Let $f: E\rightarrow B$ be a Kan fibration between pointed connected Kan complexes with fibre the Eilenberg-MacLane space $\mathrm{K}(M, n), n\geq 2, M$ an abelian group. Assume $f$ induces an isomorphism on $\pi_1$ with common value $G$, and so $G$ acts naturally on $\pi_n\mathrm{K}(M, n)=M$ hence on (the simplicial abelian group) $\mathrm{K}(M, n)$. Let $\tilde{B}$ be the universal cover of $B$, on which $\pi_1B=G$ acts.

Is there a fibration equivalence $\tilde{B}\times_G\mathrm{K}(M, n)\sim E$ over $B$?

  • 4
    $\begingroup$ The question becomes a lot simpler if you consider the special case $\pi_1(E) = \pi_1(B) = 0$. Then you are asking if for every fibration with fiber $K(M,n)$ is trivial (i.e. $E \sim B \times K(M,n)$). $\endgroup$ – Najib Idrissi May 27 '18 at 22:40
  • 1
    $\begingroup$ Good point to give counterexample! $\endgroup$ – Lao-tzu May 29 '18 at 5:51

No. If this were the case then there would be a section $s: B \to E$ to $f$ induced by the $G$-equivariant map $\widetilde{s}:\widetilde{B} \to \widetilde{B} \times {\rm K}(M,n)$ sending $x$ to $(x,0)$ (where $0 \in {\rm K}(M,n)$ is the neutral element, which is fixed by the action of $G$). However, there exists fibrations with fibers ${\rm K}(M,n)$ for $n \geq 2$ which do not admit a section, such as the quaternionic Hopf fibration $S^7 \to S^4$. In general fibrations $E \to B$ with fiber ${\rm K}(M,n)$ with a fixed $\pi_1(B)$-action on $M$ are classified by the cohomology group with local coefficients $H^{n+1}(B,M)$, where the ones of the form you describe correspond to the $0$ element.


As pointed in the remarks below, the quaternionic Hopf fibration is indeed not a counter-example for what the OP is asking since its fibers are $S^3$, which is not an Eilenberg-MacLane space. One way to fix this is to take the relative Postnikov truncation (which still doesn't have a section). Alternatively, there are probably other more natural examples.

  • $\begingroup$ Thanks a lot! My question is in fact stupid, $\tilde{B}\times_G\mathrm{K}(M, n)$ of course represents only one fibration, given the $G$-action on $M$. Could you indicate where can I find your last general statement which I don't figure out now? $\endgroup$ – Lao-tzu May 27 '18 at 9:01
  • $\begingroup$ I think you are using a result of Robinson about cohomology with twisted coefficients. $\endgroup$ – Lao-tzu May 27 '18 at 9:30
  • 1
    $\begingroup$ How does the Hopf fibration have fibres an Eilenberg-Mac Lane space? Sp(1) has highly nontrivial homotopy groups... $\endgroup$ – David Roberts May 27 '18 at 11:09
  • 2
    $\begingroup$ @Lao-tzu It is a classical result in obstruction theory, I don't know who proved it first. See for example chapter VI in Goerss, Jardine Simplicial Homotopy Theory. $\endgroup$ – Denis Nardin May 27 '18 at 11:12
  • $\begingroup$ @DavidRoberts, right! added an edit to the answer accordingly. $\endgroup$ – Yonatan Harpaz May 27 '18 at 13:05

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.