logo

Convergence of Sequences in Metric Spaces 📂MetricSpace

Convergence of Sequences in Metric Spaces

Definitions1

If there exists a point pXp \in X such that the sequence {pn}\left\{ p_{n} \right\} of points in a metric space (X,d)(X,d) satisfies the following condition, the sequence {pn}\left\{ p_{n} \right\} is said to converge to pp, and it is denoted by pnpp_{n} \rightarrow p or limnpn=p\lim \limits_{n\to \infty}p_{n}=p.

ε>0, NN s.t nN    d(pn,p)<ε \forall \varepsilon >0,\ \exists N\in \mathbb{N}\ \mathrm{s.t}\ n\ge N \implies d(p_{n},p)<\varepsilon

If {pn}\left\{ p_{n} \right\} does not converge, it diverges. Furthermore, the set of all pnp_{n}s is called the range of {pn}\left\{ p_{n} \right\}. If the range of {pn}\left\{ p_{n} \right\} is bounded, the sequence {pn}\left\{ p_{n} \right\} is said to be bounded.

Theorems

Let {pn}\left\{ p_{n} \right\} be a sequence in a metric space (X,d)(X,d).

(a) A necessary and sufficient condition for pnpp_{n}\to p is that every neighborhood of pp contains all terms of {pn}\left\{ p_{n} \right\} except for a finite number.

(b) If pnpp_{n} \to p and pnpp_{n} \to p^{\prime}, then p=pp=p^{\prime}.

(c) If {pn}\left\{ p_{n} \right\} converges, it is bounded.

(d) Given EXE\subset X. If pp is a limit point of EE, there exists a sequence {pn}\left\{ p_{n} \right\} from EE that satisfies p=limnpnp=\lim \limits_{n \to \infty}p_{n}. Furthermore, if {pn}\left\{ p_{n} \right\} is a set of distinct points, the converse also holds.

Proofs

(a)

  • (    )(\implies)

    Suppose pnpp_{n} \to p. Given any positive number ε>0\varepsilon >0, let VV be a neighborhood of pp with radius ε\varepsilon. According to the definition of a neighborhood, the following holds:

    d(p,q)<ε    qV d(p,q)<\varepsilon\quad \implies q\in V

    However, by assumption, there exists a NN that satisfies the following condition for the given ε\varepsilon:

    nN, d(pn,p)<ε \forall n \ge N,\ d(p_{n},p) <\varepsilon

    Therefore, all points pnp_{n}, except for a finite number, are included in VV.

  • (    )(\impliedby)

    Assume every neighborhood of pp contains all but a finite number of points {pn}\left\{ p_{n} \right\}. Given any positive number ε>0\varepsilon>0, let VV be a neighborhood of pp with radius ε\varepsilon. Then, by assumption, a NN exists that satisfies the following condition:

    nN    pnV n \ge N \implies p_{n}\in V

    Since VV is a neighborhood of pp, the following holds:

    nN,d(pn,p)<ε \forall n \ge N,\quad d(p_{n},p)<\varepsilon

    Therefore, pnpp_{n}\to p.

(b)

Given any positive number ε>0\varepsilon >0. By assumption, there exist two positive numbers NN, NN^{\prime} that satisfy the following condition:

nN    d(pn,p)<ε2nN    d(pn,p)<ε2 \begin{align*} n\ge N & \implies d(p_{n},p) <\frac{\varepsilon}{2} \\ n\ge N^{\prime} & \implies d(p_{n},p) <\frac{\varepsilon}{2} \end{align*}

Then, for nmax(N,N)n \ge \max(N,N^{\prime}), the following equation holds:

d(p,p)d(p,pn)+d(pn,p)<ε d(p,p^{\prime}) \le d(p,p_{n}) + d(p_{n},p^{\prime})<\varepsilon

Since ε\varepsilon is any positive number,

d(p,p)=0 d(p,p^{\prime})=0

and by the definition of distance, p=pp=p^{\prime}.

(c)

Assume that {pn}\left\{ p_{n} \right\} converges to pp. By assumption, there exists a positive number NN that satisfies the following equation:

nN    d(pn,p)<1 n \ge N \implies d(p_{n},p)<1

Now, let

r=max{1, d(p1,p), , d(pN,p)} r=\max \left\{ 1,\ d(p_{1},p),\ \cdots,\ d(p_{N},p) \right\}

Then, for all nn,

d(pn,p)r d(p_{n},p)\le r

so {pn}\left\{ p_{n} \right\} is bounded.

(d)

  • (    )(\implies)

    Assume EXE\subset X and that pp is a limit point of EE. By definition of a limit point, for each nn,

    d(pn,p)<1n d(p_{n},p) < \frac{1}{n}

    exists a pnEp_{n}\in E. Now, given any positive number ε>0\varepsilon >0 and Nε>1N\varepsilon>1 satisfying NN, the following holds for n>N n >N:

    d(pn,p)<1n<Nnε<ε d(p_{n},p)< \frac{1}{n}<\frac{N}{n}\varepsilon<\varepsilon

    Therefore, {pn}\left\{ p_{n} \right\} converges to pp.

  • (    )(\impliedby)

    Assume there exists a sequence {pn}\left\{ p_{n} \right\} of distinct points from EE that satisfies p=limnpnp=\lim \limits_{n\to\infty}p_{n}. Then, for all positive numbers ε>0\varepsilon >0,

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

    exists a NN. Let VεV_{\varepsilon} be a neighborhood of pp with radius ε\varepsilon. Thus, VεV_{\varepsilon} includes pnE(nN)p_{n} \in E (n\ge N) which is not pp, proving pp is a limit point of EE.


  1. Walter Rudin, Principles of Mathematical Analysis (3rd Edition, 1976), p47-48, 55-58 ↩︎