3

I was wondering if the singular homology functor preserve injectivity and surjectivity? I've been trying to figure out a proof or counterexample for ages now and I just can't.

This came up when I was looking at the reduced homology $H_p(S^{n},S^{n-1})$. To calculate it, I have looked at the canonical injection $$\iota: S^{n-1} \longrightarrow S^{n}.$$ I'm viewing $S^{n-1}$ as the equator of $S^n$, in this case does the functor preserve injectivity? That is, I want to say that $\iota_*$ is injective. Is this true? Thanksarinos

Najib Idrissi
  • 56,269
AlgebraicTopologyfourlyfffffff
  • 49
  • 2
    It's false, there is an injection $i : \mathbb S^1 \to \mathbb R^2$ but the induced map is of course not injective. For the surjectivity, think to the quotient map $ q : [0,1] \to \mathbb S^1$ which glue 0 and 1 together. –  May 12 '15 at 09:27
  • 2
    Based on your comments below, I have to ask: what is $\iota_$? Apparently it's not just $\iota_ : H_(S^{n-1}) \to H_(S^n)$? – Najib Idrissi May 12 '15 at 14:44

2 Answers2

3

I don't understand what you mean by relative homology preserving injectivity and surjectivity, so consider instead the following general point.

No interesting homotopy-invariant functor on spaces can preserve injections or surjections. The reason is that every map is homotopy equivalent to an injection and also to a surjection! There are very nice explicit constructions accomplishing this called the mapping cylinder and mapping cocylinder respectively.

In other words, "surjection" and "injection" aren't homotopy-theoretically meaningful concepts.

Qiaochu Yuan
  • 468,795
1

Your example already answers your question. $\iota_*: H_*(S^{n-1})\to H_*(S^n)$ is not injective. Also, $\pi: S^2\to\mathbb{RP}^2$ is surjective but the induced map on homology is not.

Alex Fok
  • 5,256