Broadcasting compatibility check in NumPy - Step-by-Step Execution

Concept Flow - Broadcasting compatibility check

Start with two arrays

↓

Compare shapes from right to left

↓

For each dimension:

↓

If all dims compatible

↓

Else

Check if two arrays can be broadcast together by comparing their shapes from right to left, dimension by dimension.

Execution Sample

NumPy

import numpy as np
shape1 = (4, 1, 3)
shape2 = (5, 1)
np.broadcast_shapes(shape1, shape2)

Check if arrays with shapes (4,1,3) and (5,1) can broadcast together.

Execution Table

Step	Dimension (from right)	Size in shape1	Size in shape2	Compatibility Check	Result
1	Rightmost (dim -1)	3	1	3 vs 1: one is 1, compatible	Compatible
2	Next (dim -2)	1	5	1 vs 5: one is 1, compatible	Compatible
3	Next (dim -3)	4	No dim (treated as 1)	4 vs 1: one is 1, compatible	Compatible
4	No more dims	-	-	All dims checked	Broadcasting possible

💡 All dimensions compatible by broadcasting rules

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	Final
shape1	(4,1,3)	(4,1,3)	(4,1,3)	(4,1,3)	(4,1,3)
shape2	(5,1)	(5,1)	(5,1)	(5,1)	(5,1)
compatibility	Unknown	Compatible	Compatible	Compatible	Compatible

Key Moments - 2 Insights

Why do we treat missing dimensions as size 1 when shapes have different lengths?

Because broadcasting rules say shorter shapes are left-padded with 1s to match lengths, as shown in step 3 of execution_table.

What happens if neither dimension sizes are equal nor one is 1?

Broadcasting fails immediately; this is shown by the compatibility check in execution_table where both sizes must be equal or one must be 1.

Visual Quiz - 3 Questions

Look at the execution table, what is the compatibility result at step 2?

A Not compatible

B Compatible

C Error

D Unknown

💡 Hint

Check the 'Result' column at step 2 in the execution_table.

At which step do we treat a missing dimension as size 1?

A Step 1

B Step 2

C Step 3

D No step

💡 Hint

Look at the 'Size in shape2' column in execution_table where shape2 has no dim at step 3.

If shape1 was (4,2,3) instead of (4,1,3), what would happen at step 2?

A Not compatible

B Error in step 1

C Still compatible

D Broadcasting possible

💡 Hint

Compare sizes at step 2: 2 vs 5, neither equal nor 1, see execution_table logic.

Concept Snapshot

Broadcasting compatibility check:
- Compare shapes from right to left
- Dimensions must be equal or one must be 1
- Missing dims treated as 1
- If all dims compatible, broadcasting works
- Else, error occurs

Full Transcript

Broadcasting compatibility check compares two array shapes from right to left. For each dimension, if sizes are equal or one is 1, they are compatible. If shapes differ in length, missing dimensions are treated as 1. If all dimensions pass these checks, broadcasting is possible. Otherwise, it fails. This process helps numpy combine arrays of different shapes safely.