5
$\begingroup$

This an exercise 3 on Terence Tao's blog:

A real number $x$ is Diophantine if for every $\varepsilon > 0$ there exists $c_\varepsilon > 0$ such that $|x - \frac{a}{q}| \geq \frac{c_\varepsilon}{|q|^{2+\varepsilon}}$ for every rational number $\frac{a}{q}$. Show that the set of Diophantine real numbers has full measure but is meager.

I have no idea how to solve this problem. Any help, please.

Thanks very much!

$\endgroup$
  • $\begingroup$ The headline at the top of the blog posting seems to be a hint at how to solve this problem, or at least half of it: use the Baire category theorem. $\endgroup$ – Michael Hardy Aug 18 '14 at 1:57
8
$\begingroup$

It suffices to consider the numbers in some bounded interval, such as $[0,1]$. For $\epsilon>0$ and $q\in \mathbb N$ let $$A(\epsilon,q)=\left\{x\in [0,1]: \left|x-\frac{a}{q}\right|<\frac{1}{q^{2+\epsilon}} \ \text{ for some } a \right\}$$ The set $A(\epsilon,q)$ consists of about $q$ intervals of size $2/q^{2+\epsilon}$. So, its measure is about $2/q^{1+\epsilon}$. Since $\sum_{q\in\mathbb N}|A(\epsilon,q)|<\infty$, the Borel-Cantelli lemma implies that the set $A(\epsilon)$ of numbers $x$ that belong to infinitely many of $A(\epsilon,q)$ has measure zero. Consequently, the set $$\mathcal A = \bigcup_{n\in\mathbb N} A(1/n)$$ also has measure zero. By definition, if $x$ is non-Diophantine, there exists $\epsilon>0$ and sequences $\{q_k\}$, $\{a_k\}$ such that $|x-a_k/q_k|q^{2+\epsilon}\to 0$. Hence, $x\in \mathcal A$. This proves that the set of non-Diophantine numbers has measure zero.

Concerning meagerness: for $k\in \mathbb N$ let $$B(k)=\left\{x\in\mathbb R: \left|x-\frac{a}{q}\right|\ge \frac{1}{kq^3} \ \text{ for all } a,q\right\}$$ This set is closed and has empty interior (as it contains no rational numbers). Therefore, the union $\bigcup_{k\in\mathbb N} B(k)$ is meager. This union contains all Diophantine numbers.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Nice solution! Without your solution, it's very difficult for me to construct $A(\epsilon,q)$ and $B(k)$ by myself. When I first saw this exercise, I feel hard to control $\epsilon$, $q$ and $C_{\epsilon}$ simultaneous. Your nice construction solve this problem. Thank you very much! $\endgroup$ – Danielsen Aug 13 '13 at 6:51
  • $\begingroup$ Can you tell more details why $\bigcup_{k\in\mathbb N} B(k)$ contains all Diophantine numbers? Thank you! $\endgroup$ – Danielsen Aug 13 '13 at 7:06
  • $\begingroup$ @Danielsen Because a Diophantine number must satisfy its definition with $\epsilon=1$, among other values of $\epsilon$. I use only $\epsilon=1$. Any positive number $c$ in $\ge \frac{c}{q^3}$ is greater than $1/k$ for some $k$. Hence $x\in B(k)$. $\endgroup$ – user Aug 13 '13 at 12:33
  • $\begingroup$ I got it. Thank you very much! $\endgroup$ – Danielsen Aug 13 '13 at 15:00
  • $\begingroup$ @user Why does $B(k)$ contains no rational number? $\endgroup$ – saurav90 Feb 23 '15 at 19:42

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.