3
$\begingroup$

Let $\pi$ is a $\Bbb{P}$-name for a partial order, i.e. there is a name $\pi'$ and $\pi''$ such that $$1\Vdash_\Bbb{P} \pi '' \in \pi\land (\text{$\pi'$ is a partial order of $\pi$ with largest element $\pi''$}).$$ We call $\pi$ is full for $\searrow$ $\omega$-sequences if whenever $p\in \Bbb{P}$, $\rho_n\in\operatorname{dom} \pi$ for each $n<\omega$ and $$p\Vdash \rho_n\in\pi \land \rho_{n+1} \le \rho_n$$ for each $n$ then there is a $\sigma\in\operatorname{dom} \pi$ s.t. $p\Vdash \sigma\in\pi$ and $p\Vdash \sigma\le \rho_n$ for each $n$.


I try to prove following statement: if $\Bbb{P}$ is $\omega_1$-closed and $\pi$ is full for $\searrow$ $\omega$-sequences, then $1\Vdash \text{$\pi$ is $\omega_1$-closed}.$

I made two 'proofs' which the former one is regarded as false and the latter one is incomplete. The 'false proof' goes as follow:

Let $\rho$ is a $\Bbb{P}$-name satisfying $$1\Vdash \forall n<\check{\omega}:\rho(n+1)\le \rho(n) \,\land\, \rho(n)\in \pi .$$ so for each $n$, $1\Vdash \rho(n+1)\le \rho(n)$ and $1\Vdash \rho(n)\in\pi$. (*)

Therefore, since $\pi$ is full for $\searrow$ $\omega$-sequences, we can find a name $\sigma$ such that $1\Vdash \sigma\le \rho(n)$ for all $n<\omega$ and $1\Vdash \sigma\in\pi$, so $1\Vdash \forall n<\check{\omega}: \sigma\le\rho(n)$ (**) and therefore $1\Vdash \text{$\pi$ is $\omega_1$-closed}$.

I doubt the part (*) and (**) are problematic. Is it right? I didn't get why they are problematic so I need an explanation for them.

I guess the 'proof' given below goes to right direction:

Let $p\in \Bbb{P}$ be arbitrary and $\rho$ be a $\Bbb{P}$-name satisfying $$p\Vdash \forall n<\check{\omega}: \rho(n+1) \le\rho(n)\,\land\, \rho(n)\in\pi.$$ Since $p\Vdash \exists x \in \pi : \rho(\check{0}) = x$, there is $p_0\le p$ and $\Bbb{P}$-name $\sigma_0$ s.t. $$p_0 \Vdash \sigma_0\in \pi\,\land\, \rho(\check{0}) = \sigma_0.$$ Similarly, we can take $p_1\le p_0$ and a $\Bbb{P}$-name $\sigma_1$ satisfying $$p_1 \Vdash \sigma_1\in \pi\,\land\, \rho(\check{1}) = \sigma_1.$$ and so on. Since $\langle p_n : n<\omega\rangle$ is decreasing $\Bbb{P}$-sequence, we can find some $q\in \Bbb{P}$ with $q\le p_n$ for all $n<\omega$. For such $q$ should satisfy $$q\Vdash \sigma_{n+1}\le\sigma_n\,\land\, \sigma_n\in\pi$$ for each $n<\omega$, so we can find a $\Bbb{P}$-name $\sigma$ s.t. $q\Vdash \sigma\le\sigma_n$ and $q\Vdash \sigma\in\pi$ for each $n$.

In above proof, however, I wonder can I conclude that $q\Vdash \forall n<\check{\omega} : \sigma\le \rho(n)$ holds. I would appreciate any help, thanks!

$\endgroup$
  • 1
    $\begingroup$ What comes to my mind concerning the first proof is: In the assumption needed to apply "fullness", we need a name $\rho_n$ for every $n<\omega$, but in your proof, you only use one name $\rho$. (*) is correct, but it doesn't enable you to apply the consequence for "full". $\endgroup$ – user35359 Aug 29 '15 at 3:42
3
$\begingroup$

The answer to your last question is affirmative: Any extension by a generic filter that contains $q$ will satisfy the formula on the right side. (For the same reason, your steps (*) and (**) are correct, but the problem with the first proof is a different one I think, see my comment.)

Your proof is almost done. To complete the argument, we could assume for a contradiction that there is $p$ so that

$p\Vdash (\rho : \check{\omega} \to \pi) \land (\forall n<\check{\omega} \ (\rho(n)\in\pi \land \rho(n+1)\leq \rho(n)))\land (\forall x\in\pi \, \exists n<\check{\omega} \ x \not \leq \rho(n))$

Then by your argument we get $q\leq p$ with $q\Vdash \sigma\in\pi \land \forall n<\check{\omega}\ \sigma \leq \rho(n)$

so $q$ forces a contradiction, which is impossible.

$\endgroup$

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.