AI summary by TwinMind
Lecture 4: Compact Metric Spaces
MIT OpenCourseWare
1h 7m 5s
Review of Previous Material
- Last time: showed metric spaces being compact on Euclidean space (R^n)
- Proved sequentially compact = closed and bounded = topologically compact on R^n
- Used Bolzano-Weierstrass theorem and Heine-Borel theorem
- Correction: sequentially compact ≠ closed and bounded on general metric spaces
- Not all metric spaces have Heine-Borel property
General Metric Space Compactness Relationships
- On general metric spaces: sequentially compact and topologically compact still imply closed and bounded
- Today's goal: prove sequentially compact = topologically compact on metric spaces
- Will show four equivalent properties of compact metric spaces
Lebesgue Number Lemma
- Statement: If X is sequentially compact metric space with open cover {U_i}, then ∃ r > 0 such that ∀ x ∈ X, ball B_r(x) ⊆ U_i for some i
- Proof by contradiction:
- Assume negation: ∀ r > 0, ∃ x such that B_r(x) not contained in any U_i
- For r = 1/n, choose x_n satisfying above property
- Take convergent subsequence x_{n_k} → x (using sequential compactness)
- x ∈ U_{i_0} for some i_0, so ∃ r_0 > 0: B_{r_0}(x) ⊆ U_{i_0}
- Choose n large enough: 1/n < r_0/2 and d(x, x_n) < r_0/2
- Show B_{1/n}(x_n) ⊆ B_{r_0}(x) ⊆ U_{i_0} (contradiction)
- Also called "Lebesgue number lemma"
Total Boundedness Definition
- Definition: Metric space X is totally bounded if ∀ ε > 0, ∃ y_1, ..., y_k such that X ⊆ ∪{i=1}^k Bε(y_i)
- Cover by finitely many balls of same radius ε
Sequentially Compact Implies Totally Bounded
- Proof by contradiction:
- Assume not totally bounded: for some ε > 0, need infinitely many ε-balls to cover X
- Construct sequence: x_1 ∈ X, x_2 ∈ X \ B_ε(x_1), x_n ∈ X \ ∪{i=1}^{n-1} Bε(x_i)
- By construction: d(x_i, x_j) ≥ ε for all i ≠ j
- No Cauchy subsequences exist (distances always ≥ ε)
- But sequentially compact implies convergent subsequence exists, which must be Cauchy
- Contradiction
Main Theorem: Sequentially Compact ⟺ Topologically Compact
Forward Direction (Sequential → Topological)
- Given open cover {U_i} of sequentially compact X
- By Lebesgue number lemma: ∃ r > 0 such that B_r(x) ⊆ U_i for some i
- By total boundedness with ε = r: ∃ y_1, ..., y_k such that X ⊆ ∪_{i=1}^k B_r(y_i)
- Each B_r(y_i) ⊆ U_{j_i} for some j_i
- Therefore X ⊆ ∪{i=1}^k U{j_i} (finite subcover)
Reverse Direction (Topological → Sequential)
- Proof by contradiction: assume topologically compact but not sequentially compact
- ∃ sequence {x_m} with no convergent subsequence
- Key observations:
- None of x_i appear infinitely often (else trivial convergent subsequence)
- ∀ j, ∃ εj > 0: B{ε_j}(x_j) = {x_j} (else could extract convergent subsequence)
- Construct open cover:
- U_j = B_{ε_j}(x_j) for j = 1, 2, ...
