Notices
Results 1 to 10 of 10

Thread: Connected but not path-connected

  1. #1 Connected but not path-connected 
    Forum Freshman Faldo_Elrith's Avatar
    Join Date
    Jul 2008
    Posts
    76
    Can you give an example of a topological space that's connected but not path-connected?


    Reply With Quote  
     

  2.  
     

  3. #2  
    Forum Professor serpicojr's Avatar
    Join Date
    Jul 2007
    Location
    JRZ
    Posts
    1,069
    http://en.wikipedia.org/wiki/Connect..._connectedness

    The extended long line and the topologists's sine curve are examples listed on this page.


    Reply With Quote  
     

  4. #3  
    Moderator Moderator
    Join Date
    Jun 2005
    Posts
    1,628
    Yeah, I looked at this question, but the only counter-examples I could come up with with were somewhat pathological. Like the topologists' sine curve that serpicojr mentioned.
    Can I explain why it's a counter-example? . Um, no, not without a lot of work....

    P.S; And even then I wonder if this is true if our space has a stricter separation axiom than that of Hausdorff, i.e. Tn > T2?

    Anyway, I would say it is safe to assume, for all useful purposes, that, in topological terms, connected means path-connected.

    Now this is not to say a (path) connected space is always arc-wise connected, this is a much stronger requirement. That is, a path-connected space may have "holes", an arc-wise connected space may not
    Reply With Quote  
     

  5. #4  
    Forum Freshman Faldo_Elrith's Avatar
    Join Date
    Jul 2008
    Posts
    76
    How to prove that the topologist's sine curve is connected but not path-connected? That's what I wanna know!
    Reply With Quote  
     

  6. #5  
    Moderator Moderator
    Join Date
    Jun 2005
    Posts
    1,628
    Actually, the proof is not as hard as I had at first thought. I'm not going to do it for you (at least not yet). Try it using the following heavy hints and show us what you get.

    1) Construct the standard topology on R<sup>2</sup>.

    2) Explain why this means this space is connected

    3) Identify a subset of R<sup>2</sup> that will give you the curve we seek. Well, OK, I'll give you this; noting that the open sets in R<sup>2</sup> are the open intervals (a, b),...., you will want A = (0, sin 1/x) as a subset of R<sup>2</sup>. (What restrictions do you need to place on this set?)

    4) Note that any subset of a connected space is connected

    5) Find the closure B of A; identify the boundary of B; give an algebraic interpretation of this guy

    6) Note that the closure of a connected set is connected

    7) finally argue that, even though B is connected, it cannot be path connected (easily seen if you get (3) right!)

    8) Good luck!
    Reply With Quote  
     

  7. #6  
    Forum Professor serpicojr's Avatar
    Join Date
    Jul 2007
    Location
    JRZ
    Posts
    1,069
    Quote Originally Posted by Guitarist
    4) Note that any subset of a connected space is connected
    So the disjoint union of the intervals [0,1] and [2,3] is connected because it sits inside of R?
    Reply With Quote  
     

  8. #7  
    Moderator Moderator
    Join Date
    Jun 2005
    Posts
    1,628
    Yeak, Ok, clever clogs, I was wrong. Scrub (4), and use this

    if A is connected, then if A ⊆ B and B⊆ A<sup>-</sup> (this is the closure of A), then B is connected, and A<sup>-</sup> is connected.

    Rotter. Not talking to you again........

    PS, point (4) was not essential for the proof, only the usual definition of connectedness
    Reply With Quote  
     

  9. #8  
    Forum Professor serpicojr's Avatar
    Join Date
    Jul 2007
    Location
    JRZ
    Posts
    1,069
    Just trying to help.
    Reply With Quote  
     

  10. #9  
    Forum Ph.D.
    Join Date
    Apr 2008
    Posts
    956
    Okay folks, this is how you show connectedness.

    Suppose (contrariwise) that the set S = {(0,0)} ∪ {(x,sin(1⁄x))*:*x ℝ<sup>+</sup>} is disconnected. Let A and B nonempty disjoint open (relative to S) sets such that AB = S with (0,0) ∊ A.

    Now let t = inf{x > 0*:*(x,sin(1⁄x)) ∊ B} (which exists). Since (0,0) ∊ A and A is open and disjoint with B, there is an ε > 0 such that the intersection of B<sub>ε</sub>(0,0) (the open ball with centre (0,0) and radius ε) with S is a subset of A and thus disjoint with B. Moreover, B<sub>ε</sub>(0,0) ∩ S contains points in S other than (0,0) due to the behaviour of the sine curve in the vicinity of (0,0). It follows that t > 0. Hence the point s = (t,sin(1⁄t)) ∊ S.

    So either sA or sB. If A, then the openness of A means there is an ε<sub>1</sub> > 0 such that B<sub>ε1</sub>(s) ∩ SA. But since t+ε<sub>1</sub> is not a lower bound, there must be a t<sub>1</sub> with t < t<sub>1</sub> < t+ε<sub>1</sub> such that (t<sub>1</sub>,sin(1⁄t<sub>1</sub>)) ∊ B – which is a contradiction. OTOH if sB, we have an ε<sub>2</sub> > 0 such that B<sub>ε2</sub>(s) ∩ SB – meaning that there are points t<sub>2</sub> between t−ε<sub>2</sub> and t such that (t<sub>2</sub>,sin(1⁄t<sub>2</sub>)) ∊ B and contradicting the fact that t is a lower bound.

    So we get a contradiction either way. Hence S cannot be disconnected.

    As for S not being path connected, sorry, I don’t know how to prove that.
    Reply With Quote  
     

  11. #10  
    Forum Professor serpicojr's Avatar
    Join Date
    Jul 2007
    Location
    JRZ
    Posts
    1,069
    Quote Originally Posted by JaneBennet
    As for S not being path connected, sorry, I don’t know how to prove that.
    Hint: Suppose a path exists between (0,0) and a point (x,sin(1/x)), i.e. some continuous map f(t) from [0,1] to the curve so that f(0) = (0,0) and f(1) = (x,sin(1/x)). First, show that the path must go through every point (x',sin(1/x')) for each 0 < x' < x. Then, say, project the path to the y-axis--i.e., if f(t) = (x(t),sin(1/x(t)), then the projection is the function sin(1/x(t)). Finally, show that this can't be continuous at t = 0.
    Reply With Quote  
     

Bookmarks
Bookmarks
Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •