Skip to content

The Fundamental Theorems

Theorem 4.1 (Hahn-Banach, Analytic Form). Let XX be a real vector space, p:X→Rp : X \to \mathbb{R} a sublinear functional (p(x+y)≤p(x)+p(y)p(x + y) \leq p(x) + p(y) and p(tx)=tp(x)p(tx) = tp(x) for t≥0t \geq 0), and f:M→Rf : M \to \mathbb{R} a linear functional on a subspace M⊆XM \subseteq X with f(x)≤p(x)f(x) \leq p(x) for all x∈Mx \in M. Then there exists a linear extension F:X→RF : X \to \mathbb{R} with F∣M=fF|_M = f and F(x)≤p(x)F(x) \leq p(x) for all x∈Xx \in X.

Theorem 4.2 (Hahn-Banach, Normed Form). Let XX be a normed space and M⊆XM \subseteq X a subspace. Every bounded linear functional f∈M∗f \in M^* extends to F∈X∗F \in X^* with ∥F∥=∥f∥\|F\| = \|f\|.

Corollary 4.3. For every x0∈Xx_0 \in X, x0≠0x_0 \neq 0, there exists φ∈X∗\varphi \in X^* with ∥φ∥=1\|\varphi\| = 1 and φ(x0)=∥x0∥\varphi(x_0) = \|x_0\|.

Theorem 4.4 (Open Mapping Theorem). If T∈B(X,Y)T \in \mathcal{B}(X, Y) is a surjective bounded operator between Banach spaces, then TT maps open sets to open sets.

Corollary 4.5 (Bounded Inverse Theorem). If T∈B(X,Y)T \in \mathcal{B}(X, Y) is bijective and X,YX, Y are Banach, then T−1∈B(Y,X)T^{-1} \in \mathcal{B}(Y, X).

Theorem 4.6 (Closed Graph Theorem). Let T:X→YT : X \to Y be a linear operator between Banach spaces. Then TT is bounded if and only if its graph Γ(T)={(x,Tx):x∈X}\Gamma(T) = \{(x, Tx) : x \in X\} is closed in X×YX \times Y.

Theorem 4.7 (Uniform Boundedness Principle / Banach-Steinhaus). Let XX be a Banach space and {Tα}α∈A⊆B(X,Y)\{T_\alpha\}_{\alpha \in A} \subseteq \mathcal{B}(X, Y) such that sup⁡α∥Tαx∥<∞\sup_\alpha \|T_\alpha x\| < \infty for each x∈Xx \in X. Then sup⁡α∥Tα∥<∞\sup_\alpha \|T_\alpha\| < \infty.

Proof sketch. Consider E={x∈X:sup⁡α∥Tαx∥≤n}E = \{x \in X : \sup_\alpha \|T_\alpha x\| \leq n\}. By the hypothesis, X=⋃nEnX = \bigcup_n E_n. By Baire category, some EnE_n has nonempty interior. Rescaling shows sup⁡α∥Tα∥<∞\sup_\alpha \|T_\alpha\| < \infty. ■\blacksquare

Application 1: Extension of linear functionals. The Hahn-Banach theorem guarantees that the dual space X∗X^* is rich enough to separate points: for any x≠0x \neq 0, there exists f∈X∗f \in X^* with f(x)≠0f(x) \neq 0.

Application 2: Banach limits. Hahn-Banach can be used to construct a Banach limit: a translation-invariant linear functional LIM:ℓ∞→R\mathrm{LIM} : \ell^\infty \to \mathbb{R} that extends the ordinary limit.

Application 3: Goldstine’s theorem. The unit ball of a Banach space XX is weak∗*-dense in the unit ball of the bidual X∗∗X^{**}, a consequence of the Hahn-Banach theorem.

4.6 Applications of the Open Mapping Theorem

Section titled “4.6 Applications of the Open Mapping Theorem”

Application 1: Equivalent norms. If XX is a Banach space under two norms ∥⋅∥1\|\cdot\|_1 and ∥⋅∥2\|\cdot\|_2 and there exists C>0C > 0 with ∥x∥1≤C∥x∥2\|x\|_1 \leq C\|x\|_2 for all xx, then the norms are equivalent.

Application 2: Sum of Banach spaces. If XX and YY are closed subspaces of a Banach space ZZ with X∩Y={0}X \cap Y = \{0\} and Z=X+YZ = X + Y, then the sum is a topological direct sum if and only if both XX and YY are closed.

4.7 Applications of the Closed Graph Theorem

Section titled “4.7 Applications of the Closed Graph Theorem”

Application. If T:X→YT : X \to Y is a linear operator between Banach spaces with the property that xn→0x_n \to 0 and Txn→yTx_n \to y implies y=0y = 0, then TT is bounded.

