2
$\begingroup$

I am trying to compute $H^*(X)$ for a (potentially large, finite, finitely filtered) simplicial complex $X$ using a cover $U_i$ of $X$.

I am building chain complexes for $X$ with a simplex that appears in the $p$th stage of the complex filtration generating a basis element of $C_*X$ with degree $p$. In cochains, this means that a simplex $\sigma$ that appears in the $p$th stage of a complex with $N$ filtration steps generates a relation $t^{N-p}\sigma=0$ in the cochain complex.

I want to use the generalized Mayer-Vietoris spectral sequence: $E_0$ is a double complex where the $p$-th column is $\oplus_{J\subset [n], |J|=p+1} C^*(\cap_{i\in J}U_i)$ with the usual coboundary, and the row coboundary is the restriction maps with signs from the nerve complex of the covering. In particular, I want the version of the spectral sequence of a double complex that starts with vertical cohomology and goes from there.

I already know how to compute the spectral sequence itself (even with algorithms!) -- my problem is with inferring $H^*(X)$ from the $E_\infty$-page. I know that $E_\infty=\mathrm{Gr} H^*(\mathrm{Tot}(E_0))$ and that $H^*(\mathrm{Tot}(E_0)) = H^*(X)$. Furthermore, I know that there are maps $H^n(X)\to E_\infty^{0,n}$ and $E_\infty^{n,0}\to H^n(X)$.

My questions are:

  1. Can I always find a representative in $\mathrm{Gr}H^*(\mathrm{Tot}(E_0))$ of any element of $E_\infty^{p,q}$ that has the form $(a_0,\dots,a_p,0,\dots,0)$ with all the $a_i$ nonzero (on a chain level)? How?
  2. Can I always find a representative in $\mathrm{Gr}H^*(\mathrm{Tot}(E_0))$ of any element of $E_\infty^{p,q}$ that has the form $(0,\dots,0,a_p,a_{p+1},\dots,a_n)$ with all the $a_i$ nonzero (on a chain level)? How?
  3. Can I always find a representative in $\mathrm{Gr}H^*(\mathrm{Tot}(E_0))$ of any element of $E_\infty^{p,q}$ that has the form $(a_0,\dots,a_{p-1},a_p,a_{p+1},\dots,a_n)$ with all the $a_i$ nonzero (on a chain level)? How?
  4. Is there a reasonably easy way to concretely write down the map $E_\infty^{n,0}\to H^*(X)$? In particular, if I have a $0$-cocycle in a $k$-fold intersection of cover elements, can I then write down a $k$-cocycle in $X$ directly from that $0$-cocycle? How?
$\endgroup$
2
  • $\begingroup$ I don't know the answer to your question, but it sounds like you might know the answer to mine mathoverflow.net/questions/91237/… $\endgroup$ Feb 5, 2015 at 19:13
  • $\begingroup$ David: This might be worth checking out: arxiv.org/abs/1112.1245 We never ended up nailing down low-level details for making this work for distributed computing in that preprint, which is why I'm trying to figure out how to do the same things in a cohomological sequence now. $\endgroup$ Feb 6, 2015 at 15:46

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.