Boosting Performance: Repeating 2D Arrays in Python with NumPy

Problem:

You want to take a 2D array (matrix) and create a new 3D array where the original 2D array is repeated N times along a specified axis. This is useful in various machine learning and image processing tasks where you need to create stacked or batched versions of data.

Solution using NumPy:

NumPy provides several efficient ways to achieve this:

Method 1: Using np.repeat

1. Reshape the input array:
2. Repeat along the desired axis:
   • Use np.repeat to repeat the reshaped array N times along the first axis (axis=0).

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6]])
N = 3

# Reshape and repeat
repeated_arr = np.repeat(np.expand_dims(arr, axis=0), N, axis=0)

print(repeated_arr)

This will output:

[[[1 2 3]
  [4 5 6]]

 [[1 2 3]
  [4 5 6]]

 [[1 2 3]
  [4 5 6]]]

Method 2: Using list comprehension or generator expression

1. Create a list or generator expression that expands the 2D array N times.
2. Convert the list or generator expression to a NumPy array.

repeated_arr = np.array([arr] * N)