Let $T$ be a stable $L$-theory with elimination of imaginaries. We work in the monster model $\mathfrak C$ of $T$. Let $A$ be a small (infinite) set of the monster, $\phi(x,y)$ be a $L(A)$-formula and $a_i:i\in \omega$ be a Morley sequence over $A$ such that $\mathfrak C\models \phi(a_0,a_1)$.

Can we find a Morley sequence $b_i:i\in \omega$ over a finite set $A_0$ such that $\mathfrak C\models \phi(b_0,b_1)$?

Note:This is of course true for a superstable theory and if one replaces $\phi$ by a type then this fails in non-superstable theories.


Your question comes down to asking whether given a formula $\phi(x,b)$ which does not fork over $A$, there is a finite subset of $A$, $A_0$, such that $\phi(x,b)$ does not fork over $A_0$.

The answer is yes. The reason depends on your preferred definition of forking, of course. Using the definability of types definition, we just have to take $a$ independent of $b$ over $A$ and take $A_0$ to be the parameters appearing in the $\phi$-definition of $\text{tp}^\phi(a/Ab)$.

  • $\begingroup$ Thanks, in my notation take the $\phi$-type $tp^\phi(a_0/Aa_1)$. It is definable over some finite set $A_0$ hence non-forking over $A_0$. Now any non-forking extension of this $\phi$-type does the job, we choose a Morley sequence $/A_0$ in one of this nf-extension. $\endgroup$ – TimZ Mar 6 '14 at 11:33
  • $\begingroup$ Exactly. I just rephrased it because I wanted to emphasize that the nice property of superstable theories (nonforking always descends to a finite base) is true for stable theories if you only care about a single (or finitely many) formulas. $\endgroup$ – Alex Kruckman Mar 6 '14 at 17:11

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.