Spectral Theory of Linear Operators

Motivation: Generalization of Eigenvalues

In matrix algebra, the spectrum of a matrix is its set of eigenvalues. In infinite-dimensional spaces, the concept is richer: the spectrum can be continuous (multiplication operator), point (eigenvalues), or residual. Spectral theory is the language of quantum mechanics: observables correspond to self-adjoint operators; their spectrum consists of the possible measurement results.

Spectrum of an Operator

Resolvent set ρ(T): λ ∈ ρ(T) if (T−λI)⁻¹ exists as a bounded operator on all of X.

Spectrum σ(T) = ℂ \ ρ(T). It has three parts:

• Point spectrum σₚ(T): ∃ x ≠ 0: (T−λI)x = 0 → λ is an eigenvalue.
• Continuous spectrum σ_c(T): (T−λI) is injective with a dense, but not everywhere defined, inverse.
• Residual spectrum σ_r(T): (T−λI) is injective, but its range is not dense.

Multiplication operator: T: L²[a,b]→L²[a,b], Tf = g·f. σ(T) = closure of the image of g. σₚ(T) = ∅ (no eigenvalues) — continuous spectrum.

Shift operator: T: l²→l², T(x₁,x₂,...) = (0,x₁,x₂,...). σₚ = ∅, σ_r = open unit disk, σ_c = unit circle.

Self-Adjoint Operators

Self-adjointness: ⟨Tx,y⟩ = ⟨x,Ty⟩ for all x,y ∈ D(T). In finite dimension: T = Tᵀ.

Properties: spectrum σ(T) ⊆ ℝ; eigenvectors for distinct λ are orthogonal; the quadratic form Q(x) = ⟨Tx,x⟩ determines T.

Spectral theorem (compact self-adjoint T): σ(T) ⊆ ℝ, σₚ is countable, |λₙ| → 0. There exists an orthonormal basis {φₙ} of eigenfunctions, T = Σₙ λₙ⟨·,φₙ⟩φₙ.

Unitary Operators

Unitary U: UU = UU = I. Isometry: ‖Ux‖ = ‖x‖. Spectrum lies on the unit circle. Examples: Fourier transform on L²(ℝ), rotation operators.

Numerical Example

Problem: Find the eigenvalues and eigenfunctions of the operator T = −d²/dx² on L²[0,π] with Dirichlet conditions u(0) = u(π) = 0.

Step 1. Problem: −u'' = λu, u(0) = u(π) = 0.

Step 2. General solution for λ > 0: u = A cos(√λ x) + B sin(√λ x). Condition u(0)=0 → A=0. Condition u(π)=0 → sin(√λ π) = 0 → √λ π = nπ → λₙ = n², n = 1,2,3,...

Step 3. Eigenfunctions: φₙ(x) = sin(nx). Normalize: ‖sin(nx)‖² = ∫₀^π sin²(nx)dx = π/2. Normalized: φₙ = √(2/π) sin(nx).

Step 4. Check self-adjointness: ⟨−u'',v⟩ = ∫₀^π (−u'')v dx = ∫₀^π u'v' dx (integration by parts, boundary conditions) = ⟨u, −v''⟩ ✓.

Step 5. Physical meaning: eigenvalues λₙ = n² are the frequencies of the natural vibrations of a string of length π. The higher the n, the higher the overtone. The expansion f(x) = Σₙ cₙ sin(nx) is the expansion into eigenfunctions of the string.

Real-World Application

Quantum mechanics: "particle in a potential box" — a potential well with infinite walls — is described by the same operator T = −(ℏ²/2m)d²/dx². Eigenvalues Eₙ = n²ℏ²π²/(2mL²) are discrete energy levels. This is energy quantization: quantum mechanics explained by spectral theory.

Additional Aspects

A linear operator T: X → Y between Banach spaces is called bounded if ‖T‖ = sup_{‖x‖≤1} ‖Tx‖ < ∞. The set B(X,Y) of bounded operators itself forms a Banach space. The spectrum of an operator σ(T) is the set of λ for which T − λI does not have a bounded inverse — it generalizes eigenvalues. For a self-adjoint operator in a Hilbert space, the spectral theorem provides an integral representation T = ∫λ dE(λ), where E is a projection-valued measure. This is the foundation of quantum mechanics (observables ↔ self-adjoint operators) and numerical methods of spectral analysis of differential operators.

Operator theory turns functional analysis into a practical tool: the spectral theorem describes observables in quantum mechanics, the theory of compact operators is at the core of numerical methods for solving integral equations, and semigroup theory gives a general framework for evolution equations and stochastic processes.

Connection with Other Branches of Mathematics

In the theory of differential equations, the spectral approach arises in the separation of variables method and the Sturm–Liouville theory of eigenfunctions. The classic result of Courant and Hilbert on the completeness of eigenfunctions for a second-order self-adjoint operator ensures the decomposition of solutions as a series over the spectrum. For evolutionary problems, the semigroup apparatus is used: the generator of a strongly continuous semigroup on a Banach space describes the spectral properties of the Cauchy problem; this is formalized in the Hille–Yosida and Lumer–Phillips theorems.

With algebra, spectral theory is linked through representations of operator algebras. The spectral theorem for commutative C*-algebras (Gelfand–Naimark) translates a self-adjoint element into a continuous function on the maximal ideal space, and the spectrum of the operator coincides with the range of the corresponding function. In Connes' noncommutative geometry, spectral triples use the spectrum of the Dirac operator to recover the metric on a "noncommutative space".

Topological aspects appear in the theory of the index of Fredholm operators and K-theory. The index of an operator, introduced by Atiyah and Singer, links the analytic properties of the spectrum with the characteristic classes of manifolds. For elliptic operators, spectral invariants such as the η-invariant of Atiyah–Patodi–Singer relate the asymmetry of the spectrum to global topology.

In probability theory, spectral analysis of Markov chain transition operators and diffusion generators describes the rate of convergence to the stationary distribution. The work of Dunford–Schwartz on linear operators formalized the connection between spectrum and ergodic properties. In numerical analysis, spectral methods (Kantorovich, Kollatz) use the approximation of the spectrum of differential operators by matrices; the stability of iterative eigenvalue estimation schemes is studied via the spectral radius.

Historical Note and the Development of the Idea

The first ideas about eigenvalues appeared with Euler and Lagrange in the 18th century while studying vibrations of strings and plates. The term "eigenwert" was introduced by Hilbert at the turn of the 19th–20th centuries, as he developed the theory of integral equations in his works "Grundzüge einer allgemeinen Theorie der linearen Integralgleichungen" (1912). It was here that the concept of the spectrum as a generalization of the set of eigenvalues arose. In the 19th century, Sturm and Liouville studied spectra of differential operators on an interval, introducing boundary value problems that today are seen as classical self-adjoint operators. Fredholm and Riesz developed the theory of compact operators and laid the foundation for the spectral theorem in Hilbert spaces. In 1929–1932, von Neumann formalized self-adjoint and unitary operators as the mathematical framework of quantum mechanics, describing in "Mathematische Grundlagen der Quantenmechanik" the connection between spectrum and measurement. In the second half of the 20th century, spectral theorems for unbounded operators appeared (work by Stone, Nelson), abstract theory of C*-algebras (Gelfand, Naimark), and semigroup theory.