The partial derivative of a vector $V^\lambda , _\nu$ is not a tensor. Neither is a Christoffel symbol $\Gamma^\lambda _{\mu \nu}$.

Is the addition of these two objects a tensor? If they were tensors, the addition woulnd't even be defined.

Using $\Gamma^\lambda _{\mu \nu}=w^\lambda(\partial_\mu e_\nu)$ how can I prove that $\partial_\nu V^\lambda+w^\lambda(\partial_\mu e_\nu)$ is a tensor? It looks like the covariant derivative, wich is a tensor, but it isn't.

I have tried to find the tranformation rule:

\begin{equation} \partial_\gamma V^\alpha+w^\alpha(\partial_\beta e_\gamma)=\frac{\partial x^\nu}{\partial x^\gamma}\frac{\partial}{\partial x^\nu}\frac{\partial x^\alpha}{\partial x^\lambda}V^\lambda+\frac{\partial x^\alpha}{\partial x^\lambda}w^\lambda\left(\frac{\partial x^\mu}{\partial x^\beta}\frac{\partial}{\partial x^\mu}\frac{\partial x^\nu}{\partial x^\gamma}e_\nu\right) \end{equation}

But I am unsure on how to operate the different derivatives.

  • 2
    $\begingroup$ You seem to be adding terms with different index structures?? $\endgroup$
    – Qmechanic
    Apr 13, 2019 at 19:53
  • $\begingroup$ @Qmechanic yes! I thought that that was a result of neither the Christoffel symbol or the derivative of the vector being tensors. Is the addition even possible? $\endgroup$
    – IchVerlore
    Apr 13, 2019 at 19:59
  • $\begingroup$ Have you tried to see how it transforms under coordinate transformation? That will tell you if the thing is a tensor or not... eqn's 5-7 here mathworld.wolfram.com/Tensor.html $\endgroup$ Apr 13, 2019 at 21:05
  • $\begingroup$ This seems extremely confused, like your last question. You should read a basic introduction to tensors from the beginning before trying to do anything else... $\endgroup$
    – knzhou
    Apr 13, 2019 at 21:05

1 Answer 1


It's not that since the Christoffel symbol and the partial derivative are not tensors you cannot add them. Think of them as sets of values: the fact that are not tensors means just that they do not transform covariantly (or contravariantly) when you do a change of coordinates; it does not forbid you to sum them and as a matter of fact the covariant derivative maps tensors (of a certain rank) in tensors (of a different rank).

In your post you are not writing the Christoffel symbol as applied to the field you are deriving in the partial derivative. The covariant derivative would be: $$\nabla_{\mu}V_{\nu}:=\partial_{\mu}V_{\nu}-\Gamma_{\mu\nu}^{\lambda}V_{\lambda}$$ Now if I understand correctly you really mean to sum the three index Christoffel symbol with the two index partial derivative right?

In that case you do not have a tensor. You can see it by proceeding as you tried but fixing the fact that you are changing coordinates so $x\rightarrow x'$. The transformation law becomes: \begin{equation} \partial_\gamma V^\alpha+w^\alpha(\partial_\beta e_\gamma)\;\;\rightarrow\;\;\frac{\partial x^\nu}{\partial x'^\gamma}\frac{\partial}{\partial x^\nu}(\frac{\partial x'^\alpha}{\partial x^\lambda}V^\lambda)+\frac{\partial x'^\alpha}{\partial x^\lambda}w^\lambda\frac{\partial x^\mu}{\partial x'^\beta}\frac{\partial}{\partial x^\mu}(\frac{\partial x^\nu}{\partial x'^\gamma}e_\nu) \end{equation} Despite you can still see this as a three index object (the first term is simply multiplied by $I_{\beta}\equiv 1 \;\forall\;\beta$), you can see that now the spurious second derivatives do not cancel each others and therefore you do not get something that transform as a tensor.


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.