11

Let $\mathbb{I}$ be the set of all irrational real numbers and $\mathbb{Q}$ be the set of all rational numbers as usual. As a subspace of the Euclidean plane $\mathbb{R}^2$, is the set $\mathbb{Q} \times \mathbb{Q} \cup \mathbb{I} \times \mathbb{I}$ disconnected?

The story is the following.

I was making a typical undergraduate level topology problem which concerns connectedness. The question which I concerned is the following. As a subspace of the Euclidean plane $\mathbb{R}^2$, is the set $\mathbb{Q} \times \mathbb{I} \cup \mathbb{I} \times \mathbb{Q}$ connected? What about the set $\mathbb{Q} \times \mathbb{Q} \cup \mathbb{I} \times \mathbb{I}$ ?

The former is relatively well known. For second, the solution which I had in mind turned out to be wrong. I guess lines of the form $y=qx$ where $q$ is some positive rationals will connect the set which is apparently wrong. Now I have no solution. Please give some help to resolve this problem.

seoneo
  • 1,963
  • I couldn't find the former one. Maybe I must have missed this interesting problem. But is it disconnected by the line $y=x$? – Sungjin Kim Oct 23 '19 at 03:49
  • @Sungjin Kim Yes the former is disconnected by the diagonal. – seoneo Oct 23 '19 at 03:52
  • https://math.stackexchange.com/questions/55638/no-continuous-function-switches-mathbbq-and-the-irrationals this solution shows that continuous function $f:\mathbb{R}\rightarrow\mathbb{R}$ switches rational to irrational, irrational to rational, does not exist. But, I am not sure if this helps. – Sungjin Kim Oct 23 '19 at 03:57
  • @SungjinKim Thanks for good comment. The non-existence of such a function is also well known and was a good exercise for undergraduate analysis.(BTW I have seen it on AOPS) I have also reminded the fact and now I suspect it may be connected. But the required logic is missing for me... – seoneo Oct 23 '19 at 04:05
  • See https://math.stackexchange.com/questions/2025632/show-that-bbb-qn-cup-bbb-r-bbb-qn-is-connected – mr_e_man Feb 15 '23 at 22:05

3 Answers3

8

The set is connected.

Suppose that the set is disconnected by disjoint open sets $U$ and $V$. So, we assume that $\mathbb{Q}\times\mathbb{Q} \cup \mathbb{I}\times\mathbb{I}\subseteq U\cup V$ and $U\cap V=\phi$.

Find open balls $B_U \subseteq U$ and $B_V\subseteq V$. Then we can find a line of the form $y=q_1x + q_2$ with $q_1\in \mathbb{Q}-\{0\}$, $q_2\in\mathbb{Q}$ such that the line goes through both balls. Note that this line takes rational to rational, irrational to irrational. Thus, $\{(x,y)|y=q_1x+q_2\}\subseteq\mathbb{Q}\times\mathbb{Q} \cup \mathbb{I}\times\mathbb{I}$.

However, since $U$ and $V$ disconnects the set $\mathbb{Q}\times\mathbb{Q} \cup \mathbb{I}\times\mathbb{I}$, so we can find a point on the line that does not belong to any of $U$ or $V$. This is a contradiction.

Sungjin Kim
  • 20,850
2

Here's a slightly different way to think about it in terms of components:

Suppose $(a,b), (c,d) \in \mathbb{Q}^2.$ Then the lines $$x+y = a+b \\ x-y = c-d$$ are continuous paths in $\mathbb{Q}^2 \cup \mathbb{I}^2 \subset \mathbb{R}^2$ containing $(a,b), (c,d)$ respectively and intersecting at $\left(\frac{1}{2}(a+b+c-d), \frac{1}{2}(a+b-c+d)\right).$ In particular, this shows that $\mathbb{Q}^2$ lies in a single path-component of $\mathbb{Q}^2 \cup \mathbb{I}^2.$ Therefore (since path components refine connected components) there is a connected component $C \subseteq \mathbb{Q}^2 \cup \mathbb{I}^2$ such that $\mathbb{Q}^2 \subseteq C.$

Finally, since $\mathbb{Q}^2$ is dense in $\mathbb{R}^2,$ it is also dense in any subspace containing it, showing that $$C \subseteq \mathbb{Q}^2 \cup \mathbb{I}^2 = \text{cl}(\mathbb{Q}^2) \subseteq \text{cl}(C) = C$$ so $\mathbb{Q}^2 \cup \mathbb{I}^2 = C$ is connected.

2

Let $S = \mathbb{Q}^2 \cup \mathbb{I}^2$. For each nonzero rationals $q$, let $L_q$ be the line through the origin with slope $q$. Then $L_q \subseteq S$. Let $T=\bigcup_{q \in \mathbb{Q}\setminus \left\{0 \right\}} L_q$. Then $T$ is connected in $S$. Moreover $\overline{T}$ relative to $S$ itself is equal to $S$. Therefore $S$ is connected.

seoneo
  • 1,963
  • I have somewhat confused. Now I realize that the original idea using $y=qx$ was not that far from the proof of connectedness by the help of others. Thanks for everyone. – seoneo Oct 23 '19 at 05:29
  • @BrianMoehring Oops! you are right! – seoneo Oct 23 '19 at 05:37