numpy.roll in Python

In Python, numpy.roll is a function that allows you to shift the elements in an array along a specific axis. This can be useful for a variety of applications, such as circularly shifting an image, or for creating a sliding window over an array.

Syntax

The syntax for using the numpy.roll function is as follows:

numpy.roll(array, shift, axis=None)

Here, array is the input array to be shifted, shift is the number of places to roll the elements along the axis, and axis (optional) specifies the axis along which the elements will be rolled. If axis is not specified, the array is flattened before rolling.

Examples

Let's look at some examples to see how the numpy.roll function works.

Example 1: Circularly shifting an image

import numpy as np
import matplotlib.pyplot as plt

# Create a 2D array containing a square
square = np.zeros((10, 10))
square[3:7, 3:7] = 1

# Show the original square
plt.imshow(square, cmap='gray')
plt.title('Original Square')
plt.show()

# Roll the elements along the y-axis
rolled = np.roll(square, shift=3, axis=1)

# Show the rolled square
plt.imshow(rolled, cmap='gray')
plt.title('Circularly Shifted Square')
plt.show()

In this example, we create a 2D array containing a square and display it using matplotlib. We then apply the numpy.roll function to shift the elements in the array 3 places along the y-axis, effectively circularly shifting the image. We then display the shifted image using matplotlib.

Example 2: Creating a sliding window over an array

import numpy as np

# Create a 2D array
array = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

# Roll the elements along the y-axis
rolled = np.roll(array, shift=1, axis=1)

# Create sliding window using numpy.lib.stride_tricks.as_strided
window_shape = (2, 2)
strides = array.strides
windowed = np.lib.stride_tricks.as_strided(rolled, shape=(3, 2, 2, 3), strides=(strides[0], strides[1], strides[1], strides[0]))

print(windowed)

In this example, we create a 2D array and apply the numpy.roll function to shift the elements in the array 1 place along the y-axis. We then use numpy.lib.stride_tricks.as_strided to create a sliding window over the rolled array. The resulting 4D array contains a window of size (2,2) at each position in the rolled array.

Conclusion

The numpy.roll function in Python is a useful tool for shifting the elements in an array along a specific axis. This function can be used for a variety of applications, such as circularly shifting an image or creating a sliding window over an array.