Condition Number Calculator

Calculate matrix condition number to assess numerical stability and error sensitivity

Matrix Input & Configuration

Quick Examples

Condition Number Analysis

7895.000000
Condition Number
0.010000
Determinant
Ill-conditioned
Status

Interpretation

❌ The matrix is ill-conditioned. Numerical computations may be unreliable.
Error Magnification: Input errors in linear systems Ax = b can be magnified by up to 7895 times.

Condition Number Guide

Well-Conditioned (< 10)

Stable numerical computations, low error sensitivity.

Moderate (10-100)

Some sensitivity to input errors, generally acceptable.

Poor (100-1000)

High error sensitivity, use caution in computations.

Ill-Conditioned (> 1000)

Very unstable, results may be unreliable.

Matrix Norms

||A||₁ = max_j Σᵢ |aᵢⱼ|

1-norm (column sum)

||A||₂ = σ_max(A)

2-norm (spectral)

||A||∞ = max_i Σⱼ |aᵢⱼ|

∞-norm (row sum)

Condition Number Formula

cond(A) = ||A|| × ||A⁻¹||

Product of matrix and inverse norms

• If det(A) = 0, then cond(A) = ∞

• Minimum value: cond(A) = 1

• Identity matrix: cond(I) = 1

• Error bound: ||δx||/||x|| ≤ cond(A) × ||δb||/||b||

Understanding Condition Numbers

What is a Condition Number?

The condition number of a matrix measures how sensitive the solution of a linear system is to small changes in the input. It's defined as cond(A) = ||A|| × ||A⁻¹||, where ||·|| represents a matrix norm.

Physical Interpretation

  • Measures the ratio of maximum stretching to maximum shrinking
  • Indicates how "close" a matrix is to being singular
  • Determines error magnification in linear systems

Applications

Condition numbers are crucial in numerical analysis for assessing the reliability of computational results and predicting how errors propagate through calculations.

Linear Systems

For Ax = b, the condition number determines how errors in b affect the solution x.

Matrix Inversion

High condition numbers indicate numerical instability in matrix inversion.

Least Squares

Affects the reliability of regression and curve fitting solutions.

Example: Error Magnification

Problem Setup

A = [3.7 0.9]
    [7.8 1.9]

b = [7.4]
    [15.6]

Solution: x = [2, 0]ᵀ

With Small Error

b_err = [7.41]
        [15.6]

x_err = [3.9]
        [-7.8]

Tiny change in b causes huge change in x!

Result: With cond(A) ≈ 7,895, a 0.13% error in b produces a 400% error in x.