MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have two questions based on exercises in Kunen's set theory. Let $\kappa = cf(\lambda) > \omega$. Why is there a c.u.b. $C \subseteq \lambda$ of order type $\kappa$? I thought we just choose $C$ as the image of an increasing unbounded function $\kappa \to \lambda$, but I doubt that this has to be closed.

Also, why can we use this $C$ to get an isomorphism of boolean algebras $P(\kappa)/Cub^*(\kappa) \cong P(\lambda)/Cub^*(\lambda)$? If we just pull back with $\kappa \to \lambda$, I don't see why this will be well-defined. Note that $Cub^*$ is the ideal of non-stationary subsets.

The second problem is the following: Let $\kappa$ be an uncountable regular cardinal. I want to prove that there is a decreasing sequence of stationary sets $S_\alpha, \alpha < \kappa$, whose diagonal intersection is $\{0\}$. This is an exercise in Kunen's set theory, and there is a hint that one should use the preceding exercise, which says that the boolean algebra $B=P(\kappa)/Cub^*(\kappa)$ has infima indexed over $\kappa$, which correspond to the diagonal intersection in $P(\kappa)$.

Here's what I've done so far: Construct a decreasing sequence in $B$: Let $x_0=1$. If $x_\alpha$ is already defined, define $x_{\alpha+1} = x_\alpha$ if $x_\alpha$ is minimal and otherwise choose some $x_{\alpha+1} < x_\alpha$. If $\alpha$ is a limit and $x_\gamma$ is defined for all $\gamma < \alpha$, let $x_\alpha$ be the infimum of these $x_\gamma$.

Now if $x_\alpha = [S_\alpha]$, then $S_\alpha$ is stationary iff $x_\alpha \neq 0$; is this the case? For $\alpha < \beta < \kappa$, we have $x_\beta \leq x_\alpha$, i.e. there is a c.u.b. $C_{\alpha,\beta}$ such that $S_\beta \cap C_{\alpha,\beta} \subseteq S_\alpha$. Now perhaps there is some double-index diagonal intersection $C$ of these $C_{\alpha,\beta}$ (?) which is c.u.b. again and which we may intersect with every $S_\alpha$, so that we may assume $S_\beta \subseteq S_\alpha$, as desired.

Finally we have to ensure the infimum of the $x_\alpha$ is $0$. I wonder if this is true at all with this naive construction.

share|cite|improve this question

First question: Take the image of your increasing function from $\kappa$ to $\lambda$ and close it off. This will be of order type $\kappa$.

For the second part of the question, indeed, pullback along the embedding $e$ from $\kappa$ to $\lambda$ (with club range). Why does this give an isomorphism of the quotient algebras? Because the complement of the range of the embedding is in the nonstationary ideal on $\lambda$.

share|cite|improve this answer
But the image of $\kappa \to \lambda$ already has type; why does that not change after taking the closure? – Martin Brandenburg Sep 3 '10 at 14:35
By "close off" I mean add all sups of bounded subsets of the image. It is straight forward that this set is closed and unbounded in $\lambda$. Let $S$ be a proper initial segment of the image. This is of some size $\mu<\kappa$. How many proper initial segments $T$ of your initial segment $S$ are there? Well, $\mu$-many. You add a sup to every such $T$. Hence, below the sup of $S$ you only add $\mu<\kappa$ many new points. Hence, every proper initial subset of the club is of size $<\kappa$. Hence the club is of order type $\kappa$. – Stefan Geschke Sep 3 '10 at 14:45

Second question: You pick a stationary $S \subset \kappa$. Define a sequence recursively like this:

$S_0 := S$, $S_{\alpha +1}:= S_{\alpha} - A_{\alpha}$ where $A_{\alpha}$ is defined as the set of the first $\alpha +1$ many elements of $S_{\alpha}$ (Note here that $S_{\alpha +1}$ remains stationary). And finally $S_{\gamma} := \bigcap_{\beta < \gamma} S_{\beta}$ for each limit $\gamma < \kappa$.

This gives you the desired sequence.

share|cite|improve this answer
You might want to add $0$ to all those sets, note that the intersection is not supposed to be empty but rather a singleton. But then of course you can add to each $S_\alpha$ anything you'd like to keep in the intersection :) – Asaf Karagila Sep 3 '10 at 17:43
I don't understand; $0$ is always in the diagonal intersection. – Martin Brandenburg Sep 3 '10 at 21:23
Yes you're right :). 0 is always in the diagonal intersection. But still 1 must not be an element of $S_0$. – Stefan Hoffelner Sep 3 '10 at 22:12

Reluctantly (not wanting to be more explicit than the book is) I decided to sketch some proof. 1st, get a maximal almost disjoint (modulo non-stationary) family of $\kappa$ stationary subsets of $\kappa$. (Easy to get: take 1st a disjoint partition of $\kappa$ into $\kappa$ stationary sets. – May need to add the transversal set of the minimal ordinals from this family, if this set happens to be stationary). Let $\mathcal{S} = { S_\alpha : \alpha < \kappa }$ be such a family.

Then define $\mathcal{T} = \{ T_\delta : \delta < \kappa \}$ by $T_\delta = \kappa \setminus \bigcup \{S_\alpha : \alpha < \delta \}$, and let $T = \triangle \{ T_\delta : \delta < \kappa \}$.

Since for every $ \delta$ $T \setminus T_\delta \subseteq \delta + 1$, $T$ is almost disjoint from all $S_\alpha$’s. Therefore, by maximality of $\mathcal{S}$, $T$ is non-stationary.

Finally, let $R_\delta = T_\delta \setminus T$. Then $\mathcal{R} = \{R_\delta : \delta < \kappa \}$ is as required, i.e. $\triangle \mathcal{R} =$ {0}.

share|cite|improve this answer
Apparently the OP already knew (from an earlier exercise) that the Boolean algebra $\mathcal P(\kappa)/$club is $\kappa$-complete. So once you have an antichain $\{s_\alpha:\alpha<\kappa\}$ of size $\kappa$, your proof can be formulated succinctly by working entirely in this Boolean algebra. The (Boolean) joins of tails $t_\xi=\bigvee_{\alpha>\xi}s_\alpha$ form a strictly decreasing $\kappa$-sequence. If its infimum (which is known to exist) isn't 0, subtract it from all the $t_\xi$'s. – Andreas Blass Feb 16 '12 at 17:25
Dear Andreas, if you are suggesting that the maximality of $\mathcal{S} $ is irrelevant, I agree! – Isaac Gorelic Feb 16 '12 at 19:36

Just a brief remark that the next to last solution is false.

share|cite|improve this answer
This does not provide an answer to the question. To critique or request clarification from an author, leave a comment below their post - you can always comment on your own posts, and once you have sufficient reputation you will be able to comment on any post. – Myshkin Sep 30 '15 at 11:10
Next to last in which order? Newest? Oldest? Vote count in some order? I will move this to a comment if you give me the author or time stamp. – Todd Trimble Sep 30 '15 at 11:35

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.