Optimizing Transposition of large NumPy arrays

NumPy: Performance Optimization Exercise-6 with Solution

Write a NumPy program that creates a function to transpose a large 2D NumPy array using nested for loops. Optimize it using NumPy's transpose() function.

Sample Solution:

Python Code:

import numpy as np

# Create a large 2D NumPy array with shape (1000, 1000)
large_array = np.random.rand(1000, 1000)

# Function to transpose a 2D array using nested for loops
def transpose_using_loops(array):
    rows, cols = array.shape
    transposed_array = np.zeros((cols, rows))
    for i in range(rows):
        for j in range(cols):
            transposed_array[j, i] = array[i, j]
    return transposed_array

# Transpose the array using the nested for loops
transposed_loop = transpose_using_loops(large_array)
print("Transposed array using for loop (first 5x5 block):\n", transposed_loop[:5, :5])

# Optimize the transposition using NumPy's transpose() function
transposed_numpy = np.transpose(large_array)
print("Transposed array using NumPy's transpose() function (first 5x5 block):\n", transposed_numpy[:5, :5])

Output:

Transposed array using for loop (first 5x5 block):
 [[0.40551353 0.40558167 0.86467457 0.58567322 0.8239284 ]
 [0.3031671  0.45184241 0.5199638  0.52295596 0.52545074]
 [0.40211577 0.25829343 0.6505094  0.60260168 0.49010591]
 [0.42364773 0.85953907 0.24050983 0.50997478 0.54401922]
 [0.94613512 0.94341257 0.30700166 0.25956794 0.17429532]]
Transposed array using NumPy's transpose() function (first 5x5 block):
 [[0.40551353 0.40558167 0.86467457 0.58567322 0.8239284 ]
 [0.3031671  0.45184241 0.5199638  0.52295596 0.52545074]
 [0.40211577 0.25829343 0.6505094  0.60260168 0.49010591]
 [0.42364773 0.85953907 0.24050983 0.50997478 0.54401922]
 [0.94613512 0.94341257 0.30700166 0.25956794 0.17429532]]

Explanation:

• Create a large array: A 2D NumPy array with shape (1000, 1000) is created using np.random.rand().
• Function with nested for loops: A function transpose_using_loops transposes the array using nested for loops.
• Transpose with loops: The array is transposed using the nested for loops, and the first 5x5 block of the result is printed.
• Optimize with NumPy: The transposition is optimized using NumPy's built-in transpose() function, and the first 5x5 block of the result is printed.

Python-Numpy Code Editor:

Have another way to solve this solution? Contribute your code (and comments) through Disqus.

Previous: Optimizing Standard Deviation calculation of large NumPy arrays.
Next: Optimizing element-wise multiplication of large 2D NumPy arrays.