Let $\{A_i\}_{i\in I}$ be a family of subsets of $\Bbb R^2$ (where $I=\Bbb N$ or $\Bbb Z$; I don't know if it makes a difference) such that

  • $\bigcup_{i\in I} A_i=\Bbb R^2$
  • $i\ne j\implies A_i\cap A_j=\emptyset$
  • $A_i\ne\emptyset$
  • $A_i$ is connected
  • $A_i\cup A_{i+1}$ is connected

How often can it happen that $A_i\cup A_j$ is connected for $j\notin\{i-1,i,i+1\}$?

Definition. Let's say that an index $n\in I$ is infinitely linked/almost completely linked/completely linked if $A_n\cup A_i$ is connected for infinitely many/almost all/all $i\in I$.

One possibility configuration is to let the $A_i$ be vertical stripes, in which case no $n\in I$ is infinitely linked. With another configuration one can achieve that there exists exactly one $n\in I$ that is completely linked (let $A_n=\{(0,0)\}$ and all other $A_i$ suitable sectors of $\Bbb R^2\setminus\{(0,0)\}$). Can more than that be achieved?

  • I.e., are there configurations with more (two, three, arbitrarily many, infinitely many, almost all, all) completely linked indices?
  • Or maybe at least more almost completely linked indices?
  • For $I=\Bbb Z$, I can find configurations two infinitely linked indices. Are three or more infinitely linked indices possible?
  • Are two infinitely linked indices possible with $I=\Bbb N$?

Note that with $\Bbb R^3$ instead of $\Bbb R^2$ one can easily achieve that all indices are completely linked.

  • $\begingroup$ It doesn't matter whether the index set is $N$ or $Z$ as they are both countably infinite. You could just say " $A$ is a countably infinite cover of $R^2$ ... ("cover" means $ \cup A\supset R^2$)... of pair-wise disjoint non-empty connected subsets of $R^2$ , such that the union of any 2 of them is connected". I like the q. $\endgroup$ – DanielWainfleet Nov 5 '15 at 21:01
  • 1
    $\begingroup$ @user254665 If my condition about $A_i\cup A_{i+1}$ were important, it would matter. It is only by the (currently two) answers that all $A_i\cup A_j$ can be made connected, thus making the order type imposed irrelavant. $\endgroup$ – Hagen von Eitzen Nov 5 '15 at 21:09

Using the axiom of choice, you can partition $\mathbb{R}^2$ into $2^{\aleph_0}$ sets $A_i$ such that any union of the $A_i$ is connected. Indeed, there are only $2^{\aleph_0}$ uncountable subsets of $\mathbb{R}^2$ that are either open or closed and each of them has cardinality $2^{\aleph_0}$, so by a straightforward diagonalization argument (very similar to the argument here, for instance), you can partition $\mathbb{R}^2$ into $2^{\aleph_0}$ disjoint sets $A_i$ with the property that each of them intersects every uncountable open or closed subset of $\mathbb{R}^2$.

Now suppose some $A_i$ were disconnected. Then there are open subsets $U,V\subset\mathbb{R}^2$ such that $A_i\subset U\cup V$, $U\cap A_i\neq\emptyset$, $V\cap A_i\neq\emptyset$, and $U\cap V\cap A_i=\emptyset$. Since $A_i$ intersects every nonempty open set, $U\cap V$ must be empty. But then $U\cup V$ is disconnected, and hence $\mathbb{R}^2\setminus (U\cup V)$ is an uncountable closed set (as the complement of any countable subset of the plane is connected). So $A_i$ must intersect $\mathbb{R}^2\setminus (U\cup V)$, which is a contradiction. By the same argument, any union of the $A_i$ (indeed, any set containing any $A_i$) is also connected.

(This is essentially counterexample 124 in Counterexamples in Topology, though there they only construct two such sets $A_i$.)

  • $\begingroup$ Although it is probably elementary, how do you show that $R^2\backslash (U\cup V)$ is uncountable$ And the remark in brackets ("as the complement of any countable...") seems to me to be irrelevant. $\endgroup$ – DanielWainfleet Nov 5 '15 at 20:53
  • $\begingroup$ If $\mathbb{R}^2\setminus(U\cup V)$ were countable, then its complement, $U\cup V$, would be connected. I have added a reference for the parenthetical remark. $\endgroup$ – Eric Wofsey Nov 5 '15 at 21:28
  • $\begingroup$ Thanks for clearing that up. I expected it would be simple. $\endgroup$ – DanielWainfleet Nov 5 '15 at 21:42
  • 1
    $\begingroup$ “…by a simple diagonalisation argument…” Maybe elaborate on this a little? The simplest version I can see relies two fairly non-trivial facts: that there are only $2^{\aleph_0}$-many uncountable open or closed sets, so that they can be enumerated correctly (not hard to prove, but not obvious to state); and that any uncountable closed set must have cardinality $2^{\aleph_0}$, so that it does not get exhausted at any stage (not at all trivial or obvious, if you don’t already know it). $\endgroup$ – Peter LeFanu Lumsdaine Nov 7 '15 at 12:33
  • $\begingroup$ @PeterLeFanuLumsdaine: Yes, that's the argument I had in mind. I'll add some more detail. $\endgroup$ – Eric Wofsey Nov 7 '15 at 17:02

In On a connected dense proper subgroup of $\mathbb R^2$ whose complement is connected, Ryuji Maehara constructs a connected dense proper subgroup of $\mathbb R^2$. If we take the collection of all cosets of this subgroup, we get a partition into (continuum many) disjoint connected dense subsets. It is standard that the union of any number of these is connected. In particular, we can lump them together into $\mathbb N$-many subsets which are still dense and connected, and the union of any two of these will be connected.

  • $\begingroup$ So I take it that Maehara's group as constructed in the ref is of continuum index? $\endgroup$ – Hagen von Eitzen Nov 5 '15 at 18:57
  • 1
    $\begingroup$ @HagenvonEitzen I should have said. The group is constructed as the graph of an additive function, so it has continuum index. $\endgroup$ – John Gowers Nov 5 '15 at 19:06
  • $\begingroup$ @Donkey_2009. I take it that the additive group is $\{(x,f(x) :x\in R\}$ where $f$ is any additive nowhere-continuous function? $\endgroup$ – DanielWainfleet Nov 5 '15 at 21:09
  • 1
    $\begingroup$ @user254665 I don't know that any additive nowhere-continuous $f$ will do. The $f$ given in the paper is constructed so that its graph intersects every closed subset of the plane whose projection on to the $x$-axis contains some open interval. $\endgroup$ – John Gowers Nov 5 '15 at 22:00
  • 1
    $\begingroup$ In fact, the additiveness of the function isn't necessary (since we can just take translates up and down). So maybe this answer is in fact just adding a whole lot of useless machinery to Eric's answer. I'll have a think, though. $\endgroup$ – John Gowers Nov 6 '15 at 0:35

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.