Spectral Theory and Applications

Spectral theory studies the eigenvalue decomposition of linear operators, connecting abstract algebra to geometry and analysis. Applications range from differential equations to quantum mechanics.

TheoremSpectral Theorem (Preview)

A real symmetric matrix $A$ (or complex Hermitian matrix) can be orthogonally diagonalized: $A = Q\Lambda Q^T$

where $Q$ is orthogonal ( $Q^TQ = I$ ) with columns being orthonormal eigenvectors, and $\Lambda$ is diagonal with eigenvalues.

Moreover, all eigenvalues of a real symmetric matrix are real.

This theorem, explored fully in later chapters, is fundamental to many applications including principal component analysis, quantum mechanics, and optimization theory.

ExampleDynamical Systems

Consider the discrete dynamical system $\mathbf{x}_{n+1} = A\mathbf{x}_n$ with initial condition $\mathbf{x}_0$ .

If $A = PDP^{-1}$ is diagonalizable, then: $\mathbf{x}_n = A^n\mathbf{x}_0 = PD^nP^{-1}\mathbf{x}_0$

The long-term behavior is determined by the eigenvalues:

If all $|\lambda_i| < 1$ , then $\mathbf{x}_n \to \mathbf{0}$ (stable equilibrium)
If any $|\lambda_i| > 1$ , the system grows unbounded in that direction
The dominant eigenvalue (largest $|\lambda|$ ) controls asymptotic behavior

TheoremDiagonalization of Symmetric Matrices

Let $A$ be a real $n \times n$ symmetric matrix. Then:

All eigenvalues of $A$ are real
Eigenvectors corresponding to distinct eigenvalues are orthogonal
$A$ has $n$ orthonormal eigenvectors forming a basis for $\mathbb{R}^n$
$A$ is orthogonally diagonalizable: $A = Q\Lambda Q^T$ where $Q$ is orthogonal

These properties make symmetric matrices the "nicest" class for computations and applications.

ExampleDifferential Equations

The system of ODEs $\mathbf{x}'(t) = A\mathbf{x}(t)$ has solution: $\mathbf{x}(t) = e^{At}\mathbf{x}_0$

where the matrix exponential is $e^{At} = I + At + \frac{(At)^2}{2!} + \frac{(At)^3}{3!} + \cdots$

If $A = PDP^{-1}$ , then: $e^{At} = Pe^{Dt}P^{-1}$

where $e^{Dt}$ is diagonal with entries $e^{\lambda_i t}$ . The eigenvalues $\lambda_i$ determine stability: $\text{Re}(\lambda_i) < 0$ for all $i$ implies stability.

TheoremPower Method

To find the dominant eigenvalue (largest $|\lambda|$ ) and its eigenvector:

Start with random vector $\mathbf{v}_0$
Iterate: $\mathbf{v}_{k+1} = \frac{A\mathbf{v}_k}{\|A\mathbf{v}_k\|}$ (normalize)
As $k \to \infty$ , $\mathbf{v}_k$ converges to the dominant eigenvector
The Rayleigh quotient $\frac{\mathbf{v}_k^TA\mathbf{v}_k}{\mathbf{v}_k^T\mathbf{v}_k}$ converges to the dominant eigenvalue

This algorithm is the basis for PageRank and many numerical eigenvalue methods.

Remark

The spectral perspective—viewing operators through their eigenvalues and eigenvectors—is transformative across mathematics and physics. In quantum mechanics, observables are represented by operators, and measurements yield eigenvalues. In data science, principal component analysis finds eigenvectors of covariance matrices to identify patterns. The eigenvalue decomposition is truly one of the most powerful tools in applied mathematics.