Important Types of Spectra

Why This Matters

In spectral theory, understanding the different types of spectra isn't just about memorizing definitions—it's about recognizing what each spectrum type tells you about an operator's fundamental behavior. You're being tested on your ability to distinguish how eigenvalues are distributed, whether eigenfunctions exist and are normalizable, and what physical systems exhibit each spectral type. These distinctions are crucial for analyzing everything from quantum mechanical systems to dynamical stability problems.

The spectrum of an operator decomposes into pieces that reveal different aspects of the system's structure. When you encounter a self-adjoint operator, you need to know whether its spectrum is discrete (isolated eigenvalues), continuous (a continuum of values), or some mixture of both. Don't just memorize which spectrum is which—understand what mathematical and physical conditions produce each type, and be ready to identify examples that illustrate the underlying principles.

---

The Primary Decomposition: Point, Continuous, and Residual

Every spectrum of a bounded operator can be partitioned into three disjoint pieces based on whether $\lambda$ is an eigenvalue and whether $(T - \lambda I)$ has dense range. This trichotomy is the foundation for all spectral classification.

Point Spectrum

• Set of all eigenvalues—values $\lambda$ where $T\psi = \lambda\psi$ has a non-zero solution $\psi$
• Operator acts like scalar multiplication on the corresponding eigenspace, making these the most "well-behaved" spectral points
• Critical for stability analysis in dynamical systems, since eigenvalues determine growth/decay rates of solutions

Residual Spectrum

• Values where $(T - \lambda I)$ is injective but lacks dense range—no eigenvector exists, yet $\lambda$ still obstructs invertibility
• Absent for self-adjoint operators on Hilbert spaces, making it primarily relevant for non-normal operators
• Signals non-compactness issues and appears in spectral decomposition of certain unbounded operators

Continuous Spectrum

• Values where $(T - \lambda I)$ is injective with dense but not closed range—approximate eigenvalues exist but no true eigenvector
• Characterized by approximate eigenvectors: sequences $\psi_n$ with $\|T\psi_n - \lambda\psi_n\| \to 0$ but no convergent subsequence
• Typical for multiplication operators on $L^2$ spaces, where the spectrum equals the essential range of the multiplier function

---

Discrete vs. Essential: Stability Under Perturbations

This classification asks: which parts of the spectrum are stable under compact perturbations? The essential spectrum captures the "robust" part, while discrete eigenvalues can shift or disappear.

Discrete Spectrum

• Isolated eigenvalues with finite multiplicity—each eigenvalue is separated from others and has a finite-dimensional eigenspace
• Typical for compact operators and boundary value problems, such as the Laplacian on bounded domains with Dirichlet conditions
• Eigenvalues can be ordered and counted, enabling explicit spectral expansions like Fourier series

Essential Spectrum

• Invariant under compact perturbations—adding any compact operator $K$ to $T$ leaves $\sigma_{\text{ess}}(T+K) = \sigma_{\text{ess}}(T)$
• Contains accumulation points of the spectrum plus points of infinite multiplicity, capturing the operator's "asymptotic" behavior
• Key for Fredholm theory: $\lambda \notin \sigma_{\text{ess}}(T)$ if and only if $T - \lambda I$ is Fredholm

---

The Lebesgue Decomposition: Absolutely Continuous, Singular Continuous, and Pure Point

For self-adjoint operators, the continuous spectrum further decomposes based on the spectral measure's relationship to Lebesgue measure. This classification connects directly to the long-time behavior of quantum systems.

Absolutely Continuous Spectrum

• Spectral measure is absolutely continuous with respect to Lebesgue measure—$d\mu = f \, d\lambda$ for some density $f$
• Associated with scattering states in quantum mechanics, where particles escape to infinity as $t \to \infty$
• RAGE theorem implication: states in the absolutely continuous subspace have $\langle \psi(t), A\psi(t) \rangle \to 0$ for compact $A$

Singular Continuous Spectrum

• Spectral measure is continuous but supported on a set of Lebesgue measure zero—no eigenvalues, yet concentrated on a "thin" set
• Eigenfunctions are not square-integrable, leading to neither bound-state nor scattering behavior
• Arises in fractal and quasiperiodic systems, such as the almost Mathieu operator at critical coupling

Pure Point Spectrum

• Spectrum consists entirely of eigenvalues with normalizable eigenfunctions spanning the Hilbert space
• Indicates complete localization in quantum systems—particles remain bound and don't spread
• Typical for systems with strong disorder (Anderson localization) or confining potentials (harmonic oscillator)

---

Composite Spectral Types

Real operators often exhibit multiple spectral behaviors simultaneously. Recognizing mixed spectra is essential for analyzing physically realistic systems.

Mixed Spectrum

• Combination of point, absolutely continuous, and/or singular continuous components—the Hilbert space decomposes into orthogonal subspaces for each type
• Common in quantum mechanics: hydrogen atom has discrete bound states (point) plus ionization continuum (absolutely continuous)
• Spectral decomposition theorem guarantees this orthogonal splitting for any self-adjoint operator

Lebesgue Spectrum

• Spectrum describable via Lebesgue measure on the real line, often with uniform multiplicity
• Central to ergodic theory: a measure-preserving transformation has Lebesgue spectrum if its unitary operator has purely absolutely continuous spectrum
• Indicates mixing behavior in dynamical systems—correlations decay and the system "forgets" initial conditions

---

Quick Reference Table

---

Self-Check Questions

1. What distinguishes point spectrum from continuous spectrum in terms of the existence of eigenvectors versus approximate eigenvectors?

2. Why is the residual spectrum always empty for self-adjoint operators on Hilbert spaces, and what class of operators can have non-empty residual spectrum?

3. Compare and contrast absolutely continuous spectrum and singular continuous spectrum—what does each imply about the long-time behavior of a quantum state?

4. If you add a compact perturbation to a self-adjoint operator, which part of the spectrum remains unchanged, and which part might shift? Give an example illustrating this distinction.

5. The hydrogen atom Hamiltonian exhibits mixed spectrum. Identify which physical states correspond to the point spectrum component and which correspond to the absolutely continuous component, and explain the physical interpretation of each.