5
$\begingroup$

how to prove this result.

$\alpha$ is limit ordinal if and only if $\beta<\alpha$ implies $\beta+1<\alpha$ for any $\beta$.

I am getting confused because of the definition of limit ordinal. it says

$\alpha=\sup\{\beta: \beta<\alpha\}=\bigcup\alpha$

for sup I am ok but for $\alpha=\bigcup\alpha$ how ??

$\endgroup$
4
$\begingroup$

To show $\alpha = \bigcup \alpha$ implies ($\beta < \alpha$ implies $\beta + 1 < \alpha$) assume that $\alpha = \bigcup \alpha$ and not ($\beta < \alpha$ implies $\beta + 1 < \alpha$):

If $\beta < \alpha$ and $\beta + 1 \ge \alpha$ then $\beta < \alpha \le \beta + 1$. Since there are no ordinals between $\beta$ and $\beta + 1$ it follows that $\alpha = \beta + 1$. But then $\alpha = \bigcup \alpha = \bigcup \beta + 1 = \beta < \alpha$ which is a contradiction.

For the other direction, assume ($\beta < \alpha$ implies $\beta + 1 < \alpha$). The inequality $\bigcup \alpha \le \alpha$ always holds. To show $\alpha \subseteq \bigcup \alpha$ let $\beta \in \alpha$. Then by assumption $\beta + 1 \in \alpha$. Since $\beta \in \beta + 1$ it follows that $\beta \in \bigcup \alpha$.

$\endgroup$
7
  • $\begingroup$ Note how you can exchange $\le$ and $\subseteq$ and $\in$ and $<$ because the sets involved are ordinals. I hope I understood your question correctly. $\endgroup$ Apr 6 '13 at 15:21
  • $\begingroup$ It is ok that's what I want, but the last statement is not clear " $\beta\in\bigcup\alpha$ $\endgroup$ Apr 6 '13 at 15:27
  • $\begingroup$ @ZirkoTammado The last statement is a consequence of $\beta + 1 \in \alpha$. Then $\bigcup \alpha = \beta + 1 \cup \dots$ and since $\beta \in \beta + 1$ then also $\beta \in \bigcup \alpha$. $\endgroup$ Apr 6 '13 at 15:30
  • $\begingroup$ I know what does this mean but we want to show that $\beta+1\in\bigcup\alpha=\alpha$ $\endgroup$ Apr 6 '13 at 15:35
  • $\begingroup$ @ZirkoTammado Apply the assumption to $\beta + 2$ to get $\beta + 2 \in \alpha$. $\endgroup$ Apr 6 '13 at 15:37
3
$\begingroup$

Recall that ordinals are sets. $\bigcup\alpha$ is the union of all the members of $\alpha$. That is, $x\in\bigcup\alpha$ if and only if there is some $\beta\in\alpha$ such that $x\in\beta$.

Recall that ordinals are transitive sets. It turns out that if $A$ is a set of ordinals then $\sup A=\bigcup A$.

Recall that ordinals are sets of ordinals, and every ordinal is the set of those ordinals strictly smaller than itself. Therefore $\sup\{\beta:\beta<\alpha\}=\bigcup\alpha$.


Here is an outline of the equivalence,

If $\alpha$ is not a limit ordinal then $\alpha=\beta+1$ for some $\beta$, therefore $\sup\{\gamma:\gamma<\alpha\}=\beta$, because $\beta$ is in fact the maximum of that set. Therefore if $\alpha$ is not a limit ordinal, $\alpha\neq\sup\{\beta:\beta<\alpha\}$.

If there is $\beta<\alpha$ such that $\beta+1\nless\alpha$, then by the fact ordinals are linearly ordered $\alpha\leq\beta+1$, and therefore $\alpha=\beta+1$. It follows that $\{\beta:\beta<\alpha\}$ has a maximal element, $\beta$.

(Some minor details and small arguments might need to be added, but the idea is there.)

$\endgroup$
4
  • $\begingroup$ Excellent Asaf for this nice explanation. Could you please give me a proof of the above result as well. $\endgroup$ Apr 6 '13 at 15:17
  • $\begingroup$ @Zirko: Which one? $\sup A=\bigcup A$? $\endgroup$
    – Asaf Karagila
    Apr 6 '13 at 15:18
  • $\begingroup$ $\alpha$ is limit ordinal if and only if $\beta<\alpha$ implies $\beta+1<\alpha$ for any $\beta$. $\endgroup$ Apr 6 '13 at 15:20
  • $\begingroup$ Thanks alot I will try to catch it. You are always the best. $\endgroup$ Apr 6 '13 at 15:29
1
$\begingroup$

For each ordinal $\alpha$, the set $\beta:=\bigcup \alpha$ is also an ordinal. As it does not contain $\alpha$ itself it is either $=\alpha$ or $<\alpha$. The case $\beta<\alpha$ occurs iff $\alpha=\beta+1$, why?

$\endgroup$

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.