Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

Arthur's answer; (ZF) Prove 'the set of all subsequential limits of a sequence in a metric space is closed.

Let $\{p_n\}$ be a sequence in a metric space $X$.

Let $B=\{p_n|n\in\mathbb{N}\}$ and $p$ be a limit point of $B$.

It first seemed obvious that there exists a subsequence convergent to $p$, but i realized that

I can construct a 'infinite subset' of $\{p_n\}$ and form a new sequence convergent to $p$, but can't construct a 'subsequence'.

Help me how to construct a subsequence.

Additional Question; What is the precise definition of convergent? I think the definition should mention ordering of $\mathbb{N}$, but every definition i saw doesn't mention this. For example, let $p_n = 1/n$. Say $G$ is the usual ordering of $\mathbb{N}$ (i.e. Well-ordered by $\in$) If i follow this ordering, $\{p_n\}$ is convergent to 0. But if i follow $G^{-1}$, it's convergent to 1.

Or is it defined by the point where is the most, unofficially speaking, 'dense'?

share|cite|improve this question
up vote 0 down vote accepted

Let $B = \{p_n : n \in\Bbb N\}$. Let $p$ be a limit point of $p$. Define by recursion (possible in ZF):

Suppose you have $p_{n_1}, p_{n_2}, ..., p_{n_k}$ such that $n_1 < n_2 < ... < n_{k}$ and $d(p, n_j) < \frac{1}{j}$ for all $1 \leq j \leq k$, then find the least $l > n_k$ such that $d(p, p_l) < \frac{1}{k + 1}$. Let $l = n_{k + 1}$.

This does not require any choice. You just start computing $d(p, p_l)$ for $l > n_k$. By assuming $p$ is a limit point of $B$, you sure to find an $l$.

Then $(p_{n_k})_{k \in\Bbb N}$ is a subsequence converging to $p$.

The definition of convergence does use the ordering of $\mathbb{N}$. The definition of $(p_n)_{n \in \mathbb{N}}$ converging to $p$ is :

For all $\epsilon > 0$, there exists a $N \in \mathbb{N}$ such that for all $n > N$, $d(p_n, p) < \epsilon$.

The part "$n > N$" refers to the usual well-ordering of ordering of $\mathbb{N}$.

share|cite|improve this answer

The definition of convergence is the same as with the axiom of choice:

Let $\{x_n\}$ be a sequence; we say that $x_n$ converges to $x$ if for every $\varepsilon>0$ there exists some $n_0$ such that for all $n>n_0$ we have $d(x,x_n)<\varepsilon$.

Now it is clear that if $\{p_n\}$ is a sequence and $p$ is a limit point of the sequence there is a subsequence converging to it:

We define by induction over $\mathbb N$:

  • Let $q_0=p_0$.
  • Suppose that $q_n=p_k$ was defined, let $q_{n+1}=p_m$ be such that $m=\min\{i\in\mathbb N\mid i>k, d(p,p_i)<\frac12d(p,q_n)\}$. We did not use Dependent Choice because we use the well-ordering of $\mathbb N$ to choose $q_{n+1}$.

    We therefore just need to verify that such $i$ exists, but this is obvious since $p$ is a limit point of $\{p_n\}$, for every $\varepsilon>0$ there is some $k$ such that $p_k\in B(p,\varepsilon)$. In particular for $\varepsilon=\frac12d(p,q_n)$.

I claim that $\{q_n\}$ converges to $p$, since for every $\varepsilon>0$ there is some $n$ such that $d(p,q_n)<\varepsilon$ (note that we take $q_n$'s to have distance which closes in on $p$ exponentially or faster), and moreover all the choices of $q_k$ for $k>n$ are such that $d(p,q_k)<d(p,q_n)<\varepsilon$, as wanted.

share|cite|improve this answer

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.