The Python Matrix Gotcha That Silently Corrupts Your Data

#python #bugs #programming #tutorial

Here's innocent-looking code to create a 2×3 matrix of zeros:

>>> row = [0, 0, 0]
>>> matrix = [row for _ in range(2)]
>>> matrix
[[0, 0, 0], [0, 0, 0]]

Looks perfect. Now let's change one cell:

>>> matrix[0][0] = 99
>>> matrix
[[99, 0, 0], [99, 0, 0]]    # BOTH rows changed!

We modified matrix[0][0], but matrix[1][0] changed too. What happened?

The Problem: Aliasing

When we wrote [row for _ in range(2)], we didn't create two rows. We created two references to the same row.

matrix[0] ──┐
            ├──► [99, 0, 0]  (same list object)
matrix[1] ──┘

Every "row" in our matrix is the same object in memory. Change one, change all.

The Fix: Create New Lists

Each row must be a new list object:

>>> matrix = [[0, 0, 0] for _ in range(2)]
>>> matrix[0][0] = 99
>>> matrix
[[99, 0, 0], [0, 0, 0]]    # Only first row changed ✓

Or more generally:

>>> rows, cols = 2, 3
>>> matrix = [[0 for _ in range(cols)] for _ in range(rows)]

The inner comprehension [0 for _ in range(cols)] runs fresh for each row, creating a new list each time.

Why This Trips People Up

The original code reads correctly: "make a list of row, two times."

But in Python, row is a reference. Repeating a reference doesn't copy the underlying object—it just creates more pointers to it.

Quick Diagnostic

Suspicious your matrix has this bug? Check with is:

>>> bad_matrix = [row for _ in range(2)]
>>> bad_matrix[0] is bad_matrix[1]
True    # Same object! Bug confirmed.

>>> good_matrix = [[0, 0, 0] for _ in range(2)]
>>> good_matrix[0] is good_matrix[1]
False   # Different objects. Safe.

The Rule

When creating nested structures with comprehensions:

Never repeat a reference. Always create fresh.

# WRONG - reuses same list
template = [0, 0, 0]
matrix = [template for _ in range(n)]

# RIGHT - creates new list each time
matrix = [[0, 0, 0] for _ in range(n)]

This applies to lists of lists, lists of dicts, lists of any mutable object.

This is adapted from my upcoming book, Zero to AI Engineer: Python Foundations.
I share excerpts like this on Substack → https://substack.com/@samuelochaba

DEV Community