Skip to content

Condition Number and Sensitivity Visualizer

Run the Condition Number Visualizer Fullscreen

Edit the MicroSim in the p5.js Editor

About This MicroSim

This visualization demonstrates how the condition number affects the sensitivity of linear system solutions. When solving Ax = b:

  • Well-conditioned (κ ≈ 1): Small changes in b cause small changes in x
  • Ill-conditioned (κ large): Small changes in b cause large changes in x

Key Concepts

The condition number κ(A) = σ₁/σ₂ bounds how much errors amplify:

\[\frac{\|\delta \mathbf{x}\|}{\|\mathbf{x}\|} \leq \kappa(A) \cdot \frac{\|\delta \mathbf{b}\|}{\|\mathbf{b}\|}\]

Visual Elements

  • Two lines: Represent the two equations in the 2×2 system
  • Green point: The solution x
  • Orange cloud: Perturbed solutions when b is slightly changed
  • Bar chart: Shows relative sizes of σ₁ and σ₂

How to Use

  1. Select a preset to see different conditioning levels
  2. Adjust ε to control perturbation magnitude
  3. Toggle perturbations to show/hide the solution cloud
  4. Observe how the cloud grows with condition number

Presets

Preset κ Lines Behavior
Well-conditioned ~1 Perpendicular Stable
Moderate ~10 Angled Some spread
Ill-conditioned ~1000 Nearly parallel Large spread
Nearly Singular ~∞ Almost same Unstable

Learning Objectives

After using this MicroSim, students will be able to:

  • Interpret condition number geometrically
  • Predict solution sensitivity from κ
  • Recognize ill-conditioned systems visually
  • Connect singular values to conditioning

References

  • Chapter 7: Matrix Decompositions - Condition Number section
  • Numerical stability in scientific computing