Let $(X,d_X),(Y, d_Y)$ be two metric spaces. We define:

\begin{equation} B(X,Y)=\{f:X\rightarrow Y, f\text{ bounded}\} \end{equation} Introducing the metric (of uniform convergence): \begin{equation} d(f,g)=\underset{x\in X}{\sup}d_Y(f(x),g(x)) \qquad\forall f,g\in B(X,Y) \end{equation}

Now, if $Y$ is complete, you can prove that $B(X,Y)$ is complete. I found this proof on a book, but there's a part I don't understand.


Let $\{f_n\}_{n\geqslant1}$ be a Cauchy sequence in $B(X,Y)$ and $x\in X$. The sequence $\{f_n(x)\}$ is a sequence in $Y$. $\{f_n(x)\}$ is trivially (so I'll be omitting some steps about this) a Cauchy sequence by definition of $d$: \begin{equation} d_Y(f_n(x),f_m(x))\leqslant\underset{x\in X}{\sup}d_Y(f_n(x),f_m(x))=d(f_n,f_m) \end{equation}

Since $Y$ is complete, $\{f_n(x)\}$ is convergent \begin{equation} \lim_{n\to+\infty} f_n(x)=f(x) \tag{1} \end{equation}

Let's write down the definition of Cauchy sequence for $\{f_n\}$: \begin{equation} \forall\epsilon>0 \quad \exists\nu\in\mathbb{N}: \forall n,m>\nu\quad n,m\in\mathbb{N}\implies d(f_n,f_m)<\epsilon \end{equation} Thus, we have: \begin{equation} \forall x\in X\quad\forall\epsilon>0 \quad \exists\nu\in\mathbb{N}: \forall n,m>\nu\quad n,m\in\mathbb{N}\implies d_Y(f_n(x),f_m(x))<\epsilon \tag{2} \end{equation}

Well, this is the critical point. The book now takes the limit as $m\to+\infty$ and using $(1)$:

\begin{equation} d_Y(f_n(x), f(x))\leqslant\epsilon \quad \forall x\in X \quad \forall n>\nu \tag{3} \end{equation}

Since this is valid $\forall x\in X$, it follows that $d(f_n, f)\leqslant\epsilon$ and the theorem is proved.

As I said, I don't understand the step between $(2)$ and $(3)$. It looks like we are assuming $d$ is continuos when we take the limit. My guess is that he used the reverse triangle inequality and then took the limit but I'm not sure about that. So how do we get from $(2)$ to $(3)$?


Yes, they used that $d$ is continuous. In a metric space, the metric itself is always continuous.

This fact can be shown (as you suspected) with the reverse triangle inequality.

  • $\begingroup$ The thing that gets me astonished is that the book never mentions reverse triangle inequality, nor continuity of distance. $\endgroup$
    – Feynman_00
    Nov 20 '20 at 11:34

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.