0

Let $T:\mathbb{R^2} \to \mathbb{R^2}$ be Lipschitz function.

Then,

(a) If $E$ is a set in $\mathbb{R}^2$ with Lebesgue measure zero, then $T(E)$ has measure zero in $\mathbb{R}^2$.

(B) If $A$ is a measurable set in $\mathbb{R}^2$, then $T(A)$ is also a measurable in $\mathbb{R}^2$.

How I can approach to this problem?

Seongqjini
  • 1,261
  • 6
    Oh, come on! It would have been enough to look at the questions suggested by MSE while you were typing yours, in order to find the answer: (a) http://math.stackexchange.com/questions/1504487/a-lipschitz-transform-maps-measurable-set-to-measurable; (b) http://math.stackexchange.com/questions/139883/why-does-a-lipschitz-function-f-mathbbrd-to-mathbbrd-map-measure-zero-s and http://math.stackexchange.com/questions/1780404/image-of-lipschitz-map-measure-zero and many others. – Alex M. Jun 14 '16 at 20:28
  • If $T$ maps $\mathbb R$ to $\mathbb R$ then what sense does $T(E)$ make? – zhw. Jun 14 '16 at 21:00

1 Answers1

3

This is true in general for $T:\mathbb R^{d}\to R^{d}$. Let us do this for $d=1$; the extension to $R^{d}$ is more or less obvious. Let $T$ have Lipschitz constant $C>0.$

a). For $\epsilon >0 $ we can find a disjoint union of intervals $\left \{ I_n \right \}_{n}$ such that $E\subseteq \cup^{m}_{n=1} I_n$ and $m(\cup^{m}_{n=1} I_n)=\sum ^{m}_{n=1} m(I_n)<\epsilon$. But then

$m(T(E))\leq m(T(\cup^{m}_{n=1} I_n))=m(\cup^{m}_{n=1} T(I_n))=\sum ^{m}_{n=1} m(T(I_n))\leq C\sum ^{m}_{n=1} m(I_n)=C\epsilon$, so $T(E)$ has measure zero.

b). If $E$ is Lebesgue measureable then its regularity implies that it is the union of a null set $N$ and an $F_{\sigma } $ set $F$. Indeed, $F$ can easily be constructed as a countable union of compact sets.

So by part a) it will suffice to prove that $m(F)$ is measureable. By the comment, write $F=\bigcup^{\infty }_{n=1}K_n$ where each $K_n$ is compact. Now since $T$ is Lipschitz, it is continuous, so $T(K_n)$ is compact, hence closed, which implies that $T(F)=T(\bigcup^{\infty }_{n=1}K_n)=\bigcup^{\infty }_{n=1}T(K_n)$, is an $F_{\sigma} $, being a countable union of closed sets.

David
  • 143
Matematleta
  • 30,081
  • 1
    The case $d>1$ needs a bit more argumentation - you can not a priori say that $m(T(I_n))\leq Cm(I_n)$ (you have to essentially bound a rectangle under a Lipschitz function). You could use intervals with equal side lengths and find an upper bound that involves the diagonal. – David Nov 21 '24 at 07:37
  • @David I see. Yes, I need to do a bit more work. Thanks! – Matematleta Nov 28 '24 at 00:16