3
$\begingroup$

Is there a "standard" construction to get a Vitali set in $\mathbb R$ of full outer Lebesgue measure?

$\endgroup$
2
  • 1
    $\begingroup$ By "Vitali set" you mean a set of representatives for the equivalence relation "have rational difference"? $\endgroup$
    – Goldstern
    Jan 29, 2013 at 8:08
  • $\begingroup$ yes, exactly . $\endgroup$ Jan 29, 2013 at 15:20

2 Answers 2

2
$\begingroup$

See e.g. https://groups.google.com/forum/?fromgroups=#!topic/sci.math/ofkao7iugNg

$\endgroup$
0
2
$\begingroup$

Take $V$ a Vitali set invariant under multiplication by rational numbers $V = \mathbb{Q}^{\times} V$. Then the set of fractional parts of elements in $V$ is a Vitali set contained in $[0,1)$ of exterior measure $1$.

You can find such a $V$ in a naive way: first choose a system of representatives $S$ under the action of the group $x \mapsto a x + b$, $a \in \mathbb{Q}^{\times}$, $b\in \mathbb{Q}$; take $V = \bigcup_{q\in \mathbb{Q}^{\times}} \ q\cdot S$. Or you can take $V$ a complement of $\mathbb{Q}$ inside $\mathbb{R}$, that is $\mathbb{R} = V\oplus \mathbb{Q}$.

Proof of the statement:

Since $\bigcup_{t\in \mathbb{Q}}(V + t) = \mathbb{R}$, we have $\mu^{*}(V)>0$. Therfore, for every $\epsilon > 0$ there exists an interval $[\frac{m}{n}, \frac{m+1}{n}]$ such that $\mu^*(V\cap I) > (1-\epsilon) \mu (I)$. Since $V$ is invariant under multiplication by $n$, we have the above inequality for the interval $I=[m,m+1]$. It should be easy now.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.