1
$\begingroup$

Ok, sort of as a follow up to my previous question, let's recall the de Rham-Weil theorem: Let $F$ be a sheaf on a topological space $X$ and let $\mathcal{L}^{\bullet}$ be an acyclic resolution of $F$. Then $H^{q}(X,\mathcal{F}) \cong H^{q}(\mathcal{L}^{\bullet}({X}))$.

Now let's say I want to compute the cohomology of the complex $\mathcal{L}^{\bullet}({X})$ but my cochain complex

$0\rightarrow \mathcal{F} \rightarrow \mathcal{L}^{1} \rightarrow \mathcal{L}^{2} \rightarrow \mathcal{L}^{3} \cdots $

is not exact, i.e. $\mathcal{L}^{\bullet}$ is not a resolution of $\mathcal{F}$. In my problem (the problem I'm working on) the sheaf $\mathcal{F}$ and all the sheaves $\mathcal{L}^{q}$ are acyclic. How do I go about solving computing the cohomology of $\mathcal{L}^{\bullet}({X})$? I'm a little bit lost here, any tips, ideas or techniques as to how to proceed?

$\endgroup$
  • $\begingroup$ Can you be more specific about the complex $\mathcal{L}^\bullet$? The answer depends on the differentials. If all the differential are zero the answer is the complex itself. For other differentials you can expect a different answer. $\endgroup$ – Liviu Nicolaescu Mar 19 '12 at 21:24
  • $\begingroup$ That's all the information I have, it's a complex of acyclic sheaves, thanks $\endgroup$ – Louis A Apr 4 '12 at 0:20
3
$\begingroup$

Given the assumed acyclicity, what you are computing is the hypercohomology of the complex ${\cal L}^*$.

$\endgroup$
3
$\begingroup$

To complement Steven Landsburg's answer, you have a spectral sequence $$E_2^{pq}= H^p(X,\mathcal{H}^q(\mathcal{L}^\bullet)) \Rightarrow \mathbb{H}^{p+q}(X,\mathcal{L}^\bullet)$$ where the thing on the right is hypercohomology, i.e. the group you're after, and $$\mathcal{H}^q(\mathcal{L}^\bullet) =\ker[\mathcal{L}^q\to \mathcal{L}^{q+1}]/image(\ldots)$$ are the cohomology sheaves. This should get you started. Also keep in mind that $\mathcal{H}^0(\mathcal{L}^\bullet)=\mathcal{F}$ assuming exactness at the first step.

$\endgroup$

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.