1
$\begingroup$

This question is a kind of dual to this earlier one. Note that if we replace $\mathsf{FOL}$ with $\mathcal{L}_{\omega_1,\omega}$, things trivialize since we can use the theory $\{\varphi^A\leftrightarrow\varphi^B: \varphi\in\mathcal{L}_{\omega_1,\omega}\}$. The situation would also trivialize if we replaced $\mathsf{FOL}$ with $\mathsf{SOL}$, but not trivially: the point is that the game characterization by Vaananen and Wang is appropriately second-order expressible. Throughout, all signatures are finite.

Fix two unary relation symbols $A,B$ and a binary relation symbol $E$ (I'll call $\{E\}$-structures "graphs" for simplicity). For every first-order theory $T$ in a language $\supseteq\{A,B,E\}$, let $\sim_T$ be the relation on graphs given by $$\mathfrak{A}\sim_T\mathfrak{B}\quad\iff\quad\exists\mathfrak{M}\models T(A^\mathfrak{M}\upharpoonright\{E\}\cong\mathfrak{A}\wedge B^\mathfrak{M}\upharpoonright\{E\}\cong\mathfrak{B}).$$

Letting $\sim$ be the intersection of all $\sim_T$s which are equivalence relations coarsening $\equiv_{\omega_1,\omega}$, we get that $\sim$ is an equivalence relation on graphs which is isomorphism-invariant and (by a result of Farmer S.) strictly coarsens $\equiv_{\omega_1,\omega}$ on graphs in general.

Question: Is $\sim$ strictly coarser than $\equiv_{\omega_1,\omega}$ (or equivalently, $\cong$) on countable graphs?

That is, are there non-isomorphic countable graphs which are $\sim$-equivalent? Note that the obvious theory describing an isomorphism (or even potential isomorphism) between the $A$- and $B$-parts does not satisfy the $\equiv_{\omega_1,\omega}$-coarsening property.

$\endgroup$
10
  • 1
    $\begingroup$ Does the notation $A^M\upharpoonright\{E\}$ mean the structure $(A^M,E\upharpoonright A^M)$, i.e. with universe $A^M$ and binary relation the restriction of $E=E^M$ to this set? $\endgroup$
    – Farmer S
    Commented Feb 8, 2023 at 6:36
  • $\begingroup$ @FarmerS Yes, that's right. $\endgroup$ Commented Feb 8, 2023 at 15:52
  • $\begingroup$ I don't see why $\sim_T$ is an equivalence relation. What if e.g. $T$ satisfies "$A\neq\emptyset=B$"? Isn't $\sim_T$ then non-reflexive? $\endgroup$
    – Farmer S
    Commented Feb 8, 2023 at 21:26
  • $\begingroup$ @FarmerS That's not $\equiv_{\omega_1,\omega}$-invariant; $\sim$ is just the intersection of the $\equiv_{\omega_1,\omega}$-invariant $\sim_T$s. $\endgroup$ Commented Feb 9, 2023 at 1:00
  • $\begingroup$ (@FarmerS But I did forget to add transitivity. Now fixed!) $\endgroup$ Commented Feb 9, 2023 at 1:41

0

You must log in to answer this question.