Keep an eye: I think there are two versions of Maurer-Cartan form: for Lie groups and for principal bundles. Probably they can be seen like the same, considering the Lie group as a principal bundle over a point...

Maurer-Cartan form of a Lie group

Given a Lie group G with Lie algebra g, we call (left-invariant) Maurer-Cartan form to the map:

θg:=d(Lg1)g:TgGTeGg

for every gG, and where Lg denote the left-multiplication by g. It is a g-valued 1-form: θTGg. Its meaning is that identifies every TgG with g (via left invariants vector fields), and so does with every two TgG and ThG. At this way, we have a sense of parallel transport.

Another point of view for Maurer-Cartan form:
mcform.png
Given a basis {Vi} for TeG=g we can extend it to a global frame of G by means of left translations (see examples here). Now, given XTgG, the components μi(X) of X in this frame are the Maurer-Cartan forms respect to the chosen basis. And the vector μi(X)Vig is also called the result of the Maurer-Cartan form. The individual Maurer-Cartan forms depend on the basis, but the "joint" one don't!

If the group is a matrix group then it can be computed with the expression μ=g1dg.

It is relevant for the method of moving frames.

Interpretation

See MC form for a matrix group#Interpretation.

Example

See example left invariants vector fields and Maurer-Cartan form.

MC form and left and right actions

What is the relation between left and right actions and the Maurer-Cartan form?
Suppose an element gG. For a vector VTeG we have two vectors in TgG: the left translated V^l=d(Lg)e(V) and the right translated V^r=d(Rg)e(V).
Pasted image 20220105100432.png
Since d(Lg)e is a vector space isomorphism, we can wonder what vector in TeG corresponds to V^r be means of it, i.e., the vector W such that:

d(Lg)e(W)=V^r=d(Rg)e(V)

so obviously

W=(d(Lg)e)1d(Rg)e(V)=d(Lg1Rg)e(V)==Adg1(V),

the adjoint representation.

So since the Maurer-Cartan form θg is an inverse for d(Lg)e:

  1. θgd(Lg)e=idg
  2. θgd(Rg)e=Adg1

From 1 it can be deduced that

Proposition
The Maurer-Cartan form is a left-invariant differential form. Moreover, it is the only left-invariant form such that θe=id.

Proof
Let VTa1gG,

La(θg)(V)=θg(((La))a1g(V))=θg(((Lg))e(θa1g(V))=θa1g(V)

Pasted image 20220925200416.png|700
So it is left-invariant.

Now, if α is another left-invariant 1-form with αe=id=Lg(θg) then

Lg1(αe)=Lg1(Lg(θg))

and so

αg=θg.

From 2, we can write, treating θ like a 1-form (g-valued):

(1)Rg(θ)=Adg1θ

for any hG.

This can be seen from the following picture:
Pasted image 20220105105255.png
Let V~l=d(Lh)e(V)ThG (it could have been an arbitrary WThG, but I found more visual this way). The right hand side of (1) would be telling to us that

V~lθhVAdg1Adg1(V)

and the left hand side:

V~ld(Rg)hd(Rg)h(V~l)θhgAdg1(V)

But that would be true if d(Rg)h(V~l)=d(Rhg)e(V), but this is true because

d(Rg)h(V~l)=d(Rg)h(d(Lh)e(V))=d(RgLh)e(V)==d(Rhg)e(V)

Keep an eye: I am not sure about how to proof the last equality, but it must be true...

On the other hand, the Maurer-Cartan form

Structural equation/MC equation

Given a Lie group G with Lie algebra g, the Maurer-Cartan form satisfies the following condition

dθ(X,Y)+[θ(X),θ(Y)]=0

where X,Y are vector fields (any) on G. It is called also the Maurer-Cartan equation.

It is also written

dθ+12[θ,θ]=0

for a Lie bracket defined in this way: If ω,θ are two g-valued 1-forms, define the g-valued 2-form [ω,θ] by

[ω,θ](X,Y)=[ω(X),θ(Y)]+[ω(Y),θ(X)].

Proof
See [Sharpe 1991] page 108.

It is similar to the curvature of a principal connection. Indeed, this structural equation is saying that a Klein geometry is a flat Cartan geometry. To see in what sense this is a kind of generalization of curvature (zero curvature, indeed) see Generalization of the flatness of R3.

Maurer-Cartan form on a principal bundle

Suppose PM is a H-principal bundle. Then, given vh we can consider the fundamental vector field v. For pP, vp is a vertical vector since the action of G leaves the fibres invariant. So for every p we have a linear isomorphism (need to be proven)

ip:hVpPTpPvvp=ddt|t=0(petv)

The inverse (ip)1, is like a Maurer-Cartan form, which acts only on vertical vector fields. That is, we can define a kind of h-valued 1-form θP by

θpP(V)=v

where V=ddt|t=0(petv)VpP, linear.

In the particular case of a Klein geometry seen as a principal bundle, the Maurer-Cartan form on the principal bundle in the sense described in this subsection coincides with the one of the beginning. See here for details.

(By the way, I think that giving a connection to P is "the same as" extending this 1-form θP to any vector, not only the vertical ones. In the case of a Cartan connection we are doing the same, but the values are taken in an extended Lie algebra...)