3
$\begingroup$

I have a theory with finitely many relations, and would like to find a model of it with continuum-many 1-types realized, and one 2-type omitted. Is there a version of the Omitting Types Theorem that would help me in this case? The proof in Marker's "Model Theory: An Introduction" explicitly uses countability of the language, so I can't for instance introduce continuum many constant symbols for the types I want to realize.

Specifically, I'm working with the weak monadic second-order theory of 1 successor (treated as a two-typed first order structure, i.e. Henkin semantics). That is, the two-typed structure $(\mathbb{N}, \mathcal{F}(\mathbb{N}), \in, S)$, where $\mathcal{F}(\mathbb{N})$ is the set of finite subsets of $\mathbb{N}$, and $S$ is the successor operation on $\mathbb{N}$. One can describe $\leq$ in this structure in terms of sets which are closed under $S^{-1}$, so it's often more intuitive to think of this as the structure $(\mathbb{N}, \mathcal{F}(\mathbb{N}), \in, \leq)$.

A nonstandard model of this structure (up to isomorphism) consists of a nonstandard model of $(\omega, \leq)$ (call it $A$) for its first-order part, then a collection of subsets of $A$ for its "second-order" part. This collection of subsets $B$ has to, for instance, be closed under union and set-subtraction, and each "second order" element $X \in B$ has to have a $\leq$-maximal element.

I'm curious if there is a nonstandard model whose "first order part" is $\omega + \zeta$ (thus omitting the 2-type of having two points infinitely far from each other and infinitely far from 0), but whose "second order part" $B$ satisfies: $\{X \cap \omega | X \in B\} = \mathcal{P}(\omega)$. That is, for every subset of the natural numbers, there is a "second order element" in the nonstandard model containing exactly those natural numbers. Since every element of the second type must have a $\leq$-maximal element, naturally all of these second order elements will have to also contain nonstandard first order elements.

If one drops the $\omega + \zeta$ requirement, one can prove the existence of a nonstandard model whose second order part "restricts" to $\mathcal{P}(\omega)$ by a simple compactness argument. For each $K \subset \omega$, add in a constant symbol $C_K$ and add $\{S^n(0) \in C_K\}_{n \in K}$ and $\{S^n(0) \notin C_K\}_{n \notin K}$ to the theory. These are the continuum many 1-types I'd like to realize.

I proved countably many such types can be realized at once in an $\omega+\zeta$ model in section 4.6 of my Ph.D. Thesis: https://ecommons.cornell.edu/handle/1813/44393.

$\endgroup$
9
  • $\begingroup$ Can you say a little more about what exactly the theory is? Do you want your model to be an elementary extension of the full monadic second-order standard model? $\endgroup$ Commented Jul 26, 2023 at 22:13
  • $\begingroup$ Thanks for clarifying. It seems you want a nonstandard model of the theory of $\langle\mathbb{N},F(\mathbb{N}),\in,\leq\rangle$ with order type $\omega+\zeta$ in the first-order part. Is that right? I find it odd to speak of a nonstandard model of a structure (which you do twice), but I guess you just mean a nonstandard model of the theory of that structure. It seems that "weak" means you only allow bounded monadic classes into the second-order part—is that right? $\endgroup$ Commented Jul 26, 2023 at 23:20
  • $\begingroup$ Some people may not know that $\zeta$ you mean the order-type of the integers $\mathbb{Z}$. So you are looking for a model of order tyle $\mathbb{N}+\mathbb{Z}$. $\endgroup$ Commented Jul 27, 2023 at 1:34
  • 1
    $\begingroup$ When I use "weak monadic second-order" here I'm just distinguishing between the theory of $(\mathbb{N}, \mathcal{F}(\mathbb{N}), \in, \leq)$ and $(\mathbb{N}, \mathcal{P}(\mathbb{N}), \in, \leq)$ where $\mathcal{F}$ is finite subsets and $\mathcal{P}$ is the full powerset. The theories are very similar, but only in the former do sets necessarily have largest elements, and the latter requires Buchi automata to analyze. On the other hand, my question is just about whether there is an omitting types-like theorem that I should be looking into. $\endgroup$
    – TomKern
    Commented Jul 27, 2023 at 1:39
  • 2
    $\begingroup$ Omitting types in uncountable languages is pretty subtle. I don't think it's going to be easy to find a relevant general theorem. $\endgroup$ Commented Jul 27, 2023 at 5:49

0

You must log in to answer this question.

Browse other questions tagged .