Let $(X, d_X)$ be a metric space, and let $(Y, d_Y)$ be a complete metric space. The space $(C(X \to Y), d_{\infty})$ is a complete subspace of $(B(X \to Y), d_\infty)$. In other words, every Cauchy sequence of functions in $C(X \to Y)$ converges to a function in $C(X \to Y)$. ($C(X \to Y)$ is the space of continuous functions; $B(X \to Y)$ is the space of bounded functions; $d_\infty$ is a supnorm metric)

We have a complete metric space $(Y, d_Y)$. For each $x \in X$, let $(f_n(x))_{n=1}^\infty$ be any Cauchy sequence in $C(X \to Y)$, and it converges to $f(x)$ (i.e., for every $\epsilon >0$, there exists $N$ s.t. for $n \ge N$, $d_Y(f_n(x), f(x)) < \epsilon$).

I have the result that $C(X \to Y)$ is closed in $B(X \to Y)$. Thus, if I show that $f_n$ converges to $f$ with respect to $d_\infty$-metric, $f$ becomes automatically in $C(X \to Y)$.

The book suggests to use triangular inequality but I am not sure how to use this. I appreciate if you give me some help.


I will use the notation $C$ for $(C(X\to Y), d_{\infty})$. Let $\{f_n\}$ be a Cauchy sequence in $C$.

Then for each $\epsilon \gt 0$, there is $N\in \mathbb Z^+$ s.t. for all $m,n\gt N$, $||f_n-f_m||_{\infty} \lt \epsilon$.

Thus for all $x\in X$, $\{f_n(x)\}$ is a Cauchy sequence in $Y$. Since $Y$ is complete define $f:X\to Y$ as

$f(x)=\lim\limits_{n\to\infty}f_n(x)$. Then clearly $f_n$ converges to $f$.

The only thing we have to show is the continuity of $f$.

Let $t\in X$ and let $\epsilon\gt0$.

Since $f(t)=\lim\limits_{n\to\infty}f_n(t)$ there is $N\in \mathbb Z^+$ s.t for all $n\gt N$, $||f_n-f||_{\infty}\lt\frac{\epsilon}{3} \qquad (*)$

Choose $n_0\gt N$. Then since $f_{n_0}$ is continious there exist $\delta\gt0$ s.t for all $x\in X$ $|x-t|\lt \delta\implies |f_{n_0}(x)-f_{n_0}(t)|\lt \frac{\epsilon}{3} \qquad (**)$

then by $(*),(**)$

for all $x\in X$, if $|x-t|\lt \delta$ then

$\begin{split}|f(x)-f(t)|& =|f(x)-f_{n_0}(x)+f_{n_0}(x)-f_{n_0}(t)+f_{n_0}(t)-f(t)|\\ &\lt |f(x)-f_{n_0}(x)|+|f_{n_0}(x)-f_{n_0}(t)|+|f_{n_0}(t)-f(t)|\\ &\lt \frac{\epsilon}{3}+\frac{\epsilon}{3}+\frac{\epsilon}{3} =\epsilon\\ \end{split}$.

Thus $f$ is continuous.

  • $\begingroup$ Thanks for the answer. In my book, $d_{\infty}(f_n, f) = \sup \{d_Y(f_n(x), f(x)) : x \in X\}$, but in $(*)$, it seems that you write it as $d_{\infty} (f_n(t), f(t))$. Do we have such $d_\infty$ definition? $\endgroup$ – shk910 Mar 25 '20 at 11:30
  • $\begingroup$ @shk910 edited the answer. thare is no deffirence $\endgroup$ – Gune Mar 25 '20 at 17:37
  • $\begingroup$ I am still confusing with $(*)$. Why $f(t) = \lim_{n\to \infty} f_n(t) \implies ||f_n - f||_\infty<\epsilon/3$? I think it should be the opposite. $\endgroup$ – shk910 Mar 25 '20 at 22:13

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.