# Completely Hausdorff Space is Preserved under Closed Bijection

## Theorem

Let $T_A = \left({S_A, \tau_A}\right), T_B = \left({S_B, \tau_B}\right)$ be topological spaces.

Let $\phi: T_A \to T_B$ be a closed bijection.

If $T_A$ is a $T_{2 \frac 1 2}$ (completely Hausdorff) space, then so is $T_B$.

## Proof

Let $T_A$ be a $T_{2 \frac 1 2}$ (completely Hausdorff) space.

Then:

- $\forall x, y \in S_A, x \ne y: \exists U_A, V_A \in \tau_A: x \in U_A, y \in V_A: U_A^- \cap V_A^- = \varnothing$

where $U_A^-$ signifies the closure of $U_A$.

Suppose that $T_B$ is not completely Hausdorff.

Then:

- $\exists a, b \in S_B: a \ne b: \forall U_B, V_B \in \tau_B: a \in U_B, b \in V_B \implies U_B^- \cap V_B^- \ne \varnothing$

That is, there exists at least one pair of points $a$ and $b$ for which the closures of all the open sets containing $a$ and $b$ are not disjoint.

From Topological Closure is Closed, both $U_B^-$ and $V_B^-$ are closed.

From Topology Defined by Closed Sets, it follows that $W_B = U_B^- \cap V_B^-$ is also a closed set.

From Preimage of Intersection under Mapping, we have:

- $\phi^{-1} \left({U_B^- \cap V_B^-}\right) = \phi^{-1} \left({U_B^-}\right) \cap \phi^{-1} \left({V_B^-}\right)$

Let $H_A = \phi^{-1} \left({U_B^-}\right), K_A = \phi^{-1} \left({V_B^-}\right), W_A = \phi^{-1} \left({W_B}\right)$.

Then we have:

- $H_A \cap K_A = W_A \ne \varnothing$

By Bijection is Open iff Closed, $\phi$ is an open bijection.

By Bijection is Open iff Inverse is Continuous, it follows that $\phi^{-1}$ is continuous.

From Continuity Defined from Closed Sets, we have that $H_A$, $K_A$ and $W_A$ are all closed.

Also, let $U_A = \phi^{-1} \left({U_B}\right)$ and $V_A = \phi^{-1} \left({V_B}\right)$.

From Continuity Defined by Closure, we have that:

- $\phi^{-1} \left({U_B^-}\right) \subseteq \left({\phi^{-1} \left({U_B}\right)}\right)^-$
- $\phi^{-1} \left({V_B^-}\right) \subseteq \left({\phi^{-1} \left({V_B}\right)}\right)^-$

that is:

- $H_A \subseteq U_A^-$
- $K_A \subseteq V_A^-$

As $H_A \cap K_A \ne \varnothing$ it follows that $U_A^- \cap V_A^- \ne \varnothing$.

From the bijective nature of $\phi$, we have that:

- $a \in U_B \implies \phi^{-1} \left({a}\right) \in U_A$

- $b \in V_B \implies \phi^{-1} \left({b}\right) \in V_A$

Let $x = \phi^{-1} \left({a}\right), y = \phi^{-1} \left({b}\right)$.

Thus we have that:

- $\exists x, y \in X_A: x \ne y: \forall U_A, V_A \in \tau_A: x \in U_A, y \in V_A \implies U_A^- \cap V_A^- \ne \varnothing$

contradicting the fact that $T_A$ is a $T_{2 \frac 1 2}$ (completely Hausdorff) space.

Hence $T_B$ must after all be a $T_{2 \frac 1 2}$ (completely Hausdorff) space.

$\blacksquare$

## Sources

- 1970: Lynn Arthur Steen and J. Arthur Seebach, Jr.:
*Counterexamples in Topology*... (previous) ... (next): $\text{I}: \ \S 2$: Functions, Products, and Subspaces