The Spectral Theorem

Definition

The Spectral Theorem is a fundamental result in linear algebra stating that every real symmetric matrix ($S^T = S$) can be diagonalized by an orthogonal matrix $Q$. $S = Q \Lambda Q^T$

• How to read: “S equals Q times Lambda times Q-transpose; S-transpose equals S.”
• Meaning: Any real symmetric matrix diagonalizes orthogonally—$Q$ holds eigenvectors, $\Lambda$ holds real eigenvalues on the diagonal.

Why It Matters

The Spectral Theorem is the ‘orthonormal bridge’ of linear algebra; it provides the mathematical guarantee that symmetric systems can be decomposed into their most fundamental, independent components, making it indispensable for everything from quantum mechanics to vibration analysis.

Core Concepts

• Key Conclusions for Symmetric Matrices:
  1. All eigenvalues are real (not complex).
  2. Eigenvectors corresponding to distinct eigenvalues are orthogonal.
  3. There exists an orthonormal basis of eigenvectors.
• Spectral Decomposition: The matrix $S$ can be written as a sum of projections onto its eigenvectors: $S = \lambda_1 \mathbf{q}_1 \mathbf{q}_1^T + \dots + \lambda_n \mathbf{q}_n \mathbf{q}_n^T$
  • How to read: “The matrix S equals lambda one times q one times q one transpose, and so on, up to lambda n times q n times q n transpose.”
  • Meaning: Each term is a rank-one projection along one eigenvector, scaled by its eigenvalue. The full matrix action is the sum of independent stretchings along mutually perpendicular directions.
• Comparison to General Diagonalization: While general diagonalization ($A = X \Lambda X^{-1}$) requires $n$ independent eigenvectors, the Spectral Theorem guarantees that for symmetric matrices, these eigenvectors are not just independent but orthonormal ($Q^T = Q^{-1}$).
  • How to read: “Matrix A equals the eigenvector matrix X times the diagonal matrix Lambda times the inverse of X; for a symmetric matrix S, the transpose of Q equals the inverse of Q.”
  • Meaning / when to use: General matrices may need a non-orthogonal change-of-basis matrix $X$. Symmetric matrices get the stronger guarantee: an orthonormal eigenbasis, so $Q^{-1}$ is simply $Q^T$ (no matrix inversion needed).

Connected Concepts

• Diagonalization
• Eigenvalues and Eigenvectors
• Hermitian Matrices
• Principal Component Analysis