Course contentsShow
← Mathematics
Lesson 2296 of 3,12147. Linear Algebra — Applications and Advanced TopicsFree lesson

Diagonalization of Quadratic Forms

Transform quadratic forms into canonical form using eigenvalue decomposition and change of variables.

Diagonalization of Quadratic Forms

What You'll Learn

You'll learn how to transform a quadratic form into its simplest canonical form by using eigenvalue decomposition and a change of variables.

The Core Idea

A quadratic form like Q(x) = x^T A x (where A is symmetric) can often look complicated with mixed terms like xy or xz. Diagonalization eliminates these cross-terms, transforming Q into a sum of pure squares.

Here's how it works:

  1. Find eigenvalues and eigenvectors of the symmetric matrix A
  2. Form the orthogonal matrix P whose columns are orthonormal eigenvectors
  3. Change variables by setting x = Py, where y is a new coordinate system
  4. The result: Q(x) = x^T A x = y^T (P^T A P) y = y^T D y, where D is diagonal with eigenvalues on the diagonal

In the new coordinate system y, the quadratic form becomes simply:

Q(y) = λ₁y₁² + λ₂y₂² + ... + λₙyₙ²

No more cross-terms! This is the canonical form.

Why It Matters

Think of it like rotating and stretching a coordinate system until an ellipse (or hyperbola) aligns perfectly with the axes. The principal axes theorem tells us this is always possible for symmetric matrices. The eigenvalues tell you how much the form stretches along each principal axis, while eigenvectors point in those principal directions.

This reveals the geometric nature of the quadratic form: whether it's an ellipsoid (all positive eigenvalues), hyperboloid (mixed signs), or saddle (indefinite).

Example Insight

If Q(x,y) = 3x² + 4xy + 3y², diagonalization might reveal it's actually 5u² + v² in rotated coordinates (u,v) — a stretched ellipse aligned with new axes.

Key Takeaway: Diagonalizing a quadratic form uses eigenvalue decomposition to eliminate cross-terms via an orthogonal change of basis, revealing the form's true geometric structure as a sum of squares.