1
$\begingroup$

I have this problem and I don't know how to attack certain parts of it, so I would appreciate if anyone could bring me hints of the parts where I'm stuck:

  • Prove that for any integer, $\Sigma S^n$ is homeomorphic to $S^{n+1}$

  • Let $g\colon X\longrightarrow Y$ continuous. Prove that the function $g\times id_I\colon X\times I\longrightarrow Y\times I$ induces a continuous function $\Sigma g\colon \Sigma X\longrightarrow \Sigma Y$

  • Deduce that for any pair of positive integers $m,n>1$, exists a continuous surjective function $f\colon S^n\longrightarrow S^{m}$.

I see very clear that for the first part I have to use the the norm of the vector $(x,t)\in S^n\times I$, but I can't see why it's well defined (it's needed to prove that is well defined since $\Sigma S^n$ is a quotient of spaces, no?).

For the sencond part I suppose I have the same case of the first part, because I can define $\Sigma g=g$ if $t\neq 1,0$ in $(x,t)$, but I can't see how correct the part of the extremes of the suspension because of the quotient.

The third one sounds logic that when $n>m$, using the suspension of the inclusion $i\colon S^n\longrightarrow S^m$, it will give a continuous and surjective function, but I have the question: It's possible when $n

  • 1
    For the third question you might need to use some space filling curves.2017-01-30
  • 0
    The second one is quite trivial. For any relations and any map, if the function maps related elements to related elements then it induces a continous function.2017-01-30
  • 0
    @JohnMa I googled it and the only thing I found was that the Peano curve is useful for the induction basis, but I can't advance more with that, and I think that I must not need that heavy tool.2017-01-30
  • 0
    @freakish my main problem in the second and in the first is what happens to the related elements.2017-01-30
  • 1
    The first two parts is a hint that you start with a surjective continuous map $\mathbb S^1 \to \mathbb S^2$ and then do the suspension $\Sigma$. But at least you need one such surjective map first. (by the way there is no $m$ in part 3?)2017-01-30
  • 0
    By the Peano curve, I can do a surjective continuous function from $S^1$ to $S^2$, so doing the suspension any number of steps it will give me a chain of surjective functions from $S^n$ to $S^m$?2017-01-30
  • 0
    From $\mathbb S^1$ to $\mathbb S^{m-(n-1)}$. Then apply suspension $n-1$ times.2017-01-30
  • 0
    That idea lets me some doubts: Can I see the explicit function from $S^1$ to $S^2$ or is only an abstract existance? Why the suspension of that function is still surjective?2017-01-30
  • 0
    I do not think that is explicit. But on the other hand, I cannot think of any ways to write down a surjective continuous function $\mathbb S^1 \to \mathbb S^2$ explicitly.2017-01-30
  • 0
    If $f:X\to Y$ is surjective , then so are $\Sigma f$. That can be checked quite easily.2017-01-30
  • 0
    One final thing: Do I have to care about the elements in the quotient of the points collapsed in the suspension?2017-01-30

1 Answers 1

4

Take $I=[-1,1]$ for simplicity and define $f:S^n\times I\to S^{n+1}$ by

$$f(x_1,\ldots, x_n, z)=(rx_1,\ldots,rx_n, z)$$ $$r=1-z^2$$

So I've picked $r$ in such a way that $\lVert f(x_1,\ldots,x_n, z)\rVert=1$. Also you can easily check that

$$f(X, z)=f(X',z')\mbox{ if and only if }(X=X'\mbox{ and } z=z')\mbox{ or }z=z'\in\{-1, 1\}$$

Now this function obviously yields a continous bijection for a normal suspension $SS^{n}\to S^{n+1}$. And thus it is a homoemoprhism (since the domain is compact).

