8
$\begingroup$

I've come upon this MO post, and I was wondering if there is a way to generalize what is said in the answer at the very bottom. Specifically, we have that given continuous maps $f: \mathbb{S}^n \to \mathbb{S}^n$, and $g: \mathbb{S}^n \to \mathbb{R}^n$, if $f$ is an involution, then there is a point $x \in \mathbb{S}^n$ such that $g(x) = g(f(x))$, i.e. the value at $x$ is fixed under the transformation of the sphere by $f$.

We know from the answer of the post that for a general $f$, there need not be such a fixed point. But now what if instead $f$ is such that there exists a $k \in \mathbb{N}$ where $f^k$ is the identity map, do we still get such a fixed point? Furthermore, do we know about other spaces besides the sphere that have this property when $f$ is an involution?

$\endgroup$

1 Answer 1

3
$\begingroup$

A first observation is that a map $f:S^n\to S^n$ with $f^k=\operatorname{Id}_{S^n}$ generates a topological action of the cyclic group $\mathbb{Z}/k$ on $S^n$. So we are talking about generalising from $\mathbb{Z}/2$ actions to $\mathbb{Z}/k$ actions.

There are many generalisations of the Borsuk-Ulam theorem of this kind in the literature. Most of them concern estimates of the dimension of the set $$ A(g)=\{x\in S^n \mid g(x)=g(f^i(x))\mbox{ for }i=1,\ldots , k\}, $$ for various flavours of map $g:S^n\to M^m$.

Interpreted this way, the answer to your first question appears to be negative in general, by a theorem of Munkholm in

Munkholm, H. J., On the Borsuk-Ulam theorem for (Z_{p^ a}) actions on (S^{2n-1}) and maps (S^{2n-1} \to R^ m), Osaka J. Math. 7, 451-456 (1970). ZBL0211.55701.

In particular, let $k=p^a\neq 9$ be an odd prime power with $a>1$, and let $\omega=\operatorname{exp}(2\pi i/p^a)$ be a primitive $p^a$-th root of unity. Let $f:S^{2n-1}\to S^{2n-1}$ be given in complex coordinates by $$ f(z_1,\ldots , z_n)=(\omega z_1,\ldots ,\omega z_n). $$ Then Munkholm's "mod $p^a$ Borsuk-Ulam anti-theorem" seems to imply that there exists a continuous map $g:S^{2n-1}\to \mathbb{R}^{2n-1}$ with $A(g)=\emptyset$.

For a summary of positive results in the literature, you could do worse than Chapter 1 of Yuri Turygin's thesis at the Unviersity of Florida.

$\endgroup$
7
  • $\begingroup$ This is interesting. But what if I don't want other powers of $f$ to fix the value, and just want $g(x)= g(f(x))$? We could still have $A(g)$ empty in this case, if for every $x$ some other power doesn't fix the value. $\endgroup$ Dec 21, 2019 at 6:02
  • $\begingroup$ In particular, $g : S^1 \to \mathbb{R}$ such that $g(\theta) = \cos(\theta)$, and $f:S^1 \to S^1$ such that $f(\theta) = \theta + 1/n$ for some natural $n$ large enough. The map $f$ is such that $f^n$ is the identity on $S^1$, and we should get $A(g)$ empty, but $g(S^1)$ and $g(f(S^1))$ have two points of intersections, so two such "fixed points". $\endgroup$ Dec 22, 2019 at 21:08
  • $\begingroup$ Sorry, @Mark I forgot to ping you $\endgroup$ Dec 23, 2019 at 0:25
  • $\begingroup$ @PaulCusson: You are right, I think, and your example above has $A(g)$ empty whilst having points for which $g(x)=g(f(x))$. (I suppose you meant to write $f(\theta)=\theta + 2\pi/n$ or something similar?) There doesn't seem to be much in the literature about your precise question. $\endgroup$
    – Mark Grant
    Dec 23, 2019 at 19:26
  • $\begingroup$ About your second question, there has been some work done on extending Borsuk-Ulam to homology spheres, see the references in the linked thesis (Conner-Floyd, Munkholm,...) $\endgroup$
    – Mark Grant
    Dec 23, 2019 at 19:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.