NumPy — Broadcasting

What is broadcasting?

Automatic expansion of arrays to compatible shapes.

Basic example

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6]])
result = arr + 10  # Adds 10 to each element

Rules

# Rule 1: Arrays with different ndim
a = np.array([1, 2, 3])
b = np.array([[1], [2], [3]])
result = a + b  # (3,) + (3,1) -> (3,3)

Compatible shapes

# Same shape
a = np.array([1, 2, 3])
b = np.array([4, 5, 6])
c = a + b  # Works

# Scalar
a = np.array([[1, 2], [3, 4]])
c = a + 10  # Works

# Incompatible
a = np.array([1, 2, 3])
b = np.array([1, 2])
# c = a + b  # Error

Practical uses

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

# Subtract column mean
col_mean = arr.mean(axis=0)
result = arr - col_mean  # Broadcasting

Mini Practice

1. Practice basic broadcasting
2. Add scalar to array
3. Subtract column mean
4. Test compatible shapes

Up Next

Continue with Functions - Useful functions.