nLab
groupoid of Lie-algebra valued forms

Contents

Idea

For G a Lie group, its delooping BG is a smooth groupoid, which we may think of as a groupoid internal to smooth spaces.

According to the theory of models for ∞-stack (∞,1)-toposes this is to be thought of as modeling the corresponding smooth ∞-stack GBund() of smooth G-principal bundles obtained after ∞-stackification.

The cohomology with coefficients in BG is degree 1 smooth nonabelian cohomology.

Let g:=Lie(G) be the Lie algebra of G.

The groupoid of g-valued differential forms , denoted here B¯G, is the refinement of BG in differential nonabelian cohomology.

Definition

Definition

For G a Lie group the groupoid of Lie(G)-valued differential forms is as a groupoid internal to smooth spaces, the sheaf of groupoids

B¯G:=[P 1(),BG]\bar \mathbf{B}G := [P_1(-), \mathbf{B}G]

that to a smooth test space UDiff assigns the functor category [P 1(U),BG] of smooth functors (functors internal to smooth spaces) from the path groupoid P 1(U) of U to the one-object delooping groupoid BG.

Theorem

The groupoid B¯G is canonically equivalent to the smooth groupoid where

  • objects are smooth g-valued 1-forms AΩ 1(U,g);

  • morphisms h:AA are given by smooth G-valued functions hC (U,G) such that

    A=Ad h(A)h *θ¯A' = Ad_h(A) - h^* \bar \theta

Here θ¯ is the right invariant canonical g-valued 1-form on G. A more sloppy but common way to write this is A=Ad h(A)+hdh 1.

Differential nonabelian cohomology

Theorem

The cohomology with coefficients in B¯G classifies G-principal bundles connection on a bundle with connection.

More is true: there is a natural canonical equivalence of groupoids

H Diff(X,B¯G)GBund (X).\mathbf{H}_{Diff}(X, \bar \mathbf{B}G) \simeq G Bund_\nabla(X) \,.

Remarks

  • There is the obvious projection

    B¯GBG.\bar \mathbf{B}G \to \mathbf{B}G \,.
  • Lifting a G-cocycle through this projection to a differential G-cocycle means equipping it with a connection.

For G=U(n) these differential cocycles model the Yang-Mills field in physics.

References

The idea goes back to John Baez. For a detailed history see the discussion at Differential Nonabelian Cohomology.

The details are in

  • U. Schreiber, Konrad Waldorf, Parallel Transport and Functors (arXiv)

The definition in terms of differential forms is def 4.6 there. The equivalence to [P 1(),BG] is proposition 4.7.