# Definition:Convergent Sequence/Metric Space

## Definition

Let $M = \struct {A, d}$ be a metric space or a pseudometric space.

Let $\sequence {x_k}$ be a sequence in $A$.

### Definition 1

$\sequence {x_k}$ **converges to the limit $l \in A$** if and only if:

- $\forall \epsilon \in \R_{>0}: \exists N \in \R_{>0}: \forall n \in \N: n > N \implies \map d {x_n, l} < \epsilon$

### Definition 2

$\sequence {x_k}$ **converges to the limit $l \in A$** if and only if:

- $\forall \epsilon > 0: \exists N \in \R_{>0}: \forall n \in \N: n > N \implies x_n \in \map {B_\epsilon} l$

where $\map {B_\epsilon} l$ is the open $\epsilon$-ball of $l$.

### Definition 3

$\sequence {x_k}$ **converges to the limit $l \in A$** if and only if:

- $\ds \lim_{n \mathop \to \infty} \map d {x_n, l} = 0$

### Definition 4

$\sequence {x_k}$ **converges to the limit $l \in A$** if and only if:

- for every $\epsilon \in \R{>0}$, the open $\epsilon$-ball about $l$ contains all but finitely many of the $x_k$.

We can write:

- $x_n \to l$ as $n \mathop \to \infty$

or:

- $\ds \lim_{n \mathop \to \infty} x_n \to l$

This is voiced:

**As $n$ tends to infinity, $x_n$ tends to (the limit) $l$.**

If $M$ is a metric space, some use the notation

- $\ds \lim_{n \mathop \to \infty} x_n = l$

This is voiced:

**The limit as $n$ tends to infinity of $x_n$ is $l$.**

Note, however, that one must take care to use this alternative notation only in contexts in which the sequence is known to have a limit.

It follows from Sequence Converges to Point Relative to Metric iff it Converges Relative to Induced Topology that this definition is equivalent to that for convergence in a topological space.

### Comment

The sequence $x_1, x_2, x_3, \ldots, x_n, \ldots$ can be thought of as a set of approximations to $l$, in which the higher the $n$ the better the approximation.

The distance $\map d {x_n, l}$ between $x_n$ and $l$ can then be thought of as the error arising from approximating $l$ by $x_n$.

Note the way the definition is constructed.

**Given***any*value of $\epsilon$, however small, we can always find a value of $N$ such that ...

If you pick a smaller value of $\epsilon$, then (in general) you would have to pick a larger value of $N$ - but the implication is that, if the sequence is convergent, you will *always* be able to do this.

Note also that $N$ *depends on* $\epsilon$. That is, for each value of $\epsilon$ we (probably) need to use a different value of $N$.

### Note on Domain of $N$

Some sources insist that $N \in \N$ but this is not strictly necessary and can make proofs more cumbersome.