7
$\begingroup$

Let $n$ be a positive integer and let $K\subseteq \mathbb{R}^n$ be compact. Pick $x^* \in \mathbb{R}^n\setminus K$.

Let $E$ be the connected component of $\mathbb{R}^n\setminus K$ that contains $x^*$. Let ${\cal C}$ be the collection of connected components of $K$. For each $C\in {\cal C}$ let $E_C$ be the connected component of $\mathbb{R}^n\setminus C$ that contains $x^*$.

Is it true that $E=\bigcap\{E_C: C\in{\cal C}\}$?

$\endgroup$
4
  • 2
    $\begingroup$ Note that this statement is false if one replaces "connected component" by "path-connected component". $\endgroup$ Feb 11, 2018 at 9:51
  • $\begingroup$ @BenoîtKloeckner isn't for open subsets of locally path-connected space all notions related to connectivity and path-connectivity coincide? $\endgroup$
    – erz
    Feb 12, 2018 at 13:22
  • $\begingroup$ @erz: yes, but the question includes connected components of the compact set $K$. $\endgroup$ Feb 12, 2018 at 21:21
  • $\begingroup$ @BenoîtKloeckner oh yes, of course, sorry. $\endgroup$
    – erz
    Feb 12, 2018 at 23:29

1 Answer 1

5
$\begingroup$

The answer to this problem is Yes.

Indeed, the inclusion $E\subset \bigcap_{C\in\mathcal C}E_C$ is trivial, so it remains to prove that for any point $x\in\mathbb R^n\setminus E$ there exists a connected component $C\in\mathcal C$ of $K$ such that $x\notin E_C$. By Zorn's Lemma, the compact set $K$ contains a minimal closed subset $S\subset K$ such that the points $x^*$ and $x$ belong to distinct components of $X\setminus S$. Such a minimal compact set $S$ is called an irreducible barrier between $x$ and $x^*$. The Claim below implies that the irreducible barrier $S$ is connected. Then for the connected component $C$ of $K$ containing the irreducible barrier $S$ we get $x\notin E_C$.

Claim. The irreducible barrier $S$ is connected.

Proof. The proof involves some tools of algebraic topology. I do not know if there is a more elementary proof. Assuming that $S$ is disconnected, we can write $S$ as the disjoint union $A\cup B$ of two non-empty compact sets. Since $S$ is a minimal barrier between $x$ and $x^*$ the points $x,x^*$ can be linked by a path in $\mathbb R^n\setminus A$ and $\mathbb R^n\setminus B$. Since $x,x^*$ belong to different connected components, the 0-dimensional singular cycle $\alpha=x-x^*$ detemines a non-zero-element in the homology group $H_0(\mathbb R^n\setminus S)$. Now observe that $(\mathbb R^n\setminus A)\cup(\mathbb R^n\setminus B)=\mathbb R^n\setminus(A\cap B)=\mathbb R^n$ and write the exact Mayer-Vietoris sequence $$H_0(\mathbb R^n)\to H_0(\mathbb R^n\setminus S)\to H_0(\mathbb R^n\setminus A)\oplus H_0(\mathbb R^n\setminus B).$$ By the minimality of $S$, the image of the singular cycle $\alpha$ in $H_0(\mathbb R^n\setminus A)\oplus H_0(\mathbb R^n\setminus B)$ is trivial. Now the exactness of the Mayer-Vietros sequence implies that $\alpha$ is zero in $H_0(\mathbb R^n\setminus S)$, which is a desired contradiction.

$\endgroup$
4
  • 1
    $\begingroup$ Wow - amazing, thanks Taras!! This question has been open on MO for almost 3 years. I didn't get a handle on it myself at all. Thanks again! $\endgroup$ Feb 11, 2018 at 15:38
  • 1
    $\begingroup$ @DominicvanderZypen I noticed that the highest probability to get an answer is in several days after posing the problem. Concerning this your question, I think we have not so much good continuum theorists here in mathoverflow (irreducible barriers should be well-known to specialists, I hope). The same concerns general topology, it turns out that this field is far from being overcrowded and the number of topologists seems to decrease in time, unfortunately. $\endgroup$ Feb 11, 2018 at 16:34
  • 1
    $\begingroup$ @DominicvanderZypen Dominic, I am afraid that you have not noticed that I wrote anwers to other two problems posed by you: (mathoverflow.net/questions/289740/…) and (mathoverflow.net/questions/223909/…). The last problem was really difficult! $\endgroup$ Feb 12, 2018 at 18:34
  • $\begingroup$ Oh - I haven't noticed, sorry for my mistake @TarasBanakh! Will have time to read your answers later this afternoon $\endgroup$ Feb 13, 2018 at 12:54

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.