Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am reading through the proof of Borsuk - Ulam in Bredon, I have appended the proof below for reference. There are two things I don't understand in this proof:

1) In the first picture below, he writes " For any simplex $\sigma : \Delta_p \to X$, the simplex $g \circ \sigma$ is distinct from $\sigma \ldots $..... There are exactly two such liftings of the form $\sigma$ and $g \circ \sigma$." Now I understand why there are exactly two liftings but why are they of the form $\sigma$ and $g \circ \sigma$? I guess this comes to knowing why the simplices of $X$ fall into two types, which I don't understand. Is there something extra I'm not knowing?

2) I understand what the proof says, but what is boggling me is where has the assumption that $n > m$ been used? It seems to me that the exact same proof could be used to derive a contradiction if we assume $m > n$.


share|cite|improve this question
If $\tau$ is a simplex in $Y$ and $\sigma$ is a lifting of $\tau$ to $X$, then $g\circ\sigma$ is another lifting of $\tau$. Don't think there's any more to it than that. – Einar Rødland Sep 12 '12 at 0:57
Sorry, but not only is my homology rusty after years of disuse, I was never used to doing it over $\mathbb{Z}_2$. Maybe, if my mind clears up a little, I'll see it. However, I note that it suffices to prove the results for $n=m+1$ since the result will always apply to an $S^{m+1}\subset S^n$. – Einar Rødland Sep 12 '12 at 15:41
You have a pdf copy of Brendon? I can't believe I paid so much for the hard cover now. – Holdsworth88 Sep 12 '12 at 22:32
@benjalim I would really appreciate that. My email is in my profile. – Holdsworth88 Sep 13 '12 at 5:56
oh, sorry, @benjalim, it's Sorry it took so long, I was in the middle of nowhere in Bavaria. Thanks again. – Holdsworth88 Sep 17 '12 at 13:52
up vote 3 down vote accepted

I have worked on 2) and this is why I think it does not work if we try to derive a contradiction from assuming $m > n$. Now in the long exact sequence above we would have

$$H_m(P^n;\Bbb{Z}/2\Bbb{Z}) = \ldots = H_{n+1}(P^n;\Bbb{Z}/2\Bbb{Z}) = 0$$

and no information from these extra terms is extracted. So without loss of generality we are looking at the long exact sequence

$$\begin{eqnarray*} 0 &\to& H_n(P^n;\Bbb{Z}/2\Bbb{Z}) \stackrel{\cong}{\to} H_n(\Bbb{S}^n;\Bbb{Z}/2\Bbb{Z}) \stackrel{0}{\to} H_n(P^n;\Bbb{Z}/2\Bbb{Z}) \stackrel{\cong}{\to} H_{n-1}(P^n;\Bbb{Z}/2\Bbb{Z}) \to \ldots \\ \\ &&\ldots H_1(P^n;\Bbb{Z}/2\Bbb{Z}) \to \ldots \to H_0(P^n;\Bbb{Z}/2\Bbb{Z}) .\end{eqnarray*}$$

However if we follow the steps in the proof as above we end up with the diagram

$$\begin{array}{ccc} H_n(P^m;\Bbb{Z}/2\Bbb{Z}) &\stackrel{t_\ast}{\longrightarrow}&H_n(\Bbb{S}^m,\Bbb{Z}/2\Bbb{Z}) = 0 \\ \uparrow \psi_\ast&& \uparrow \phi_\ast \\ H_n(P^n;\Bbb{Z}/2\Bbb{Z}) &\stackrel{t_\ast}{\longrightarrow} & H_n(\Bbb{S}^n,\Bbb{Z}/2\Bbb{Z})\end{array}$$

with the arrows on the left column and bottom row isomorphisms. But this tells us nothing because going round one way and going round the other we both get zero.

share|cite|improve this answer
This is correct. For 1) just note that you can choose a fundamental domain for the action of $\mathbb{Z}/2\mathbb{Z}$ given by $g$ on $X$ which gives the desired decomposition. For 2) it might be interesting to exhibit an equivariant map $S^n \to S^m$ for $n \leq m$... – t.b. Sep 12 '12 at 2:06
For 1) Yes, if you specialize to the covering map of the sphere (you need to be a bit careful about what you do at the equator). I don't understand your explanation of 2) in the comment. – t.b. Sep 12 '12 at 2:23

Despite somewhat rusty homology knowledge, I think I can follow the proof.

1) Just to repeat my comment on this, there's actually not much too it. If $\tau$ is a simplex in $Y$ and $\sigma$ is a lifting of $\tau$ to $X$, then $g\circ\sigma$ is another lifting of $\tau$.

