# Nonabelian cohomology via crossed modules

Let $G$, $H$ be topological groups and let $t \colon G \to H$ be a homomorphism, such that $t \colon G \to H$ is a topological crossed module. For a topological space $X$ we can define the nonabelian cohomology set $\check{H}^1(X, G \to H)$. There is a map of crossed modules from $1 \to H$ to $G \to H$ and this induces a map $$\check{H}^1(X, H) \to \check{H}^1(X,G \to H)$$ What are the conditions for this map to be injective? If I express this problem in terms of classifying spaces, then I think I am asking for the fiber of the map $$BH \to B(G \to H)\ .$$ In particular, my vague hope was that if $G$ is contractible, then the above map is injective. Is this true?

-
(i) Would perhaps the homotopy fibre of the map be more informative? (ii) There are results in Larry Breen's work on Bitorsors that give some interpretation of what the elements of the non-Abelian cohomology with coefficients in a crossed module 'look like'. That source may give you the answer. (I have a summary of some of it in one of the versions of the Menagerie, but it is better to look at the original.) –  Tim Porter Jun 21 '11 at 18:17
@Tim: I meant the homotopy fiber, sorry. –  Ulrich Pennig Jun 21 '11 at 20:06

For results on the classifying space of (discrete) crossed modules, or more generally, crossed complexes, and in relation to homotopy classification and fibrations, see R. Brown, Exact sequences of fibrations of crossed complexes, homotopy classification of maps, and nonabelian extensions of groups, J. Homotopy and Related Structures 3 (2008) 331-343.

However the topological case has not been worked up in this format, as far as I know.

-
The recent paper of Murray, Roberts and Stevenson (arxiv.org/PS_cache/arxiv/pdf/1102/1102.4388v1.pdf) may be relevant to this. In any case it deserves a mention. :-) –  Tim Porter Jun 22 '11 at 11:56
The fiber of that map is well known to be $BG$, so what you expect is true.