First of all, I know that my question is quite long. I still would really appreciate any help because I am kind of stuck with this topic, I really try to express my question as clear as possible! Also, sorry for the language!

Let $(V,g), (M,h)$ be semi-Riemannian manifolds, $u: V \rightarrow M$ continous. Let $dim \ V=n+1, dim \ M=d$. Let $U \subset V$ be an open set with coordinates $x_{\alpha}$, s.t. $u(U)\subset M$ with coordinates $y^A$.

Next, let $E= T^{*}V \otimes TM$ be the vector bundle such that $E_x=T_xV \otimes T_{u(x) M}$.

In a paper is says that "The covariant derivative ∇ acting on sections of E has for coefficients the connexion coefficients of the Riemannian connexion of g and the pull back by u of the Riemannian connexion of h" and there is a local expression for the covariant derivative, where $f \in \Gamma(E)$:

$\nabla_{\alpha} f^A_{\beta}(x)= \partial_{\alpha} f^A_{\beta}(x)- \Gamma^{\mu}_{\alpha \beta}(x) f^{A}_{\mu}(x)+ \partial_{\alpha}u^B(x) \Gamma^A_{BC}(u(x))f^C_{\beta}(x)$

where $\Gamma^{\mu}_{\alpha \beta}$ are the components of the covariant derivative of $g$ and $\Gamma^A_{BC}(u(x))$ are the components of the covariant derivative of $h$.

Now I am trying to understand how this works. My thoughts so far:

Let $f \in E$. Then $f$ is of the form

$f_x=\underbrace{(\sum\limits_{\alpha=1}^{n+1} f_{\alpha}(x) \partial_{\alpha}|_x)}_{s} \otimes \underbrace{(\sum\limits_{A=1}^{d} \tilde{f}_A(x) \partial_A|_{u(x)})}_{t}$

where $f_{\alpha}, \tilde{f}_{A}:U \rightarrow \mathbb{R}$.

Is that correct so far?

Now it also holds that $\nabla (s \otimes t)= (\nabla s) \otimes t + s \otimes (\nabla t)$

So now I only need to understand how $\nabla s$ and $\nabla t$ looks like.

  1. $\nabla_{\alpha} s= \sum\limits_{\beta=1}^{n+1} (\partial_{\alpha}(f_{\beta}) \partial_{\beta} + f_{\beta} \sum\limits_{\mu=1}^{n+1} \Gamma^{\mu}_{\alpha \beta} \partial_{\gamma})$ Is that correct?

  2. Now for $\nabla t$, I don't really know what to do with it...I know the definition of a pullback bundle but I don't know how to express its covariant derivative in local coordinates. Can someone help me with this step?

Thanks in advance!


Since $f$ is a section of $T^{*} V \otimes u^{*}(TM)$, it can be written locally as

$$ f = f_{\beta}^{A} \, dx^{\beta} \otimes u^{*}(\partial_{A}) $$

where the $x^{\beta}$ are coordinates on $V$ and the $x^{A}$ are coordinates on $M$ and we are using Einstein summation so that the sums won't clutter our expression (note that this is not what you wrote as you need to use the $dx^{\beta}$ and not $\partial_{\beta}$).

Then the covariant derivative in the direction $\partial_{\alpha}$ is given by

$$ \nabla_{\partial_\alpha} f = \nabla_{\partial_\alpha} \left( f_{\beta}^{A} \, dx^{\beta} \otimes u^{*}(\partial_{A}) \right) = \\ \left( \partial_{\alpha} f^A_{\beta} \right) dx^{\beta} \otimes u^{*}(\partial_{A}) + f^A_{\beta} \left( \nabla_{\partial_\alpha} dx^{\beta} \right) \otimes u^{*}(\partial_A) + f^A_{\beta} dx^{\beta} \otimes \left( \nabla_{\partial_{\alpha}} u^{*}(\partial_A) \right) $$

where we used the product rule to move from the second expression to the third. Now use the definition of the connection on $T^* V$ and on $u^{*}(TM)$ to write the second and third expression in terms of the Christoffel symbols and gather coefficients to get formula you want.

  • $\begingroup$ BTW, I suspect there is a sign error in your formula. The last term should be plus and not minus. $\endgroup$ – levap Jun 13 '20 at 19:05
  • $\begingroup$ Thank you very much for your answer, it already cleared up a lot of questions. I also fixed the sign error. I now tried to continue: It holds, that $\nabla_{\partial_{\alpha}} d x ^{\beta} = \sum\limits_{\mu=1}^{n+1} \Gamma^{\mu}_{\alpha \beta} d x^{\mu} $. Now I don't see where the $f^A_{\mu}$ in the expression from the paper comes from here. (Sorry for not using the Einstein convention, it still kind of confuses me) $\endgroup$ – User1 Jun 16 '20 at 10:50
  • $\begingroup$ My result would be $\nabla_{\alpha} f^A_{\beta}= \partial_{\alpha} f^A_{\beta}(dx^{\beta} \otimes u^{*}(\partial_A)) + f^A_{\beta} (\Gamma^{\mu}_{\alpha \beta} d x ^{\mu}) \otimes u^{*}(\partial_A)+ f^A_{\beta} (dx^{\beta}) \otimes (\partial_{\alpha} u^{\beta} \Gamma^C_{BA} (u(x)) u^{*}(\partial_C))$. Is there already a mistake here or do I just miss why it is the same as in the paper? $\endgroup$ – User1 Jun 16 '20 at 11:22

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.