2) We need some slight prior knowledge: for any $p>0$, $$ H_p(S^p,\mathbb{Z}_2)\approx H_0(S^p,\mathbb{Z}_2)\approx\mathbb{Z}_2 \textrm{ while } H_q(S^p,\mathbb{Z}_2)\approx0\textrm{ for }q=1,\ldots,p-1. $$

2a) The first step is to prove that $H_q(P^p,\mathbb{Z}_2)\approx\mathbb{Z}_2$ for $q=0,\ldots,p$. To do this, we use the long exact sequence $$ \cdots \rightarrow H_q(P^p,\mathbb{Z}_2) \rightarrow H_q(S^p,\mathbb{Z}_2) \rightarrow H_q(P^p,\mathbb{Z}_2) \rightarrow H_{q-1}(P^p,\mathbb{Z}_2) \rightarrow H_{q-1}(S^p,\mathbb{Z}_2) \rightarrow\cdots $$ where $H_q(P^p,\mathbb{Z}_2)\rightarrow H_q(S^p,\mathbb{Z}_2)$ maps a simplex to the sum of both liftings of the simplex, while $H_q(S^p,\mathbb{Z}_2)\rightarrow H_q(P^p,\mathbb{Z}_2)$ just maps a simplex to the image in the ordinary way. If $2<q<p$, we get $H_q(P^p,\mathbb{Z}_2)\approx H_{q-1}(P^p,\mathbb{Z}_2)$ since $H_q(S^p,\mathbb{Z}_2)\approx H_{q-1}(S^p,\mathbb{Z}_2)\approx0$.

At the lower end of the sequence we get $$ 0 \rightarrow H_1(P^p,\mathbb{Z}_2) \rightarrow H_0(P^p,\mathbb{Z}_2) \rightarrow H_0(S^p,\mathbb{Z}_2) \rightarrow H_0(P^p,\mathbb{Z}_2) \rightarrow 0 $$ where the first and the last maps are isomorphisms, and get $H_1(P^p,\mathbb{Z}_2)\approx H_0(P^p,\mathbb{Z}_2)\approx\mathbb{Z}_2$.

At the upper end of the sequence, we get $$ 0\rightarrow H_p(P^p,\mathbb{Z}_2) \rightarrow H_p(S^p,\mathbb{Z}_2) \rightarrow H_p(P^p,\mathbb{Z}_2) \rightarrow H_{p-1}(P^p,\mathbb{Z}_2) \rightarrow 0 $$ where the middle map must be zero, hence $H_p(P^p,\mathbb{Z}_2)\approx H_{p-1}(P^p,\mathbb{Z}_2)\approx\mathbb{Z}_2$. The reason the middle map is zero is that the composition $H_p(S^p,\mathbb{Z}_2)\rightarrow H_p(P^p,\mathbb{Z}_2)\rightarrow H_p(S^p,\mathbb{Z}_2)$, which are first two maps of the long sequence just composed in the opposite order, has to be zero (since it maps a simplex to twice itself); but $H_p(P^p,\mathbb{Z}_2)\rightarrow H_p(S^p,\mathbb{Z}_2)$ is injective (monomorphism), so for the composition to be zero the map $H_p(S^p,\mathbb{Z}_2)\rightarrow H_p(P^p,\mathbb{Z}_2)$ must be zero.

2b) The next step is to apply the assumed equivariant map $\phi:S^n\rightarrow S^m$ for $n>m$ to obtain a contradiction.

The map $\phi_*:H_k(P^n,\mathbb{Z}_2)\rightarrow H_k(P^m,\mathbb{Z}_2)$ is an isomorphism for $k=0,\ldots,m$. This follows e.g. by induction, starting with $k=0$ and using $H_k(P^n,\mathbb{Z}_2)\approx H_{k-1}(P^n,\mathbb{Z}_2)$ and $H_k(P^m,\mathbb{Z}_2)\approx H_{k-1}(P^m,\mathbb{Z}_2)$ (commutative diagram in the book).

The final commutative diagram is now $$ \begin{array}{ccc} \mathbb{Z}_2\approx H_m(P^n,\mathbb{Z}_2)&\rightarrow &H_m(S^n,\mathbb{Z}_2)\approx 0\\ \downarrow&\circ&\downarrow\\ \mathbb{Z}_2\approx H_m(P^m,\mathbb{Z}_2)&\rightarrow &H_m(S^m,\mathbb{Z}_2)\approx\mathbb{Z}_2\\ \end{array} $$ where going in one direction should produce an isomorphism, while going in the other should map to zero since $H_m(S^n,\mathbb{Z}_2)\approx 0$. I.e., we have a contradiction.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.