In order to generate a continous bijection for the reduced suspension one more step has to be done. We have to collapse all elements along the arc of the sphere joining two polar poles. So let $(a_1, \ldots, a_n)\in S^n$ be a fixed point. Pick

$$g:S^{n+1}\to S^{n+1}$$

in such a way that $g(x_1,\ldots, x_n, z)=g(y_1,\ldots, y_n, z')$ if and only if ($x_i=y_i$ and $z=z'$) or ($x_i=y_i=a_i$). The existence of such map does not seem to be trivial or at least I'm not sure how to prove it yet. Perhaps it can be filled by someone else.

Anyway $g\circ f$ now induces a continous bijection $\sum S^n\to S^{n+1}$ which is a homeomorphism since the domain is compact.


For the second one try something simplier first.

Lemma. Let

$$f:X\to Y$$

be a continous map. Consider 2 relations: $\sim_X$ on $X$ and $\sim_Y$ on $Y$. Now assume that for all $x, y\in X$ we have: if $x\sim_X y$ then $f(x)\sim_Y f(y)$. Then we have a well defined continous function

$$F:X/\sim_X\to Y/\sim_Y$$ $$F([x]_{\sim_X})=[f(x)]_{\sim_Y}$$

Proof. $F$ is well defined directly from the assumption that $f$ maps related elements to related elements. We will show that $F$ is continous. Indeed, let $U$ be open in $Y/\sim_Y$. Consider two projections

$$\pi_X: X\to X/\sim_X$$ $$\pi_Y: Y\to Y/\sim_Y$$

By definition $\pi_Y^{-1}(U)$ is open in $Y$. And thus $f^{-1}(\pi_Y^{-1}(U))$ is open in $X$. To complete the proof we only need to show that

$$f^{-1}(\pi_Y^{-1}(U))=\pi_X^{-1}(F^{-1}(U))$$

because by definition of quotient map this will mean that $F^{-1}(U)$ is open. But this is trivial since for any $x\in X$ we have

$$\pi_Y(f(x))=[f(x)]_{\sim_Y}=F([x]_{\sim_X})=F(\pi_X(x))$$

$\Box$

Back to your task. Consider $g\times id$. Note that in order to talk about reduced suspension you need to pick fixed points in $X$ and $Y$. These will be $x_0\in X$ and $g(x_0)\in Y$. Pick $(x,y)\in X\times\{0\}\cup X\times\{1\}\cup\{x_0\}\times I$. It is clear that

$$(g\times id)(x, y)=(g(x), y)\in Y\times\{0\}\cup Y\times\{1\}\cup \{g(x_0)\}\times I$$

Thus the induced map is well defined and continous.


Pick any surjective map $f:S^1\to S^2$ (space filling curve). By reduced suspension you have a surjective map $f^2:S^2\to S^3$ (indeed, suspension takes surjective maps to surjective maps). You can continue this construction to get any surjective map $f^n:S^n\to S^{n+1}$. Thus for any $n > m$ you get a surjective map $f^{n, m}:S^m\to S^n$ being a composition of those defined earlier.

The other way around is analogous. First define surjective map $f:S^2\to S^1$ by collapsing $S^2$ to a ball $B^2$ (by projection), then projecting $B^2$ to $I$ (since $I=B^1$) and finally wrapping $I$ on $S^1$ (by $(\cos(x), \sin(x))$ mapping). Each of these transformations is surjective and so is their composition. By suspension and composition you get a surjective map $f^{n, m}:S^m\to S^n$ where $n < m$.

  • 0
    But my main problem is still there, that is how to proceed with the points that are relationated by the quotient.2017-01-30
  • 0
    @MonsieurGalois I'm not sure I understand. Proceed where? What is the problem exactly?2017-01-30
  • 0
    Perhaps I wasn't clear enough. Whenever $x\sim y$ you just have to make sure that given function maps them to related elements as well. Note that the first task is a special case. On the right side $S^{n+1}$ you also have an implicit relation $x\sim y$ iff $x=y$. The core to work with these induced mappings is the lemma I've defined in the middle.2017-01-30