5
$\begingroup$

As MO questions go, this one might be borderline - I'm guessing it could be a homework problem in a suitably advanced differential geometry class. I tried asking on math.stackexchange yesterday and it has scarcely received 20 views let alone an answer, so I'm trying it here instead. If it gets 4 votes to close, I'll give the 5th.

I'm trying to decipher a differential geometric comment on page 23-24 of Berline, Getzler, and Vergne's "Heat Kernels and Dirac Operators".

Take a trivial vector bundle $E \times M$ on a manifold $M$ with connection $\nabla = d + \omega$ where $\omega$ is an $End(E)$-valued 1 form. Let $g: GL(E) \to End(E)$ be the tautological map sending a linear map in $GL(E)$ to itself as an element of $End(E)$. The claim is that the connection 1-form on the (trivial) frame bundle for $E \times M$ is given by $g^{-1} \pi^* \omega g + g^{-1} d g$. In particular, if $\omega = 0$ then we get that the trivial connection on the trivial bundle is the Maurer-Cartan 1-form. Unfortunately, I don't see how to give a convincing proof of this - can someone help?

$\endgroup$

1 Answer 1

2
$\begingroup$

A concrete explanation: If you have a trivial connection, then a constant frame has covariant derivative zero. If you have a non-constant frame, then it can be written as a $GL(n)$-valued function multiplied by the constant section. Then the covariant derivative of the non-constant frame relative to itself can be obtained by differentiating this using the product rule. The Maurer-Cartan forms appear when you do this.

Of course, after you do this, you want to translate this into a much more abstract and sophisticated proof.

(But I do consider this to be a reasonable exercise for an advanced graduate student in geometric analysis and therefore at best borderline for MathOverflow. I'm surprised that no one on math.stackexchange.com has helped.)

$\endgroup$
1
  • 1
    $\begingroup$ If one wants to see an "abstract and sophisticated" proof, I think there is one in Morita's Geometry of differential forms. $\endgroup$ Jan 25, 2011 at 19:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.