Compactness and Covering Properties in Metric Spaces






Compactness — Teaching Notes (Template Style)




MATH 621 — Real Analysis I

Compactness in Metric Spaces — Teaching Notes


Purpose, Intuition & Outcomes

Compactness distills the idea that a set is small enough to be controlled by finitely many local pieces. In metric spaces, compact sets behave like closed intervals: they are closed, bounded, every sequence admits a convergent subsequence, and continuous functions attain maxima and minima.

Learning outcomes. After this note, you should be able to: (i) state and use the open-cover definition of compactness; (ii) diagnose non-compactness by building an open cover with no finite subcover; (iii) prove compact ⇒ closed and bounded; (iv) apply Heine–Borel in $\mathbb R^n$; (v) use basic consequences (sequential compactness; extreme value theorem).

Definition: Cover, Open Cover, Subcover

Let $(X,d)$ be a metric space and $S\subseteq X$. A family $\{E_i:i\in I\}$ is a cover of $S$ if $\displaystyle S\subseteq\bigcup_{i\in I}E_i$. It is an open cover if each $E_i$ is open (in the subspace topology on $X$). A subcover is a subfamily that still covers $S$; if the index set is finite, it is a finite subcover.

Try it. Give an open cover of $(0,1]$ by intervals that has no finite subcover.

Definition: Compact Subset

$K\subseteq X$ is compact if every open cover of $K$ admits a finite subcover.

Pitfall. In general metric spaces, “closed and bounded” does not imply compact. That equivalence is special to Euclidean spaces (Heine–Borel).

Examples: Two Natural Non‑Compact Sets

(a) $(0,4]$ in $\mathbb R$. For $n\in\mathbb N$, set $E_n=(\tfrac1n,5)$. Then $(0,4]\subseteq\bigcup_n E_n$, but any finite subfamily is $(\tfrac1N,5)$ for some $N$, missing points near $0$.

(b) $B_1(0)$ in $\mathbb R^2$. For $n\ge 2$, set $E_n=B_{1-1/n}(0)$. Then $B_1(0)=\bigcup_{n\ge2}E_n$, but any finite subfamily misses a thin annulus near the boundary.

Proposition: In Any Metric Space, Compact $\Rightarrow$ Closed and Bounded

Strategy. For boundedness, cover by concentric balls $\{B_r(x)\}_{r\in\mathbb N}$ and shrink to finitely many. For closedness, fix $y\notin K$; for each $z\in K$ separate $z$ and $y$ by disjoint open sets $(U_z,V_z)$ and then intersect finitely many $V_z$’s.

Boundedness. Fix $x\in X$. The family $\{B_r(x): r\in\mathbb N, r\ge 1\}$ covers $K$. By compactness, choose a finite subcover $\{B_{r_i}(x)\}_{i=1}^m$. With $R=\max_i r_i$ we have $K\subseteq B_R(x)$.

Closedness. Let $y\in X\setminus K$. For each $z\in K$, the Hausdorff property provides disjoint open sets $U_z\ni z$ and $V_z\ni y$. The family $\{U_z\}$ covers $K$. By compactness, pick $z_1,\dots,z_N$ with $K\subseteq\bigcup_{i=1}^N U_{z_i}$. Then $V=\bigcap_{i=1}^N V_{z_i}$ is open, contains $y$, and is disjoint from $K$. Hence $K$ is closed.

Check. Where did we use Hausdorffness? Why does compactness let us take finitely many $U_z$?

Heine–Borel (Euclidean Spaces)

In $\mathbb R^n$ (equivalently $\mathbb C^n$ with its Euclidean metric), a set is compact iff it is closed and bounded.

Scope. This equivalence fails in general metric spaces.

Proof Sketch: Closed & Bounded $\Rightarrow$ Compact

Strategy. Prove $H=[-r,r]^n$ is compact by contradiction: assume an open cover with no finite subcover; recursively choose a nested sequence of closed subcubes with no finite subcover; use $\operatorname{diam}(H_m)\to 0$ and the Cantor intersection property to obtain a point $x$; openness then covers a whole small cube $H_m$, a contradiction.

Assume $H$ has an open cover $\mathcal C$ with no finite subcover. Bisect each side to partition $H$ into $2^n$ congruent closed subcubes; pick one, $H_1$, that is not finitely coverable by $\mathcal C$. Iterate to obtain nested closed cubes $H\supset H_1\supset H_2\supset\cdots$ with $\operatorname{diam}(H_m)\to 0$. Then $\bigcap_m H_m=\{x\}\ne\varnothing$. Pick $U\in\mathcal C$ with $x\in U$. There exists $r>0$ such that $B_r(x)\subset U$. For $m$ large, $H_m\subset B_r(x)\subset U$, contradicting the choice of $H_m$.

Try it. Adapt the argument to show that every closed rectangle $\prod_{i=1}^n[a_i,b_i]$ is compact.

Core Consequences in Metric Spaces

Sequential compactness. In metric spaces, $K$ is compact iff every sequence in $K$ has a convergent subsequence with limit in $K$.

Continuity & images. If $f:X\to Y$ is continuous and $K\subset X$ is compact, then $f(K)$ is compact. Consequently, continuous $f:K\to\mathbb R$ with $K$ compact are bounded and attain their extrema (Extreme Value Theorem).

Completeness & total boundedness. In metric spaces: $K$ is compact $\iff$ $K$ is complete and totally bounded.

Apply. Prove that every continuous function $f:[0,1]\to\mathbb R$ attains its maximum and minimum.

Further Examples

Compact sets. Finite sets; $[a,b]$; closed rectangles in $\mathbb R^n$; the unit sphere $S^{n-1}$ (closed and bounded $\Rightarrow$ compact via Heine–Borel).

Non-compact sets. $(0,1)$, $\mathbb R$, the open unit ball in $\mathbb R^n$, graphs of polynomials in $\mathbb R^2$ (unbounded).

Construct. Give an explicit open cover of $[0,1)$ with no finite subcover. Why does the same construction fail for $[0,1]$?

Common Pitfalls & How to Avoid Them

Confusing $X$ and $S$. A cover of a subset $S$ requires $S\subseteq\cup E_i$ (not $X$).

“Closed and bounded” everywhere. Only valid in Euclidean spaces with the Euclidean metric.

Forgetting subspace openness. In a metric subspace, a set is open if it is the intersection of an open set of the ambient space with the subspace.

Practice: Proof‑Guided Exercises

Guided proof. Show that if $K$ is compact and $F\subseteq K$ is closed, then $F$ is compact. Hint: Intersect an arbitrary open cover of $F$ with $K$ and use compactness of $K$.

Counterexample builder. Produce an open cover of $\mathbb R$ with no finite subcover. Why does this show $\mathbb R$ is not compact?

Sequential compactness. Let $(x_n)$ lie in a compact $K$. Explain how to extract a convergent subsequence using total boundedness (finite $\varepsilon$-nets) and completeness.

© 2025 — V. Oussa. Prepared for students of MATH 621 (BSU). Template styling aligned with course notes.