3

Working on the real line $(\mathbb{R})$, let $\mu : \mathscr{M} \rightarrow [0, +\infty]$ represent the Lebesgue measure ($\mathscr{M}$ is the set of measurable subsets of $\mathbb{R}$).

For $E \in \mathscr{M}$ and $x \in \mathbb{R}$, we define $$\rho(E, x) = \lim_{\delta \to 0+} \frac{\mu(E \cap (x - \delta, x + \delta))}{2\delta},$$ if the limit exists. The above limit is called the metric density of $E$ at $x$.

$(1)$ Given $E=(1,2)\cup(2,5]\cup\{6\}$, find the metric density of $E$ for all $x \in E$.

$(2)$ Let $\alpha \in (0, 1)$. Construct a set $E\subset\mathbb{R}$ such that $\rho(E, 0) = \alpha$.


My work and thoughts:

$(1)$ Since $x$ can be any real number, the intersection $E \cap (x - \delta, x + \delta)$ can be empty.

Also, if $(x - \delta, x + \delta) \subset E$, as $\delta \rightarrow 0+$ the intersection $E \cap (x - \delta, x + \delta) = x$.

In either case the measure equals zero. Is this correct?

For $(2)$ I have no idea.

glpsx
  • 2,202

2 Answers2

1

Let $x=\frac{3}{2}$ then, for small $\delta$ we have $$E\cap \left(\frac{3}{2}-\delta,\frac{3}{2}+\delta\right)=\left(\frac{3}{2}-\delta,\frac{3}{2}+\delta\right)$$

So, $\mu(E\cap(x-\delta,x+\delta))=2\delta$..

So??

  • As $\delta \rightarrow 0+$ the limit tends to zero. What I don't understand is why $E\cap (\frac{1}{2}-\delta,\frac{1}{2}+\delta)=\left(\frac{1}{2}-\delta,\frac{1}{2}+ \delta \right)$. The intersection should be equal to zero in my opinion since $\frac{1}{2} \notin E$. Could you explain this? – glpsx Mar 03 '16 at 14:05
  • @VonKar : CHeck the edit... I mean $\frac{3}{2}$... –  Mar 03 '16 at 14:07
  • $\rho(E,x)=0$, if $x<1$; $\rho(E,x)=1/2$, if $x=1$; $\rho(E,x)=1$, if $1<x<5$; $\rho(E,x)=1/2$, if $x=5$ and $\rho(E,x)=0$, if $x>5$ – George Mar 04 '16 at 16:26
  • @GogiPantsulaia : What is your point... You can write it as answer... –  Mar 04 '16 at 16:53
1

For Q(1). For $x\in E, $ if $x\in (1,2)\cup (2,5)$ then for all sufficiently small $d>0$ we have $$E\cap (x-d,x+d)= (x-d,x+d)$$ and $$(2 d)^{-1}\mu (E\cap (x-d,x+d))=(2 d)^{-1}\mu ((x-d,x+d))=(2 d)^{-1}\cdot 2 d=1.$$ You should be able to to show that $\rho (E,5)=1/2$ and $\rho (E,6)=0.$

  • Thank you very much! I was able to show that $\rho(E, 5) = 1/2$. Indeed, for $x = 5$, $E \cap (x -\delta,x + \delta)= (x - \delta)$ and $\mu(x - \delta) = \delta$. Therefore, $\rho(E, 5) = \lim_{\delta \to 0+} \delta / 2\delta = 1/2$. The case where $x = 6$ is the simplest. – glpsx Jun 13 '16 at 18:57