7
$\begingroup$

Suppose $M$ is a countable transitive model of some fragment of $\mathbf{ZFC}$, $\mathbb{P}\in M$ is a forcing notion and $G, H$ are $\mathbb{P}$-generic such that $M[G]=M[H]$. Does it then follow that there is some automorphism $\pi:\mathbb{P}\longrightarrow\mathbb{P}$ such that $\pi\in M$ and $\pi[G]=H$?

If the answer is no, are there natural restrictions one could impose on $\mathbb{P}$ (maybe apart from $\mathbb{P}$ being finite) such that the above sentence holds?

$\endgroup$
2
  • $\begingroup$ I have recently found Lemma 25.5 in an old version of "Set Theory" by Thomas Jech, which states the exact result im looking for, but only in a version for complete boolean algebras (i.e. when we take $\mathbb{P}$ to be a complete boolean algebra). However, i am unsure how to adapt it to my case, even when $\mathbb{P}$ is separative and we can use the representation theorem. $\endgroup$ – Hannes Jakob Jan 28 at 14:29
  • 1
    $\begingroup$ The adaptation to posets other than complete Boolean algebras is, as I said in my answer, messy. One problem is that two posets can produce isomorphic complete Boolean algebras even when the posets are nowhere near isomorphic. Another is that the $b$ and $b'$ in my answer cannot generally be found in the posets (you can get one or the other in the poset you want, but not generally both). $\endgroup$ – Andreas Blass Jan 30 at 16:48
12
$\begingroup$

This works much better in terms of complete Boolean algebras. If $\mathbb B$ and $\mathbb B'$ are complete Boolean algebras and a $\mathbb B$-generic filter $G$ and a $\mathbb B'$-generic filter $G'$ generate the same forcing extension, then there are $b\in G$ and $b'\in G'$ such that the part of $\mathbb B$ below $b$ is isomorphic to the part of $\mathbb B'$ below $b'$ by an isomorphism in the ground model, that sends the restriction of $G$ to the restriction of $G'$.

I believe this fact is in Serge Grigorieff's paper "Intermediate submodels and generic extensions in set theory" [Ann. Math. Second Series, Vol. 101, No. 3 (May, 1975), pp. 447-490].

It's certainly possible to rewrite this in terms of posets instead of complete Boolean algebras, but the result looks messy to me.

$\endgroup$
1
  • $\begingroup$ I hope you don't mind, i have "unaccepted" the answer because i managed to find a more suitable result in an old Version of Jechs "Set Theory" that is more closely related to my question. $\endgroup$ – Hannes Jakob Jan 28 at 14:29

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.