I am reading Jech's book on Set Theory just now and I had a question about Ramsey ultrafilters.

Jech defines a Ramsey ultrafilter to be: "A non-principal ultrafilter $D$ on $\omega$ is a Ramsey ultrafilter if for every partition $\{A_n | n\in \omega \}$ of $\omega$ into $\aleph_0$ pieces such that $A_n \not\in D$ for all $n$, there exists $X \in D$ such that $X \cap A_n$ has one element for all $n \in \omega$."

Now if $2^{\aleph_0} = \aleph_1$, then we can enumerate all the partitions of $\omega$ by $\mathcal{A}_{\alpha}$ for $\alpha < \omega_1$, and then construct an $\omega_1$-sequence of infinite subsets of $\omega$ as follows: Given $X_{\alpha}$, let $X_{\alpha+1} \subseteq X_{\alpha}$ be such that either $X_{\alpha+1} \subseteq A$ for some $A \in \mathcal{A}_{\alpha}$, or that $|X_{\alpha+1} \cap A| \le 1$ for all $A \in \mathcal{A}_{\alpha}$. If $\alpha$ is a limit ordinal, let $X_{\alpha}$ be such that $X_{\alpha} \setminus X_{\beta}$ is finite for all $\beta < \alpha$.

Now I understand the above, but he defines the set $D = \{ X | X \supseteq X_{\alpha} \text{ for some } \alpha \in \omega_1\}$ and says it is a Ramsey ultrafilter. But I cannot see how it is nonprincipal, an ultrafilter, or Ramsey.

I would appreciate any help with those three points.


Because the $X_\alpha$ are almost decreasing, we have that $D$ is closed under intersections and it is clearly closed under enlargements, and all members are infinite.
We can assume that all finite partitions of $\omega$ were included in our enumeration as well.

To see it is Ramsey (i.e. the partition property part):

Let $\{ A_n \mid n \in \omega \}$ be a partition of $\omega$, and assume no $A_n$ is in $D$. This partition occurs in the enumeration as some $\cal{A}_{\alpha_0}$. If $X_{\alpha_0+1}$ would be a subset of one of the $A_n (\in \cal{A}_{\alpha_0})$, then this $A_n$ would be in $D$ by definition, so this cannot happen. By construction, we have that $X_{\alpha_0 + 1} \in D$ meets every $A_n$ in at most one point, which is what needed to be shown for the partition property.

To see it is an ultrafilter consider for a subset A, the partition $\cal{A}_\alpha = \{ A , \omega\setminus A \}$, and note that the second option cannot hold for $X_{\alpha+1}$, as this makes the set finite, so one of the others hold and makes one of them a member of $D$.

  • $\begingroup$ There I was slaving over an answer, and just as I was having this understand that the construction only chooses infinite sets you posted this answer which is quite cleaner than what I intended to post. :-) $\endgroup$ – Asaf Karagila Feb 1 '12 at 13:04
  • $\begingroup$ Thanks for your response. But why can we assume that we can include finite partitions? $\endgroup$ – Paul Slevin Feb 1 '12 at 14:07
  • $\begingroup$ And how do I know that the sequence is almost decreasing? $\endgroup$ – Paul Slevin Feb 1 '12 at 14:18
  • $\begingroup$ @PaulSlevin this is by construction, isn't it? Almost decreasing means that $A_\alpha \setminus A_\beta$ is finite for all $\alpha > \beta$, and for successors we have just standard decreasing, and for limits we construct a predecessor. And Jech enumerates all partitions, so why not the finite ones too? It helps the proof later, and makes no real difference in the construction... $\endgroup$ – Henno Brandsma Feb 1 '12 at 14:53
  • 1
    $\begingroup$ I looked in Jech, and he doesn't seem to say he just enumerates the infinite partitions, just "partitions". So I think my assumption is justified. I looked into other books that construct Ramsey ultrafilters under CH (or $\frak{t} = 2^{\aleph_0}$, or MA) and the others also take care of the requirement by considering finite partitions (or partitions with just 2 sets) as well, one way or another. $\endgroup$ – Henno Brandsma Feb 4 '12 at 7:24

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.