1

So the question was:

For all $E\subset \mathbb R$ compact, $\lim_{x\to y} \frac{d(x, E)}{|x − y|} = 0$ for $a.e.y\in E.$

This is pretty travial if $E$ is open since for every point in $E$ you can create a ball around it and have the ball be in $E$. I was first thinking of proofing the fact on $\text{int}(E)$ (which is true since it is open), however it is not guaranteed that $m(\partial E)=0$. How would I fix this?

zhw.
  • 105,693

1 Answers1

2

This is true for any Lebesgue point $y$ of $E.$ First, recall the definition: $y$ is a Lebesgue point of $E$ provided

$$\lim_{x\to y^-}\frac{m((x,y)\cap E)}{y-x}=1,$$

with a similar statement from the right. Let's also recall Lebesgue's result: a.e. $y\in E$ is a Lebesgue point of $E.$

Fix such a $y.$ Think of $x$ as approaching $y$ from the left. Suppose the closest point of $E\cap [x,y]$ to $x,$ is $x+\delta,$ where $\delta\ge 0.$ As $x$ moves toward $y,$ $\delta$ will likely vary.

Now $(x,x+\delta)\cap E=\emptyset.$ It follows that

$$1=\lim_{x\to y^-}\frac{m((x,y)\cap E)}{y-x} \leq \lim_{x\to y^+}\frac{y-(x+\delta)}{y-x} = 1-\lim_{x\to y^+}\frac{\delta}{y-x}.$$

This implies $\dfrac{\delta}{y-x} \to 0$ which implies $\dfrac{d(x,E)}{y-x} \to 0.$

zhw.
  • 105,693