I have a concern regarding maps of serre fibrations and the induced map into the respective spectral sequence. The situation is the following.

Consider two serre fibrations $F \rightarrow E \rightarrow B$ and $F' \rightarrow E' \rightarrow B'$. Where $B$ and $B'$ are simply connected to make things easier.

Suppose that we have maps $f: E' \rightarrow E$, and $g: B' \rightarrow B$ that commute with the respective projections. (I couldn't figure out how to make a commutative diagram here) and let $h: F' \rightarrow F$ the map induced in the fibers.

If we consider cohomology with coefficients in a field for instance, the $E_2$ page decomposes as $$E_2 \cong H^*(B) \otimes H^*(F)$$

My first question is, do we have an induced map between the spectral sequences given by

$$g^* \otimes h^* : E_2 \rightarrow E'_2$$

if this is true, is also true that the map induced on the infinite term $E_\infty \rightarrow E'_\infty$ coincides with $f^*: H^*(E)\rightarrow H^*(E')$.

I appreciate any comment.


You do get an induced map between the $E_r$ pages for $2 \leq r \leq \infty$, but a priori the map on the $E_\infty$ page coincides with $f^*$ only up to filtration.

Here's a (somewhat silly) example. Pick your favorite nontrivial space $X$, and consider the map of fibrations $$\begin{array}{ccc} * & \rightarrow & X \\ \downarrow & & \downarrow \\ X & \xrightarrow{=} & X \\ \downarrow & & \downarrow \\ X & \rightarrow & * \end{array}$$

The maps of fibers and base spaces are essentially trivial, so on each $E_r$ page we have an isomorphism of $E_r^{0,0}$ but otherwise the induced map is zero. But the actual map between total spaces is the identity. This is the maximally worst case: almost all of the information contained in $f^*$ is hidden in the filtration, and the spectral sequence exposes almost none of it.


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.