1D and 2D broadcasting in NumPy

Introduction

Broadcasting lets you do math on arrays of different shapes easily. It saves time and code by stretching smaller arrays to match bigger ones.

Adding a list of numbers to each row of a table of data.

Multiplying each column of a matrix by a single number.

Subtracting a 1D array from each row of a 2D array.

Scaling rows or columns of data without writing loops.

Syntax

NumPy

result = array1 + array2

Arrays must have compatible shapes for broadcasting.

Smaller arrays are 'stretched' without copying data.

Examples

The 1D array b is added to each row of A by stretching b.

NumPy

import numpy as np

# 1D array added to each row of 2D array
A = np.array([[1, 2, 3], [4, 5, 6]])
b = np.array([10, 20, 30])
result = A + b
print(result)

Each column of A is multiplied by the corresponding element in b.

NumPy

import numpy as np

# 2D array multiplied by 1D array (broadcast on columns)
A = np.array([[1, 2, 3], [4, 5, 6]])
b = np.array([2, 3, 4])
result = A * b
print(result)

Subtracts b from each row of A using broadcasting.

NumPy

import numpy as np

# 1D array subtracted from each row of 2D array
A = np.array([[5, 6, 7], [8, 9, 10]])
b = np.array([1, 2, 3])
result = A - b
print(result)

Sample Program

This program adds a 1D array to each row of a 2D array. Broadcasting stretches the 1D array across rows so addition works element-wise.

NumPy

import numpy as np

# Create a 2D array (3 rows, 4 columns)
matrix = np.array([[1, 2, 3, 4],
                   [5, 6, 7, 8],
                   [9, 10, 11, 12]])

# Create a 1D array with 4 elements
vector = np.array([10, 20, 30, 40])

# Add the 1D array to each row of the 2D array using broadcasting
result = matrix + vector

print("Original 2D array (matrix):")
print(matrix)
print("\n1D array (vector):")
print(vector)
print("\nResult of broadcasting addition:")
print(result)

Important Notes

Broadcasting works when trailing dimensions match or one of them is 1.

It avoids slow loops by doing operations fast in C code inside numpy.

If shapes are not compatible, numpy will raise an error.

Summary

Broadcasting lets you do math on arrays with different shapes easily.

1D arrays can be added or multiplied across rows or columns of 2D arrays.

It saves time and code by stretching smaller arrays without copying data.

Practice

1. What does broadcasting in NumPy allow you to do with arrays of different shapes?

easy

A. Only add arrays if they have the exact same shape

B. Sort arrays in ascending order

C. Convert arrays into lists automatically

D. Perform element-wise operations without manually reshaping arrays

Solution

1. Step 1: Understand broadcasting concept

   Broadcasting lets NumPy perform math on arrays with different shapes by stretching smaller arrays.

2. Step 2: Identify what broadcasting enables

   This means you can do element-wise operations without manually reshaping arrays to the same shape.

3. Final Answer:

   Perform element-wise operations without manually reshaping arrays -> Option D

4. Quick Check:

   Broadcasting = element-wise operations on different shapes [OK]

Hint: Broadcasting stretches smaller arrays to match bigger ones [OK]

Common Mistakes:

• Thinking arrays must have the same shape to operate
• Confusing broadcasting with sorting or reshaping
• Assuming broadcasting copies data

2. Which of the following is the correct way to add a 1D array b = np.array([1, 2, 3]) to each row of a 2D array a with shape (4, 3)?

easy

A. a + b

B. a + b.T

C. a + b.reshape(3, 1)

D. a + b.reshape(1, 4)

Solution

1. Step 1: Check shapes of arrays

   Array a has shape (4, 3), and b has shape (3,). Adding a + b broadcasts b across rows.

2. Step 2: Understand broadcasting rules

   Since b matches the number of columns, it adds to each row correctly without reshaping.

3. Final Answer:

   a + b -> Option A

4. Quick Check:

   1D array added across rows = a + b [OK]

Hint: Add 1D array directly if its length matches columns [OK]

Common Mistakes:

• Trying to transpose 1D array which has no effect
• Reshaping to wrong dimensions causing errors
• Confusing rows and columns in shapes

3. What is the output of the following code?

import numpy as np
a = np.array([[1, 2, 3], [4, 5, 6]])
b = np.array([10, 20, 30])
result = a + b
print(result)

medium

A. [[11 22 33]
[14 25 36]]

B. [[11 12 13]
[14 15 16]]

C. [[10 20 30]
[10 20 30]]

D. Error due to shape mismatch

Solution

1. Step 1: Analyze shapes and operation

   a shape is (2,3), b shape is (3,). Adding broadcasts b across rows.

2. Step 2: Calculate element-wise addition

   Row 1: [1+10, 2+20, 3+30] = [11, 22, 33]
   Row 2: [4+10, 5+20, 6+30] = [14, 25, 36]

3. Final Answer:

   [[11 22 33] [14 25 36]] -> Option A

4. Quick Check:

   Broadcast add across rows = [[11 22 33], [14 25 36]] [OK]

Hint: Add 1D array to 2D rows matches columns [OK]

Common Mistakes:

• Expecting error due to different shapes
• Adding element-wise without broadcasting
• Confusing rows and columns in addition

4. Identify the error in this code snippet:

import numpy as np
a = np.array([[1, 2], [3, 4], [5, 6]])
b = np.array([10, 20, 30])
result = a + b
print(result)

medium

A. Using wrong operator for addition

B. Syntax error in array creation

C. Shapes of a and b are incompatible for broadcasting

D. No error, code runs fine

Solution

1. Step 1: Check shapes of arrays

   a shape is (3, 2), b shape is (3,). They differ in second dimension.

2. Step 2: Understand broadcasting rules

   Broadcasting requires trailing dimensions to match or be 1. Here, 2 != 3, so addition fails.

3. Final Answer:

   Shapes of a and b are incompatible for broadcasting -> Option C

4. Quick Check:

   Shape mismatch causes broadcasting error [OK]

Hint: Check trailing dimensions match or are 1 for broadcasting [OK]

Common Mistakes:

• Assuming any shapes can broadcast
• Ignoring dimension order in arrays
• Thinking 1D arrays always broadcast to 2D

5. You have a 2D array data of shape (5, 4) representing 5 samples with 4 features each. You want to subtract the mean of each feature (given as a 1D array mean_vals of length 4) from the corresponding feature column in data. Which code correctly applies broadcasting to do this?

hard

A. data - mean_vals.reshape(4, 1)

B. data - mean_vals

C. data - mean_vals.reshape(1, 5)

D. data - mean_vals.T

Solution

1. Step 1: Understand shapes and operation

   data shape is (5, 4), mean_vals shape is (4,). Subtracting mean_vals from data broadcasts mean_vals across rows.

2. Step 2: Confirm correct broadcasting

   Since mean_vals length matches columns, data - mean_vals subtracts feature-wise means correctly.

3. Final Answer:

   data - mean_vals -> Option B

4. Quick Check:

   Subtract 1D array from 2D rows = data - mean_vals [OK]

Hint: Subtract 1D array matching columns directly [OK]

Common Mistakes:

• Reshaping mean_vals incorrectly causing errors
• Transposing 1D array which has no effect
• Trying to subtract mean_vals across wrong axis