Skip to Content
Math4201Topology I (Lecture 28)

Math4201 Topology I (Lecture 28)

Compact spaces

Extreme value theorem

Definition of diameter

Let (X,d)(X,d) be a metric space and AXA\subseteq X. The diameter of AA is defined as

diam(A)=sup{d(x,y):x,yA}\operatorname{diam}(A) = \sup\{d(x,y):x,y\in A\}

Lebesgue number lemma

Let XX be a compact metric space and {Uα}αI\{U_\alpha\}_{\alpha\in I} be an open cover of XX. Then there is δ>0\delta>0 such that for every subset AXA\subseteq X with diameter less than δ\delta, there is αI\alpha\in I such that AUαA\subseteq U_\alpha.

Proof

Consider xXx\in X, there is an element UαU_\alpha in the open covering such that xUαx\in U_\alpha.

In particular, there is rxr_x such that Brx(x)UαB_{r_x}(x)\subseteq U_\alpha.

Then the collection {Brx2(x)}xX\{B_{\frac{r_x}{2}}(x)\}_{x\in X} is an open covering of XX. (each xXx\in X is contained in some Brx2(x)B_{\frac{r_x}{2}}(x))

Since XX is compact, there is a finite subcover {Brxi2(xi)}i=1n\{B_{\frac{r_{x_i}}{2}}(x_i)\}_{i=1}^n of XX. Such that i=1nBrxi2(xi)=X\bigcup_{i=1}^n B_{\frac{r_{x_i}}{2}}(x_i)=X.

Let δ=min{rrx12,...,rrxn2}>0\delta = \min\{r_{\frac{r_{x_1}}{2}}, ..., r_{\frac{r_{x_n}}{2}}\}>0.

Let AXA\subseteq X be a subset with diameter less than δ\delta.

Take yAy\in A, then ABδ(y)A\subseteq B_\delta(y).

Take xix_i such that yBrxi2(xi)y\in B_{\frac{r_{x_i}}{2}}(x_i). (such cover exists by definition of the subcover)

And then α\alpha such that Brxi2(xi)UαB_{\frac{r_{x_i}}{2}}(x_i)\subseteq U_\alpha.

We claim that Bδ(y)UαB_\delta(y)\subseteq U_\alpha, which would imply that AUαA\subseteq U_\alpha.

yBrxi2(xi)y\in B_{\frac{r_{x_i}}{2}}(x_i), and we know that Brxi(xi)UαB_{r_{x_i}}(x_i)\subseteq U_\alpha.

Since δ<rxi2\delta < \frac{r_{x_i}}{2}, it suffices to show that Brxi2(y)UαB_{\frac{r_{x_i}}{2}}(y)\subseteq U_\alpha.

For any zBrxi2(y)z\in B_{\frac{r_{x_i}}{2}}(y), we have d(z,y)<rxi2d(z,y)<\frac{r_{x_i}}{2}, and d(y,xi)<rxi2d(y,x_i)<\frac{r_{x_i}}{2}.

So d(z,xi)d(z,y)+d(y,xi)<rxi2+rxi2=rxid(z,x_i)\leq d(z,y)+d(y,x_i)<\frac{r_{x_i}}{2}+\frac{r_{x_i}}{2}=r_{x_i}, so zBrxi(xi)Uαz\in B_{r_{x_i}}(x_i)\subseteq U_\alpha.

So Brxi2(y)UαB_{\frac{r_{x_i}}{2}}(y)\subseteq U_\alpha.

Definition of finite intersection property

A collection {Cα}αI\{C_\alpha\}_{\alpha\in I} of subsets of a set XX has finite intersection property if for every finite subcollection {Cα1,...,Cαn}\{C_{\alpha_1}, ..., C_{\alpha_n}\} of {Cα}αI\{C_\alpha\}_{\alpha\in I}, we have i=1nCαi\bigcap_{i=1}^n C_{\alpha_i}\neq \emptyset.

