I am little bit confused on Riemann curvature tensor,

Riemann curvature tensor written in component form as; $$R^d_{cab}=\partial_a\Gamma^d_{bc}-\partial_b\Gamma^d_{ac}+\Gamma^i_{bc}\Gamma^d_{ai}-\Gamma^i_{ac}\Gamma^d_{bi},$$ where $\Gamma^d_{bc}$ is affine connection.

My question is for the Riemann curvature tensor $R^d_{cab}$ to be a truly tensor there should be covariant derivative instead of partial derivative in those expression ? Such as $$R^d_{cab}=\nabla_a\Gamma^d_{bc}-\nabla_b\Gamma^d_{ac}+\Gamma^i_{bc}\Gamma^d_{ai}-\Gamma^i_{ac}\Gamma^d_{bi}.$$

  • 3
    $\begingroup$ $\Gamma^a_{bc}$ doesn't transform as a tensor field, so you can't act covariant derivative on it $\endgroup$
    – KP99
    Aug 3 at 11:17

2 Answers 2


The formula $$R^d{}_{cab}=\partial_a\Gamma^d{}_{bc}-\partial_b\Gamma^d{}_{ac} +\Gamma^i{}_{bc}\Gamma^d{}_{ai}-\Gamma^i{}_{ac}\Gamma^d{}_{bi} \tag{1}$$ actually is covariant, even though it doesn't look like this.

An equivalent (evidently covariant) definition of the Riemann curvature tensor $R^d{}_{cab}$ is the Ricci identity $$\nabla_b\nabla_a A_c-\nabla_a\nabla_b A_c=A_d R^d{}_{cab} \tag{2}$$ where $A_c$ is an arbitrary field and $\nabla_a$ is the covariant derivative.

From equation (2) together with the definition of the covariant derivative $\nabla_a$ (in terms of $\partial_a$ and $\Gamma^d{}_{bc}$) you can derive equation (1).

  • $\begingroup$ Are you sure about the sign? The formula you have linked expanded to the left and right is $\nabla_\sigma\nabla_\rho A_\nu-\nabla_\rho\nabla_\sigma A_\nu=A_{\nu;\rho\sigma} - A_{\nu;\sigma\rho} = A_\beta{R^\beta}_{\nu\rho\sigma}=-A_\beta{R^\beta}_{\nu\sigma\rho}$? $\endgroup$ Aug 3 at 13:49
  • 1
    $\begingroup$ @SamuelAdrianAntz You are right. I have fixed the mistake. $\endgroup$ Aug 3 at 14:17

To expand on the answer of Thomas Fritsch, this is how the calculation looks like in detail: \begin{align*} [\nabla_\mu,\nabla_\nu] A_\lambda &=(\nabla_\mu\nabla_\nu-\nabla_\nu\nabla_\mu)A_\lambda =\nabla_\mu(\partial_\nu A_\lambda-\Gamma_{\nu\lambda}^\rho A_\rho) -\nabla_\nu(\partial_\mu A_\lambda-\Gamma_{\mu\lambda}^\rho A_\rho) \\ &=\partial_\mu(\partial_\nu A_\lambda-\Gamma_{\nu\lambda}^\rho A_\rho) -\Gamma_{\mu\nu}^\tau(\partial_\tau A_\lambda-\Gamma_{\tau\lambda}^\rho A_\rho) -\Gamma_{\mu\lambda}^\tau(\partial_\nu A_\tau-\Gamma_{\nu\tau}^\rho A_\rho) \\ &-\partial_\nu(\partial_\mu A_\lambda-\Gamma_{\mu\lambda}^\rho A_\rho) +\Gamma_{\nu\mu}^\tau(\partial_\tau A_\lambda-\Gamma_{\tau\lambda}^\rho A_\rho) +\Gamma_{\nu\lambda}^\tau(\partial_\mu A_\tau-\Gamma_{\mu\tau}^\rho A_\rho) \\ &=-\partial_\mu(\Gamma_{\nu\lambda}^\rho A_\rho) -\Gamma_{\mu\lambda}^\tau(\partial_\nu A_\tau-\Gamma_{\nu\tau}^\rho A_\rho) +\partial_\nu(\Gamma_{\mu\lambda}^\rho A_\rho) +\Gamma_{\nu\lambda}^\tau(\partial_\mu A_\tau-\Gamma_{\mu\tau}^\rho A_\rho) \\ &=-\left(\partial_\mu\Gamma_{\nu\lambda}^\rho -\partial_\nu\Gamma_{\mu\lambda}^\rho +\Gamma_{\mu\tau}^\rho\Gamma_{\nu\lambda}^\tau -\Gamma_{\nu\tau}^\rho\Gamma_{\mu\lambda}^\tau \right)A_\rho =-R_{\lambda\mu\nu}^\rho A_\rho. \end{align*} The left side transforms like a tensor, therefore the right side does as well and since $A$ is arbitary, so does the Riemann tensor. You can also prove that directly using the transformation formula of the Christoffel symbol, but that is more laborious. You can look at it here.


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.