Normed Spaces and Banach Spaces

Motivation: structure compatible with linearity

A metric space is "bare" distance. A normed space adds linear structure: you can add elements and multiply them by scalars, and the norm is consistent with these operations. Banach spaces—complete normed spaces—are the main object of functional analysis and numerical methods. Here live linear operators, integral equations, and optimization methods.

Norms and Normed Spaces

A norm ‖·‖: V → ℝ on a vector space V (over ℝ or ℂ):

1. ‖x‖ ≥ 0; ‖x‖ = 0 ⟺ x = 0.
2. ‖αx‖ = |α|·‖x‖ (homogeneity).
3. ‖x+y‖ ≤ ‖x‖ + ‖y‖ (triangle inequality).

A metric from a norm: d(x,y) = ‖x−y‖. Not every metric arises from a norm.

Examples of norms on ℝⁿ:

• ‖x‖₁ = Σ|xᵢ| ("Manhattan" norm).
• ‖x‖₂ = √(Σxᵢ²) (Euclidean norm).
• ‖x‖_∞ = max|xᵢ| ("Chebyshev" norm).

Equivalence of norms: ‖·‖_a ≈ ‖·‖_b if ∃C₁,C₂: C₁‖x‖_a ≤ ‖x‖_b ≤ C₂‖x‖_a. In finite-dimensional spaces, all norms are equivalent. In infinite-dimensional spaces—not so.

Banach space: normed space, complete with respect to the metric ‖x−y‖.

Examples: (ℝⁿ, ‖·‖_p), (lᵖ, ‖·‖p), (C[a,b], ‖·‖∞), (Lᵖ[a,b], ‖·‖_p) for p ∈ [1,∞].

Key Theorems About Banach Spaces

Banach–Steinhaus Theorem (Principle of Uniform Boundedness): If {Aₙ} is a family of bounded linear operators X→Y (X is Banach) and sup_n ‖Aₙx‖ < ∞ for each x, then sup_n ‖Aₙ‖ < ∞.

Open Mapping Theorem: A surjective bounded linear operator T: X→Y between Banach spaces is an open mapping. Corollary: a bounded bijective operator has a bounded inverse (Inverse Operator Theorem).

Closed Graph Theorem: A linear operator T: X→Y with a closed graph Gr(T) = {(x,Tx)} is bounded.

Series in Banach Spaces

In a Banach space: Σ‖xₙ‖ < ∞ ⟹ Σxₙ converges (absolute convergence implies conditional). Cauchy criterion: Σxₙ converges ⟺ ‖xₙ₊₁ + ... + xₙ₊ₘ‖ → 0.

Numerical Example

Problem: Find the norm of the operator T: C[0,1] → C[0,1], Tf(x) = ∫₀ˣ f(t) dt.

Step 1. Estimate |Tf(x)| = |∫₀ˣ f(t)dt| ≤ ∫₀ˣ |f(t)|dt ≤ x·‖f‖∞ ≤ 1·‖f‖∞. Therefore, ‖Tf‖∞ ≤ ‖f‖∞ → ‖T‖ ≤ 1.

Step 2. Show ‖T‖ = 1: take f ≡ 1. Then Tf(x) = x. ‖Tf‖∞ = 1 = ‖f‖∞. Thus, ‖T‖ ≥ 1.

Step 3. Result: ‖T‖ = 1. The operator is bounded.

Step 4. Comparison of norms on C[0,1]: ‖f‖∞ ≥ ‖f‖{L²} ≥ ‖f‖{L¹}. For fₙ(x) = xⁿ: ‖fₙ‖∞ = 1, ‖fₙ‖{L²} = 1/√(2n+1) → 0, ‖fₙ‖{L¹} = 1/(n+1) → 0. The norms are not equivalent on C[0,1] (infinite-dimensional space) ✓.

Step 5. The equality ‖T‖ = 1 corresponds to the fact that integration "does not increase" the uniform norm. The differentiation operator d/dx: C¹[0,1] → C[0,1] is unbounded: fₙ(x) = sin(nx)/n → 0 uniformly, but fₙ'(x) = cos(nx) does not → 0. This explains why numerical differentiation is unstable.

Practical Application

Numerical analysis: estimates of errors of numerical methods are estimates of norms of residuals. The Inverse Operator Theorem guarantees that if a linear system is well-conditioned (inverse operator is bounded), then small changes in the right-hand side produce small changes in the solution.

Additional Aspects

The norm ‖x‖ generalizes the concept of length: homogeneity ‖αx‖ = |α|·‖x‖, triangle inequality ‖x+y‖ ≤ ‖x‖ + ‖y‖, positive definiteness. Different norms on the same space can give different topologies (in infinite-dimensional spaces). Concrete Banach spaces are everywhere: C[a,b] with sup-norm (continuous functions, Weierstrass approximation theorems), L^p (functions integrable to the p-th power, foundation of probability theory and quantum mechanics for p=2), ℓ^p (sequences), Sobolev spaces W^{k,p} (basis of PDE theory). The Hahn–Banach theorem guarantees that in any Banach space there exists "enough" linear functionals—the dual space X* is non-trivial.

Connection With Other Areas of Mathematics

Banach spaces naturally arise in the theory of differential equations. Partial differential equations are often rewritten as operator equations in Lp or Sobolev spaces Wk,p. A classical example is the formulation of the Dirichlet problem via weak solution in H1,0(Ω) and application of the Riesz representation theorem for continuous linear functionals in Hilbert spaces. The Lax–Milgram theorem (1954) uses completeness and continuity of the bilinear form for existence and uniqueness of solutions to elliptic problems.

In algebra, Banach algebras combine linear and multiplicative structure. Spectral theory in such algebras generalizes matrix decomposition to the infinite-dimensional case. Gelfand's theorem (1941) links commutative Banach algebras with compact topological spaces via maximal ideals and continuous functions on the spectrum.

Topology is present through locally convex spaces: many results (Hahn–Banach, open mapping principle) have versions for broader classes of topological vector spaces. Compactness of operators, studied by Fredholm and Riesz, relies on the metric structure of normed space and the concept of relatively compact sets.

In probability theory, spaces Lp(Ω, F, P) are used to describe random variables and processes. The Banach–Steinhaus principle of uniform boundedness resonates with the Lebesgue Dominated Convergence Theorem and plays a role in the study of convergence of sequences of random variables in norms. In numerical methods, a priori estimates in the norm, scheme stability, and the Lipschitz condition for fixed-point operators (Banach’s contraction mapping theorem) underpin the simple iteration method and Newton’s method.

Historical Reference and Development of the Idea

The first clear formulations of normed spaces appeared with Fredholm and F. Riesz in the early 20th century during the study of integral equations. Stefan Banach, in his book "Théorie des opérations linéaires" (Monografie Matematyczne, 1932), systematized the concept of a complete normed space and laid the foundation of functional analysis. The Hahn–Banach theorem was independently obtained by Hans Hahn (1927) and Banach (1929), motivated by the problem of extending linear functionals to spaces of functions. The open mapping theorem and closed graph principle were formulated by Banach and his students (Mazur, Orlicz, Kistenevliut) in the 1930s in the Lwów mathematical school. Development in the mid-20th century is associated with the works of L. Schwartz on distributions, where normed spaces are supplemented by more general topological constructions, and with Gelfand’s theory of Banach algebras. John von Neumann and Marshall Stone applied Hilbert and Banach spaces to quantum mechanics and spectral theory of operators.