3.+0.j, 3.+7.j])

torch.is_complex(input=my_tensor)
# True

my_tensor = torch.tensor([[-5, 8., True],
                          [float('nan'), float('inf'), float('-inf')]])
my_tensor = torch.tensor([[-5, 8., float('nan')],
                          [float('inf'), float('-inf'), True]])
torch.is_complex(input=my_tensor)
# False

my_tensor = torch.tensor([[[3.+0.j, 3.+7.j], # complex64
                           [-5, 8.]],
                          [[True, float('nan')],
                           [float('inf'), float('-inf')]]])
my_tensor = torch.tensor([[[3.+0.j, 3.+7.j], # complex64
                           [-5, 8.]],
                          [[float('nan'), float('inf')],
                           [float('-inf'), True]]])
torch.is_complex(input=my_tensor)
# True

is_nonzero() can check if the 0D or more D tensor of only one element is a nonzero, getting the scalar of a boolean value as shown below:

*Memos:

is_nonzero() can be used with torch or a tensor.
The 1st argument with torch or using a tensor is input(Required-Type:tensor of int, float, complex or bool).
There must be only one element in a tensor.

import torch

my_tensor = torch.tensor(-5)
my_tensor = torch.tensor(8.)
my_tensor = torch.tensor([float('nan')])
my_tensor = torch.tensor([float('inf')])
my_tensor = torch.tensor([[float('-inf')]])
my_tensor = torch.tensor([[3.+0.j]])
my_tensor = torch.tensor([[[3.+7.j]]])
my_tensor = torch.tensor(True)

torch.is_nonzero(input=my_tensor)
my_tensor.is_nonzero()
# True

my_tensor = torch.tensor(0)
my_tensor = torch.tensor([0.0])
my_tensor = torch.tensor([[0.+.0j]])
my_tensor = torch.tensor([[[False]]])

torch.is_nonzero(input=my_tensor)
# False