AI summary by TwinMind
Lecture 4: The Open Mapping Theorem and the Closed Graph Theorem
MIT OpenCourseWare
1h 14m 57s
Baire Category Theorem Recap
- Reviewed Baire category theorem: if M is complete metric space, Ck collection of closed sets, M = union of closed sets, then at least one closed set has interior point/contains open ball
Open Mapping Theorem Statement
- Theorem: If B1, B2 are Banach spaces, T bounded linear operator B1→B2 which is surjective, then T is open map
- Open map means: for all open U ⊂ B1, T(U) is open in B2
- Surjective linear bounded operators send open sets to open sets
Open Mapping Theorem Proof - Step 1: Special Case
- Goal: Prove T(B₀₁) contains open ball centered at zero in B2
- B₀₁ = open ball radius 1 centered at 0 in B1
- Since T surjective: B2 = ∪_{n∈ℕ} T(closure(B₀ⁿ))
- By Baire theorem: ∃n₀ such that T(closure(B₀^{n₀})) contains interior point
- Using linearity and scaling: T(closure(B₀₁)) contains open ball radius 4r centered at v₀
- Found v₁ = T(u₁) where u₁ ∈ B₀₁, ||v₀ - v₁|| < 2r
Open Mapping Theorem Proof - Step 2: Removing Closure
- Claim: Ball radius r/2 ⊂ T(B₀₁) (without closure)
- For v with ||v|| < r/2, constructed sequence {bₖ} with:
- ||bₖ|| < 2^{-k}
- ||v - Σₖ₌₁ⁿ T(bₖ)|| < 2^{-n-1}r
- Series Σbₖ absolutely summable since B1 complete
- ∃b ∈ B1 with b = Σbₖ, ||b|| ≤ 1
- By continuity: T(b) = v, so v ∈ T(B₀₁)
Open Mapping Theorem Proof - Step 3: General Case
- Extended to arbitrary open sets using translation and scaling
- If U ⊂ B1 open, v₂ ∈ T(U), then ∃ε > 0: ball(v₁, ε) ⊂ U
- Using established result: ball(v₂, εδ) ⊂ T(U) for some δ > 0
Corollary: Inverse Operator Theorem
- Corollary: If B1, B2 Banach spaces, T bounded linear bijective operator B1→B2, then T⁻¹ is bounded linear operator B2→B1
- Proof: T⁻¹ continuous ⟺ T maps open sets to open sets (by open mapping theorem)
Closed Graph Theorem Setup
- Product Space: B1 × B2 with norm ||(b₁,b₂)|| = ||b₁|| + ||b₂|| is Banach space when B1, B2 are Banach
- Graph of T: Γ(T) = {(u,v) : v = T(u)} ⊂ B1 × B2
Closed Graph Theorem Statement
- Theorem: B1, B2 Banach spaces, T linear operator B1→B2
- T is bounded linear operator ⟺ Γ(T) is closed in B1 × B2
- Advantage: Easier to check than continuity - assume T(uₙ) converges, only verify limit equals T(limit of uₙ)
Closed Graph Theorem Proof
- Direction 1: If T bounded ⟹ Γ(T) closed (straightforward by continuity)
- Direction 2: If Γ(T) closed ⟹ T bounded
- Γ(T) closed subspace of B1 × B2, hence Banach space
- Defined π₁: Γ(T) → B1, π₂: Γ(T) → B2 (projections)
- π₁ bijective bounded linear operator
- By inverse operator theorem: π₁⁻¹ bounded
- T = π₂ ∘ π₁⁻¹ composition of bounded operators, hence bounded
Introduction to Hahn-Banach Theorem
- Question: For general non-trivial normed space V, is dual space V* = {0}?
- Examples with non-trivial duals:
- (ℓᵖ)* ≅ ℓᵖ' where 1/p + 1/p' = 1 (1 ≤ p < ∞)
- (c₀)* ≅ ℓ¹
- Elements of dual space called "functionals"
Set Theory Prerequisites: Partial Orders
- Partial order on set E: relation ≤ satisfying
- Reflexivity: e ≤ e ∀e ∈ E
- Antisymmetry: e ≤ f, f ≤ e ⟹ e = f
- Transitivity: e ≤ f, f ≤ g ⟹ e ≤ g
- Upper bound of D ⊂ E: element e ∈ E such that d ≤ e ∀d ∈ D
- Maximal element: e ∈ E such that e ≤ f ⟹ e = f
- Chain: subset C where any two elements comparable
Zorn's Lemma
- Zorn's Lemma: If every chain in non-empty partially ordered set E has upper bound, then E has maximal element
- Applications mentioned:
- Proves axiom of choice
- Every vector space has Hamel basis
- Will be used to prove Hahn-Banach theorem
Hamel Basis Definition
- Hamel basis H ⊂ V (vector space): linearly independent set such that every element of V is finite linear combination of elements from H
- Examples: standard basis for ℝⁿ, sequence of unit vectors for ℓ¹