- U_0 = X \ {x_1, x_2, ...} (open since complement is closed with no limit points)
- X = U_0 ∪ ∪_{j=1}^∞ U_j is open cover
- Any finite subcover omits infinitely many points (contradiction with topological compactness)
Applications of Compactness
Continuous Images of Compact Sets
- Lemma: If K ⊆ X is compact and f: X → Y continuous, then f(K) is compact
- Proof: Given open cover {U_i} of f(K), then {f^{-1}(U_i)} covers K
- f^{-1}(U_i) are open (continuity), find finite subcover
- Corresponding {U_i} form finite subcover of f(K)
Extreme Value Theorem Generalization
- Corollary: Continuous f: X → ℝ achieves max and min on compact sets
- Proof: f(K) compact in ℝ ⟹ f(K) closed and bounded ⟹ has max and min
Boundedness of Continuous Functions
- Corollary: If X compact and f: X → ℝ continuous, then f is bounded
- Proof: f(X) is compact, hence bounded
Cantor's Intersection Theorem
- Statement: Given K_1 ⊇ K_2 ⊇ K_3 ⊇ ... where each K_i is non-empty and compact, then ∩_{i=1}^∞ K_i ≠ ∅
- Proof:
- Let x_i ∈ K_i for all i
- {x_i} ⊆ K_1, so ∃ convergent subsequence x_{n_k} → a ∈ K_1
- Similarly show a ∈ K_j for all j
- Used extensively in complex analysis
Finite Intersection Property
- Definition: Collection of closed sets has finite intersection property if every finite subcollection has non-empty intersection
Four Equivalent Definitions of Compactness
Theorem: On metric space X, the following are equivalent:
- X is topologically compact
- X is sequentially compact
- X is totally bounded and Cauchy complete
- Every collection of closed sets with finite intersection property has non-empty intersection
Proof of (1) ⟺ (4)
(1) ⟹ (4)
- Assume collection {C_i} of closed sets with finite intersection property but ∩C_i = ∅
- Then X = ∪C_i^c (De Morgan's law)
- {C_i^c} is open cover, has finite subcover {C_{i_1}^c, ..., C_{i_k}^c}
- So X = ∪{j=1}^k C{i_j}^c ⟹ ∩{j=1}^k C{i_j} = ∅
- Contradicts finite intersection property
(4) ⟹ (1)
- Given open cover {U_i}, let C_i = U_i^c (closed sets)
- Assume no finite subcover exists
- Then {C_i} has finite intersection property
- By (4): ∩C_i ≠ ∅, but ∩C_i = (∪U_i)^c = X^c = ∅
- Contradiction
Proof of (2) ⟺ (3)
(2) ⟹ (3)
- Already showed: sequentially compact ⟹ totally bounded
- Cauchy completeness: Let {x_i} be Cauchy sequence
- ∃ convergent subsequence x_{n_k} → x (sequential compactness)
- For ε > 0, ∃ N: ∀ n,m ≥ N, d(x_n, x_m) < ε/2
- ∃ K: d(x, x_{n_k}) < ε/2 for k large enough
- Then d(x, x_n) ≤ d(x, x_{n_k}) + d(x_{n_k}, x_n) < ε
- So x_i → x
(3) ⟹ (2)
- Given arbitrary sequence {x_i}, construct convergent subsequence
- Construction using total boundedness:
- For each n, ∃ finitely many y_1^{(n)}, ..., y_{r(n)}^{(n)} such that X ⊆ ∪B_{1/n}(y_j^{(n)})
- By pigeonhole principle: ∃ ball B_1(y_{j_1}^{(1)}) containing infinitely many x_i
- Choose z_1 from this ball
- Similarly find z_2 in intersection of B_1(y_{j_1}^{(1)}) and some B_{1/2}(y_{j_2}^{(2)})
- Continue: z_k in intersection of B_{1/n}(y_{j_n}^{(n)}) for n = 1, ..., k
- Cauchy property: For ε > 0, choose n: 1/n < ε
- For all k, ℓ ≥ n: z_k, z_ℓ both in some B_{1/n}(y_j^{(n)})
- So d(z_k, z_ℓ) ≤ 2/n < 2ε
- By Cauchy completeness: {z_k} converges
- {z_k} is subsequence of original sequence
Practical Applications
- Union of finitely many compact sets is compact
- Choose appropriate definition based on proof context:
- Topological compactness for open/closed set arguments
- Sequential compactness for convergence arguments
- Total boundedness + completeness for metric properties
- Finite intersection property for set-theoretic arguments
Preview
- Next lecture: history and motivation of compact metric spaces
- Dirichlet problem and development of compactness theory
