Approximate comparisons (pytest.approx)

Introduction

Sometimes numbers are close but not exactly equal due to small differences. Approximate comparisons help check if values are close enough.

When testing calculations with floating point numbers that may have tiny rounding errors.

When comparing results from different machines or environments that produce slightly different outputs.

When checking values that come from measurements or sensors with small variations.

When you want to allow a small margin of error in your test assertions.

Syntax

PyTest

assert actual == pytest.approx(expected, rel=relative_tolerance, abs=absolute_tolerance)

expected is the value you want to compare against.

You can set rel for relative tolerance and abs for absolute tolerance to control how close values must be.

Examples

Checks if 0.1 + 0.2 is approximately equal to 0.3, allowing for small floating point errors.

PyTest

assert 0.1 + 0.2 == pytest.approx(0.3)

Allows a 1% relative difference between values.

PyTest

assert 100.0 == pytest.approx(100.1, rel=1e-2)

Allows an absolute difference of 0.001, useful for very small numbers.

PyTest

assert 0.0001 == pytest.approx(0.0, abs=1e-3)

Sample Program

This test checks three cases where values are close but not exactly equal. It uses pytest.approx to allow small differences.

PyTest

import pytest

def test_approximate_comparison():
    result = 0.1 + 0.2
    expected = 0.3
    assert result == pytest.approx(expected)

    # Test with relative tolerance
    value = 100.0
    expected_value = 100.1
    assert value == pytest.approx(expected_value, rel=1e-2)

    # Test with absolute tolerance
    small_value = 0.0001
    zero = 0.0
    assert small_value == pytest.approx(zero, abs=1e-3)

Important Notes

Use pytest.approx to avoid test failures caused by tiny floating point errors.

You can combine rel and abs to fine-tune how close values must be.

Approximate comparisons work for floats and sequences like lists or tuples of floats.

Summary

Approximate comparisons help test values that are close but not exactly equal.

Use pytest.approx with optional relative and absolute tolerances.

This avoids false test failures due to small rounding differences.