In Jensen's The fine structure of the constructible hierarchy, it is stated that Solovay proved the consistency of $\neg\square_{\omega_1}$ by collapsing a Mahlo cardinal to $\omega_2$. I was wondering where I can find a proof of this result.
Thanks
-
2$\begingroup$ In case you are not looking for the original proof, there is a paper by Harrington and Shelah (titled "Some exact equiconsistency results in set theory") where it is shown that stationary reflection for $E_{\omega}^{\omega_2}$ is consistent from a Mahlo cardinal. This principle implies the failure of $\square_{\omega_1}$ (see for example Lemma 23.06 in Jechs book). $\endgroup$– Hannes JakobCommented May 20 at 15:59
-
$\begingroup$ @HannesJakob Thanks! No I'm not looking for the original proof, even though it would be nice to have a proof that directly shows the failure of $\square_{\omega_1}$ from collapsing a Mahlo to $\omega_2$ (if there is one, I don't know what Solovay did). $\endgroup$– LorenzoCommented May 20 at 16:05
-
1$\begingroup$ That was Solovay's proof, IIRC. It's pretty direct. $\endgroup$– François G. DoraisCommented May 20 at 17:48
1 Answer
I'll give the proof, which was told to me by Martin Zeman.
Let $G \subseteq \mathrm{Col}(\omega_1,{<}\kappa)$ be generic, where $\kappa$ is Mahlo. Suppose towards contradiction that $\square_{\omega_1}$ holds in $V[G]$, let $\mathcal C$ be a square sequence, and let $\dot{\mathcal C}$ be a name for it. By the $\kappa$-c.c., there is a club of $\alpha<\kappa$ such that $\mathcal C \restriction \alpha$ is computed in $V[G \restriction \alpha]$. By Mahloness, let $\mu<\kappa$ be such a point that is inaccessible in $V$. So $\mu = \omega_2^{V[G \restriction \mu]}$, and $\mathcal C \restriction \mu$ is a square sequence in $V[G \restriction \mu]$.
Now in $V[G]$, $\mathcal C \restriction \mu$ gets threaded by $\mathcal C(\mu)$, and $\mu$ has cofinality $\omega_1$ in $V[G]$. But it is easy to show that there is only one thread, so it is definable whenever it exists. (This seems to me to be a curious example that may go against some philosophical intuitions, a mathematical object that is uniquely definable, but may or may not exist. Please give your thoughts in the comments! ;-P)
Finally, the tail of the forcing, $\mathrm{Col}(\omega_1,{<}\kappa)/ (G \restriction \mu)$, is homogeneous. Since $\mathcal C(\mu)$ is a subset of $\mu$ definable from a parameter from $V[G \restriction \mu]$, general facts about homogeneous forcing show that it is an element of $V[G \restriction \mu]$. (Definable as the set of $\alpha<\mu$ such that $1 \Vdash \alpha \in \dot{\mathcal C}(\mu)$.) But this means $\mu$ is collapsed to $\omega_1$ in $V[G \restriction \mu]$, which is false because $\mathrm{Col}(\omega_1,{<}\mu)$ is $\mu$-c.c.