Computing Orthogonal Diagonalization
What you'll learn: You'll master the complete step-by-step process of finding an orthogonal matrix Q that diagonalizes a symmetric matrix A, putting all the pieces together into one concrete algorithm.
The Core Idea
Orthogonal diagonalization takes a symmetric matrix A and finds an orthogonal matrix Q and diagonal matrix D such that A = QDQ^T (equivalently, D = Q^T AQ). Unlike general diagonalization, the magic here is that Q's columns form an orthonormal set of eigenvectors. This is always possible for symmetric matrices thanks to the Spectral Theorem.
The Complete Algorithm
Step 1: Find Eigenvalues
Compute the characteristic polynomial det(A - λI) = 0 and solve for all eigenvalues λ₁, λ₂, ..., λₙ.
Step 2: Find Eigenvectors for Each Eigenvalue
For each λᵢ, solve (A - λᵢI)x = 0 to get the eigenspace. If the eigenvalue has multiplicity greater than 1, find a basis for the entire eigenspace.
Step 3: Orthogonalize Within Each Eigenspace
If any eigenspace is multi-dimensional, apply the Gram-Schmidt algorithm to obtain an orthogonal basis for that eigenspace. (Eigenvectors from different eigenvalues are already orthogonal!)
Step 4: Normalize All Eigenvectors
Divide each eigenvector by its norm to create unit vectors (orthonormal set).
Step 5: Form Q and D
Arrange the orthonormal eigenvectors as columns of Q. Place corresponding eigenvalues along the diagonal of D in the same order.
Step 6: Verify
Check that Q^T Q = I (confirming orthogonality) and Q^T AQ = D.
Key Takeaway: Computing orthogonal diagonalization systematically combines eigenvalue computation, eigenvector finding, Gram-Schmidt orthogonalization within eigenspaces, and normalization to produce an orthogonal matrix Q that cleanly diagonalizes any symmetric matrix.