3
$\begingroup$

I'm having trouble proving that covariant differentiation is an associative operation. Essentially I'll have to show

$$\nabla_\mu( \nabla_\nu \nabla_\sigma) = (\nabla_\mu\nabla_\nu) \nabla_\sigma. $$

But is it enough to show that both LHS and RHS yield the same result when acted up on a scalar or a contravariant vector?. Will this hold for any general tensor? Is there any other method to show this ?

$\endgroup$
2

2 Answers 2

1
$\begingroup$

If you see the covariant derivation as a bundle-map between the relevant vector bundles of tensors, associativity is automatically guaranteed by the usual associativity of composition of functions.

$\endgroup$
0
$\begingroup$

Frankly it boils down to function/operator composition, which is associative. Take a general tensor $T^\alpha_\beta$. $$(\nabla_\nu \nabla_\sigma)T^\alpha_\beta = \nabla_\nu (\nabla_\sigma T^\alpha_\beta )$$ so $$\nabla_\mu( \nabla_\nu \nabla_\sigma)T^\alpha_\beta = \nabla_\mu(\nabla_\nu (\nabla_\sigma T^\alpha_\beta ))$$ Similarly for the right hand side

$$ (\nabla_\mu\nabla_\nu) (\nabla_\sigma T^\alpha_\beta)= \nabla_\mu(\nabla_\nu (\nabla_\sigma T^\alpha_\beta ))$$ So both sides of your equation are the same, when expanded.

$\endgroup$

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.