9
$\begingroup$

Let $\lambda$ be a singular cardinal of countable cofinality.

Is there necessarily a sequence $\{A_\alpha\mid\alpha<\lambda^+\}$ of countable subsets of $\lambda$, such that $\alpha<\beta$ if and only if $A_\alpha\setminus A_\beta$ is finite?

In other words, we know that for $\omega$, there is an uncountable sequence of subsets which is increasing modulo finite changes. Is the same true for other cardinals of countable cofinality?

$\endgroup$
7
  • $\begingroup$ I've put tower in quotation marks since I would expect towers to be in $\mathcal P(\lambda)$ and modulo the bounded ideal, rather than countable subsets modulo finite. I've also tentatively put the pcf tag since I feel it might have something to do with the answer. $\endgroup$
    – Asaf Karagila
    Commented Jan 17, 2018 at 11:22
  • $\begingroup$ Can you please give an increasing modulo finite changes uncountable sequence of subsets of $\omega$. Thanks. $\endgroup$
    – Joel Adler
    Commented Aug 14, 2018 at 9:43
  • $\begingroup$ @Joel: No, since it requires the axiom of choice. But even without choice you can prove that a sequence of coinfinite subsets of $\omega$ has a coinfinite set almost containing all of them. Therefore by Zorn's Lemma there is an uncountable tower. $\endgroup$
    – Asaf Karagila
    Commented Aug 14, 2018 at 9:57
  • $\begingroup$ @AsafKaragila Assuming that by co-infinite you mean "complement is infinite". Let $X_n$ be the set of all odd numbers for $n$ odd, and let $X_n$ be the set of all even numbers for $n$ even. Then there is no co-infinite set containing almost all of the $X_n$, right? Or did you mean by "... you can prove that a sequence...": "you can prove that there is a sequence of co-infinite sets such that..."? $\endgroup$ Commented Aug 16, 2018 at 10:02
  • $\begingroup$ @Dominic: I meant an increasing sequence. Not just any sequence. And in the case where you talk about partitions, then you want to talk about something almost disjoint from all the parts. $\endgroup$
    – Asaf Karagila
    Commented Aug 16, 2018 at 10:08

1 Answer 1

12
$\begingroup$

For $\lambda > 2^{\aleph_0}$, there is no such sequence.

Suppose $\lambda > 2^{\aleph_0}$. Because $2^{\aleph_0}$ cannot have countable cofinality, there is some $\kappa < \lambda$ with $2^{\aleph_0} < \kappa$. Consider the sequence $\{A_\alpha \cap A_\kappa \mid \alpha < \kappa\}$. For each particular $\alpha$, the sets $A_\alpha$ and $A_\alpha \cap A_\kappa$ differ by only finitely many elements. Each set of this form is a countable subset of the countable set $A_\kappa$, so there are fewer than $\kappa$ possibilities for the sets $A_\alpha \cap A_\kappa$. Thus we may find $\alpha < \beta < \kappa$ such that $A_\alpha \cap A_\kappa = A_\beta \cap A_\kappa$. But then $A_\alpha$ and $A_\beta$ differ by only finitely many elements, and in particular we have $A_\beta \setminus A_\alpha$ finite while $\alpha < \beta$, contrary to your condition.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .