Closure and Boundedness

Drawing a bar over AA gives the closure A‾\overline{A} — the set with its escaped limit points pulled back in. Follow the round-song: closed ⟺ contains its limit points ⟺ equals its closure, and closed + bounded ⟺ compact (Heine-Borel).

x∈A‾  ⟺  ∀ε>0:(x−ε,x+ε)∩A≠∅x \in \overline{A} \iff \forall \varepsilon > 0 : (x - \varepsilon, x + \varepsilon) \cap A \neq \varnothing
A closed   ⟺  A‾=A,A bounded   ⟺  ∃M:∣x∣≤M ∀x∈AA \text{ closed } \iff \overline{A} = A, \qquad A \text{ bounded } \iff \exists M : \lvert x\rvert \leq M\ \forall x \in A

Examples & Exercises

Prove the closure of (0,1)(0,1) is [0,1][0,1].

(0,1)‾=[0,1]\overline{(0,1)} = [0,1]
The red dots 0,10, 1 — limit points added by the closure.
Show proof
  1. Points of (0,1)(0,1) are trivially in the closure. Let x=0x = 0: for every ε>0\varepsilon > 0, ε/2∈(0,ε)∩(0,1)\varepsilon/2 \in (0,\varepsilon) \cap (0,1), so every neighborhood of 00 meets (0,1)(0,1); i.e. 0∈(0,1)‾0 \in \overline{(0,1)}.
  2. Likewise every ε\varepsilon-ball at 11 contains 1−ε/2∈(0,1)1 - \varepsilon/2 \in (0,1), so 1∈(0,1)‾1 \in \overline{(0,1)}.
  3. If x<0x < 0, the ball of radius −x/2-x/2 around xx lies in (−∞,0)(-\infty, 0), disjoint from (0,1)(0,1); similarly for x>1x > 1. Hence the closure is exactly [0,1][0,1].

Prove (0,1)(0,1) is not closed, using the closure.

0∈(0,1)‾∖(0,1)0 \in \overline{(0,1)} \setminus (0,1)
M=2M = 2 — widening the bound does not make it closed.
Show proof
  1. A closed set equals its closure: AA is closed ⟺ A‾=A\overline{A} = A.
  2. By problem 1, (0,1)‾=[0,1]\overline{(0,1)} = [0,1]. But 0∈[0,1]0 \in [0,1] while 0∉(0,1)0 \notin (0,1), so the closure is strictly larger.
  3. Hence (0,1)≠(0,1)‾(0,1) \neq \overline{(0,1)}, so (0,1)(0,1) is not closed.

Show A={1/n:n≥1}A = \{1/n : n \geq 1\} is bounded and prove its closure is A∪{0}A \cup \{0\}.

{1/n:n≥1}‾={1/n:n≥1}∪{0}\overline{\{1/n : n \geq 1\}} = \{1/n : n \geq 1\} \cup \{0\}
The red dot 00 — the limit point the sequence approaches but never touches.
Show proof
  1. For all nn, 0<1/n≤10 < 1/n \leq 1, so with M=1M = 1 we have ∣x∣≤M\lvert x\rvert \leq M for x∈Ax \in A — AA is bounded.
  2. Every ε\varepsilon-ball at 00 contains 1/N∈A1/N \in A for 1/N<ε1/N < \varepsilon, so 00 is a limit point of AA: 0∈A‾0 \in \overline{A}.
  3. Points outside A∪{0}A \cup \{0\} miss AA: for x<0x < 0 the ball of radius ∣x∣/2\lvert x\rvert/2 avoids AA, similarly for x>1x > 1; a point of (0,1]∖A(0,1] \setminus A lies strictly between consecutive terms 1/(n+1)<x<1/n1/(n+1) < x < 1/n, so a ball of the nearer gap radius avoids AA. Hence the closure is exactly A∪{0}A \cup \{0\}.

4. Topology