Let $\mathbb{I}=\langle a_{i} \mid i< \omega + \omega\rangle$ be a sequence of indiscernibles over a set $A$. Then $\langle a_{i} \mid \omega \le i< \omega + \omega\rangle$ is an indiscernible sequence over $A \cup \langle a_{i} \mid i< \omega \rangle$.

This is a specific case of a stronger result: Let $(I, <)$ be an infinite linearly ordered set and suppose that exists $p \in I$ such that $\{q \in I \mid p<q\}$ is infinite, if $\langle a_{i} \mid i \in I \rangle$ is a sequence of indiscernible over $A$ then $\langle a_{i} \mid i \in I, p<i\rangle$ is a sequence of indiscernibles over $A \cup \{a_{i} \mid i \in I, i<p \}$.

I do not understand why the sets $\{q \in I \mid p<q\}$ and $\{q \in I \mid p>q\}$ must be both infinite. At first $I$ thought of using a permutation of the set I to order ascending elements of $\langle a_{i} \mid i \in I, p<i\rangle$ and use the hypothesis that $\langle a_{i} \mid i \in I \rangle$ is a sequence of indiscernible over $A$. But apparently is not the right way.


You don't need either set to be infinite. The full result should be

Let $I$ be linearly ordered, $I = I_0 \cup I_1$ a disjoint union, $I_0 < I_1$. Furthermore, let $M$ be a model, $A \subseteq M$, and $\{b_i\}_{i\in I}$ be tuples from $M$. If $\{b_i\}_{i\in I}$ is indiscernible over $A$, then both

  • $\{b_i\}_{i\in I_0}$ is indiscernible over $A \cup \{b_i\}_{i\in I_1}$ and
  • $\{b_i\}_{i\in I_1}$ is indiscernible over $A \cup \{b_i\}_{i\in I_0}$.

Proof sketch. We need to show that the truth of a formula about $\{b_i\}_{i\in I_0}$ with parameters from $\{b_i\}_{i\in I_1}$ doesn't change when we move the $\{b_i\}_{i\in I_0}$ pieces (preserving order). (The other part is the same argument with the roles of $I_0$ and $I_1$ switched.) This result follows immediately by the indiscernibility of the full sequence, together with the fact that $I_0 < I_1$ means we can't have moved any of the terms "past" the parameters (in $I$'s order).

(Exercise: find an example where the converse fails.)


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.