Here is how hatcher proves this. Let $(C_•(X),\partial_•)$ be the reduced singular chain complex where $X$ is a nonempty topological space. Since $\partial_0:C_0(X)\rightarrow \mathbb{Z}$ is an epimorphism, we have $C_0(X)/\ker(\partial_0)\cong \mathbb{Z}$. Since $im(\partial_1)\leq \ker(\partial_0)\leq C_0(X)$, by the third isomorphism theorem, we have $H_0(X)/\tilde{H}_0(X)\cong C_0(X)/\ker(\partial_0)\cong \mathbb{Z}$. Hatcher wrote that this implies that $H_0(X)\cong \tilde{H}_0(X)\oplus\mathbb{Z}$, but I do not understand why. How do I show this? Thank you in advance.

  • $\begingroup$ The short exact sequence $$0\rightarrow\tilde{H}_0(X)\rightarrow{H}_0(X)\rightarrow H_0(X)/\tilde{H}_0(X)\rightarrow0$$ splits, where the maps are the canonical inclusion and projection maps. $\endgroup$
    – Enigma
    Oct 2, 2016 at 0:55
  • 1
    $\begingroup$ @Enigma, but why does it split? $\endgroup$ Oct 2, 2016 at 1:03

2 Answers 2


We have the exact sequence

$$0 \to \widetilde{H_0}(X) \to H_0(X) \stackrel{r_*}{\to} H_0(P)\to 0,$$

where $P$ is a one-point space and $r:X \to P$ is the unique map from $X$ to $P$.

Any map $i: P \to X$ induces a splitting, since $Id_*=(r \circ i)_*=r_*\circ i_*$, from which follows that $$H_0(X) \cong \widetilde{H_0}(X) \oplus H_0(P) .$$ Note that the argument holds for any homology theory satisfying Eilenberg-Steenrod axioms ($H_0(P)$ does not need to be free, for example) , since the splitting is given by functoriality of homology.

  • $\begingroup$ Nicely formulated. $\endgroup$ Oct 2, 2016 at 1:04

There is a map $\mathbb{Z}\rightarrow H_0(X)$ inverse to $\partial$. Namely take sny point (or vertex) $v$ in X and define $n\mapsto n[v]$. The image of this map is isomorphic to $\mathbb{Z}$ and with the reduced group forms a direct sum.


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.