4
$\begingroup$

Suppose $G \subset \mathbb{P}$ is a $\mathbb{P}$-generic filter over a countable transitive model $M \ni \mathbb{P}$ (satisfying enough of ZFC). Then $G$ is an ultrafilter.

The idea of the proof, outlined in this answer is to look at $$ D = \{ q \in \mathbb{P} : q \leq p \} \cup \{ q \in \mathbb{P} : q \leq \neg p \} $$ and see that it is dense, but there's a problem with this argument when looking at a general poset $\mathbb{P}$, namely that in a general poset, we don't have negations/complements!

We can complete an arbitrary poset to a complete boolean algebra $\mathbb{B}$, but does this transformation equally turn $G \subset \mathbb{P}$ into a generic filter $G^\prime \subset \mathbb{B}$?

Furthermore, will showing that $G^\prime$ is an ultrafilter imply that $G$ itself is an ultrafilter?

Finally, if $\mathbb{P}$ is a boolean algebra in the original statement, I have an idea of how to proceed:

Take arbitrary $r \in \mathbb{P}$. Suppose $r \cdot p = 0$. Then $\neg (p \cdot r) = 1$, so $\neg p + \neg r = 1$. Hence, $\neg p \cdot r = r$, so $\neg p \leq r$. Likewise, if $r \cdot \neg p = 0$, then $r \leq p$. Therefore, either $r \leq p$ or $r \leq \neg p$. This shows that $D$ is dense.

But this argument feels wrong to me. I never actually use $D$ itself, and instead look only at $p$ and $\neg p$.

Since the argument via the boolean completion appears unnecessarily complicated, maybe it's better to show that $G \subset \mathbb{P}$ is maximal. In that case I think I should proceed by contradiction to obtain $G^\prime \supset G$ and build a dense set out of $G^\prime \setminus G$ to contradict that $G$ is a generic filter. However, I don't see how to build this dense set.

$\endgroup$
3
  • $\begingroup$ What does "ultrafilter" mean if you don't have a boolean algebra? A maximal proper filter? $\endgroup$ Apr 10 '17 at 1:11
  • $\begingroup$ Yes, a maximal proper filter. $\endgroup$ Apr 10 '17 at 1:22
  • $\begingroup$ In your argument, you showed $rp=0$ implies $r\leq\neg p$ and $r(\neg p)=0$ implies $r\leq p$. But typically neither $rp$ nor $r(\neg p)$ will be $0$. The trick is that if $rp\neq 0$, then $rp$ itself is an extension of $p$ in $D$. $\endgroup$ Apr 10 '17 at 1:26
3
$\begingroup$

For any $p\in\mathbb{P}$, let $A(p)=\{q\in\mathbb{P}:q\leq p\}$ and let $B(p)$ be the set of $q\in\mathbb{P}$ that are incompatible with $p$. Then I claim that $D(p)=A(p)\cup B(p)$ is dense (if $\mathbb{P}$ is Boolean then $B(p)=\{q:q\leq\neg p\}$, so this is your $D$). Indeed, if $r\in\mathbb{P}$ has no extension in $A(p)$, that means exactly that $r$ is incompatible with $p$, so $r\in B(p)$.

Thus if $G$ is generic, $G\cap D(p)\neq 0$ for all $p$. If $p\not\in G$, then $G\cap A(p)=\emptyset$. We thus must have $G\cap B(p)\neq\emptyset$, which means no filter containing $G$ can contain $p$. Since $p\not\in G$ was arbitrary, $G$ is maximal.

$\endgroup$

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.