1
$\begingroup$

Why does $(f_{n})_{n}$ equicontinuous and $f_{n}$ uniformly continuous not imply $ (f_{n})_{n}$ uniformly equicontinuous. I have already seen the example of the case $(f_{n})_{n}$ where $f_{n}(x)=\arctan{(nx)}$ where $x \in ]0,\infty[$. But I fail to understand why it is so?

My logic: Since $(f_{n})_{n}$ equicontinuous, for any $x$ and $\epsilon > 0$, there exists a $\delta_{\operatorname{equi}} >0$ so that for any $y: \vert x-y\vert<\delta_{\operatorname{equi}}\Rightarrow$ $\vert f_{n}(x)-f_{n}(y)\vert<\epsilon$ for all $n \in \mathbb N$. And further, since for each respective $n \in \mathbb N$, $f_{n}$ is uniformly continuous, i.e. there exists $\delta_{\operatorname{unif}}>0$ so that for $\vert x-y\vert<\delta_{\operatorname{unif}}\Rightarrow$ $\vert f_{n}(x)-f_{n}(y)\vert<\epsilon$, couldn't we get uniform equicontinuity by taking the minimum, namely $\delta:=\min\limits_{n \in \mathbb N} \{\delta_{\operatorname{unif_{n}}},\delta_{\operatorname{equi}}\}$. I can see that my idea is wrong but I would just like clarity

$\endgroup$
  • 1
    $\begingroup$ How do you guarantee $\delta>0?$ BTW, $\delta_{\text{equi}}$ depends on $x$. $\endgroup$ – saulspatz Jul 12 '19 at 15:41
2
$\begingroup$

An infinite set that is bounded below does not need to attain its minimum. Your expression $$ \delta:=\min\limits_{n \in \mathbb N} \{\delta_{\operatorname{unif_{n}}},\delta_{\operatorname{equi}}\} $$ is not well-defined. For example, says we have $\delta_{\operatorname{unif_{n}}}=\frac 1n$, then $$ \left\{\delta_{\operatorname{equi}},1, \frac 12, \frac 13, \dots \right\} $$ has $0$ as its infimum so you'd need $\delta=0$ here.

$\endgroup$
0
$\begingroup$

You could also consider a sequence of "tent" functions, f_n, defined on (0, 2] by letting f_n be the function that increases linearly from 0 to 1 on the interval (0, 1/n] and then descends down to 0 linearly on [1/n, 2/n]. Clearly they are uniformly continuous and they are pointwise equicontinuous because for any fixed x in (0, 2] all tents will eventually be to the left of x so you only have to worry about finitely many of them. However, if the f_n s are defined on a compact metric space, say $C$, you can show that pointwise equicontinuity implies uniform equicontinuity. Indeed, fix $\epsilon>0$ and for each $x\in C$ let $\delta_x>0$ be such that for each $n$, $d(x, y)<\delta_x \rightarrow d(f_n(x), f_n(y))<\epsilon$. By compactness $\{B(x; \delta_x) : x\in X\}$ has a finite subcover, say $B(x_1; delta_{x_1}), B(x_2; delta_{x_2}), \cdots, B(x_1; delta_{x_1})$. Define a continuous function, $C \rightarrow (0,\infty), x \mapsto \sup\{r : B(x;r) \text{ is contained in one of the } B(x_i; delta_{x_i})\}$. It attains a minimum value, say $\delta>0$ and this is the delta you need to prove uniform equicontinuity.

$\endgroup$

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.