Skip to content

Countability & Separation Axioms

Separation axioms

Separation axioms

Suppose \(X\) is a topological space, we have the following four separation axioms.

(i) \(T_1\) separation axiom. \(\forall x\neq y\in X\), there exsits a neighborhood \(U\) of \(x\) such that \(y\not\in U\).

(ii) \(T_2\) separation axiom. \(\forall x\neq y\in X\), there exsit neighborhood \(U\), \(V\) of \(x\), \(y\), respectively, such that \(U\cap V=\varnothing\).

(iii) \(T_3\) separation axiom (regular). \(\forall x\in X\) and closed set \(A\subset X\), there exist neighborhood \(U\), \(V\) of \(x\) and \(A\) such that \(U\cap V=\varnothing\).

(iv) \(T_4\) separation axiom (normal). For every two disjoint closed set \(A,B\subset X\), there exist neighborhood \(U\), \(V\) of \(A\) and \(B\) such that \(U\cap V=\varnothing\).

A space that satisfies \(T_2\) axiom is called a Hausdorff space, which turned out to have good properties for analysis.

Relationship of serapation axioms

(i) \(T_2\) implies \(T_1\).

(ii) If \(T_1\) axiom holds, then \(T_4\) implies \(T_3\), \(T_3\) implies \(T_2\).

Example. Show that \(X\) is \(T_1\), iff finite-number subset of \(X\) is closed.

Now we give the definition of convergence.

Definition of convergence

Suppose \(X\) is a topological space, with \(\{x_n\}\subset X\) is a sequence, and \(x\in X\). \(x_n\) is said to converge to \(x\), if for all open neighborhood \(U\) of \(x\), there exists \(N>0\), such that \(x_n\in U\) whenever \(n>N\).

Example. If \(X\) is a Hausdorff space, then any sequence in \(X\) converges to at most one point.

Countability axioms

Countability axioms

Suppose \(X\) is a topological space, we have the following two countability axioms.

(i) \(C_1\) countability axiom. \(\forall x\in X\), there exsits a countable collection \(\mathcal{B}\) of neighborhoods of \(x\), such that every neighborhood of \(x\) contains some \(B\in \mathcal{B}\).

We also call \(X\) has a countable basis at \(x\).

(ii) \(C_2\) countability axiom. \(X\) has a countable basis for its topology.

\(C_2\) axiom is stronger than \(C_1\), and some metric space could not satisfy the condition.

Example. In \(\mathbb{R}\), define metric

\[ d(x,y)=\delta_{xy}. \]

Show that every point of the above space is its open set and contains uncountable basis for its topology.

Choose open ball

\[ B(x, 1/2)=\{x\}, \]

which is the smallest element of basis for its topology, and thus is uncountable.

Example. Separable metric space is \(C_2\). Thus \(\mathbb{E}^n\), Hilbert space \(\mathbb{E}^\omega\) are \(C_2\).

Lindelöf Theorem

If a topological space satisfies \(C_2\) and \(T_3\), then it is also \(T_4\). Or in some terms, A regular space with countable basis for its topology is normal.

Choose a countable basis for its topology \(\mathcal{B}\). Suppose \(F\) and \(F'\) are disjoint closed subsets.

Since the space is \(T_3\), so for each \(x\in F\), there exist open sets \(U_x\) \(V_x\) such that \(x\in U_x\), \(F'\subset V_x\) and \(U_x\cap V_x=\varnothing\). Thus we have \(\overline{U_x}\cap F'=\varnothing\). For each \(U_x\), choose an open set \(B_x\in \mathcal{B}\) such that \(B_x\subset U_x\) and thus \(\overline{B_x}\cap F'=\varnothing\). Since \(\mathcal{B}\) is countable, we have

\[ \{B_n\}_{n=1}^\infty,\quad \{B_n'\}_{n=1}^\infty \]

such that for each \(n\), \(\overline{B_n}\cap F'=\varnothing\) and \(\overline{B_n}'\cap F=\varnothing\), and \(F\subset \bigcup_{n} B_n\) and \(F'\subset \bigcup_n B_n'\).

Define open sets

\[ U_n=B_n-\bigcup_{i=1}^n B_i', \quad V_n = B_n' -\bigcup_{i=1}^n B_i \]

So we have \(U_n\cap V_m=\varnothing\) for all \(n,m\geq 1\). Let

\[ U:=\bigcup_{n=1}^\infty U_n, \quad V:=\bigcup_{n=1}^\infty V_n, \]

so

\[ U\cap V = \bigcup_{n=1}^\infty (U_n\cap V_n)=\varnothing. \]

Easy to show that \(U\) and \(V\) are the desired open sets that separate \(F\) and \(F'\).

Urysohn Lemma & its applications

Urysohn Lemma

Assume topological space \(X\) is \(T_4\), then for any two disjoint closed sets \(A\) and \(B\), there exists a continuous function defined on \(X\) such that \(f|_A=0\) and \(f|_B=1\).