14
$\begingroup$

Suppose I have two maps of topological spaces, $f:X\rightarrow B$ and $g:Y\rightarrow B$, such that $f$ induces a homology isomorphism and $g$ is a fibration and $B$ is connected. Is it true that the natural map $X\times_{B}Y\rightarrow Y$ induces an isomorphism in homology?

$\endgroup$
20
$\begingroup$

This is not necessarily true. For example, there is a space $X$ constructed by attaching a 3-dimensional cell to $S^1 \vee S^2$, which serves as a standard counterexample to several questions. The map $X \to S^1$ produces isomorphisms $\pi_1(X) \to \pi_1(S^1)$ and $H_* X \to H_* S^1$, but $\pi_2 X \to \pi_2 S^1$ is not an isomorphism. As a result, the map of universal covers $\tilde X \to \Bbb R$ does not give an isomorphism $H_2(\tilde X) \to H_2(\Bbb R)$.

By letting $B = S^1$ and $g: \Bbb R \to S^1$ be the universal covering, we get a counterexample.

However, it turns out that this is almost completely an issue with fundamental groups. If $B$ is simply-connected, the answer to your question is yes. A heavy hammer that can be used to prove this is the Eilenberg-Moore spectral sequence.

$\endgroup$
  • 2
    $\begingroup$ The Serre spectral sequence is another heavy hammer that will also do the job. $\endgroup$ – Tom Goodwillie Oct 22 '17 at 20:50
11
$\begingroup$

Here is a positive answer to a slightly different question.

Call a map $X\to B$ "acyclic" if it induces an isomorphism in homology for every coefficient system on $B$. (If $B$ is simply connected then this is equivalent to your hypothesis.) Note that a map $F\to \ast$ is acyclic if and only if the space $F$ is acyclic in the sense that it has trivial integral homology.

Using Serre spectral sequences one can prove the following:

If $X\to B$ is acyclic then for any fibration $Y\to B$ the map $X\times_BY\to Y$ is acyclic.

A fibration $X\to B$ is acyclic if and only if for every point $b\in B$ the fiber over $B$ is acyclic.

$\endgroup$
  • $\begingroup$ By the Hurewicz Theorem a simply connected acyclic space is necessarily weakly contractible. There exist non-simply-connected acyclic spaces. If $F$ is such a space then you get a counterexample to your question by taking $B=F$ and taking $X$ and $Y$ to be contractible. $\endgroup$ – Tom Goodwillie Oct 22 '17 at 20:49

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.