Suppose $X$ is connected and locally path connected, then $X$ is path connected. Proof BWOC, let $Y$ is path component poper subset of $X$. Since $X$ is locally path connected then $Y$ is open and $X-Y$ is open. Then $Y$ is closed which is contradiction because $X$ is connected.

My question how can I prove it directly from the definition without depending on theorem.

  • $\begingroup$ I upvote, not because of the question, but because of the BWOC proof. $\endgroup$ – user5280911 Jan 8 at 4:27

Let $X$ be the space and fix $p \in X$. Let $C$ be the set of all points in $X$ which are path connected to $p$. Now, it is enough to show that $C$ is closed and open in $X$, to show that $C=X$ (given that $p \in C$ via a constant path, so $C$ is non-empty).

To show $C$ is open, let $c \in C$, then we can choose (by local path connectedness) an open subset $U$ containing $c$. For $u \in U$, $u$ is path connected to $c$ which is path connected to $p$, so by joining paths, we have that $u$ is path connected to $p$. In other words, $U \subset C$, so $C$ is open.

Look at the closure of $C$, namely $\bar{C}$. Let $d \in \bar{C}$, and choose, like above, an open path connected subset $V$ containing $d$. Note that $V \cap C \neq \phi$, because $V$ is open! Hence let $e \in V \cap C$, then $d$ is path connected to $e$ which is path connected to $p$ (because $e \in C$)! Hence $d$ is path connected to $p$ and $d \in C$, so $C=\bar{C}$, $C$ is closed.

Hence $C=X$, and $X$ is path connected.

EDIT : In the style of equivalence relations , we can create another proof. Indeed, define an equivalence relation on $X$ given by $p \sim q$ if there is a path connecting $p$ and $q$. It is easy to see that this is an equivalence relation : for reflexivity, use the constant path. For symmetry, use the reverse of a path, and for transitivity, use the concatenation of the two paths.

Finally, note that every equivalence class is open, because if I take $p$, then by local path connectedness, some neighbourhood of $p$ is path connected, but this includes $p$. Therefore, this entire neighbourhood is in the same equivalence class as $p$. The openness follows.

Now, the equivalence classes partition $X$ into disjoint open sets. Since $X$ is connected, this can't happen unless there's only one equivalence class, which is the whole of $X$. We are done!

  • $\begingroup$ Do you mean that the open sets $U$ and $V$ in your answer are path-connected? $\endgroup$ – Oskar Henriksson Mar 8 '18 at 13:22
  • $\begingroup$ Yes, they must be path connected. $\endgroup$ – Teresa Lisbon Mar 10 '18 at 2:41
  • 3
    $\begingroup$ A quite obvious remark: The connectedness of $X$ enters the argument at the very end, when concluding from the fact that $C$ is open and closed, that $C=X$. $\endgroup$ – klirk Aug 24 '19 at 19:00
  • $\begingroup$ @klirk True, thank you for that. $\endgroup$ – Teresa Lisbon Aug 25 '19 at 7:15
  • $\begingroup$ @астонвіллаолофмэллбэрг, to show $C$ is closed could we argue in the same way as previously? $X \backslash C$ is opend for if we take $q \in X\backslash C$, there is an open set $U$ containing q which is path connected. $U\cap C = \emptyset$ because otherewise we can find $r$ in the intersection which is path connected to $p$ and path connected to q which is not possible. $\endgroup$ – roi_saumon Apr 27 '20 at 18:30

I recommend looking up some of the equivalent definitions of connectedness, it is a useful exercise to prove the equivalences:

enter image description here

In particular characterization (6) is what you're looking for.

If I recall of the many definitions I found this equivalence the hardest to prove (one direction is straight-forward, the other a little tricky), it helped to draw a picture. Anyways, given for now the equivalence, the proof you're asking about is (hopefully) almost immediate. To begin, fix two points in X. Around each point in X you can take a path connected open set, this forms an open cover of X, which admits a finite subcover with the property that "adjacent" open sets are non-empty and still path-connected so you can glue together a path.

  • $\begingroup$ If this theorem is taken from book, if yes , could you give the name of this book becuse may be will ever helpfuly in the connected $\endgroup$ – Gob Aug 2 '16 at 0:02
  • $\begingroup$ It's from a set of notes from an old professor of mine. The definition of connected given is "X is called connected if X is not the union of two disjoint non-empty open sets." The notes don't supply the proofs, we were expected to do so ourselves. $\endgroup$ – mb- Aug 2 '16 at 0:05

Let $x\in X$, consider $C$ such that for every $y\in C$ there exists a path between $x\in Y$. Let's show that $C=X$, suppose $C\neq X$, remark that $C$ is open: $y\in C$, there exists $y\in U$ $U$ open subset such that for every $z\in U$ there is path $c$ between $z$ and $y$. Compose $c$ with the path between $x$ and $y$ to obtain a path between $x$ and $z$.

$C$ is closed, let $y\in \bar C$ the adherence of $C$, you have again an open subset $y\in V$ path connected $V\cap C$ is not empty, take $z\in V\cap C$ compose a path between x and z with a path between z and y.

$C$ is open and closed, $C=X$ since $X$ is connected.

  • $\begingroup$ It is obvious now thank you $\endgroup$ – Gob Aug 2 '16 at 0:06

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.