Compact space Study Guide

📖 Core Concepts Compactness – “every open cover has a finite subcover.” Intuitively, a space cannot “escape to infinity.” Sequential Compactness – every infinite sequence has a convergent subsequence whose limit lies in the space. Limit‑point Compactness – every infinite subset possesses at least one limit (accumulation) point in the space. Totally Bounded + Complete (Metric spaces) – a metric space is compact iff it is both complete (all Cauchy sequences converge) and totally bounded (for any ε > 0 the space can be covered by finitely many ε‑balls). Closed + Bounded (ℝⁿ) – by the Heine–Borel theorem, a subset of ℝⁿ is compact ⇔ it is closed and bounded. Continuous Image – the image of a compact space under a continuous map remains compact. --- 📌 Must Remember Open‑cover definition: ∀ 𝒞 ⊆ 𝒪𝑋, (⋃𝒞 = X) ⇒ ∃ finite 𝔉⊆𝒞 with ⋃𝔉 = X. Equivalences (metric): compact ⇔ sequentially compact ⇔ limit‑point compact ⇔ complete + totally bounded. Closed subsets: In any Hausdorff space, compact ⇒ closed. Finite unions: Finite union of compact sets is compact. Product theorem: Arbitrary product of compact spaces is compact (Tychonoff). Extreme Value Theorem: Continuous $f\colon K\to\mathbb{R}$ on non‑empty compact $K$ attains max & min. Co‑countable topology: Compact subsets = finite sets only. --- 🔄 Key Processes Testing compactness via open covers Identify a covering family $\{Ui\}$ of open sets. Look for a finite subfamily that still covers the whole space. Using sequential compactness Take an arbitrary sequence $\{xn\}$. Extract a convergent subsequence $\{x{nk}\}\to x\in X$. Verifying total boundedness (metric spaces) For each $\varepsilon>0$, construct finitely many $\varepsilon$‑balls covering the set. Applying Heine–Borel (ℝⁿ) Check closed + bounded → compact; otherwise non‑compact. Compactness of continuous images Given $f\colon X\to Y$ continuous and $X$ compact, conclude $f(X)$ compact. --- 🔍 Key Comparisons Compact vs. Closed (Hausdorff) Compact need not be closed in non‑Hausdorff spaces. In Hausdorff spaces: compact ⇒ closed. Sequential Compactness vs. Limit‑point Compactness In metric spaces they are equivalent. In general topological spaces they may differ. Finite vs. Infinite Discrete Spaces Finite discrete → compact (trivial finite subcover). Infinite discrete → non‑compact (cover by singletons). Open vs. Closed Disks (ℝ²) Closed disk $\overline{D}$: compact (closed & bounded). Open disk $D$: not compact (sequence approaching boundary has no limit in $D$). Co‑countable topology vs. Usual topology Usual topology on ℝ: many infinite compact subsets (e.g., $[0,1]$). Co‑countable topology on an uncountable set: only finite sets are compact. --- ⚠️ Common Misunderstandings “Bounded ⇒ compact” – false in general; need closedness (or completeness & total boundedness in metric spaces). “All closed subsets of any space are compact” – only true when the ambient space itself is compact. “Every Hausdorff space is normal” – true for compact Hausdorff spaces, not for all Hausdorff spaces. “A limit‑point compact space is automatically sequentially compact” – holds in metric spaces but not in arbitrary topological spaces. “Compactness is preserved under taking closures” – not in non‑Hausdorff spaces; closure of a compact set may fail to be compact. --- 🧠 Mental Models / Intuition “No way to run away” – Think of compactness as a room with no exits: any endless wandering (sequence) must eventually settle down inside. “Finite subcover = covering with a handful of blankets” – An open cover is a pile of blankets; compactness guarantees you can choose a few blankets that still keep the whole floor covered. “Totally bounded = can be trapped in a finite mesh” – For any mesh size you draw, only finitely many mesh cells are needed to capture the whole set. --- 🚩 Exceptions & Edge Cases Non‑Hausdorff compact subsets may be non‑closed. Example: trivial topology on a set with more than one point. Infinite product of compact spaces – requires Tychonoff’s theorem (needs the axiom of choice). Compactness in co‑countable topology – only finite subsets are compact, despite many sets being closed. Closed unit ball in infinite‑dimensional normed spaces – not compact (fails total boundedness). --- 📍 When to Use Which Open‑cover test – best for abstract spaces where a covering family is given. Sequential test – ideal in metric spaces or when dealing with sequences directly. Heine–Borel – apply in ℝⁿ (or any finite‑dimensional normed vector space). Continuous image rule – use when mapping a known compact domain (e.g., $[0,1]$) to a more complicated codomain. Product theorem – when handling product spaces (e.g., Hilbert cube, $[0,1]^I$). --- 👀 Patterns to Recognize Sequences approaching a “missing point” → non‑compact (e.g., open disk, $\mathbb{R}$). Covers by “small” open sets with no finite subcover → typical non‑compactness proof (e.g., $\{(n-1,n+1)\}{n\in\mathbb{Z}}$ for ℝ). Finite subcover arguments in proofs – often appear as “pick one open set for each point; finitely many points ⇒ finite subcover.” Closed + bounded in ℝⁿ repeatedly signals compactness. --- 🗂️ Exam Traps Distractor: “Every closed set is compact.” – Only true in compact spaces or in ℝⁿ with boundedness. Distractor: “A compact space must be bounded.” – In general topological spaces boundedness may be undefined; in metric spaces true, but not in arbitrary spaces. Near‑miss choice: “The whole real line is compact because it is closed.” – Wrong; fails the open‑cover condition (use intervals $(n-1,n+1)$). Confusing “finite union of compact sets” with “arbitrary union.” – Only finite unions preserve compactness. Co‑countable topology: Choosing “any countable set is compact” – false; only finite subsets are compact. ---