logo

Convergence of Cauchy Sequences in Metric Spaces 📂MetricSpace

Convergence of Cauchy Sequences in Metric Spaces

Definition

  • Let {pn}\left\{ p_{n} \right\} be a sequence of points in a metric space (X,d)(X,d). If for every positive number ε\varepsilon, there exists a positive number NN such that

    nN, mN    d(pn,pm)<ε n\ge N,\ m\ge N \implies d(p_{n},p_{m})<\varepsilon

    is satisfied, then {pn}\left\{ p_{n} \right\} is called a Cauchy sequence.

  • If every Cauchy sequence in a metric space XX converges to a point in XX, then XX is called a complete space.

Explanation

From the following theorem, it can be seen that every compact metric space and Euclidean space is complete.

Theorem

(a) In a metric space, every converging sequence is a Cauchy sequence.

(b) Let XX be a compact metric space and {pn}\left\{ p_{n} \right\} be a Cauchy sequence in XX. Then, {pn}\left\{ p_{n} \right\} converges to some pXp\in X.

(c) In Rk\mathbb{R}^{k}, all Cauchy sequences converge.


(a) and (b) together imply ‘In a compact metric space, converging sequences and Cauchy sequences are equivalent.’

Proof

(a)

Let pnpp_{n} \to p and ε>0\varepsilon >0 be given. Then, there exists NN satisfying nN, d(p,pn)<ε\forall n \ge N,\ d(p,p_{n})<\varepsilon. Therefore, the following is satisfied:

d(pn,pm)d(pn,p)+d(p,pm)<2ε,m,nN d(p_{n},p_{m}) \le d(p_{n},p)+d(p,p_{m})<2\varepsilon,\quad \forall m,n\ge N

Therefore, by definition, {pn}\left\{ p_{n} \right\} is a Cauchy sequence.

(b)

Let {pn}\left\{ p_{n} \right\} be a Cauchy sequence in the compact metric space XX. And let it be said for any natural number NN:

EN={pN,pN+1,pN+2,} E_{N}=\left\{ p_{N},p_{N+1},p_{N+2},\cdots\right\}

Then, the following is satisfied:

limNdiam EN=0 \begin{equation} \lim \limits_{N\to\infty}\mathrm{diam\ }\overline{E_{N}}=0 \label{eq1} \end{equation}

Moreover, EN\overline{E_{N}} is a closed subset of the compact space XX, so EN\overline{E_{N}} is compact. It is also evident that the following equation is satisfied:

ENEN+1andENEN+1 E_{N}\supset E_{N+1} \quad \text{and} \quad \overline{E_{N}}\supset \overline{E_{N+1}}

Therefore, from the above conditions, it can be seen that there exists a unique pXp \in X satisfying NN, pEN\forall N \in \mathbb{N},\ p \in \overline{E_{N}}1. Now, let ε>0\varepsilon >0 be given. Then, by (eq1)\eqref{eq1},

NN0    diam EN<ε N \ge N_{0}\implies \mathrm{diam\ }\overline{E_{N}}< \varepsilon

a N0N_{0} exists that satisfies the condition. But since diam E=diam E\mathrm{diam\ }\overline{E}=\mathrm{diam\ }E and pENp \in \overline{E_{N}}, for all qENq \in E_{N}, d(p,q)<εd(p,q)<\varepsilon is satisfied. In other words, the following is implied:

nN0    d(pn,p)<ε n \ge N_{0} \implies d(p_{n},p)< \varepsilon

This is the definition of pnpp_{n}\to p, therefore limnpn=p\lim \limits_{n\to\infty} p_{n}=p

(c)

Let {xn}\left\{ \mathbf{x}_{n} \right\} be a Cauchy sequence in Rk\mathbb{R}^{k}. Assuming ENE_{N} as the proof of (b), then

diam EN<1 \mathrm{diam\ } E_{N} <1

let a NN be chosen that satisfies the condition. And

r=max{d(xN,x1), d(xN,x2), , d(xN,xN1), 1} r=\max \left\{ d(\mathbf{x}_{N},\mathbf{x}_{1}),\ d(\mathbf{x}_{N},\mathbf{x}_{2}),\ \cdots,\ d(\mathbf{x}_{N},\mathbf{x}_{N-1}),\ 1 \right\}

since m,nN, d(xn,xm)<r\forall m,n \in \mathbb{N},\ d(\mathbf{x}_{n},\mathbf{x}_{m}) <r, {xn}\left\{ \mathbf{x}_{n} \right\} is bounded. Therefore, {xn}\overline{ \left\{ \mathbf{x}_{n} \right\}} is a closed and bounded subset of Rk\mathbb{R}^{k}, hence compact. Therefore, {xn}\left\{ \mathbf{x}_{n} \right\} is a Cauchy sequence in a compact space, and thus by (b), {xn}\left\{ \mathbf{x}_{n} \right\} converges.


  1. See theorem (b) ↩︎