How To Sort The Elements of a Tensor in PyTorch?

In this article, we are going to see how to sort the elements of a PyTorch Tensor in Python.

To sort the elements of a PyTorch tensor, we use torch.sort() method.  We can sort the elements along with columns or rows when the tensor is 2-dimensional.

Syntax: torch.sort(input, dim=- 1, descending=False)

Parameters:

• input: It is an input PyTorch tensor.
• dim: The dimension along which the tensor is sorted. It is an optional int value.
• descending: An optional boolean value used for sorting tensor elements in ascending or descending order. Default is set to False, sorting in ascending order.

Returns: It returns a named tuple of (values, indices), where values are the sorted values and indices are the indices of the elements in the original input tensor.

Example 1:

In the below example, we sort the elements of a 1-dimensional tensor in ascending and descending orders. Sort tensor in ascending or descending order. We apply the torch.sort() method to sort the elements of an input tensor. To sort in descending order pass descending=True to the method.

Python3

# importing required library
import torch

# defining a PyTorch Tensor
tensor = torch.tensor([-12, -23, 0.0, 32,
                       1.32, 201, 5.02])
print("Tensor:\n", tensor)

# sorting the tensor in ascending order
print("Sorting tensor in ascending order:")
values, indices = torch.sort(tensor)

# printing values of sorted tensor
print("Sorted values:\n", values)

# printing indices of sorted value
print("Indices:\n", indices)

# sorting the tensor in  descending order
print("Sorting tensor in  descending order:")
values, indices = torch.sort(tensor, descending=True)

# printing values of sorted tensor
print("Sorted values:\n", values)

# printing indices of sorted value
print("Indices:\n", indices)

Output:

Example 2:

In this example, we sort the elements of a 2-dimensional tensor in ascending as well as descending orders along with the columns.

Python3

# importing the library
import torch

# define a 2D torch tensor
tensor = torch.tensor([[43,31,-92],
                       [3,-4.3,53],
                       [-4.2,7,-6.2]])
print("Tensor:\n", tensor)

# sorting the tensor in  ascending order
print("Sorting tensor in \
ascending order along the column:")
values, indices = torch.sort(tensor, dim = 0)

# printing  values in sorted tensor
print("Sorted values:\n", values)

# print indices of values in sorted tensor
print("Indices:\n", indices)

# sorting the tensor in  descending order
print("Sorting tensor in \
descending order along the column:")
values, indices = torch.sort(tensor, dim = 0,
                             descending=True)

# printing  values in sorted tensor
print("Sorted values:\n", values)

# print indices of values in sorted tensor
print("Indices:\n", indices)

Output:

Example 3:

In this example, we sort the elements of a 2-dimensional tensor in ascending as well as descending orders along the rows.

Python3

# importing the library
import torch

# define a 2D torch tensor
tensor = torch.tensor([[43, 31, -92],
                       [3, -4.3, 53],
                       [-4.2, 7, -6.2]])
print("Tensor:\n", tensor)

# sorting the tensor in  ascending order
print("Sorting tensor in \
ascending order along the row:")
values, indices = torch.sort(tensor, dim=1)

print("Sorted values:\n", values)

# print indices of values in sorted tensor
print("Indices:\n", indices)

# sorting the tensor in  descending order
print("Sorting tensor in \
descending order along the row:")
values, indices = torch.sort(tensor,
                             dim=1,
                             descending=True)

# printing  values in sorted tensor
print("Sorted values:\n", values)

# printing indices of values in sorted tensor
print("Indices:\n", indices)

Output: