Let $X$ be a compact metric space satisfying the following condition: for any given positive number $\delta>0$, only finitely many components of $X$ have diameter larger than $\delta$.

For a given component $P$ of $X$, let us remove a connected compact subset $C$ of $P$ such that $P\setminus C$ is disconnected. Consider a disjoint partition $P\setminus C=A\cup B$, with $A$ and $B$ at a positive distance. Suppose that $x\in A$ and $y\in B$.

Are then $x$ and $y$ in different quasi-components of $X\setminus C$? or can we find a separation of $X\setminus C=M\cup N$ such that $x\in M$ and $y\in N$?

  • $\begingroup$ Can you clarify whether we can assume $x\in A$ and $y\in B$? Or is it only that a partition $A\cup B$ of $P\setminus C$ exists, and $x,y$ might both be in $A$? $\endgroup$ Commented Jul 30, 2020 at 10:09
  • $\begingroup$ sorry for the confuse, actually, I want to express that $x\in A$ and $y\in B$. $\endgroup$
    – Yee Neil
    Commented Jul 30, 2020 at 10:29

2 Answers 2


(This answers the previous version of the question, for components but not quasicomponents).

I think, it us true in every topological space: if $X$ is a topological space with connected components $P$ and $Q_j$: $X=P\sqcup (\sqcup_j Q_j)$, and you remove a set $C$ from a connected component $P$ so that it becomes disconnected: $P\setminus C=\sqcup A_i$, where $A_i$ are the connected components of $P\setminus C$, then the connected components of $X\setminus C$ are $A_i$'s and $Q_j$'s. The reason is that the connected components are the inclusion-maximal connected subsets, and each connected subset of $X\setminus C$ lies inside a single connected component of $X$.

  • $\begingroup$ what are the sets A_i'? $\endgroup$
    – Yee Neil
    Commented Jul 30, 2020 at 7:55
  • $\begingroup$ connected components of $P\setminus C$ $\endgroup$ Commented Jul 30, 2020 at 7:56
  • $\begingroup$ If P is a segment with length 1, and there is a sequence of segmenst with length 1 which are all parallel to $P$ converging to $P$ , $X$ is the union of these segmetns. Then we remove the midpoint $p$ of $P$, now $P\setminus p$ is disconnected, but $P\setminus p$ is thecomponent of the $X$. $\endgroup$
    – Yee Neil
    Commented Jul 30, 2020 at 8:02
  • $\begingroup$ You probably mean that $P\setminus p$ is the component of $X\setminus p$, but it is not the case: the components must be connected, and $P\setminus p$ is not. $\endgroup$ Commented Jul 30, 2020 at 8:08
  • 1
    $\begingroup$ @YeeNeil I think you might be talking about quasi-components. Your example is example 2 in this wikipedia page: en.wikipedia.org/wiki/Locally_connected_space#Quasicomponents However, connected components really are always connected. They are precisely the connected subsets which are maximal under the inclusion ordering. $\endgroup$ Commented Jul 30, 2020 at 9:05

Your compactum $X$ is finitely Suslinian (by defnition). As such, every connected subset of $X$ is locally connected (see this paper), and this implies that the components of $X\setminus C$ are the same as the quasi-components of $X\setminus C$. See this paper, Theorem 2.1. @Fedor Petrov already argued that $x$ and $y$ are in different components of $X\setminus C$, therefore they are also in different quasi-components.

Note that $C$ does not have to be connected or compact. It can be any subset of $P$ such that $P\setminus C$ is disconnected.

  • $\begingroup$ Yeah, as far as I can see, the idea is that finitely Suslinian implies hereditarily locally connected, and then hereditarily locally connected implies (q=c) according to Theorem 2.1. But I wonder, by the definition in my question, can we say that $X$ is finitely Suslinian, note that we only requirement that the components with diameter than a given positive number $\delta>0$ is finitely, and this of course different with that a sequence a pairwise disjoint subcontinua form a null sequence. $\endgroup$
    – Yee Neil
    Commented Jul 31, 2020 at 0:55
  • $\begingroup$ @YeeNeil Your objection is perfectly valid. But I think that a more elementary approach can be taken to show that $x$ and $y$ are in different quasi-components. I will update my answer when I get the chance. $\endgroup$ Commented Aug 5, 2020 at 22:15
  • $\begingroup$ that sounds good, thanks for your works! $\endgroup$
    – Yee Neil
    Commented Aug 8, 2020 at 23:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.