3
$\begingroup$

Let $\mathcal M$ be a $\mathcal L$-structure, and $A\subseteq M$. Let $p(x)\in S_1(A)$ be a complete 1-type of the theory $T_A := \mathrm{Th}(\mathcal M_A)$. We say that $p$ is algebraic if there exists an integer $k$ such that $p$ has at most $k$ realisations in every model of $T_A$.

I have to show that if $p$ is an algebraic 1-type then it is isolated. The only thing I could think of is the following: let $k$ be the upper bound on the number of realisations of $p$, and let $C = \{c_1,\ldots,c_{k+1}\}$ be new constants. We then have that $T_A\cup \bigcup_{i=1}^{k+1} p(c_i) \models \bigvee_{i\neq j} c_i=c_j$, and by compactness there exists a finite subset $\Phi$ of $p$ such that $T_A\cup \bigcup_{i=1}^{k+1} \Phi(c_i) \models \bigvee_{i\neq j} c_i=c_j$, i.e. (treating $\Phi$ as a conjunction a formulas of $p$) $$T_A \models \forall x_1,\ldots,x_{k+1} \left(\bigwedge_{i=1}^{k+1}\Phi(x_i) \Rightarrow \bigvee_{i\neq j} x_i=x_j \right)$$

This only tells me that in every model of $T_A$ at most $k$ points may satisfy $\Phi$, but I don't see why $\Phi$ would isolate $p$. On the other hand, I don't see how to extract more information from $p$ than this bound on the number of realisations. I'd really appreciate a hint!

$\endgroup$
6
$\begingroup$

Hint: if a $\Phi$ does not isolate $p$, then there is a formula $\Psi \in p$, such that $\Phi \land \Psi$ define a proper subset of the set defined by $\Phi$. Repeat this argument for $\Phi \land \Psi$. Can this process go to infinity?

$\endgroup$
  • 2
    $\begingroup$ Since $\Phi$ defines a subset of size $\leq k$ in every model, this process terminates after at most $k$ steps. Letting $\Psi_1,\ldots,\Psi_l$ be the formulas obtained during the process, $\Phi\wedge \bigwedge_{i=1}^l \Psi_i$ isolates $p$. Thanks a lot! $\endgroup$ – zarathustra Jan 29 '14 at 13:08

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.