It would be appreciated if someone could review my proof for accuracy. Thanks!

Show that every compact metrizable space has a countable basis


Let X be a compact metrizable space. Then let d be a metric that induces the topology on X. Let A$_n$ = { B$_d$(x,1\n) | x $\in$ X, n$\in$ Z$_+$ }. Then each A$_n$ is an open covering of X. Since X is compact, each A$_n$ has a finite subcovering, call this subcovering A$'$$_n$.

Then $B'$ = $\bigcup$A$'$$_n$ is a countable set as it is the countable union of finite sets. $B'$ is also our desired countable basis since:

Given the original, uncountable basis, $B$, we have that for any basis element $B_i$ $\in$ $B$, and for any x $\in$ $B_i$, there must be some open $\epsilon$-ball such that B(x,$\epsilon$) $\subset$ $B_i$.

Now we can take $N$ such that 1/$N$ $\lt$ $\epsilon$/2. Then for all $A'_n$ for n $\gt$ N we have that each $A'_n$ must be an open covering of X and hence must have some open ball centered around some y $\in$ X such that x$\in$ B(y,1/n). Then B(y,1/n) $\subset$ $B_i$ since suppose some z $\in$ X - $B_i$ was contained within B(y,1/n). Then we would have:

$$ d(x,z) \le d(x,y) + d(y,z) $$ $$ = d(x,z) \lt \epsilon/2 + \epsilon/2 = \epsilon $$

But B(x,$\epsilon$) was chosen to be completely contained within $B_i$. Hence z cannot be an element of B(x,$\epsilon$), so we have a contradiction and so z cannot be an element of B(y,1/n). Hence B(y,1/n) $\subset$ $B_i$.

So, $B'$ is finer than the original basis $B$. The reverse inclusion is clear since the collection of $\epsilon$ balls for $\epsilon$ $\lt$ 1 form an equivalent basis for X. These $\epsilon$ balls are clearly contained within our proposed countable basis $B'$.

Hence $B'$ is a countable basis for X.


I dont understand why you use the original $B$. You have to show that $\bigcup_nA'_n$ is a basis. This is equivalent to saying that for every open subset $U$ and $x\in U$, there exists $n$ such that $x\in B(x_n,1/n)$ where $B(x_n,1/n)$ is an element of $A'_n$ and $B(x_n,1/n)\subset U$.

To show that since $U$ is open, there exists $r>0$ such that $B(x,r)\subset U$, take ${1\over{n}}<{r\over 4}$. There exists $B(x_n,1/n)$ such that $x\in B(x_n,1/n)$. For every $y\in B(x_n,1/n)$, $d(x,y)\leq d(x,x_n)+d(x_n,y)\leq {2\over n}<r/2$ since ${1\over n}<{r\over 4}$.We deduce that $y\in B(x,r)$ and $B(x_n,1/n)\subset B(x,r)\subset U$.

  • $\begingroup$ I use the original basis B to show they generate the same basis elements, hence establishing equivalent topologies $\endgroup$ – H_1317 Apr 21 '19 at 22:47

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.