14
$\begingroup$

Let $G$ be a nice topological group and $E\to B$ a universal $G$-bundle. I'm interested in a proof of contractibility of $E$ using only the universal property of it. I also know that if there is a contractible $G$-bundle, then all of others are also contractible, but does there exist a direct proof not using a special construction of contractible universal $G$-bundles?

$\endgroup$
  • 4
    $\begingroup$ If $B$ represents the functor «$X$ → principal $G$-bundles on $X$», then $E$ represents the functor «$X$ → principal $G$-bundles on $X$ with a fixed section». This functor is trivial, so $E$ is contractible. $\endgroup$ – Grigory M Jun 4 '15 at 21:57
  • 3
    $\begingroup$ @GrigoryM I know that a map to $E$ gives a section of a $G$-bundle but I'm having trouble to prove your correspondence for homotopy classes of maps. Could you please write a complete answer? thanks! $\endgroup$ – Mostafa Jun 5 '15 at 18:43
  • $\begingroup$ Nobody has an answer for this important fact? $\endgroup$ – Eduardo Longa Jun 10 at 18:49
1
$\begingroup$

First note that the projection $\pi: E \rightarrow B$ is nullhomotopic by the universal property of $B$ since the pullback of a principal bundle by its projection is always trivial.

It follows from clutching and the loop-suspension adjunction that both $\Omega B$ and $G$ represent $G$-bundles over the suspension of your space.

This means that we can use the following rectangle which is commutative up to homotopy (formatting taken from https://mathoverflow.net/a/132340/134512) . $$\begin{matrix} G & \to & E & \to & B \\ \downarrow \simeq && \downarrow * && \downarrow = \\ \Omega B & \to & PB & \to & B \end{matrix}$$

This square induces a morphism of long exact sequences, which by the five lemma is an isomorphism from the homotopy groups of $B$ to the trivial group, as desired.

$\endgroup$

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.