8
$\begingroup$

Can we prove the consistency of $\mathsf{ZF+SVC}$ + "There is a Reinhardt cardinal?" (Preferably from the consistency of $\mathsf{ZF}$ with a Reinhardt cardinal, but using a stronger assumption is also okay.)

Here $\mathsf{SVC}$ means the Small Violation of Choice, claiming the axiom of choice is forcible by a set forcing.

Here are some easy observations.

  • Let $\kappa$ be a critical point of $j\colon V\to V$. Then no forcing $\mathbb{P}\in V_\kappa$ can force $\mathsf{AC}$. Similarly, no forcing $\mathbb{P}\in V_{j^\omega(\kappa)}$ can force $\mathsf{AC}$ (as $j^n(\kappa)$ is also a critical point of some elementary embedding $V\to V$.)

  • If there is a super Reinhardt cardinal $\kappa$, then $\mathsf{SVC}$ fails: if there is a $\mathbb{P}$ forcing $\mathsf{AC}$, then we can pull it back to $V_\kappa$ by using $j\colon V\to V$ satisfying $j(\kappa)>\operatorname{rank} \mathbb{P}$.

$\endgroup$
20
  • 1
    $\begingroup$ If this doesn't work, it would be cool to know if adding WISC is relatively consistent over ZF+Reinhardt, since that's a pretty much minimal choice principle that is needed for various things in category theory. $\endgroup$
    – David Roberts
    Commented Dec 6, 2022 at 6:27
  • 1
    $\begingroup$ @Noah It is a name for a workshop we held this September. $\endgroup$
    – Hanul Jeon
    Commented Dec 6, 2022 at 19:17
  • 1
    $\begingroup$ @DavidRoberts I think the answer to your question is yes (starting with even stronger assumptions). For example, a super Reinhardt should outright imply WISC (if I’m not misunderstanding what WISC says). In fact, doesn’t WISC follow from a proper class of Lowenheim-Skolem cardinals? Also I think a Reinhardt implies WISC for wellordered codomains. $\endgroup$ Commented Dec 8, 2022 at 0:01
  • 1
    $\begingroup$ @GabeGoldberg I don't know! There are two ways to state WISC that I know of: "For every set X there is a set of surjections to it such that [etc]", or the version François Dorais figured out and gives in mathoverflow.net/a/99934/4177 It would be very cool if WISC and SVC are separated by large cardinal axioms like this. $\endgroup$
    – David Roberts
    Commented Dec 8, 2022 at 1:11
  • 1
    $\begingroup$ @DavidRoberts Yes, the Dorais version is (close to) the one I had in mind. The proof that WISC holds for wellorderable codomains under a Reinhardt cardinal is what I called the wellordered collection lemma in a paper named something like “Measurable cardinals and choiceless axioms.” Showing that full WISC follows from Lowenheim-Skolem cardinals is actually much easier. (It’s open whether Reinhardt is lower in consistency strength than Reinhardt plus a proper class of Lowenheim-Skolems.) $\endgroup$ Commented Dec 8, 2022 at 1:35

1 Answer 1

9
$\begingroup$

No, a Reinhardt cardinal implies SVC is false.

First, if there is a Reinhardt cardinal, then by Woodin's proof of the Kunen inconsistency theorem, for sufficiently large regular cardinals $\delta$, the set $S^\delta_\omega$ of ordinals of cofinality $\omega$ cannot be partitioned into $\delta$-many disjoint stationary sets.

On the other hand, SVC implies that for all sufficiently large regular cardinals $\delta$, every stationary subset of $\delta$ can be partitioned into $\delta$-many stationary sets: to see this, let $\mathbb P$ be a partial order that forces choice, and let $\delta$ be a regular cardinal such that $|\mathbb P| < \delta$ in $V^{\mathbb P}$. Fix any stationary subset $S$ of $\delta$. In $V^\mathbb P$, $S$ remains stationary (by a standard argument, a club in the extension contains a club in the ground model) and there is a partition of $S$ into stationary sets $\langle S_\alpha\rangle_{\alpha < \delta}$ by Solovay's theorem. For each $p\in \mathbb P$, let $S^p_\alpha = \{\xi < \delta : p\Vdash \xi\in S_\alpha\}$ (fixing a name, etc, etc). We claim there is some $p\in \mathbb P$ such that $A_p = \{\alpha < \delta : S^p_\alpha\text{ is stationary}\}$ has cardinality $\delta$. If not, $|\bigcup_{p\in \mathbb P} A_p| < \delta$, so there is some $\alpha\in \delta\setminus \bigcup_{p\in \mathbb P} A_p$. But then in $V^{\mathbb P}$, $S_\alpha \subseteq \bigcup_{p\in \mathbb P} S^p_\alpha$ is a stationary set which is contained in the union of fewer than $\delta$ nonstationary sets, which is a contradiction. So fix $p$ such that $|A_p| = \delta$. Then in $V$, we have partitioned $S$ into $\delta$-many stationary sets of the form $S_\alpha^p$.

$\endgroup$
2
  • $\begingroup$ That's a cool argument. It would have been helpful to have you around in cheese. $\endgroup$
    – Asaf Karagila
    Commented Dec 6, 2022 at 20:24
  • $\begingroup$ Wish I could have made it! $\endgroup$ Commented Dec 7, 2022 at 3:38

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .