I am trying to solve the following two problems:

1) if $\beta <\omega_1$, $2^{\aleph_1}<\aleph_{\omega_1}$, and $\aleph_\alpha^{\aleph_0} \leq \aleph_{\alpha +\beta}$ for a stationary set of $\alpha$'s, then $\aleph_{\omega_1}^{\aleph_1} \leq \aleph_{\omega_1+\beta}$

2) if $2^{\aleph_{\omega_1+\alpha}} < \aleph_{\omega_1+\alpha+\alpha}$ for all $\alpha <\omega_1$, then $2^{\aleph_{\omega_1 + \omega_1}} < \aleph_{\omega_1 +\omega_1 +\omega_1}$.

I was trying to prove 2) by modifying Galvin-Hajnal theorem, but I get stuck at a certain point. As far as 1) is concerned, I suppose one has to go by contradiction, but I don't see how to get one.


The first proof appear in Introduction to Cardinal Arithmetic by Holz, Steffens, Weitz. The first question appears as a corollary to the Galvin-Hajnal theorem.

The first question is Corollary 2.3.7, appearing on page 124 in the edition open in front of me. While the second does not appear in this form I deduced it from similar lemmas/exercises appearing in the same chapter as well common cardinal arithmetic (appearing in the first chapter of the book)

I have given the proof in a relatively self-contained form here:

First question: Suppose $2^{\aleph_1}<\aleph_{\omega_1}$, and $\beta<\omega_1$ such that the set $\{\alpha<\omega_1\mid\aleph_\alpha^{\aleph_0}\le\aleph_{\alpha+\beta}\}$ is stationary in $\omega_1$ then $\aleph_{\omega_1}^{\aleph_1}\le\aleph_{\omega_1+\beta}$.

By the assumption $2^{\aleph_1}=\aleph_\xi<\aleph_{\omega_1}$ for some $\xi<\omega_1$, therefore for $\alpha>\xi\cdot\aleph_0=\xi'$ we have $\aleph_\alpha^{\aleph_1}=2^{\aleph_1}\cdot\aleph_\alpha^{\aleph_0} = \aleph_\alpha^{\aleph_0}\le\aleph_{\alpha+\beta}$.

This means that both $\aleph_{\omega_1}$ is closed under $\aleph_1$ powers, as well the following equality: $$\{\xi'<\alpha<\omega_1\mid\aleph_\alpha^{\aleph_0}\le\aleph_{\alpha+\beta}\} = \{\xi'<\alpha<\omega_1\mid\aleph_\alpha^{\aleph_1}\le\aleph_{\alpha+\beta}\}$$

We can, if so, prove the same fact about the RHS of this equality. We do so by defining the function $\Phi(\alpha)$ to be such that $\aleph_\alpha^{\aleph_1}=\aleph_{\alpha+\Phi(\alpha)}$.

Since we have that $\aleph_{\omega_1}$ is closed under powers of $\aleph_1$ this is a function from $\omega_1$ to $\omega_1$, and by the assumption we have that $\{\alpha<\omega_1\mid\Phi(\alpha)\le\beta\}$ is stationary, therefore the rank of $\Phi$ is at most $\beta$.

We have the needed conditions for the Galvin-Hajnal lemma, taking the normal sequence simply as $\langle \alpha\mid\alpha<\omega_1\rangle$, we have that $\aleph_\alpha^{\aleph_1} = \aleph_{\alpha+\Phi(\alpha)}$ and therefore $\aleph_{\omega_1}^{\aleph_1} \le \aleph_{\omega_1+\beta}$.

Second question: If $2^{\aleph_{\omega_1+\alpha}}<\aleph_{\omega_1+\alpha+\alpha}$ for all $\alpha<\omega_1$, then $2^{\aleph_{\omega_1+\omega_1}}<\aleph_{\omega_1+\omega_1+\omega_1}$. (The assumption cannot be true if $\alpha<2$. I will assume that as well, and will conveniently ignore these cases when I say "for all $\alpha$")

By the assumption that if $\alpha<\omega_1$ then $2^{\aleph_{\omega_1+\alpha}}<\aleph_{\omega_1+\alpha+\alpha}<\aleph_{\omega_1+\omega_1}$ we have that latter is a strong limit with cofinality $\aleph_1$, and therefore $\aleph_{\omega_1+\omega_1}^{\aleph_1} = 2^{\aleph_{\omega_1+\omega_1}}$.

We can, if so, reduce back to the first question, apply the same argument with $\Phi(\alpha)$ defined to be such that $\aleph_{\omega_1+\alpha}^{\aleph_1}=\aleph_{\omega_1+\alpha+\Phi(\alpha)}$.

Since we have $\aleph_{\omega_1+\alpha}^{\aleph_1}\le 2^{\aleph_{\omega_1+\alpha}} < \aleph_{\omega_1+\alpha+\alpha}$ we have that $\Phi(\alpha)<\alpha$. for all $\alpha<\omega_1$. Therefore the rank of $\Phi$ over the non-stationary ideal of $\omega_1$ is some $\rho\le\omega_1$.

As before we have by the Galvin-Hajnal lemma that $\aleph_{\omega_1+\omega_1}^{\aleph_1} \le \aleph_{\omega_1+\omega_1+\rho}$.

If $\rho<\omega_1$ we are done.

Otherwise, we have $2^{\aleph_{\omega_1+\omega_1}}= \aleph_{\omega_1+\omega_1}^{\aleph_1} \le \aleph_{\omega_1+\omega_1+\omega_1}$.

This cannot be an equality since by Konig's lemma we have that $\operatorname{cf}(2^\kappa)>\operatorname{cf}\kappa$, and both $\aleph_{\omega_1+\omega_1}$ and $\aleph_{\omega_1+\omega_1+\omega_1}$ are of the same cofinality.

We have if so that $2^{\aleph_{\omega_1+\omega_1}}<\aleph_{\omega_1+\omega_1+\omega_1}$ as needed.

| cite | improve this answer | |
  • $\begingroup$ I apologize in advance for any repeated editing, after two and a half hours of trying to solve this question directly into the answer box there are bound to be typos and extraneous statements. $\endgroup$ – Asaf Karagila May 23 '11 at 21:38

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.