It is instructed in Munkre's topology that proof of imbedding theorem is almost the copy of step 1 of this post.It is instructed to just replace $n$ by $\alpha$ and $\mathbb R^{\omega}$ by $\mathbb R^J$.

Imbedding theorem:Let X be a space in which one point sets are closed.Suppose that {$f_{\alpha}$}_{$\alpha\in J$}} is an indexed family of continuous functions $f_{\alpha}:X\in \mathbb R$ satisfying the requirement that for each point $x_0$ of X and each nbhd U of $x_0,$ there is an index $\alpha$ such that $f_{\alpha}$ is positive at $x_0$ and vanish outside U.then the funcion $F:X\rightarrow \mathbb R^J$ defined by $F(x)=(f_{\alpha})_{\alpha\in J}$ is an imbedding of X in $\mathbb R^J.$if $f_{\alpha}$ maps X into $[0,1]$ for each $\alpha,$ then $F$ imbeds $X$ in $[0,1]^J$We shall prove that X is metrizable by imbedding X into a metrizable space Y;that is by showing that X is homeomorphic to some subspace of Y.

Step 1: We prove the following:-

There exists an indexed collection of continuous functions $f_{\alpha}:X\rightarrow [0,1]$ having the property that given any point $x_0$ of $X$ and any neighborhood $U$ of $x_0$,there exists an index $\alpha$ such that $f_{\alpha}(x_0)>0$ and $f_\alpha$ vanishes outside $U$.


Let {$B_{\alpha}$} be the countable basis for $X$.For each pair $\alpha,\beta$ of indices for which $\overline{B_{\alpha}} \subset B_{\beta},$apply Urysohn's lemma to choose a continuous function $g_{\alpha,\beta}:X\rightarrow[0,1] $ such that $g_{\alpha,\beta}[\overline{B_{\alpha}}]=\{1\}$ and $g_{\alpha,\beta}[X-B_{\beta}]=\{1\}$. Then the collection {$g_{\alpha,\beta}$} satisfies our requirement:

Given $x_0$ and given a neighbourhood $U$ of $x_0,$ one can choose a basis element $B_m$ containing $x_0$ that is contained in $U$. Using Regularity,one can then choose $B_{\alpha}$ so that $x_0 \in B_{\alpha}$ and $\overline{B_{\alpha}} \subset B_{\beta}$.Then $\alpha,\beta$ is a pair of indices for which function $g_{\alpha,\beta}$ is defined; and it is positive at $x_0$ and vanishes outside of $U$. Because the collection {$g_{\alpha,\beta}$} is indexed with a subset of indexing set $J$, gives us the desired collection {$f_{\alpha}$}.

Given a function $f_{\alpha}$ of step 1,take $\mathbb{R}^{J} $ in the product topology and define a map $F: X\rightarrow \mathbb{R}^{J}$ by the rule- $F(x)=(f_{\alpha})_{\alpha\in J}$

We assert that $F$ is an embedding-

$F$ is continuous

Because $\mathbb{R}^{\omega}$ has a product topology and each $f_{\alpha}$ is continuous.

$F$ is injective

Let $x,y\in X$ such that $x\neq y\implies$ there is some basic element $B_{\alpha}$ that contains $x$ and misses y.By applying Urysohn's lemma there exists a function $f_{\delta}:X\rightarrow [0,1]$ such that $f_{\delta}(x)=1,f_{\delta}(y)=0$.So,$F(X)\neq F(y)$(because the image differs atleast in one coordinate)

$F$ is a homeomorphism of $X$ onto its image, the subspace $Z=F[X]$ of $\mathbb{R}^{J} $.

Because each component of $F$ is continuous,so $F$ is continuous.**

$F:X\rightarrow F[X]$ is a bijection. So,we need only show that for each open set $U$ in $X,$ the set $F[U]$ is open in $Z=F[X]$.

Let $z_0\in F(U).$

We shall find an open set $W$ of Z such that $z_0\in W\subset F(U)$.

Let $x_0\in U$ such that $F(x_0)=z_0.$Choose an index $\theta$ for which $f_{\theta}(x_0)>0$ anf $f_{\theta}(X-U)=${$0$.}

Take the open ray $(0,\infty) $ in $\mathbb R$ and let $V$ be the open set $V=\pi_{\theta}^{-1}((0,\infty))$ of $\mathbb R^{J}$.

