Why doesn't this construction make the rational numbers a $G_\delta$ set?

240 Views Asked by At

A $G_\delta$ set is a countable intersection of open sets. The set of rational numbers is not a $G_\delta$ set.

But I'm wondering why the following construction doesn't make it a $G_\delta$ set. Let $(q_n)$ be a sequence enumerating the rational numbers. For any natural number $m$, let $U_m = \cup_n (q_n-\frac{1}{2^{m+n}},q_n+\frac{1}{2^{m+n}})$, and let $U = \cap_m U_m$. Then it seems to me that $U_m$ is an open set for all $m$, and $U$ is the set of rational numbers. So the set of rational numbers is a $G_\delta$ set.

Where is the flaw in my reasoning?

1

There are 1 best solutions below

5
On BEST ANSWER

There are many ways to enumerate the rational numbers. Since you did not specify an enumeration, let me pick one for you.

For each positive integer $m$, choose a rational number $q_{2m}$ so that $|q_{2m}-\sqrt2|\lt\frac1{2^{3m}}$. In this way we get a sequence $q_2,q_4,q_6,\dots$ of rational numbers converging to $\sqrt2$. Enumerate the rest of the rational numbers any way you like in a sequence $q_1,q_3,q_5,\dots$. Conbining these, we get an enumeration $q_1,q_2,q_3,\dots$ of the set of all rational numbers.

Note that, for each positive integer $m$, we now have $$\sqrt2\in\left(q_{2m}-\frac1{2^{m+2m}},\ q_{2m}+\frac1{2^{m+2m}}\right)\subseteq\bigcup_{n=1}^\infty\left(q_n-\frac1{2^{m+n}},\ q_n+\frac1{2^{m+n}}\right)=U_m,$$ so $\sqrt2\in\bigcap_{m=1}^\infty U_m$, although $\sqrt2$ is not a rational number.