4
$\begingroup$

What happens to the constructible universe $L$ if we build it in the first of infinitary languages $\mathcal L_{\omega_1, \omega}$?

Would the usual limitation of $L$ not satisfying existence of a measurable cardinal still hold?

$\endgroup$
2

1 Answer 1

7
$\begingroup$

It seems to me that if you build the constructible universe using $\mathcal{L}_{\omega_1,\omega}$ logic, you will get the inner model $L(\mathbb{R})$. The reason is that every $\mathcal{L}_{\omega_1,\omega}$ formula is coded by a real, and so $L(\mathbb{R})$ has all the formulas and can compute truth correctly. (Note that this would not be true for $\mathcal{L}_{\omega_1,\omega_1}$, since $L(\mathbb{R})$ is not necessarily closed under $\omega$-sequences.) So the resulting model is contained within $L(\mathbb{R})$. But since every subset of $\omega$ is definable by an $\mathcal{L}_{\omega_1,\omega}$ formula specifying the members, the resulting model will contain $\mathbb{R}$ and thus contain $L(\mathbb{R})$. So they are equal.

It is indeed possible that $L(\mathbb{R})$ has a measurable cardinal, if the ambient universe has sufficiently strong large cardinals, since if AD holds in $V$, then it holds in $L(\mathbb{R})$, and this implies that $\omega_1$ is measurable, with the club filter being an ultrafilter.

If $L(\mathbb{R})$ satisfies ZFC, however, then it can have no measurable cardinals.

$\endgroup$
8
  • $\begingroup$ In the definition of the $\mathcal{L}_{\omega_1,\omega}$-constructible universe, does the OP not allow formulas involving infinitely many parameters? It seems you're assuming this, is that correct? $\endgroup$
    – Farmer S
    Commented May 3 at 13:19
  • 2
    $\begingroup$ Usually in $\mathcal{L}_{\omega_1,\omega}$ we only allow formulas with finitely many free variables. (The others are considered ill formed.) You are right that this is important, since it would change the answer. $\endgroup$ Commented May 3 at 13:19
  • $\begingroup$ I see. (Adding extra characters to make comment appear...) $\endgroup$
    – Farmer S
    Commented May 3 at 13:21
  • $\begingroup$ I think allowing formulas with infinitely many free variables and then allowing them to be occupied by parameters would result in the Chang model $L([\text{Ord}]^\omega)$, since it would enable one to specify any countable sequence of ordinals. So this would in effect enable the power of $\mathcal{L}_{\omega_1,\omega_1}$. $\endgroup$ Commented May 3 at 13:31
  • $\begingroup$ @FarmerS, Yes! Infinite parameters are not allowed. $\endgroup$ Commented May 3 at 14:41

You must log in to answer this question.

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