For this periodic continuous $g:\Bbb R\to \Bbb R$, and $f_n(x):=g(x/n)$, does $\{f_n\}_{n=1}^\infty$ converge uniformly?

80 Views Asked by At

I can not find a counterexample although I have the feeling it is not true.

Let $\ g: \mathbb{ R} \rightarrow \mathbb{R}$ continuous function

$ \forall x \in \mathbb{R} \ g(x+1) = g(x)$

$g(0) = 0$

Let $(f_n)_{n=1}^{\infty} \ f_n : \mathbb{R} \rightarrow \mathbb{R},\forall n \in \mathbb{N} \ f_n(x) = g(\frac{x}{n})$

Then $ \ (f_n)_{n=1}^{\infty} \ $ is uniform convergence in $ \ \mathbb{R} \ $ to $0$

So clearly we can see that there is a pointwise convergence $\ \forall x \in \mathbb{R} \ $ to $0$

And because there is a cycle I need to check only what is happening in $[0,1]$

foreach $ \epsilon \ \exists x\in [0,1] \ g(x) = \epsilon $ so there must be a such a $N \in \mathbb{N}$ so that even for the "worst" $x$ we can say that $|g(\frac{x}{n})| < \epsilon$

This is my unformal proof.

So, is the proof good ?

Am I right ?

If it is not true, please explain to me what am I doing wrong and give a counterexample.

Thanks in advanced !!

1

There are 1 best solutions below

3
On BEST ANSWER

With the definition

$$f_n(x) = g\left(\frac{x}{n}\right),$$

is is clear that

$$f_n(\mathbb{R}) = \{ f(x) : x \in \mathbb{R}\} = \{ g(y) : y \in \mathbb{R}\} = g(\mathbb{R}) = g([0,\,1]).$$

So, for all $n$, we have $\sup\limits_{x\in\mathbb{R}} \lvert f(x) - g(0)\rvert = \max\limits_{y\in [0,\,1]} \lvert g(y) - g(0)\rvert$.

Hence the sequence converges uniformly on all of $\mathbb{R}$ if and only if $g$ is constant.

$f_n$ converges uniformly to $g(0)$ on all bounded subsets of $\mathbb{R}$, however.