This is a question which I asked on StackExchange first, but might be more suited here.

I got stuck on the proof of Theorem 5.5.5 in Weibel's book. Not only that, but I also could not even find the statement of said theorem in any other source, so I am completely at a loss how to proceed.

The result is called the Eilenberg-Moore Filtration Sequence for complete complexes. Interestingly, I was unable to find this result under this name anywhere else.

Suppose the filtered chain complex $C$ is complete with respect to it's filtration $F_*C$, i.e $C=\lim\limits_{\leftarrow} C/F_pC.$ In this case the exact sequence $$0\longrightarrow\mathbin{\lim\limits_{\longleftarrow}}^1H_{n+1}(C/F_pC)\longrightarrow H_n(C)\longrightarrow\mathbin{\lim\limits_{\longleftarrow}} H_n(C/F_pC)\longrightarrow0$$ takes the form $$0\longrightarrow\bigcap F_pH_n(C)\longrightarrow H_n(C)\longrightarrow H_n(C)/\bigcap F_pH_n(C)\longrightarrow0$$ and $${\lim\limits_{\longleftarrow}} H_n(C/F_pC)\cong {\lim\limits_{\longleftarrow}} H_n(C)/F_pH_n(C). $$

The first exact sequence is Milnor's $\mathbin{\lim\limits_{\longleftarrow}}^1$ exact sequence (Theorem 3.5.8):

Let $C_i$ be a tower of chain complexes satisfying the Mittag-Leffler condition, and set $C=\mathbin{\lim\limits_{\longleftarrow}} C_i$. Then there is an exact sequence for each n: $$0\longrightarrow\mathbin{\lim\limits_{\longleftarrow}}^1H_{n+1}(C_i)\longrightarrow H_n(C)\longrightarrow\mathbin{\lim\limits_{\longleftarrow}} H_n(C_i)\longrightarrow0$$

The proof then proceeds as follows:

Taking the limit of the exact sequence of towers $$0\longrightarrow F_pH_*(C)\longrightarrow H_*(C)\longrightarrow H_*(C)/F_pH_*(C)\longrightarrow0$$ we find that there is a monomorphism $H_*(C)/\bigcap F_pH_*(C)\to {\lim\limits_{\longleftarrow}} H_*(C)/F_pH_*(C)$. Also, there is an exact sequnce $$0\longrightarrow H_*(C)/F_pH_*(C)\to H_*(C/F_pC),$$ which also gives a monomorphism ${\lim\limits_{\longleftarrow}} H_*(C)/F_pH_*(C)\to \lim\limits_{\longleftarrow} H_*(C/F_pC)$. This combines into a monomorphism $H_*(C)/\bigcap F_pH_*(C)\to\lim\limits_{\longleftarrow} H_*(C/F_pC)$.

Then Weibel says to "combine this with the $\mathbin{\lim\limits_{\longleftarrow}}^1$ sequence of 3.5.8". And I don't get this last step at all. It seems we have not yet connected $\mathbin{\lim\limits_{\longleftarrow}}^1$ and $\bigcap F_pH_n(C)$.

This result is used quite a bit in later sections, so I would really like to understand it. I would be thankful for any help here.


You have constructed, up to this point, a monomorphism $H_n C / \bigcap F_p H_n C \to \lim H_n(C/F_p C)$, and this is compatible with the map from $H_n C$. This allows you to construct the right-hand square, and then assemble all of, the following map of exact sequences: $$ \require{AMScd} \begin{CD} 0 @>>> \bigcap F_pH_n(C) @>>> H_n(C) @>>> H_n(C)/\bigcap F_pH_n(C) @>>> 0\\ @. @VVV @| @VVV @.\\ 0 @>>> \mathbin{\lim\limits_{\longleftarrow}}^1H_{n+1}(C/F_pC) @>>> H_n(C) @>>> \mathbin{\lim\limits_{\longleftarrow}} H_n(C/F_pC) @>>>0. \end{CD}$$

The middle map is an isomorphism and you have shown that right-hand vertical map is injective. Applying the snake lemma, we find that all the vertical maps are isomorphisms.

| cite | improve this answer | |

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.