# If two classes of models are disjoint then there exists a pair of sentences that aren't satisfiable.

Given a theory $S$, $M(S)$ denotes the class of models of $S$, that is, if $A\in M(S)$, then $A\vDash S$.

The theorem goes like this: Given two theories $S$ and $T$ of the same first order language such that $M(S)\cap M(T)=\emptyset$, there exists $\theta\in S$ and $\phi\in T$ such that no models of $\theta$ is a model of $\phi$.

To prove the contrapositive statement, suppose that for every pair of $\theta\in S$ and $\phi\in T$, there exists a model $A$ such that $A\vDash\theta\land\phi$.

Let $\Pi\subset S\cup T$ be a finite subset. If I can show that $\Pi$ has a model, then by compactness theorem we can conclude that $S\cup T$ has a model, which finishes the proof.

If $\Pi\subset S$ or $\Pi\subset T$, then the result follows immediately (since $M(S)\neq\emptyset$ and $M(T)\neq\emptyset$). So suppose $\Pi=\{\theta_1,...\theta_m,\phi_1,...\phi_n\}$ with each $\theta_i\in S$ and each $\phi_i\in T$. Then for each pair of $\theta_i$ and $\phi_j$ there is a model $A_{i,j}$ such that $A_{i,j}\vDash\theta_i\land\phi_j$.

Here is where I'm stuck, I want to construct a model $A$ that satisfies $\theta_i\land\phi_j$ for each $i\leq m$ and $j\leq n$. But I'm not sure how/if am I able to do that.

• What definition of "theory" are you using? Is a theory just any set of sentences, or is it deductively closed? – bof Sep 28 '17 at 10:15
• $S$ is a theory in a first order language $L$ if $S=Ent(\Sigma)$ for some $\Sigma\subset L$. $Ent(\Sigma):=\{\phi\mid\phi$ is a $L$ sentence, and if $A\vDash\Sigma$ then $A\vDash\phi\}$. – Sid Caroline Sep 28 '17 at 10:23
• @bof yes, a theory is deductively closed. A theory is a set of sentences that are consequences of a set of axioms. – Sid Caroline Sep 28 '17 at 10:26
• So, if $\theta_1,\dots,\theta_m\in S,$ then $\theta_1\land\cdots\land\theta_m\in S?$ – bof Sep 28 '17 at 10:28
• Why can't you just say: if $S\cup T$ is inconsistent, then some finite subset $S_0\cup T_0\ (S_0\subseteq S,\ T_0\subseteq T)$ is inconsistent; let $\theta$ be the conjunction of the sentences in $S_0$ and let $\phi$ be the conjunction of the sentences in $T_0?$ – bof Sep 28 '17 at 10:33

To prove the contrapositive statement, suppose that for every pair of $\theta\in S$ and $\phi\in T$, there exists a model $A$ such that $A\vDash\theta\land\phi$.
Suppose for a contradiction $S\cup T$ is not satisfiable. Then by compactness theorem there exists a finite subset $S_0\cup T_0\subset S\cup T$ that is not satisfiable. Let $S_0\cup T_0$ be such, with $S_0=\{\theta_1,...,\theta_m\}\subset S$ and $T_0=\{\phi_1,...,\phi_n\}\subset T$. Then $S_0\cup T_0=\{\theta_1,...,\theta_m,\phi_1,...,\phi_n\}$.
Let $\theta=\theta_1\land...\theta_m$, $\phi=\phi_1\land...\phi_n$. Then by definition of theory, $\theta\in S$ and $\phi\in T$. By our hypothesis, there exists a model $A$ such that $A\vDash\theta\land\phi$. Hence $A\vDash\theta_1\land...\land\theta_m\land\phi_1\land...\land\phi_n$. Hence $A\vDash S_0\cup T_0$, which contradicts that $S_0\cup T_0$ being not satisfiable.
Hence $S\cup T$ is satisfiable. So $M(S)\cap M(T)\neq\emptyset$.