21
$\begingroup$

I have recently been wondering if the following is consistent with ZFC:

For every infinite ordinal $\alpha$: $|V_\alpha\cap L|=|\alpha|$.

Intuitively, this states that for $L$ is very "thin", in that it doesn't branch off too much from the set of ordinals as we climb cumulative hierarchy.

I have heard that it is possible to have it for $V_{\omega+1}$, which is essentially having $\Bbb R^L$ countable. I suspect that this is consistent, and we can construct a model of this starting from $V=L$ and adding a proper class of generic sets giving bijections between parts of $L$ and ordinals, but I have no real background in forcing to see if that makes any sense.

Thanks in advance.

$\endgroup$

2 Answers 2

29
$\begingroup$

Claim: $|V_\alpha \cap L| = |\alpha|$ for every $\alpha \geq \omega$ implies that $0^\#$ exists.

Proof: Let's assume, toward contradiction, that $0^\#$ doesn't exist that $|V_\alpha \cap L| = |\alpha|$ for every $\alpha \geq \omega$.

Let $\mu$ be a singular cardinal in $V$. Let's apply the assumption $|V_\alpha \cap L| = |\alpha|$ for $\alpha = \mu + 1$. In $V$, $\mu = |\mu + 1|=|V_{\mu + 1} \cap L| \geq |\mathcal{P}^L(\mu)| \geq |(\mu^{+})^L|$.

In particular, $\mu^{+} > (\mu^{+})^L$. Therefore, the cofinality of $(\mu^{+})^L$ in $V$ is strictly below $\mu$. Applying Jensen's covering theorem, we conclude that the cofinality of $(\mu^{+})^L$ in $L$ is strictly below $\mu$ - a contradiction to the regularity of $(\mu^{+})^L$ in $L$.

By the way, if we restrict the values of $\alpha$ for which we want $|V_\alpha \cap L| = |\alpha|$ to be $\leq \aleph_{\omega}$, the consistency strength drops considerably:

Claim: $|V_\alpha \cap L| = |\alpha|$ for every $\omega \leq \alpha \leq \aleph_\omega$ is equiconsistent with the existence of $\omega$ inaccessible cardinals.

Proof: For the first direction, force over $L$ with an iteration of Levi collapses in order to make the $n$-th inaccessible in $L$ equal to the $\aleph_n$ of the generic extension. For the second direction, note that the assumption implies that for all $1\leq n < \omega$, $\aleph_n$ is a $\beth$-fixed point in $L$, since for every infinite $\beta < \aleph_n$, $|V_\beta \cap L| = |\beth_\beta^L| < \aleph_n$. In particular, $\aleph_n$ is a limit cardinal in $L$. It is regular in $V$ and thus also in $L$ and therefore it is inaccessible.

$\endgroup$
1
  • $\begingroup$ That a very straightforward argument, I also like the result about limiting the question to $\alpha\leq\aleph_\omega$. $\endgroup$
    – Wojowu
    Oct 7, 2015 at 11:29
23
$\begingroup$

This situation is a consequence of the existence of $0^\sharp$.

To see this, suppose that $0^\sharp$ exists. It follows that every uncountable cardinal is a Silver indiscernible and a limit of Silver indiscernibles. If $\alpha$ is any infinite ordinal, then $\beth_{\omega+\alpha}^L$, which is definable from $\alpha$, must appear before the next indiscernible (since they are closed under definability in $L$), which is before $(|\alpha|^+)^V$. Since $V_\alpha\cap L=(V_\alpha)^L$ has size $\beth_{\omega+\alpha}^L$, it follows that this will have size strictly less than $|\alpha|^+$ and hence size $|\alpha|$, as desired.

$\endgroup$
3
  • 10
    $\begingroup$ It is equivalent to $0^\#$: Take $\alpha = \mu + 1$ where $\mu = \aleph_{\omega}^V$ one can see that $|P(\mu )^L| = |{\mu^{+}}^L| = \mu$ so $V$ doesn't agree with $L$ on the successor to this singular and thus $0^\#$ exists. $\endgroup$
    – Yair Hayut
    Oct 6, 2015 at 20:11
  • $\begingroup$ @Yair: I was going to make that into an answer, but then I saw your comment. So you should make that into an answer. $\endgroup$
    – Asaf Karagila
    Oct 6, 2015 at 23:01
  • $\begingroup$ I also was writing up that as an extension to my answer when you commented. Please go ahead and post as an answer! $\endgroup$ Oct 6, 2015 at 23:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.