For a Lie group $G$ let $EG \to BG$ denote the universal bundle. A Lie group homomorphism $\rho: G \to H$ determines a map $B \rho: BG \to BH$ as the classifying map for the principal $H$-bundle $EG \times_\rho H \to BG$. While this argument yields existence of $B \rho$ (and uniqueness up to homotopy) it is far from being explicit. Does there exist a true construction of $B \rho$? If $\rho$ is a embedding of a subgroup, then on gets the classifying map by a quotient procedure. I was hoping for a similar result for arbitrary group homomorphism. Assume anything you want on $H$ (compactness ect).

Some background motivation: It is well-known that universal bundles characterize equivalence classes of principal bundles. A further, not so common characterization is given by 'homomorphisms $\rho$' between the loop space of $M$ and $G$. The principal bundle $P$ is then the to the path bundle associated bundle $PM \times_\rho G$. Hence there are two classification of principal bundles with connections and I want to better understand the interplay between them.

  • 2
    $\begingroup$ Constructions of classifying spaces tend to be functorial (e.g., Milnor or Milgram bar construction), so take a homomorphism $\rho$ of topological groups to continuous maps $B\rho$ explicitly. $\endgroup$ – Todd Trimble Jul 1 '14 at 14:55
  • 1
    $\begingroup$ Following on Todd's comment, note that the simplicial bar construction gives a model for $BG$ when $G$ is a Lie group (see Segal's paper "Classifying spaces and spectral sequences," available here: maths.ed.ac.uk/~aar/papers/segalclass.pdf). $\endgroup$ – Dan Ramras Jul 1 '14 at 17:42
  • 3
    $\begingroup$ Alternatively, if you want to stick with the model for $BG$ which is the quotient of a contractible space ($EG$) by a free $G$ action, note that if $EG$ is one such, then so too is $EG \times EH$, where $G$ acts on $EH$ through the homomorphism $\rho$. Then $BG = (EG \times EH) / G$, and $BH = EH / H$, so $B\rho$ can be given by the projection onto the second factor $(EG \times EH) / G \to EH / \rho(G) \to EH/H$. $\endgroup$ – Craig Westerland Jul 2 '14 at 3:28

The Milnor construction realised EG as the infinite join of copies of G, see the references in http://ncatlab.org/nlab/show/Milnor+construction

It is then obvious how to construct a $\rho$-equivariant map $EG\to EH$.


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.