Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Can someone please check whether the following simple proof is "mathematical"? Is it correct, complete, rigid? Can it be simplified? I'm a complete autodidact so I'm looking for someone to give me feedback to gain experience in writing proofs... This is also my first question on MSE.

The proposition:

Let $(X, d)$ be a metric space and $x_n \to x$ where each $x_n \in X$ and $x \in X$. Let $A$ be the subset of $X$ which consists of $x$ and all of the points $x_n$. Prove that $A$ is closed in $(X, d)$.

My tentative to prove this:

We first show that all infinite sequences in $A$ converge to $x$: Let $y \in X$, $y \ne x$. Then there is some open ball $B_\epsilon(x)$ with $\epsilon < d(x,y)$ containing all but finitely many elements of $A$. As $y \notin B_\epsilon(x)$ there can be no infinite sequence in $A$ converging to $y$. Consequently all infinite series in $A$ converge to a point in $A$ which therefore must be a closed set.

Edited: As rightly pointed out in the comments, I should have written in the first sentence "...sequences with infinitely many distinct terms and which converge to some point of $X$" and the last sentence should be "Consequently all infinite sequences...".

share|cite|improve this question
In your last sentence, you want to say "all infinite sequences". – Martin Argerami Mar 28 '12 at 2:18
Regarding your proof, the idea is certainly ok. The proof could be simpler, more complicated, or different, depending on your definition of closed set. Here you are using that a set is closed if it contains all of its accumulation points. This is ok, but my point is that you are using a previous result (unless that's your definition of closed set) and the proof would be different if you want to do it straight from a different definition of closed ("complement of open", for example). – Martin Argerami Mar 28 '12 at 2:21
By "infinite sequence", I think you mean "a sequence containing infinitely-many distinct elements". (The constant sequence $x_n = x_0$ for all $n$ is an infinite sequence.) – Austin Mohr Mar 28 '12 at 2:22
I'm impressed with this fast and friendly feedback. Thanks a lot! @Martin Argerami: Yes, you're right. I wanted to say "sequence" rather than "series". And you're right that I'm using a previous result in this case rather than doing the proof from (my) definitions. – jerico Mar 28 '12 at 2:24
@Austin Mohr: I thought that the sequence can contain duplicates without the proposition or proof being invalid... Is that wrong? – jerico Mar 28 '12 at 2:28
up vote 3 down vote accepted

As I said in my comment, your proof is ok. But here is the way I would have done it. I'll write the proof in a more formal way, because in math you can only talk loose after you master writing properly.

Let $y\in X\setminus A$. Let $\varepsilon=d(y,x)/2$. Then, by the convergence $x_n\to x$, there exists $n_0$ such that $x_n\in B_\varepsilon(x)$ for all $n\geq n_0$. So, for $n\geq n_0$, $$\tag{1} d(x_n,y)> d(x,y)-d(x_n,x)>d(x,y)-\varepsilon=d(x,y)/2. $$ Let $\delta=\min\{d(x,y),d(x_1,y),\ldots,d(x_{n_0},y)\}/2$. Then $d(y,x_n)\geq\delta$ if $n\leq n_0$, and by ($1$) $d(y,x_n)\geq\delta$ if $n\geq n_0$. This shows that $B_\delta(y)$ has no intersection with $A$, i.e. $B_\delta(y)$ is contained in $X\setminus A$. As $y$ was arbitrary, this shows that $X\setminus A$ is open, i.e. $A$ is closed.

share|cite|improve this answer
Great answer. That's exactly what I was looking for. :-) – jerico Mar 28 '12 at 19:11

What you really want to do is verify that $A$ satisfies your definition of closed set: that first of all any sequence $y_n \in A$ has a limit, i.e. that there exists $y$ so that $\lim y_n = y$, and second of all that $y \in A$. Informally speaking you want to say that every sequence in $A$ converges to a limit that is also in $A$ (note that the limit does not need to be $x$). This essentially eliminates the problems people pointed out in the comments. Based on this I bet you can write a new proof pretty quickly. Nevertheless, here is a quick outline of a solution:

If $y_n \in A$ you know that $y_n = x_{n_k}$ or $y_n = x$. From this you may conclude that either $y_n$ is eventually constant or $y_n$ converges to $x$. Either way you see that $y_n$ converges to a limit contained in $A$.

share|cite|improve this answer
Thanks @treble for this additional approach. If I could choose two answers then I would have chosen yours, too. I slightly preferred the other answer because in my special situation it was useful to have the proof spelled out in detail. – jerico Mar 28 '12 at 19:01
It is false that every sequence in A converges! Consider the sequence x_1, x_2, x_1, x_2, x_1, x_2, x_1, x_2, ... – Omar Antolín-Camarena Mar 28 '12 at 23:21
Of course, it doesn't matter that not every sequence in A converges, because the right definition of closed is not "every sequence in A converges to a point of A", but rather "every sequence in A which happens to converge to a point of X actually converges to a point of A". – Omar Antolín-Camarena Mar 28 '12 at 23:22
Yes, of course you are right- no need for the exclamation point. I should have said that every sequence has a subsequence that converges, which is what I was thinking, or I should have said that every convergent sequence converges to a point in $A$. – treble Mar 29 '12 at 1:15

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.