Theorem

A space XX is compact if and only if every collection {Zα}αI\{Z_\alpha\}_{\alpha\in I} of closed subsets of XX satisfies the finite intersection property has a non-empty intersection.

αIZα\bigcap_{\alpha\in I} Z_\alpha \neq \emptyset

Non-example

Consider X=(0,1)X=(0,1) is not compact with the standard topology.

Consider Zn=(0,1n]Z_n=(0,\frac{1}{n}], each interval is closed in XX. This satisfies the finite intersection property because i=1kZni\bigcap_{i=1}^k Z_{n_i}\neq \emptyset for any finite subcollection {Zn1,...,Znk}\{Z_{n_1}, ..., Z_{n_k}\}. We can find a smaller for any finite subcollection to get a non-empty intersection.

But n=1Zn=\bigcap_{n=1}^\infty Z_n = \emptyset.

Proof

    \implies

Let Uα=XZαU_\alpha=X-Z_\alpha is open for each αI\alpha\in I. By contradiction, suppose that αIZα=\bigcap_{\alpha\in I} Z_\alpha = \emptyset.

So XαIZα=X=αIUα=αI(XZα)=XX-\bigcap_{\alpha\in I} Z_\alpha = X=\bigcup_{\alpha\in I} U_\alpha = \bigcup_{\alpha\in I} (X-Z_\alpha)=X.

So {Uα}αI\{U_\alpha\}_{\alpha\in I} is an open cover of XX. Since XX is compact, there is a finite subcover {Uα1,...,Uαn}\{U_{\alpha_1}, ..., U_{\alpha_n}\}.

So i=1nUαi=Xi=1nZαi=X\bigcap_{i=1}^n U_{\alpha_i} = X-\bigcup_{i=1}^n Z_{\alpha_i} = X, i=1nZαi=\bigcap_{i=1}^n Z_{\alpha_i} = \emptyset. This contradicts the finite intersection property.

    \impliedby

Proof is similar.

Definition of isolated point

A point xXx\in X is an isolated point if {x}\{x\} is an open subset of XX.

Example of isolated point

X=[0,1]{2}X=[0,1]\cup \{2\} with subspace topology from R\mathbb{R}.

Then {2}\{2\} is an isolated point {2}=X(212,2+12)\{2\}=X\cap (2-\frac{1}{2}, 2+\frac{1}{2}).

Theorem of compact Hausdorff spaces without isolated points

Any non-empty compact Hausdorff space without an isolated point is uncountable.

Proof

Proof by contradiction.

Let X={xn}nNX=\{x_n\}_{n\in\mathbb{N}} be a countable set.

Since x1x_1 is not an isolated point, so there exists y1Xy_1\in X such that y1x1y_1\neq x_1. Apply the Hausdorff property, there exists disjoint open neighborhoods U1U_1 and V1V_1 such that x1U1x_1\in U_1 and y1V1y_1\in V_1.

In particular V1\overline{V_1} does not contain x1x_1, but it contains y1y_1. (Follows from disjoint open neighborhoods)

Since x2x_2 is not an isolated point, so there exists y2Xy_2\in X such that y2x2y_2\neq x_2. Apply the Hausdorff property, there exists disjoint open neighborhoods U2U_2 and V2V_2 such that x2U2x_2\in U_2 and y2V2y_2\in V_2.

If x2V1x_2\notin V_1, then we define V2V_2 as V1V_1.

If x2V1x_2\in V_1, then by the assumption, there is another point y2y_2 in V1V_1 which isn’t the same as x2x_2.

CONTINUE NEXT TIME.

Theorem real numbers is uncountable

R\mathbb{R} is uncountable, and any interval in R\mathbb{R} is uncountable.

Proof

It suffices to prove this for a closed interval [a,b][a,b] with a<ba<b. Because any interval contains such a closed interval.

The claim for a closed interval [a,b][a,b] follows from the following theorem because [a,b][a,b] is a non-empty compact Hausdorff space without an isolated point.

Last updated on