8
$\begingroup$

Suppose $\phi:\mathbb{S}^2\to\mathbb{S}^2$ is a mapping, homotopic to the identity map. Show that there is a fixed point $\phi(p)=p$.

$\endgroup$
5
$\begingroup$

Have you seen homology and degree of continuous mappings? If so, this is pretty short:

Since $f$ is homotopic to the identity, $\deg f = \deg \text{id}_{S^2} = +1$. On the other hand, if a continuous map $g:S^n\to S^n$ has no fixed points, then $\deg g = (-1)^{n+1}$. This shows that $f$ must have a fixed point, since otherwise $\deg f = (-1)^{2+1} = -1$.

Alternatively, you can prove that any map $S^n\to S^n$ with no fixed point is homotopic to the antipodal map. (This is used to prove the statement about $g$ in the previous proof.) Note that if $f$ has no fixed points, then $$h(x, t) = (1-t)f(x)-tx$$ is nonzero, so $h(x,t)/|h(x,t)|$ defines a homotopy from $f$ to the antipodal map. (This proof is in Hatcher, Algebraic Topology, section 2.2.) And the antipodal map is homotopic to the identity if and only if $n$ is odd, so this proves the claim by contraposition.

$\endgroup$
3
$\begingroup$

Alternatively, this follows from the Lefschetz fixed point theorem. As $\phi$ is homotopic to the identity map, the Lefschetz number of $\phi$ is just the Euler characteristic of $S^2$, which is 2. As $2 \neq 0$, it follows that $\phi$ has at least one fixed point.

$\endgroup$
0
$\begingroup$

Here is an idea: for each point $p \in S^2$, let $v_p$ be the result of orthogonally projecting $\phi(p)-p \in \mathbb{R}^3$ onto the tangent plane at $p$. If $\phi$ has no fixed point, this would given an everywhere nonzero continuous section of the projection $T(S^2) \to S^2$, which we know can't exist ("hairy ball theorem").

Edit: Of course the idea above is botched, as Harald pointed out. A possibly flaky second idea is to consider a homotopy $H(t, -)$ from $\phi$ to the identity. Assume that $\phi$ has no fixed point. For each $p$, let $t(p)$ be the least $t$ such that $\|H(t, p) - p\| = (1/2)\|\phi(p)-p\|$. Then define $\psi(p) = H(t(p), p)$. This $\psi$ has no fixed point either, and $\psi(p)$ is never antipodal to $p$. The remaining thing to check is that $p \mapsto t(p)$ is continuous.

$\endgroup$
  • $\begingroup$ I had that thought too, but it breaks down if $\phi(p)$ is antipodal to $p$. $\endgroup$ – Harald Hanche-Olsen Oct 25 '13 at 17:30
  • $\begingroup$ @HaraldHanche-Olsen Thanks! I tried a second idea, and I invite you to poke a hole in it. :-) $\endgroup$ – user43208 Oct 25 '13 at 18:02

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.