Let $M$ be a an oriented riemannian manifold. I have seen the following definition for the Hodge-star operator acting on a differential form. Starting with $\beta\in \Omega^p(M)$ we have $$\alpha \wedge \star \beta = \left<\alpha,\beta\right>\text{vol} ~~\forall \alpha \in \Omega^p(M)$$

where $ \left<\alpha,\beta\right> = \left<e^1\wedge\ldots\wedge e^p, f^1\wedge \ldots \wedge f^p\right> = \det[\left<e^i,f^j\right>]$.

My question is if we have a Lie algebra (of a matrix group) valued form $F\in \Omega^p(M,g)$. How do we now define the Hodge star operator ?. I am asking this question because I want to understand expressions such as $$ F \wedge \star F$$ in the context of Yang-Mills theory. Thank you very much.

EDIT: I think my confusion really comes from the fact wheter I should use the commutator or the matrix product as the first answer kindly mentioned. In the case of Yang-Mills functional is it meant as: $$ L_{YM}= \int_{M}Tr(F\wedge\star F), ~~F = \sum\limits_{j}\omega_j \otimes g_j\in \Omega^2(M,g) \\Tr(F\wedge \star F) = \sum\limits_{j,k}\left<\omega_j, \omega_k\right>\text{vol}~\text{Tr}([g_j,g_k])$$ or is it meant as $$F\in \Omega^2(M,g) \\Tr(F\wedge \star F) = \sum\limits_{j,k}\left<\omega_j, \omega_k\right>\text{vol}~\text{Tr}(g_jg_k).$$ Thank you for you help.


The Hodge star is not any different. But, depending on the notation you're using, if you have Lie-algebra valued forms, you use wedge on the forms and Lie bracket on the Lie-algebra part. To be explicit, if you have $F= \sum\limits_j \omega_j\otimes v_j$ and $G = \sum\limits_k\eta_k\otimes w_k$, with $\omega_j,\eta_k$ differential forms and $v_j,w_k\in\mathfrak g$, then $F\wedge G = \sum\limits_{j,k} (\omega_j\wedge\eta_k) \otimes [v_j,w_k]$.

Some authors will use regular matrix product rather than matrix commutator, when working with matrix-valued forms, so you need to check your particular author.

  • $\begingroup$ Thank you. I have added a small question at the end of my original post. $\endgroup$ Sep 16 '17 at 14:50
  • $\begingroup$ Well, trace of a commutator is $0$ :) $\endgroup$ Sep 16 '17 at 17:10

If I'm not misteaking, the first expression you propose vanishes as the trace is conjugaison invariant. Moreover YM action is not suppose to measure only an error of commutation (with G=U(1), YM should be different from 0).


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.