I want to show that the indicator function (aka. the characteristic function) is not a regulated function.

\begin{align} \chi : \begin{cases}[a,b] & \longrightarrow \mathbb{R} \\ x & \longmapsto \begin{cases} 0, \ x \in \mathbb{Q} \cap [a,b] \\ 1 , \ x \in (\mathbb{R} \setminus \mathbb{Q} ) \cap [a,b] \end{cases} \end{cases} \end{align}

The definition (Königsberger Analysis) uses seems to differ to other sources where such topics are discussed.

Def: A function $f: [a,b] \to \mathbb{R}$ is called a regulated function if $\exists (f_n)_{n \in \mathbb{N}}$ sequence of step functions $f_n: [a,b] \to \mathbb{R}$ such that $$ \lim_{n \to \infty}\|f_n-f \| =0 $$ where $\|. \|$ denotes the supremum norm.

My approach: I am having trouble with the 'elegant' formulation of the solution. Also I am not sure how to do the calculation.

I want to show that for all step functions $\phi: [a,b] \to \mathbb{R}$ it follows that $\| \phi -f \| \geq 1$ or some other small value.

So let $\phi : [a,b] \to \mathbb{R}$ be a step function and $a=x_0< x_1< \dots < x_n =b$ a partition of $[a,b]$, since $\mathbb{Q}$ is dense in $\mathbb{R}$ I can always find $v_1 \in (a, x_1)\cap \mathbb{Q}$ and $v_2 \in (a,x_1) \cap (\mathbb{R} \setminus \mathbb{Q})$. Since $\phi$ is a step function it would also follow that $\phi(v_1) = \phi (v_2) = c_0 \in \mathbb{R}$

Now I want to use this for a calculation with the norm $$\| \phi - \chi \| = \sup_{x \in [a,b]} |\phi (x) -\chi(x) | \geq \max_{x \in [a,b]} | \phi (x) -\chi(x) | $$

and now, sadly, I lack experience on how to argument with the maximum or the minimum. I tried to visalize it for easier functions than $\chi$ but I couldn't see how I could split up the maximum in order to make even use of my calculations above.

I'd appreciate some insight on how to continue with my calculations, i.e. since I have partition of my interval, I can maybe split up the maximum for all these cases and always end up with states such as: $$ \geq \max(|\phi(v_1)-\chi(v_1)|, \dots , |\phi(v_n)-\chi(v_n)| \\ \geq \max(|\phi(v_1)-\chi(v_1)|, | \phi(v_2)- \chi(v_2)|) = \max (|c_0|, |c_0-1|) $$ but that just doesn't seem right to me.

  • $\begingroup$ Use that any $y\in \Bbb{R}$ fulfils $\max \{|y|, |y-1|\}\geq 1/2$ to show $\| \chi- \phi \|\geq 1/2$. With the value $1$ you will not have success (consider $\phi \equiv 1/2$) $\endgroup$ – PhoemueX Dec 7 '14 at 8:47
  • $\begingroup$ Thanks for your comment @PhoemueX, so it looks like I am only one step away from the solution. I understand what you've said, but I still don't know how to correctly write that. Wlog I choose just one point with said characteristics and obtain that that $\| \phi - \chi \| \geq \max{(|c_y|,|c_y-1|)}\geq 1/2$ ? $\endgroup$ – Spaced Dec 7 '14 at 12:38
  • 1
    $\begingroup$ I posted a continuiation of the beginning your proof as an answer. Does that help you? $\endgroup$ – PhoemueX Dec 7 '14 at 18:42

Your proof is nearly perfect. Simply note that

$$ \Vert \phi - \chi \Vert \geq \max \{ |\phi(v_1) - \chi(v_1)|, |\phi(v_2) - \chi(v_2)|\}\\ = \max\{ |c_0 - 0|, |c_0 - 1|\} \geq 1/2. $$

Here, we used that $v_1 \in \Bbb{Q}$ and hence $\chi(v_1) = 0$ and $v_2 \in \Bbb{R}\setminus\Bbb{Q}$, hence $\chi(v_2) = 1$ and we also used $\phi(v_1) = \phi(v_2) = c_0$ (as you already noted yourself).

Finally, we made use of the inequality

$$ \max\{|y|, |y - 1|\} \geq 1/2, $$ which is valid for all $y \in \Bbb{R}$.


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.