The problem I'm trying is to prove whether or not the metric space of real-valued sequences $(x_n)$ such that $x_n=0$ for all but finitely many values of $n$, with the sup metric: $d((x_n),(y_n)) = max |x_n-y_n|$ is complete.

Now, I tried to find counterexamples but did not seem to find any...

Hence I'm trying to prove that if a sequence $((x_n)^{(k)}) : (x_n)^{(1)}, (x_n)^{(2)}, ... $ is Cauchy, then it converges to some sequence.

We can easily show how the sequence of the $k$th entries: $(x_k^{(n)})$ for a fixed $k$ is Cauchy in $\mathbb{R}$ and hence convergent. Moreover for all but finitely many values of $k$ we can find a subsequence of this series which is constantly zero. So for all but finitely many values of $k$, $(x_k^{(n)})\rightarrow 0$ by properties of real Cauchy sequences.

For finitely many values $k_1,k_2,...$, $(x_{k_{i}}^{(n)})\rightarrow \alpha_i$ where the limit could be nonzero.

This shows that $((x_n)^{(k)})\rightarrow (\alpha_n)$ where this limit sequence is still an element of the space we are considering, hence this space is complete.

This is my proof, but I feel the part in which I claim the existence of a constantly zero subsequence needs a little bit more care...

Any help or comment would be extremely helpful!!


Let $(X,\|\cdot\|)$ denote the space you are describing.

For each $k\in\mathbb{N}$ define $Y_k$ to be the subspace of $X$, consisting of sequences $x=(x_j)$ such that $x_j=0$ for all $j>k$. It is easy to see each $Y_k$ is closed in $X$ and $X=\cup Y_k$.

However, $Y_k$ has empty interior. To see this, take $x=(x_j)\in Y_k$ and $\epsilon>0$. Then we can always define $y=(y_j)$ such that $y_j=x_j$ for $j\le k$, $y_{k+1}=\epsilon$ and $y_j=0$ for other $j$. Then $\|y-x\|<\epsilon$ but $y$ is not in $Y_k$.

Now each $Y_k$ is nowhere dense, so their countable union $X$ cannot be complete as a result of Baire's theorem.

Or, if you prefer concrete examples. Take $x^n$ to be the sequence whose first $n$ terms are $(1,\frac{1}{2},\dots,\frac{1}{n})$ and all other terms vanishes. You can see this $x^n$ is Cauchy in $X$, but you should feel they converge to the sequence $(1,\frac{1}{2},\frac{1}{3},\dots,\frac{1}{n},\frac{1}{n+1},\dots)$, which is not in $X$.

| cite | improve this answer | |
  • $\begingroup$ thank you very much, I just started studying metric spaces so the first part of the answer is a bit obscure! $\endgroup$ – Moritzplatz Nov 18 '12 at 14:29
  • $\begingroup$ so was the problem in my "proof" that part about choosing a subsequence? $\endgroup$ – Moritzplatz Nov 18 '12 at 14:29
  • $\begingroup$ Yes, that part is a non sequitur. $\endgroup$ – Hagen von Eitzen Nov 18 '12 at 14:31
  • $\begingroup$ thank you very much i got why now! $\endgroup$ – Moritzplatz Nov 18 '12 at 14:32

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.