Application: Hellinger-Toeplitz theorem. If T:H→HT : H \to H is a linear operator on a Hilbert space satisfying ⟨Tx,y⟩=⟨x,Ty⟩\langle Tx, y\rangle = \langle x, Ty\rangle for all x,y∈Hx, y \in H, then TT is bounded.

4.8 Applications of the Uniform Boundedness Principle

Section titled “4.8 Applications of the Uniform Boundedness Principle”

Application 1: Fourier series divergence. There exists a continuous function on [−π,π][-\pi, \pi] whose Fourier series diverges at a point. The UBP shows that the set of partial sum operators {Sn}\{S_n\} is not pointwise bounded on C[−π,π]C[-\pi, \pi].

Application 2: Weak boundedness implies norm boundedness. If a set of operators is bounded in the weak operator topology, it is bounded in the operator norm.

The fundamental theorems of functional analysis all rely on the Baire category theorem:

Theorem 4.8 (Baire Category Theorem). In a complete metric space, the intersection of countably many dense open sets is dense. Equivalently, a complete metric space cannot be expressed as a countable union of nowhere dense sets.

This is the key ingredient in the proofs of the open mapping theorem and the uniform boundedness principle.

Problem 1. Use the Hahn-Banach theorem to show that for any closed subspace M⊆XM \subseteq X, the quotient map π:X→X/M\pi : X \to X/M has an isometric right inverse.

Problem 2. Show that if T:X→YT : X \to Y is a bijective bounded operator between Banach spaces, then T−1T^{-1} is bounded (without using the open mapping theorem, reconstruct the proof).

Problem 3. Let {fn}\{f_n\} be a sequence in X∗X^* such that lim⁡n→∞fn(x)\lim_{n\to\infty} f_n(x) exists for every x∈Xx \in X. Show that the limit functional f(x)=lim⁡nfn(x)f(x) = \lim_n f_n(x) is bounded.

Problem 4. Show that the differentiation operator D:C1[0,1]→C[0,1]D : C^1[0, 1] \to C[0, 1] defined by Df=f′Df = f' is unbounded when C1[0,1]C^1[0, 1] is given the ∥⋅∥∞\|\cdot\|_\infty norm.

The Uniform Boundedness Principle is also called the Banach-Steinhaus theorem or the Resonance theorem. A common formulation is:

Theorem 4.9 (Banach-Steinhaus). If {Tn}⊆B(X,Y)\{T_n\} \subseteq \mathcal{B}(X, Y) and lim⁡n→∞Tnx\lim_{n\to\infty} T_n x exists for every x∈Xx \in X, then the limit Tx=lim⁡nTnxTx = \lim_n T_n x defines a bounded linear operator TT.

Theorem 4.10 (Closed Range Theorem). If T:X→YT : X \to Y is a bounded linear operator between Banach spaces, then the following are equivalent:

  1. The range R(T)R(T) is closed in YY.
  2. The range R(T∗)R(T^*) is closed in X∗X^*.
  3. TT is relatively open: there exists c>0c > 0 such that for every y∈R(T)y \in R(T), there exists xx with Tx=yTx = y and ∥x∥≤c∥y∥\|x\| \leq c\|y\|.

4.13 Applications in Partial Differential Equations

Section titled “4.13 Applications in Partial Differential Equations”

Application: Well-posedness of elliptic PDEs. Consider the Dirichlet problem −Δu=f-\Delta u = f on Ω\Omega with u∣∂Ω=0u|_{\partial\Omega} = 0. The operator Δ:H01(Ω)→H−1(Ω)\Delta : H^1_0(\Omega) \to H^{-1}(\Omega) is bounded. By the Lax-Milgram lemma (a corollary of Hahn-Banach and Riesz representation), there exists a unique weak solution for every ff.

Application: Fourier series. The Uniform Boundedness Principle implies that there exist continuous functions whose Fourier series diverge at a point, since the Dirichlet kernels Dn(t)=∑k=−nneiktD_n(t) = \sum_{k=-n}^n e^{ikt} have ∥Dn∥1∼log⁡n→∞\|D_n\|_1 \sim \log n \to \infty as n→∞n \to \infty.

Problem 5. Prove that if XX is a Banach space and T:X→XT : X \to X is a bounded linear operator with ∥T∥<1\|T\| < 1, then I−TI - T is invertible (Von Neumann series). Use the open mapping theorem or construct the inverse explicitly.

Problem 6. Show that the closed graph theorem is equivalent to the open mapping theorem (given the bounded inverse theorem).

Problem 7. Let XX and YY be Banach spaces and suppose T:X→YT : X \to Y is a linear operator. Prove that if xn→0x_n \to 0 and Txn→yTx_n \to y implies y=0y = 0, then TT is bounded.