4
$\begingroup$

Is the following claim correct (Chapter 13 before Theorem 87 of Todorcevic's book: Notes on forcing axioms): Let $\alpha$ be an infinite countable indecomposable ordinal and $U$ be an uniform ultrafilter on $\alpha$ (namely elements in $U$ have order type $\alpha$). Then for any collection $\{B_i\in U: i<\mathfrak{m}\}$ it is true that there exists $B\subset \alpha$ of order type $\alpha$ such that $B-B_i$ is finite for all $i<\mathfrak{m}$. Here $\mathfrak{m}$ is the least cardinal such that Martin's Axiom holds below $\mathfrak{m}$ (i.e. meeting any $\beta$ many dense sets for any $\beta<\mathfrak{m}$).

In fact, it will be interesting to know if this is true at all for countable collection $\{B_i: i<\omega\}$.

It is proved there that the statement is true if we relax the conclusion asking only $B-B_i$ to be bounded in $B$. Any thoughts?

$\endgroup$
2
  • $\begingroup$ @NotMike: the mod finite part is a problem $\endgroup$
    – Jing Zhang
    Jul 14, 2018 at 23:41
  • $\begingroup$ @JingZhang Ah, I see now. $\endgroup$
    – Not Mike
    Jul 15, 2018 at 20:01

1 Answer 1

3
$\begingroup$

I believe the claim is wrong:

If the claim is right I claim I can show $\alpha\to (\alpha)^2_2$ which is obviously wrong for countable ordinal $\alpha\geq \omega+2$.

Given a coloring $f: [\alpha]^2\to 2$, for each $\beta\in \alpha$, let $A_\beta^i=\{\gamma<\alpha: f(\beta,\gamma)=i\}$ for $i<2$. Let $g: \alpha\to 2$ be such that $A_{\beta}^{g(\beta)}\in U$. Find $i<2$ and $B'\in U$ such that for each $\beta\in B'$, $g(\beta)=i$. $\{A_\beta^{i}: \beta\in B'\}$ is a countable collection of elements in $U$, hence there exists a pseudointersection $B\subset B'$ such that $B$ has order type $\alpha$ and $B-A_{\beta}^{i}$ is finite for any $\beta \in B'$.

Now decompose $B$ into $\alpha$ increasing intervals each of which has order type $\omega$, say $\{I_\xi: \xi<\alpha\}$. We inductively in $\omega$-stages pick a monochromatic subset of order type $\alpha$ such that $I_\xi$ contributes exactly one element. Fix a bijection $h: \omega\to \alpha$. Suppose $\alpha_j$ have been picked for $j<k$. We will pick $\alpha_k$ in $I_{h(k)}$. Since $\bigcup_{j<k} B-A_{\alpha_j}^i$ is finite, there exists $\alpha_{k+1}\in I_{h(k)}\cap \bigcap_{j<k} A_{\alpha_j}^i$. Then eventually we will define a monochromatic subset of order type $\alpha$ (with constant color $i$).

$\endgroup$

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.