3
$\begingroup$

Let $G$ be a compact Lie group and $\mathfrak g$ its Lie algebra. For any $f$ in the dual space $\mathfrak g^*$, we can define a skew-symmetric bi-linear form on $\mathfrak g$ by $(A,B)\mapsto f([A,B])$. Let us say that a differential 2-form on $G$ is a commutator 2-form if it has this structure at every point. We can identify the tangent space by left translations, so commutator 2-forms correspond to functions in $C^\infty(G,\mathfrak g^*)$. (The function corresponding to a given 2-form is not generally unique although every function corresponds to some form. For a silly example, consider any abelian group: all commutator 2-forms vanish identically.)

I have some questions about this concept:

  • Have commutator 2-forms been studied, perhaps under another name? If yes, where can I read more? I have not managed to find anything in this direction, but I may be missing a crucial keyword in my searches.
  • What kinds of commutator 2-forms are closed? In particular, can a non-constant commutator 2-form be closed? If exactness is better understood than closedness, answers in that case are welcome as well.
  • If the answer to the previous question is complicated in general, are there examples of non-abelian Lie groups where every non-constant commutator 2-form is closed?
$\endgroup$
  • $\begingroup$ Google Kostant-Kirillov-Souriau symplectic structure. $\endgroup$ – José Figueroa-O'Farrill Dec 16 '15 at 16:51
  • $\begingroup$ @José: doesn't that live on $\mathfrak{g}^{\ast}$, not $G$? $\endgroup$ – Qiaochu Yuan Dec 16 '15 at 17:18
  • $\begingroup$ It's symplectic on coadjoint orbits, so just pull it back? $\endgroup$ – José Figueroa-O'Farrill Dec 16 '15 at 18:27
5
$\begingroup$

Let $X$, $Y$, be left invariant vector fields, and let $f$ be a left invariant one-form. By Cartan's formula, $$(df)(X,Y)=X(\underbrace{f(Y)}_{\text{const}})-Y(\underbrace{f(X)}_{\text{const}})-f([X,Y])=-f([X,Y])\;,$$ so your constant commutator $2$-forms are exact. In the nonconstant case, Cartan's formula together with the Jacobi identity give a simple expression for $d(f([-,-]))$.

| cite | improve this answer | |
$\endgroup$
3
$\begingroup$

It seems to me more natural to dispose with the arbitrary choice of $f \in \mathfrak{g}^{\ast}$ and consider the corresponding $\mathfrak{g}$-valued $2$-form. If I'm not mistaken, this form is more or less the differential of the Maurer-Cartan $1$-form, and in particular is always exact.

| cite | improve this answer | |
$\endgroup$

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.