3
$\begingroup$

Let $G$ be a sugroup of $\mathbb{R}$ under addition and $G \neq \mathbb{R}$ and $G$ is Lebesgue measurable.Prove that $m(G)=0$.

One idea to solve this, is that if $m(G)>0$ then $G+G$(or $G-G$) contains an interval.Also because of the fact that $G$ is a group we have that $G+G \subseteq G$.

My idea is that i can construct a non measurable set and derive a contradiction,using the fact that $G \triangleleft \mathbb{R}$ nad use the quotient group $\mathbb{R} /G$.But i cannot achieve countability.le

Maybe i have to use a different approach.

Can someone give me some ideas to solve this(not the whole answer)?

Thank you in advance!

$\endgroup$
5
  • $\begingroup$ If $G$ contains an interval, then it contains an interval around $0$. If $G$ contains an interval around $0$, then it is $\mathbb{R}^2$. $\endgroup$ – Michael Burr Jan 18 '17 at 17:04
  • $\begingroup$ i did not understand that.Can you explain it a bit more? $\endgroup$ – Marios Gretsas Jan 18 '17 at 18:54
  • $\begingroup$ ok thank you.Whenever you have the time...:) $\endgroup$ – Marios Gretsas Jan 18 '17 at 20:12
  • 1
    $\begingroup$ That $G$ has positive Lebesgue measure doesn't imply it contains an interval. There are sets of positive measure that contain no non-trivial intervals. $\endgroup$ – AJY Jan 18 '17 at 20:33
  • $\begingroup$ i use the fact that $G$ is a group and it is closed under addition,thus $G+G \subseteq G$.I did not assume that every set with possitive measure contains an interval $\endgroup$ – Marios Gretsas Jan 18 '17 at 20:36
3
$\begingroup$

If one can show that $G$ contains an interval, then there is some interval $(a,b)\subseteq G$. Let $m$ be the midpoint of the interval, since $m\in G$, $-m\in G$, so we can subtract $m$ and remain in $G$ to get that $G$ contains $(a-m,b-m)$ which is an interval centered at the origin, i.e., of the form $(-c,c)$. Therefore, $(-c,c)\subseteq G$. Now, if you consider $G+G$ (which is a subset of $G$), you can conclude that $(-2c,2c)\subseteq G$. By continuing this, one gets that $(-2^nc,2^nc)\subseteq G$ for all $n$, so $G$ is all of $\mathbb{R}$.

$\endgroup$
2
  • $\begingroup$ isn't enough for jestification that $G+G \subseteq G$? We know of course that if a set $A \subseteq \mathbb{R}$ has a possitive lebesgue measure then $A+A$ contains an interval. $\endgroup$ – Marios Gretsas Jan 18 '17 at 23:33
  • $\begingroup$ I agree, it does follow, see Positive measure of a sum. $\endgroup$ – Michael Burr Jan 19 '17 at 1:20
1
$\begingroup$

If the group is numerable, the resuslt is true since it is the union of a numerable sets (the singletons) and the measure of each of these sets is zero. If the group is not numerable, a classic result shows it is dense in $R$. Let $c\in ]0,1[\cap G$, If $m(G\cap [0,c[)=c>0$ implies $m(G)=\sum_{n\in Z}m(G\cap [nc,nc+c[)=\sum_{n\in Z}m(G\cap [0,c[)=+\infty$. We deduce that $m(G\cap [0,c[)=0$ and since $\bigcup_{n\in Z}G\cap [nc,nc+c[)=G$ is a disjoint union of a numerable set of measure zero and $m(G\cap [0,c[)=m(G\cap [nc,nc+c[)$, its measure is zero.

$\endgroup$
1
  • $\begingroup$ So if $G$ is measurable then $m(G)=0.$ Can we show that $G$ is a measurable set?...........+1 $\endgroup$ – DanielWainfleet Jan 25 '17 at 16:34
1
$\begingroup$

Let $G$ be a Lebesgue measurable proper subgroup of $(\mathbb R,+).$ Assume for a contradiction that $m(G)\gt0.$

Choose an interval $I$ with $m(G\cap I)\gt\frac23m(I).$ Let $\varepsilon=\frac13m(I).$

Choose $t\in(0,\varepsilon)\setminus G.$

Observe that $m((G+t)\cap I)\gt\frac13m(I).$

Since $G\cap(G+t)=\emptyset$ it follows that $m(I)\ge m(G\cap I)+m((G+t)\cap I)\gt m(I),$ which is absurd.

$\endgroup$
2
  • $\begingroup$ This shows very clearly that if $G$ is measurable then $m(G)=0.$ Can we show that $G$ is a measurable set?..........+1 $\endgroup$ – DanielWainfleet Jan 25 '17 at 16:32
  • $\begingroup$ Yes it occurred to me later that I knew that. $\endgroup$ – DanielWainfleet Jan 26 '17 at 1:07
-1
$\begingroup$

The idea the group of $R$ must contain an interval in general it is not true since we can take the rational number is sub group of $R$ ( filed) but it doesn't contain any interval.

$\endgroup$
3
  • $\begingroup$ That should be a comment $\endgroup$ – DanielWainfleet Jan 25 '17 at 16:34
  • $\begingroup$ yes but i assumed that the group has positive measure $\endgroup$ – Marios Gretsas Jan 30 '17 at 9:22
  • $\begingroup$ the rationals have zero measure and a lebesgue measurable set $A$ with positive measure has this property : $A+A$ contains an interval $\endgroup$ – Marios Gretsas Jan 30 '17 at 9:24

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.