15
$\begingroup$

I need to show that any open interval is homeomorphic to the real line.

I know that $f(x)=a+e^x$ will work for the mapping $f:R \to (a,\infty)$ and $f(x)=b-e^{-x}$ will work for the mapping $f:R \to (-\infty,b).$

Without using two functions, how can I prove the result in general?

$\endgroup$
1
  • $\begingroup$ Why not use these two functions? $\endgroup$
    – user14972
    Nov 23, 2012 at 2:21

3 Answers 3

Reset to default
18
$\begingroup$

Let $$ f(x) = \frac{x}{x^2-1}. $$ This is a homeomorphism from $(-1,1)$ to $\mathbb R$.

Let $$ g(x) = \frac{1}{1+2^{-x}}. $$ That is a homeopmorphism from $\mathbb R$ to $(0,1)$.

For any other bounded interval $(a,b)$, just rescale and relocate.

$\endgroup$
5
  • $\begingroup$ To show that the real line is homeomorphic to any inteval,I need to show that (0,1) is homeomorphic to R and any two intervals are homeomorphic.so that I can find a composition function then that is a homeomorphism.So can you please help me with this idea? $\endgroup$
    – ccc
    Nov 24, 2012 at 8:57
  • 5
    $\begingroup$ @ccc : Let $\ell(x)=a(1-x)+bx$. Then $\ell$ is a homeomorphism from $(0,1)$ to $(a,b)$. (That is what I meant by "rescale and relocate".) Thus every open interval is homeomorphic to $(0,1)$. And since the function $g$ in the answer is a homeomorphism from $\mathbb R$ to $(0,1)$, the composition gives a homeomorphism from $\mathbb R$ to $(a,b)$. $\endgroup$
    – Michael Hardy
    Nov 24, 2012 at 15:55
  • $\begingroup$ thanks Michael,I understand the method,but do I need to define a function for the composition,i.e from R to (a,b)? $\endgroup$
    – ccc
    Nov 25, 2012 at 8:24
  • 6
    $\begingroup$ You have $g(x)=\dfrac{1}{1+2^{-x}}$ going from $\mathbb R$ to $(0,1)$ and $\ell(x)=a(1-x)+bx$ going from $(0,1)$ to $(a,b)$, so $\ell(g(x)) = a\left(1-\dfrac{1}{1+2^{-x}}\right)+b\left(\dfrac{1}{1+2^{-x}}\right)$ goes from $\mathbb R$ to $(a,b)$. With a bit of algebra you can simplify that last expression somewhat. $\endgroup$
    – Michael Hardy
    Nov 25, 2012 at 19:33
  • $\begingroup$ The first one is not even injective $\endgroup$
    – quantum
    Mar 22 at 10:16
13
+50
$\begingroup$

You've already been given two possible homeomorphisms, but how about another one?

Say that you have two maps $\varphi : A \to B$ and $\psi : B \to C$, both of which are homeomorphisms. It's clear that $\psi \circ \varphi : A \to C$ is again a homeomorphism.

Using this fact, choose your favorite finite open interval $(a,b)$, and prove it is homeomorphic to $\mathbb{R}$.

Next up, take an arbitrary open interval $(c,d)$, and construct a homeomorphism between this an $(a,b)$, and voila, you are done.

In particular, look at the interval $(0,1)$, and its image under the function $\tan(\pi(x-\frac{1}{2}))$. This is pretty clearly a homeomorphism. Now just map an open interval to $(0,1)$ (homeomorphically), and call it a day.

$\endgroup$
3
$\begingroup$

Consider the function $\log(b-x)-\log(x-a)$

$\endgroup$
2
  • $\begingroup$ how can we show that the inverse is continuous here? $\endgroup$
    – ccc
    Nov 23, 2012 at 0:21
  • 1
    $\begingroup$ @ccc: Solving $y = \log(b-x) - \log(x-a)$ for x would work. $\endgroup$
    – user14972
    Nov 23, 2012 at 2:18

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.