I want to prove that :
if $B$ is a countable boolean algebra then $St(B)$ ( = Stone space ) is separable.
Assuming $BPI(B)$ i.e. the boolean prime ideal princible for the boolean algebra $B$, there is a theorem stating that $St(B)$ is a compact, Housdorff, zero-dimensional topological space and $B \cong CLOP(St(B))$ where $CLOP(St(B))$ is the set of the Clopen sets of $St(B)$.
I have the homomorfism $N : B \to P(St(B))$ defined by $N(b) = \{U \in St(B) | b \in U\}$ and $\{N(b) | b \in B \}$ is a base of clopen set.
Since $B$ is countable i can enumerate the clopen sets : $\{N(b_i) | b_i \in B \land i \in \omega\}$. For every $i \in \omega$ we have that $\bigcap N(b_i)$ is non-empty (because every $U \in N(b_i)$ contains $b_i$) and it is a basis for a filter on $B$ that can be extended to an ultrafilter $U_i$ by $BPI(B)$. I'm pretty sure that the set $V = \bigcup_{i \in \omega} \{U_i\}$ is the countable dense set that witnesses the separability of St(B) but I can't prove that its closure is St(B). Any advice? Any misteake? Any element $N(b_i)$ of the base of clopen sets has non-empty intersection with $V$; is that enough ? if yes, why?

  • $\begingroup$ Are you trying to avoid the axiom of choice other than assuming $BPI(B)$? (Note that you don't actually need to assume $BPI(B)$ as a form of choice; you can prove $BPI$ for any well-orderable Boolean algebra in $ZF$.) $\endgroup$ – Eric Wofsey Jan 18 '18 at 20:34
  • $\begingroup$ Well, I ask because in choosing your sequence of ultrafilters $U_i$, you are seemingly using the axiom of choice. (This use can be avoided again though because $B$ is well-orderable.) $\endgroup$ – Eric Wofsey Jan 18 '18 at 20:46
  • $\begingroup$ $B$ is well-orderable (because it's countable) then there's a choice function on it! right? $\endgroup$ – M.B. Jan 18 '18 at 20:51

Indeed, the Stone space has a countable clopen base, and any second countable space has a countable dense subspace namely pick a point from each basic set, which is what you did.

A set $D\subseteq X$ is dense in $X$ iff every non-empty (basic) open set of $X$ intersects $D$.

  • $\begingroup$ That's the idea, but how can I prove that $V$ is dense in St(B)? $\endgroup$ – M.B. Jan 18 '18 at 20:27
  • $\begingroup$ @M.Bailetti Because every basic set intersects the set by construction. $\endgroup$ – Henno Brandsma Jan 18 '18 at 20:32
  • $\begingroup$ ok, I suspected it, but I was not sure! $\endgroup$ – M.B. Jan 18 '18 at 20:40
  • $\begingroup$ @M.Bailetti it’s shown (it’s pretty easy) in several answers on this site, as well. It’s a standard fact. $\endgroup$ – Henno Brandsma Jan 18 '18 at 20:58

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.