Given $X×Y$ hausdorff. Show that $X$ hausdorff.

Assume $x_1≠x_2$ in $X$.
Then $(x_1,y_0)≠(x_2,y_0)$ for some $y_0∈Y$.
Then there exists disjoint open neighborhoods in $X×Y$.
As those neigborhoods are open there exist (disjoint) basis elements of $X×Y$ such that:
$(x_1,y_0)⊂U_1×V_1$ and $(x_2,y_0)⊂U_2×V_2$
where $U_1,U_2$ are open. If I can show that $U_1,U_2$ are disjoint, then I'm done, but I don't see how. Any hints ?

  • 2
    $\begingroup$ Suppose that $z \in U_1 \cap U_2$. What can you say about $(z,y_0)$? By the way, you need to require $Y \neq \varnothing$. $\endgroup$ – Daniel Fischer Mar 22 '14 at 19:25

You need to require that $Y$ be nonempty to draw any conclusions about $X$ from properties of $X\times Y$.

If $Y\neq \varnothing$, it is frequently useful to note that one can embed $X$ as a subspace in the product, for any $y_0 \in Y$, the map

$$\iota \colon X \to X\times Y;\quad \iota(x) = (x,y_0)$$

is continuous, since both of its components are, and its inverse is the restriction of a coordinate projection, hence also continuous. Thus $\iota$ is an embedding.

Hence $X$ has all properties that $X\times Y$ has and that are inherited by subspaces, such as being $T_0$, $T_1$, Hausdorff, $T_3$, $T_{3\frac{1}{2}}$ and more. If $Y$ is a $T_1$ space (more generally, if $Y$ contains a point $y_1$ such that $\{y_1\}$ is closed), then $X$ embeds as a closed subspace into $X\times Y$, and hence also has all properties of $X\times Y$ that are inherited by closed subspaces, such as being $T_4$ or complete (in case of uniform spaces).

Of course we can also see that $X$ must be Hausdorff if $X\times Y$ is and $Y\neq \varnothing$ in the way you went:

$$\varnothing = (U_1 \times V_1) \cap (U_2 \times V_2) = (U_1\cap U_2) \times (V_1 \cap V_2),$$

and since $y_0 \in V_1 \cap V_2$, it follows that $U_1 \cap U_2 = \varnothing$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.