It's a problem from the book "Topology of Metric Spaces", written by Kumaresan:

"Show that the set $\textbf{c}$ of convergent sequences in the Normed Linear Space of all bounded real sequences under the sup norm $\|\|_\infty$ is complete. Hint: Enough to show that $\textbf{c}$ is closed. If x = $(x_n)$ is a limit point of $\textbf{c}$, it suffices to show that $x$ is Cauchy."

I got a little bit confused here... What I tried to do:

I know that to show that a metric space is complete, I have to show that every Cauchy sequence is convergent there.

About the hint: I don't understand why it is enought to show that $\textbf{c}$ is closed, and what does it mean to say that $(x_n)$ is a limit point of $\textbf{c}$? Does that mean that there is a squence $(y_k) \in \textbf{c}$ such that $(y_k) \to (x_n)$?

If $(y_k) \to (x_n)$, given $\epsilon > 0$, there are $n_0, k_0 \in \mathbb{N}$ such that $$\|y_k-x_n\|_\infty<\epsilon \forall k,n \geq \max\{n_0, k_0\}=m_0$$

As we have that $(y_n) \in \textbf{c}$, we have that $(y_n) \to y \in \mathbb{R}$. Because of the uniqueness of the limit, $(x_n) \to y$, so $(x_n) \in \textbf{c}$, therefore, $\textbf{c}$ is closed.

Now, let $(y_{n_k}) \in \textbf{c}$ be a Cauchy sequence. Well, $(y_{n_k})=(y_{n_1},y_{n_2},\dots,y_{n_j},\dots)$ is a sequence of sequences. Knowing that $(y_{n_k})$ is Cauchy's, we know that, given $\epsilon > 0, \exists j_0 \in \mathbb{N}$ such that $\forall m, p \geq j_0$

$$\|y_{n_m}-y_{n_p}\|_\infty < \epsilon$$

I don't know what to do from here. My idea is to show that, since every term $y_{n_j}$, of the sequence $(y_{n_k})$, belongs to $\textbf{c}$, therefore, every term $y_{n_j}$ is convergent, we have that $(y_{n_k})$ converges for a convergent sequence $(y_n)$. So, since $(y_n)$ is convergent, $(y_n) \in \textbf{c}$, we have that every Cauchy sequence in $\textbf{c}$ is convergent, meaning that $\textbf{c}$ is complete.

But I don't know how to do that, and I don't know if I understood the problem correctly, so I don't know if what I did before is correct...

I'd be really grateful if someone could give a hint, a solution, a coment about anything I wrote here! :)

  • 1
    $\begingroup$ If it is closed, then this means it contains all of its limit points, but then if $x=(x_n)_{n\in \mathbb{N}}$ is Cauchy then it converges (see this -u.arizona.edu/~mwalker/econ519/Econ519LectureNotes/… - for verification that Cauchy implies convergent in your case). But that's exactly what we need to do... show it converges because that would show $any$ limit point is convergent (because any limit point is Cauchy). So just show any limit point is Cauchy! Does that help? Any questions? $\endgroup$ – Squirtle Jun 30 '14 at 2:59
  • 1
    $\begingroup$ Also, look up the definition of a limit point because your definition isn't quite right but its worth noting that a limit point doesn't even need to be part of a set. $\endgroup$ – Squirtle Jun 30 '14 at 3:18
  • 1
    $\begingroup$ @Carol: Notice that the points in c are themselves sequences, and then you have a sequence of sequences that (if it converges, it)converges to a sequence. Notice also that a closed subset of a complete space is complete, and that a Cauchy sequence of sequences is one where $|| (x_n)_m-(y_n)_k ||_{\infty} < \epsilon $ , for $m,k >j$ for some positive integer $j$. $\endgroup$ – user99680 Jun 30 '14 at 3:23
  • $\begingroup$ @Squirtle I read the link you sent, but I still don't understand why Cauchy implies convergent in this case... However, I think I understood why it's enough to show that $\textbf{c}$ is closed. $\textbf{c}$ is a subset of the set A of all bounded sequences. A is complete, right? So, if $\textbf{c}$ is closed, it means $\textbf{c}$ is complete. Is that correct? $\endgroup$ – Anna Jun 30 '14 at 4:28
  • 1
    $\begingroup$ Oh I'm sorry, I feel absolutely terrible, I had a dyslexia moment the link I sent showed every convergent sequence is Cauchy. Still it's not horribly difficult to prove the converse. In response to your comment about what you call $A$, please read the following post: (is the set of all bounded sequences complete) $\endgroup$ – Squirtle Jun 30 '14 at 4:45

1)First notice that a closed subset of a complete space is closed. Another way of understanding closed sets is that a closed set contains all its limit points. Equivalently, a subset $S$ is closed, if every convergent sequence in $S$ has its limit in $S$.

2) An element $(x_n)$ is a limit point of a set $S$, if every neighborhood of $(x_n)$ intersects $S$. It is not always the case that if $(x_n)$ is a limit point, that there exists a sequence converging to $(x_n)$; this is true for 1st-countable spaces (which includes metric spaces).

3)Now, try to show c is closed by showing that if $(x_n)$ is a limit point of c, then $(x_n)$ is in $S$, or that, if you have a convergent sequence in c, then its limit is also in c. The triangle inequality should help you there.

A general observation is that you can show the subspace $c$ is closed, by showing that its complement is open: consider a point $(x_n)$ in $\mathbb R^{\mathbb R} -c$ , i.e., a sequence of Reals that does not converge . Then show that there is an $\epsilon>0$ , so that the ball $B((x_n), \epsilon)$ (ball given in the $||.||_{\infty}$ metric ), so that no sequence in that ball converges. This should not be too hard; if a sequence $(x_n)$does not converge and $Sup|x_n-y_n| < \epsilon$ , then I think it is not too hard to show that $(y_n)$ does not converge either.

| cite | improve this answer | |
  • $\begingroup$ When you say "if you have a convergent sequence in c, then its limit is also in c", what you mean? The sequences in c are real sequences, so they converge for some real number... But real numbers are not in c, right? Because c is a set of sequences... hehe sorry, I'm confused $\endgroup$ – Anna Jun 30 '14 at 4:32
  • $\begingroup$ @Carol: Please use the @ to write a comment, to make sure I get the pingback. I mean that the limit of a Cauchy sequence in c would also be a convergent sequence of Real numbers. So you have a Cauchy sequence of convergent sequence, and its limit should be a convergent sequence ( in the $||.||_{\infty}$ norm) itself. $\endgroup$ – user99680 Jun 30 '14 at 5:02
  • $\begingroup$ Sorry! I thought that since my coment was in your answer, you'd be notified. About your answer, thank you, I understood it now. $\endgroup$ – Anna Jun 30 '14 at 13:24
  • $\begingroup$ I don't understand oO I write "@user99680" in the beginning of the coment, but when I send, it desappears, and I tried more than once! $\endgroup$ – Anna Jun 30 '14 at 13:26
  • $\begingroup$ @Carol: No problem, enjoy the futbol cup. $\endgroup$ – user99680 Jun 30 '14 at 16:43

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.