Let $W=V\cap Z;$ then $W$ is open in $Z$ by definition of subspace topology.We assert that $z_0\in W\subset F(U).$

First $z_0\in W$ because $\pi_{\theta}(z_0)=\pi_{\theta}F(z_0)=f_{\theta}(x_0)>0.$

Second,$W\subset F(U).$For if $z\in W$ then $z=F(x)$ for some $x\in X$ and $\pi_N(z)\in (0,\infty)$.

Since,$\pi_{\theta}(z)=\pi_{\theta}(F(x))=f_{\theta}(x),$ and $f_{\theta}$ vanish outside $U,$the point $x$ must be in $U.$

Then,$z=F(x)$ is in $F(U)$,as desired.

Thus,$F$ is an imbedding of $X$ in $\mathbb R^J.$

Also,examine this proof critically,and if there is some scope of improvement please let me know...

  • $\begingroup$ We don't need Urysohn's lemma but just the property for the functions and the fact one point set aree closed. See my proof below. Bases are irrelevant. $\endgroup$ – Henno Brandsma Nov 24 '18 at 9:41

You don't have a countable base for $X$. So any part of the proof that refers to those base elements must be scratched.

You just know 2 things about $X$:

  • One-point sets in $X$ are closed. ($X$ is $T_1$)

  • There is a family of continuous functions $\{f_\alpha: X \to \mathbb{R}: \alpha \in J \}$ that obeys:

$(\ast)$ For every $x_0 \in X$ and every open neighbourhood $U$ of $x_0$, there exists some $\alpha_0 \in J$ such that $f_{\alpha_0}(x_0) > 0 $ and $f_{\alpha_0}(x)=0$ for all $x \notin U$.

We then define $F:X \to \mathbb{R}^J$ by $F(x)= (f_\alpha(x))_{\alpha \in J}$, or equivalently $\pi_\alpha \circ F = f_\alpha$ for all $\alpha \in J$. It is immediate that $F$ is continuous as the its compositions with all projections are continuous (universal property of maps into products, that Munkres has in its text).

So $F:X \to Z=f[X]$ is also continuous, and onto by definition.

That leaves to check that $F$ is 1-1 and $F$ is open as a map onto $Z$.

If $x \neq y$ we use the first property and take $U = X\setminus \{y\}$ which is an open neighbourhood of $x$. We apply $(\ast)$ to get the promised $f_{\alpha_0} : X \to \mathbb{R}$ with $f_{\alpha_0}(x)> 0$ and $f_{\alpha_0}(y)=0$. So $F(x)$ and $F(y)$ differ at the $\alpha_0$-th coordinate and so $F(x) \neq F(y)$ and $F$ is 1-1.

With $x_0\in U$, $U$ open in $X$, such that $F(x_0) = z_0 \in F[U]$ a you chose, we find by $(\ast)$ (Applied to $x_0$ and $U$) again some $\beta \in J$ (you have $\theta$ but I prefer to use $\beta$ after $\alpha$) such that

$$f_\beta(x_0) >0 \text{ and } f_\beta(x) =0 \text{ for } x \notin U\tag{1}$$

Then of course $$\pi_\beta(F(x_0)) = f_\beta(x_0) \in (0,\infty)$$ so that $$z_0= F(x_0) \in \pi_{\beta}^{-1}[(0,\infty)] \cap Z = W$$

The fact that $W \subseteq F[U]$ is clear: $y \in W$ implies $y \in Z =f[X]$, so there is some $x \in X$ such that $f(x) =y$.

If $x \notin U$ we know that $f_\beta(x) = 0$, by how we chose $\beta$ in $(1)$, and so $F(x) \notin \pi_\beta^{-1}[(0,\infty)] =V$, but this cannot be as $F(x) \in W= V \cap Z$. So $x \in U$ must hold and so $y=F(x) \in F[U]$ showing the inclusion $W \subseteq F[U]$.

  • $\begingroup$ :While showing injectivity I think ,by mistake you wrote $f_{\alpha_0}(y)=1$ instead of $f_{\alpha_0}(y)=0$. $\endgroup$ – P.Styles Dec 2 '18 at 13:12
  • $\begingroup$ @P.Styles already edited. $\endgroup$ – Henno Brandsma Dec 2 '18 at 13:13

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.