Let $(X,d)$ be a complete metric space, $F\subset X$. Then $F$ is closed if and only if the relative metric space $(F, d_F)$ is complete.

Since $(X,d)$ is complete, Cauchy sequences in $X$ converge in $X$ with respect to $d$. Suppose $F$ is closed, then convergent sequences contained in $F$ converge in $F$. Since Cauchy sequences contained in $F$ are convergent, they must converge in $F$ under $(F, d_f)$.

Let $(x_k)_k\subset F$ be a Cauchy sequence converging to $x\in X$. Then for every $\epsilon > 0,\exists N\in\mathbb{N}$ such that $m,n\ge N\implies d_F(x_n,x_m)<\epsilon/3$. Since $(x_k)_k$ is convergent, $\forall\epsilon>0, \exists M\in\mathbb{N}$ such that $d_F(x_k, x)<\epsilon/3$. Fix $\epsilon>0$ and let $n\ge \max\{M,N\}$. Since $F$ is closed, $\exists y_\epsilon\in B_{\epsilon/3}(x)\cap F\ne \emptyset$.

Now, $$d_F(x_n, y)\le d_F(x_n, x_m)+d_F(x_m, x)+d_F(x,y)<\epsilon/3+\epsilon/3+\epsilon/3=\epsilon$$

Thus $(x_k)_k$ converges to some $y\in F$, but then $x=y$, so that $(F,d_F)$ is complete.

I'm wondering if I'm not being somewhat redundant in my attempt to make my proof more rigorous. Would appreciate some suggestions.

For the other direction, suppose $(F,d_F)$ is complete. Then every Cauchy sequence contained in $F$ converges in $F$ w.r.t. $d_F$. Let $(x_k)_k\subset F$ be Cauchy. Then $(x_k)_k$ converges to some point $x\in F$. Suppose there exists some limit point $y$ of $(F,d_F)$ not contained in $F$. Then there exists a sequence $(y_k)_k$ which converges to $y$. But then $(y_k)_k$ must be Cauchy, so that $(y_k)_k$ converges in $F$, a contradiction. Hence, $F$ is closed.

Please let me know if you find my proof satisfactory or if something better be fixed.


If $(F,d_F)$ is closed, it is complete: suppose $(x_n)$ is a CAuchy sequence in $F$. It is still a Cauchy sequence in $(X,d)$ as $d$ and $d_F$ by definition agree on $F$. So it converges to some $x \in X$ as $X$ is complete. As $F$ is closed, a limit of a sequence from $F$ is in $F$, so $x \in F$ and $x_n \to x$ in $(F,d_F)$ as well.

If $(F,d_F)$ is complete: let $x_n \to x$ (in $(X,d)$) where all $x_n \in F$. We want to show $x \in F$. Well, any convergent sequence is Cauchy (in $(X,d)$ so also in $(F,d_F)$, so $(x_n)$ converges in $(F,d_F)$ to some $y \in F$, but as sequence limits are unique $x=y$ and $x \in F$. So $F$ is closed.

  • $\begingroup$ Thanks. Can you please let me know if the technical part of my proof is satisfactory? $\endgroup$
    – sequence
    Oct 17 '17 at 21:57
  • $\begingroup$ @sequence Your proof makes no sense to me, sorry. $\endgroup$ Oct 18 '17 at 20:16
  • $\begingroup$ Actually I used the same argument (my wording may be a little less clear), but I also attempted to make it more formal. $\endgroup$
    – sequence
    Oct 19 '17 at 22:01
  • $\begingroup$ What purpose does the $\frac{\varepsilon}{3}$ serve? What are you trying to show?@sequence $\endgroup$ Oct 20 '17 at 8:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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