It's a $2$ mark problem in a past paper $3$ years ago and I am about to kill myself for having no clue how to solve even such a basic problem. Regretting taking topology so much.

Show that $X=\{x_0,x_1,x_2\}$ with three points and the discrete topology in not contractible. Your proof should include a statement of the property of the unit line interval $I$.

Well, damn. To those used to it it may seem like a piece of cake but to me, it's a giant boulder rolling down a mountain.

So, for some $X$ to be contractible, I saw multiple equivalent statements on wikipedia (though I have no idea whatsoever as to why they are equivalent)

  1. $X$ is contractible
  2. $X$ is homotopy equivalent to $\{0\}$
  3. For any $Y$, $f,g:Y \to X$ are homotopic
  4. The identity map on $X$ is homotopic to some constant map.

Say, condition $4$ and $3$. Well, why are they the same thing?

Anyway, I've tried showing this problem by sticking to the definition I was first give in class, condition $2$. So I need to find some $f,g$ such that $f:X \to Y$, $g:Y \to X$ where $fg \simeq id_Y$ and $gf \simeq id_X$.

So, in this case $Y=\{0\}$ and I have to show that there are no such $f,g$s possible between $X,Y$, yes? But then, I don't see how I can prove this. Moreover, intuitively, I feel like it might even be possible to find such $f,g$.

Plus, where does the "property of the unit line interval $I$" come in to play? It's not related to my approach(if it can be called that).

I'm confused, does anyone know a concise way to clear this up?

  • $\begingroup$ I think the most basic definition is : the identity is homotopic to a constant map. Try to write what it means that the identity is homotopic to the map constant to some $x_i$ from the definition of a homotopy. $\endgroup$ Mar 30 '16 at 11:43

The property of the unit interval $I$ comes into play when you expand "$\simeq$".

Assume we have such $f,g$ where $X$ is any discrete space (i.e., the threeness of points is of little relevance) and $Y=\{0\}$. Then there is a homotopy $H\colon X\times I\to X$ such that $H(x,0)=gf(x)$ and $H(x,1)=x$. As $f,g$ are necessarily constant (becaue $|Y|=1$), we have $gf(x)=x_0:=g(0)$ independently of $x$. For fixed $x\in X$, the map $\gamma\colon I\to X$, $t\mapsto H(x,t)$ is continuous. As $I$ is connected and $X$ is discrete, $\gamma$ must be constant. In particular, $x=\gamma(1)=\gamma(0)=x_0$ for all $x$. We conclude that a discrete space $X$ can be contractible only if $|X|=1$.


You might have learned the fact that the fundamental group is a homotopy invariant of a path connected space.

Even more elementary than that is the fact that the cardinality of the set of path components is a homotopy invariant of any space.

A three point discrete space has three path components, whereas a one point space has one path component, so they are not homotopy equivalent. Contractibility means, by definition, that the space is homotopy equivalent to a one point space. Hence, a three point discrete space is not contractible.


Let's use condition 3 and show the two maps $f,g : X \to X$ defined by $f(x_1)=f(x_2)=f(x_3)=x_1$ and $g(x_1)=g(x_2)=g(x_3)=x_2$ are not homotopic.

If the maps were homotopic, we would have a map $H : X \times I \to X$ such that $H(x,0)=x_1$ and $H(x,1)=x_2$ for all $x \in X$. Now the domain has exactly $3$ components, namely $C_1 = \{x_1\} \times I$, $C_2 = \{x_2\} \times I$ and $C_3 = \{x_3\} \times I$. Since $H$ is continuous each $H(C_i)$ is a connected subset of $X$. But since $X$ is discrete its only connected subsets are the singletons.

We conclude $H(C_1) = \{x_i\}$ for some $i=1,2,3$. But since $(x_1, 0) \in C_1$ the assumption that $H(x,0)=x_1$ implies $i=1$. So $H(C_1) = \{x_1\}$. For the same reason $H(C_2) = \{x_1\}$. But this contradicts the assumption that $H(x,1)= \{x_2\}$.


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.