Let $(G_\alpha)$ and $(K_\alpha)$ $(\alpha<\aleph_1)$ be strictly increasing chains of countable sets such that if $\alpha$ is a limit, then $G_\alpha=\bigcup_{\beta<\alpha}G_\beta$ and $K_\alpha=\bigcup_{\beta<\alpha}K_\beta$. Assume $\bigcup_{\alpha<\aleph_1}G_\alpha=\bigcup_{\alpha<\aleph_1}K_\alpha$. Does there exist a club $C$ such that $K_\gamma=G_\gamma$ for all $\gamma\in C$?

In the paper ``The Abelianization of Almost Free Groups" (the end of the proof of Lemma 2.5 on page 1801), this assertion is made where the $G_\alpha$ and $K_\alpha$ are subgroups of a free group with some additional properties, but the author makes the assertion I want without any explanation, so I assume the reason must be simple and may not depend on the group theory.

  • 4
    $\begingroup$ From wikipedia: a club set is a subset of a limit ordinal which is closed under the order topology, and is unbounded (see below) relative to the limit ordinal. The name club is a contraction of "closed and unbounded". $\endgroup$
    – YCor
    Jul 6, 2018 at 7:59

1 Answer 1


The set $C$ of $\gamma$ such that $K_\gamma = G_\gamma$ is closed, because if $\alpha$ is a limit point of $C$, then $$G_\alpha = \bigcup_{\beta < \alpha } G_\beta = \bigcup_{\beta <\alpha, \beta \in C} G_{\beta} = \bigcup_{\beta <\alpha, \beta \in C} K_{\beta} = \bigcup_{\beta<\alpha} K_\beta= K_\alpha.$$

So it suffices to show that $C$ is unbounded. Given any $\alpha_1$, because $K_{\alpha_1}$ is countable and is contained in $\bigcup_{\alpha<\aleph_1} G_\alpha$, we have $K_{\alpha_1} \subseteq G_{\alpha_2}$ for some $\alpha_2$. Similarly we have $G_{\alpha_2}\subseteq K_{\alpha_3} \subseteq G_{\alpha_4} \subseteq \dots$ so $$G_{\lim_{n \to \infty} }\alpha_n = K_{\lim_{n \to \infty} } \alpha_n$$ and thus $C$ contains an element greater than $\alpha_1$. So $C$ is unbounded.

  • 2
    $\begingroup$ @Tri Let me add that this hasn't anything to do with groups (as you expected and Will's proof demonstrates). Any two increasing, continuous sequences (cumulative hierarchies) with a length whose cofinality is strictly bigger than the cardinality of any of their components that have the same union exhibit this property. $\endgroup$ Jul 6, 2018 at 10:33

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.