I'm looking at lemma IV.$3.10$ from Kunen's Set Theory ($M$ denotes a model of $\sf{ZFC}$:

If $\mathbb{P}\in M$ and $\tau,\mu\in M^\mathbb{P}$ (the set of names in $M$), then there is a nice name $\vartheta\in M^\mathbb{P}$ for a subset of $\tau$ such that $\mathbb{1}\Vdash (\mu\subset\tau\rightarrow\mu=\vartheta)$.

The proof starts by saying:

Work entirely in $M$, let $\vartheta=\bigcup\{\{\sigma\}\times A_\sigma:\sigma\in\text{dom}(\tau)\}$, where each $A_\sigma$ is chosen so that:

  1. $A_\sigma$ is an antichain in $\mathbb{P}$.
  2. For all $p\in A_\sigma$, $p\Vdash \sigma\in\mu$.
  3. $A_\sigma$ is maximal with respect to (1)(2).

This can be done in $M$ by using $\sf{AC}$ in $M$ and the Definability Lemma.

I'm lost as to how exactly he's forming the $A_\sigma$. I can see that, by the Definability Lemma, we can form the set $\{p\in M:p\Vdash \sigma\in\mu\}$, and this set must belong to $M$, but how to we get an antichain?

I'm guessing condition $(3)$ follows by an application of Zorn's lemma (which is valid since $\sf{AC}$ holds in $M$), although I haven't sat down to check the details since condition $(1)$ seems much more mysterious. Is this the correct path, assuming we can form an $A_\sigma$ satisfying $(1)$ and $(2)$?


Let $B_{\sigma} = \{p \in M : p \Vdash \sigma \in \mu\}$. Select $x_0 \in B_{\sigma}$. For each $\alpha > 0$, let $x_{\alpha}$ be any element of $B_{\sigma}$ that is incomparable with every $x_{\beta}$ for $\beta < \alpha$, if it exists. At some point, we cannot choose another $x_{\alpha}$; for this $\alpha$, the class $\{x_i \mid i < \alpha\}$ is a maximal antichain. By the Axiom of Choice, this class is a set (we need Choice to pick each new $x_i$, of course).

Alternatively: antichains exist because a single-element set is an antichain - pick any single element from $B_{\sigma}$. As you noted, Zorn's Lemma implies the existence of maximal antichains, provided antichains exist at all - so starting from a single-element antichain, a maximal antichain exists.

  • $\begingroup$ Ah, got it know, thank you very much! $\endgroup$ – Reveillark Feb 11 '17 at 3:52

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.