Let $(X_i,\phi_i^j)$ be a directed system of topology spaces and its direct limit is $(X,\phi_i)$
$$\lim_{\rightarrow}(X_i,\phi_i^j)=(X,\phi_i)$$
Since $H_n$ ($n^{th}\, homology \,\,group$ ) is functor so $(H_n(X_i),(\phi_i^j)_*)$ is a directed system in category of abelian groups such that $(\phi_j)_*(\phi_i^j)_*=(\phi_i)_*$ for every $i\leq j$.
I know direct limit exist for any directed system in category of groups then we can assume
$$\lim_{\rightarrow}(H_n(X_i),\phi_{i^j_*})=(G,f_i)$$
and by defination of direct limit there exit a unique homomorphism
$$h:G\rightarrow H_n(X)$$ such that $\phi_{i_*}=h(f_i)$ for every i.
If I show $h$ is an isomorphism then
$$H_n(\lim_{\rightarrow}(X_i,\phi_i^j))\cong\lim_{\rightarrow}\left(H_*(X_i),(\phi_i)_*\right)$$
Can some body help me to proving the bijection of $h$?
- 690
2 Answers
It is not true that $h$ is an isomorphism in general. For instance let $X=S^1$ and consider the directed system of all countable subspaces of $X$ (with their inclusion maps). Then $X$ is the direct limit of this system (since a subset of $X$ is closed iff it is sequentially closed). But $H_1(X_i)=0$ for each $i$ in the system, since each $X_i$ is totally disconnected. So $$\lim_{\rightarrow}\left(H_1(X_i),(\phi_i)_*\right)=0$$ whereas $$H_1(\lim_{\rightarrow}(X_i,\phi_i^j))=\mathbb{Z}$$ for this system.
- 330,363
-
What fails is then that $\varinjlim S(X_i)$ is different from $S(X)$, correct? – Pedro Mar 07 '17 at 06:02
-
1Right, this is true if you're talking about simplicial sets instead of spaces. – Eric Wofsey Mar 07 '17 at 06:03
-
More generally, if every compact subset of $X$ is contained in some $X_i$, the claim holds. – Pedro Mar 07 '17 at 06:04
-
1@PedroTamaroff: If each $X_i$ is a subspace of $X$, then yes. But in general the maps $X_i\to X$ may not be injections, so you would need to say any map from a compact subspace factors through some $X_i$. – Eric Wofsey Mar 07 '17 at 06:06
-
@EricWofsey: My directed system has been define on directed directed index set – Mojtaba Mar 07 '17 at 06:06
-
@A-M: The countable subsets of $X$ (and their inclusions) are directed, since the union of two countable subsets is countable. – Eric Wofsey Mar 07 '17 at 06:07
-
you mean homology can't commute with direct limit in general? – Mojtaba Mar 07 '17 at 06:07
-
1That's correct, homology of spaces does not commute with direct limits in general. – Eric Wofsey Mar 07 '17 at 06:16
-
why $X_i$ is totally disconnected? – Mojtaba Mar 07 '17 at 07:11
-
See http://math.stackexchange.com/questions/96541/connected-metric-spaces-with-at-least-2-points-are-uncountable – Eric Wofsey Mar 07 '17 at 07:16
-
Ok , could you please explain more about the how $X$ is direct limit of its countable subspace? – Mojtaba Mar 07 '17 at 07:26
-
1Sorry, the explanation is potentially quite complicated depending on what you already know, and I'm afraid I'm not up to it right now. I would suggest first trying to understand why any set $X$ is the direct limit of its countable subsets, in the category of sets. If you really want a detailed explanation, you might ask that as a new question. – Eric Wofsey Mar 07 '17 at 07:56
-
I think that is not true that any $X$ is direct limit of its countable subsets, – Mojtaba Mar 07 '17 at 08:10
The directed system of topoloical spaces gives a directed sistem of complexes $S_i = S(X_i)$ where $S$ is the singular chains functor. However, it is not necessarily true that $\varinjlim S_i = S$. If this were the case, the following general considerations apply.
Consider any diagram $I$, and the functor $\varinjlim : A^I \to A$. This is right exact, and then there is a canonical map
$$\theta : \varinjlim H(S_i) \to H(\varinjlim S_i)$$
which may fail to be an isomorphism. However, if your indexing set $I$ is directed, then the functor $\varinjlim : A^I \to A$ is exact, and exact "functors commute with homology". More precisely, there is a map
$$\psi : H(\varinjlim S_i)\to \varinjlim H(S_i)$$
which is an inverse to $\psi$.
- 122,002
-
My indexing set I is directed. Could you please explain more about the exactness ! and what is the $A$ and why there exist a cononical map ? – Mojtaba Mar 07 '17 at 07:09