How to compute element-wise remainder of given input tensor in PyTorch?

In this article, we are going to see how to compute the element-wise remainder in PyTorch. we have two methods to compute element-wise reminders one is torch.remainder() and the other one is torch.fmod() let's go discuss both of them one by one.

torch.remainder() method 

The PyTorch remainder() method computes the element-wise remainder of the division operation (dividend is divided by divisor). The dividend is a tensor whereas the divisor may be a scalar quantity or tensor. The values must be an integer and float only.  before moving further let's see the syntax of the given method.

Syntax: torch.remainder(input, other, out=None)

Parameters:

• input (Tensor or Scalar) : the dividend element.
• other (Tensor or Scalar) : the divisor element.

Return: This method returns a new tensor with remainder values.

Example 1:

The following program is to compute the element-wise remainder of two single-dimension tensors.

# importing torch
import torch

# define the dividend
tens_1 = torch.tensor([5., -12., 25., -10., 30])
print("Dividend: ", tens_1)

# define the divisor
tens_2 = torch.tensor([5., -5., -6., 5., 8.])
print("Divisor: ", tens_2)

# compute the remainder
remainder = torch.remainder(tens_1, tens_2)

# display result
print("Remainder: ", remainder)

Output:

Dividend:  tensor([  5., -12.,  25., -10.,  30.])

Divisor:  tensor([ 5., -5., -6.,  5.,  8.])

Remainder:  tensor([ 0., -2., -5., -0.,  6.])

Example 2:

The following program is to compute the element-wise remainder of two 2D tensors.

# importing torch
import torch

# define the dividend
tens_1 = torch.tensor([[5., -12.],
                       [-10., 30.], ])
print("\n Dividend: \n", tens_1)

# define the divisor
tens_2 = torch.tensor([[5., -5.],
                       [5., 8.], ])

print("\n Divisor: \n", tens_2)

# compute the remainder
remainder = torch.remainder(tens_1, tens_2)

# display result
print("\n Remainder: \n", remainder)

Output:

 Dividend: 
 tensor([[  5., -12.],
        [-10.,  30.]])

 Divisor: 
 tensor([[ 5., -5.],
        [ 5.,  8.]])

 Remainder: 
 tensor([[ 0., -2.],
        [-0.,  6.]])

torch.fmod() method 

This method gives also helps us to compute the element-wise remainder of division by the divisor. The divisor may be a number or a Tensor. When the divisor is zero it will return NaN. before moving further let's see the syntax of the given method.

Syntax: torch.fmod(input, other)

Parameters:

• input (Tensor) : the dividend.
• other (Tensor or Scalar) : the divisor.

Return: This method returns a new tensor with remainder values.

Example 1:

The following program is to compute the element-wise remainder of two single-dimension tensors.

# importing torch
import torch

# define the dividend
tens_1 = torch.tensor([5., -10., -17., 19., 20.])
print("\n\n Dividend: ", tens_1)

# define the divisor
tens_2 = torch.tensor([2., 5., 17., 7., 10.])

print("\n Divisor: ", tens_2)

# compute the remainder using fmod()
remainder = torch.fmod(tens_1, tens_2)

# display result
print("\n Remainder: ", remainder)

Output:

 Dividend:  tensor([  5., -10., -17.,  19.,  20.])

 Divisor:  tensor([ 2.,  5., 17.,  7., 10.])

 Remainder:  tensor([1., -0., -0., 5., 0.])

Example 2:

The following program is to compute the element-wise remainder of two 2D tensors.

# importing torch
import torch

# define the dividend
tens_1 = torch.tensor([[16., -12.],
                       [-10., 30.], ])
print("\n\n Dividend: \n", tens_1)

# define the divisor
tens_2 = torch.tensor([[5., -6.],
                       [5., 8.], ])

print("\n Divisor: \n", tens_2)

# compute the remainder using fmod()
remainder = torch.fmod(tens_1, tens_2)

# display result
print("\n Remainder:\n", remainder)

Output:

 Dividend: 
 tensor([[ 16., -12.],
        [-10.,  30.]])

 Divisor: 
 tensor([[ 5., -6.],
        [ 5.,  8.]])

 Remainder:
 tensor([[1., -0.],
        [-0., 6.]])