2
$\begingroup$

Let $F \hookrightarrow E \to B$ be a fibration, and let everything be over a field $k$. The leray hirsch theorem in its usual form says that the (homology) cohomology spectral sequence degenerates at ($E^2$) $E_2$ if $H^*(E) \to H^*(F)$ has a splitting $(H_*(F) \to H_*(E)$ has a splitting). The condition imposed, however, that $H^*(E) \to H^*(F)$ be surjective does not allow for the possibility of a nontrivial local coefficient system.

Suppose we instead impose the condition on that $H_*(E) \to H_*(B)$ is surjective. Using the edge maps $H_*(E) \to E^\infty_{p,*} \subset E^2_{p,*}=H_*(B)$, we get that $d_r^{*,0}$ is 0 for all $r$ and using the coalgebra structure on $E^r$, we get that this spectral sequence degenerates at $E_2$.

This allows for a nontrivial local system since for any fiber bundle with nontrivial local coefficient system, one can take the product with a trivial fibration over a point, to get a bundle with a nontrivial local coefficient system with a topological splitting.

Do you know any examples where this is fruitful?

$\endgroup$
1
  • 1
    $\begingroup$ I don't believe that surjectivity of $H_\ast(E)\to H_\ast(B)$ lets the spectral sequence degenerate at $E^2$. Let the group $G$ be the semidirect product of $N$ (normal) and $Q$. Then the projection $q: G \to Q$ yields a surjection $q_\ast: H_\ast(BG,-) \to H_\ast(BQ,-)$. Take for example $BC_4 \to BD_8 \to BC_2$ where $D_8$ is the dihedral group of order 8 and $C_n$ cyclic of order $n$. I have no compuation for the homology ss at hand, but I know that the cohomology ss for the fibration doesn't stop at $E_2$. Therefore I'm pretty sure that the homology ss doesn't degenerate either. $\endgroup$
    – tj_
    Commented Aug 31, 2016 at 7:56

0

You must log in to answer this question.