0
$\begingroup$

If I want to covariantly differentiate a one form then I can write:

$\nabla_\beta \tilde p = \dfrac{\partial p_\alpha}{\partial x^\beta} \tilde \omega^\alpha + p_\alpha \dfrac{\partial \tilde \omega^\alpha}{\partial x^\beta} $

So I can write, $\nabla_\beta \tilde p = \dfrac{\partial p_\alpha}{\partial x^\beta} \tilde \omega^\alpha + p_\alpha \bigg(\dfrac{\partial \tilde \omega^\alpha}{\partial x^\beta} \bigg)_\gamma \tilde \omega^\gamma $

Manipulating with the dummy indices a little bit, I get,

$\nabla_\beta \tilde p = \dfrac{\partial p_\alpha}{\partial x^\beta} \tilde \omega^\alpha + p_\gamma \bigg(\dfrac{\partial \tilde \omega^\gamma}{\partial x^\beta} \bigg)_\alpha \tilde \omega^\alpha = \bigg[ \dfrac{\partial p_\alpha}{\partial x^\beta} + p_\gamma \bigg(\dfrac{\partial \tilde \omega^\gamma}{\partial x^\beta} \bigg)_\alpha \bigg] \tilde \omega^\alpha $

This is what I get using the definition of covariant differentiation. The way my book uses is a bit different. It uses the fact that the covariant differentiation of a scalar is just simple differentiation along with the fact that contraction of a vector on a one form is a scalar. i.e.

$\phi= <\tilde p,\vec{V}>=p_\alpha V^\alpha$

Therefore, $\nabla_\beta\phi=\nabla_\beta(p_\alpha V^\alpha)$.

Therefore, $\nabla_\beta(p_\alpha V^\alpha) = \dfrac{\partial(p_\alpha V^\alpha)}{\partial x^\beta}$

A little manipulation leads to the expression

$\nabla_\beta \phi = (\dfrac{\partial p_\alpha}{\partial x^\beta} - p_\gamma \Gamma^\gamma_{\alpha\beta})V^\alpha+p_\alpha V^\alpha_{;\beta}$.

Now considering the product rule for the covariant differentiation to have the usual properties ( i.e. $\nabla_\beta(p_\alpha V^\alpha) = p_{\alpha;\beta}V^\alpha + p_\alpha V^\alpha_{;\beta} $ ), they arrive at the conclusion that the term in the bracket must be $p_{\alpha;\beta}$. And thus,

$\nabla_\beta \tilde p= (\dfrac{\partial p_\alpha}{\partial x^\beta} - p_\gamma \Gamma^\gamma_{\alpha\beta})\tilde \omega^\alpha $.

Thus, $\bigg(\dfrac{\partial \tilde \omega^\gamma}{\partial x^\beta} \bigg)_\alpha = -\Gamma^\gamma_{\alpha\beta}$.

My question is whether or not there exist a more direct way to conclude the above equality without making the a priori assumption that $\nabla_\beta(p_\alpha V^\alpha) = p_{\alpha;\beta}V^\alpha + p_\alpha V^\alpha_{;\beta} ? $

$\endgroup$
1
$\begingroup$

The first line is incorrect. The action of the connection $\nabla$ on the 1-form is:$$ \nabla_{\beta} (p_{\alpha}w^{\alpha})=(\partial_{\beta}p_{\alpha})w^{\alpha} + p_{\alpha}\nabla_{\beta}w^{\alpha}=(\partial_{\beta}p_{\alpha})w^{\alpha} + p_{\alpha}\tilde\Gamma^{\alpha}_{\beta \gamma}w^{\gamma}$$ by definition of connection (which requires the product rule with scalar functions used in my first equality) and the fact that the $w^{\alpha}$ form a local basis of the cotangent bundle ( so, since $\nabla_{\beta}w^{\alpha}$ belongs in it, it must be a linear combination of the type $\tilde\Gamma^{\alpha}_{\beta \gamma}w^{\gamma}$). By then using the method you stated, one finds that $\tilde\Gamma^{\alpha}_{\beta \gamma}= -\Gamma^{\alpha}_{\beta \gamma}$ (I have been assuming from the beginning that your $w^{\alpha}$ are the dual of $\partial_{\alpha}$).

The fact that, if X,Y are vector fields and $w$ a 1-form then $\nabla_X(w(Y))=w(\nabla_X Y) + \nabla_X(w)Y$, is not an assumption and can be demonstrated. I won't write everything down (i believe you can find it in any differential geometry book) but the idea is the following: having a connection on the tangent bundle, one extends it uniquely to all tensor bundles, asking that it satisfies some properties ( like $\nabla_X (t_1 \otimes t_2 )=(\nabla_X t_1)\otimes t_2 + t_1\otimes\nabla_X(t_2)$ ). It can then be demonstrated that it commutes with contractions, and since $w(Y)$ is a contraction, one gets the result we wanted.

p.s.:I only noticed that the question was 1 month old when I had already written most of the stuff down. If you had already solved this, sorry for the notification.

$\endgroup$

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.