I'm currently trying to prove that
The space $S_\Omega$ is sequentially compact.
where $S_\Omega$ is the first uncountable ordinal (i.e. it is well ordered, uncountable and all proper sections are countable).
So far, I've managed to see the following
- By an argument similar to the one done in the reals, sequences in $S_\Omega$ have monotone subsequences: we can say that a point of a sequence is a 'viewpoint' if it is greater than all posterior terms. If there are infinitely many viewpoints, we can construct a decreasing sequence. If on the contrary there are only finitely many, this implies that at a certain point, every term is eventually preceded by a greater one, thus allowing the construction of an increasing sequence.
- If a sequence is decreasing, it converges in $S_\Omega$. In effect, it will converge to $\min_n x_n$.
In the same fashion, if a sequence is increasing and the set $S = \{x : x \geq x_n \ (\forall n)\}$ is non empty, its first element $s$ will be the sequence's limit point. However, it may a priori be that the original sequence is unbounded. My intuition is that this cannot occur precisely because such a scenario in itself would contradict the sequential compactness of $S_\Omega$, which I'm trying to prove. So, any ideas on how to prove, if true, that no sequence is $S_\Omega$ is unbounded?
Edit: as Henno Brandsma has referenced, theorem 10.3 of Munkres proves that $S$ is non empty, thus the sequence has a supremum (to which it converges): sections of each term $S_{x_n}$ are countable so by cardinality,
$$ S_{\Omega} \setminus \bigcup_{n \geq 1}S_{x_n} \neq \emptyset. $$
Your proof idea can be made to work: Let $(x_n)$ be any sequence in $S_\Omega$.
In any linearly ordered set we have a subsequence determined by indices $n_1 < n_2 < n_3 < \ldots$, such that either
In the latter case we cannot have the sequence be strictly decreasing because $S_\Omega$ is well-ordered and the sequence stabilises at some $x$ such that all terms beyond a finite initial segment are equal to $x$. This clearly defines a (constant thus convergent) subsequence.
In the former case $C= \sup\{x_{n_k}: k=0,1,2,\ldots\}$ exists and lies in $S_\Omega$ (thm 10.3 Munkres (2nd ed.)) and by the definition of the order topology one can check that this subsequence converges to $C$.