Consider structures $(A,f)$ encoding a Boolean algebra $A$ endowed with an automorphism $f$. There is an obvious notion of isomorphism between such structures.

Consider the endomorphism $\hat{\Phi}$ of the Boolean algebra $2^\omega$ of subsets of $\omega$ given by $A\mapsto \{a\in\omega:a+1\in A\}$. It induces an automorphism $\Phi$ of the quotient Boolean algebra $2^\omega/\mathrm{fin}$, where $\mathrm{fin}$ is the ideal of finite subsets. (Under Stone duality, this corresponds to the self-homeomorphism of the Stone-Čech remainder of $\omega$ induced by $n\mapsto n+1$.)

Whether $(2^\omega/\mathrm{fin},\Phi)$ and $(2^\omega/\mathrm{fin},\Phi^{-1})$ are isomorphic is essentially unknown (see my previous related question for more details).

My question is

Are $(2^\omega/\mathrm{fin},\Phi)$ and $(2^\omega/\mathrm{fin},\Phi^{-1})$ elementary equivalent?

That is, do they satisfy the same first-order sentences (in the language of Boolean algebras endowed with an automorphism)?

Note that $\Phi^n$ has exactly $2^{|n|}$ fixed points for $n\neq 0$. In particular, $(2^\omega/\mathrm{fin},\Phi)$ and $(2^\omega/\mathrm{fin},\Phi^n)$ are not not equivalent for $|n|\ge 2$.

If the question has a positive answer, it is tempting to ask whether one can characterize simply those $(A,f)$ having the same first-order theory as $(2^\omega/\mathrm{fin},\Phi)$.

  • 2
    $\begingroup$ If they are elementary equivalent, then a back and forth argument shows that they are isomorphic under CH. (And whether they are isomorphic under CH is unknown.) So I think this is open. $\endgroup$ Feb 19, 2020 at 15:17
  • 2
    $\begingroup$ @PaulMcKenney this was indeed my hope to pass from EE to isomorphism, but at the same time if they're not EE, it should be for some "concrete"reason. So I hope this approach leads to something. $\endgroup$
    – YCor
    Feb 19, 2020 at 15:30
  • 2
    $\begingroup$ @PaulMcKenney how does the back-and-forth argument work? is $(2^\omega/\mathrm{fin},\Phi)$ $\omega_1$-saturated? since it seems to reduce to my previous question, I think this would be worth an answer. $\endgroup$
    – YCor
    Feb 19, 2020 at 16:44
  • 2
    $\begingroup$ That's what I was thinking, but it's no longer clear to me that $(2^\omega / \mathrm{fin}, \Phi)$ is saturated. $\endgroup$ Feb 19, 2020 at 22:56
  • $\begingroup$ @PaulMcKenney: I was thinking a bit more about this question today and, for what it's worth, I can show that $(2^\omega/\mathrm{fin},\Phi)$ is not $\omega_1$-saturated. I have a somewhat tedious proof of this assertion, but this comment is too short to contain it . . . but feel free to email me if you'd like some details. $\endgroup$
    – Will Brian
    Aug 11, 2020 at 14:39

1 Answer 1


Yes, these two structures are elementarily equivalent.

This is proved as a corollary to another theorem, which states

Theorem: CH implies that $\Phi$ and $\Phi^{-1}$ are conjugate to each other in the automorphism group of $\mathcal P(\omega) / \mathrm{fin}$.

You can find a proof of this on the arXiv (here), and a short overview of the proof in my answer to this MO question. Let me add a few words explaining why the theorem quoted above implies that the structures $\langle \mathcal P(\omega) / \mathrm{fin},\Phi \rangle$ and $\langle \mathcal P(\omega) / \mathrm{fin},\Phi^{-1} \rangle$ are elementarily equivalent (even without assuming CH).

The short version is: we can force CH without changing the theory of $\langle \mathcal P(\omega) / \mathrm{fin},\Phi \rangle$ and $\langle \mathcal P(\omega) / \mathrm{fin},\Phi^{-1} \rangle$, so if they're isomorphic in the forcing extension, and their theories have not changed, they must have been elementarily equivalent to begin with.

In more detail, let $\mathbb P$ denote the poset of countable partial functions $\omega_1 \times \omega \to \omega$, ordered by extension, i.e., the usual poset for forcing CH with countable conditions. Let $G$ be a $V$-generic filter on $\mathbb P$. Because $\mathbb P$ is a countably closed notion of forcing, no new subsets of $\omega$ are added by $\mathbb P$, which means that $V \cap \mathcal P(\omega) / \mathrm{fin} = V[G] \cap \mathcal P(\omega) / \mathrm{fin}$. Furthermore, because $\Phi$ is definable by the simple formula $\Phi([A]) = [A+1]$, the action of $\Phi$ is the same in $V$ and in $V[G]$. Thus the structures $\langle \mathcal P(\omega) / \mathrm{fin},\Phi \rangle^V$ and $\langle \mathcal P(\omega) / \mathrm{fin},\Phi \rangle^{V[G]}$ are one in the same. It follows (by the ``absoluteness of the satisfaction relation'') the theory of $\langle \mathcal P(\omega) / \mathrm{fin},\Phi \rangle$ is the same in $V$ and in $V[G]$. The same argument applies to $\langle \mathcal P(\omega) / \mathrm{fin},\Phi^{-1} \rangle$, so the theory of $\langle \mathcal P(\omega) / \mathrm{fin},\Phi^{-1} \rangle$ is also the same in $V$ and in $V[G]$. Because $V[G] \models$ CH, the theorem quoted above implies $\langle \mathcal P(\omega) / \mathrm{fin},\Phi \rangle$ and $\langle \mathcal P(\omega) / \mathrm{fin},\Phi^{-1} \rangle$ are conjugate in $V[G]$. In particular, they are elementarily equivalent in $V[G]$. Because the theories of these two structures is the same in $V$ and in $V[G]$, this means they have the same theory in $V$ as well.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.