torch.deg2rad(): Converting Angles to Degress

PyTorch torch.deg2rad() method converts angles from degrees to radians.

It returns a tensor of the same shape and data type as the input, containing angles converted to radians.

The mathematical formula is this:

Syntax

torch.deg2rad(input, out=None)

Parameters

Argument	Description
input (Tensor)	It represents a tensor containing angle values in degrees.
out (Tensor, optional)	It represents a tensor to store the output.

Basic conversion

import torch

degrees_tensor = torch.tensor([0.0, 90.0, 180.0, 270.0, 360.0])

print("Degrees Tensor:", degrees_tensor)
# Output: Degrees Tensor: tensor([  0.,  90., 180., 270., 360.])

radians_tensor = torch.deg2rad(degrees_tensor)

print("Radians Tensor:", radians_tensor)
# Output: Radians Tensor: tensor([0.0000, 1.5708, 3.1416, 4.7124, 6.2832])

You can see from the above output that it converted element-wise and returns the same shape sensor with radian values.

Negative angles

If you pass a tensor with negative degrees, it will give you the negative radians.

import torch

negative_tensor = torch.tensor([-45.0, -90.0, -180.0])

print("Negative Degrees Tensor:", negative_tensor)
# Output: Negative Degrees Tensor: tensor([-45., -90., -180.])

negative_radians_tensor = torch.deg2rad(negative_tensor)

print("Negative Radians Tensor:", negative_radians_tensor)
# Output: Negative Radians Tensor: tensor([-0.7854, -1.5708, -3.1416])

Handling a 2D tensor of degrees

If you pass the degrees in a 2D tensor, it will return radians in a 2D tensor. Dimensions do not matter. It works smoothly.

import torch

deg_2d_tensor = torch.tensor([[15.0, 30.0], [45.0, 60.0]])

print("2D Degrees Tensor:", deg_2d_tensor)
# Output: 2D Degrees Tensor: tensor([[15., 30.],
#                                   [45., 60.]])

rad_2d_tensor = torch.deg2rad(deg_2d_tensor)

print("2D Radians Tensor:", rad_2d_tensor)
# Output: 2D Radians Tensor: tensor([[0.2618, 0.5236],
#                                   [0.7854, 1.0472]])

Pre-allocated tensor

If you have a pre-allocated tensor created using methods like torch.empty(), you can store the radian values in this tensor.

import torch

input_tensor = torch.tensor([45.0, 135.0])

print(input_tensor)
# Output: tensor([ 45., 135.])

output_tensor = torch.empty(2)

print(output_tensor)
# Output: tensor([0., 0.])

torch.deg2rad(input_tensor, out=output_tensor)

print(output_tensor)
# Output: tensor([0.7854, 2.3562])

Integrate with Trigonometric Functions

Let’s calculate the sine of 30 degrees.

import torch

degrees = torch.tensor(30.0)

radians = torch.deg2rad(degrees)

sine = torch.sin(radians)

print(f"Sine of {degrees.item()} degrees is {sine.item()}")

# Output: Sine of 30.0 degrees is 0.5

That’s all!