2
$\begingroup$

I came across this definition of the Ricci Scalar on its Spanish Wikipedia page:

$$R=-g^{\mu\nu}\left(\Gamma_{\mu\nu}^{\lambda} \Gamma_{\lambda\sigma}^{\sigma} - \Gamma_{\mu\sigma}^{\lambda}\Gamma_{\nu\lambda}^{\sigma}\right) - \partial_{\nu}\left(g^{\mu\nu}\Gamma_{\mu \sigma}^{\sigma} - g^{\mu\sigma}\Gamma_{\mu \sigma}^{\nu}\right).$$

Does anyone know where it came from or how to demonstrate it?

I'll leave the link for the Wikipedia page here: https://es.wikipedia.org/wiki/Curvatura_escalar_de_Ricci

$\endgroup$
0

2 Answers 2

2
$\begingroup$

The indices in one of the terms in OP's expression might not be correct. The rest are. The following is a direct derivation.

The following two identities (which aren't immediately obvious) are needed: $$\partial_\mu g^{\sigma\lambda} = -g^{\sigma\rho}g^{\lambda\nu} \partial_\mu g_{\rho\nu} \\ g^{\mu\sigma}\Gamma^\nu_{\rho\sigma} - g^{\mu\sigma}g^{\nu\lambda}\partial_\rho g_{\sigma\lambda} = -g^{\nu\lambda}\Gamma^\mu_{\rho\lambda}.$$

The standard formula for the Ricci scalar obtained by contracting the Riemann tensor is $$R = g^{\mu\nu}\partial_\rho\Gamma^\rho_{\nu\mu} - g^{\mu\nu}\partial_\nu\Gamma^\rho_{\rho\mu} + g^{\mu\nu}\Gamma^\rho_{\rho\lambda}\Gamma^\lambda_{\mu\nu} - g^{\mu\nu}\Gamma^\rho_{\mu\lambda}\Gamma^\lambda_{\rho\nu}.$$ Integrating the first two terms by parts, $$R = \color{red}{\partial_\rho(g^{\mu\nu}\Gamma^\rho_{\nu\mu})} - \color{blue}{(\partial_\rho g^{\mu\nu})\Gamma^\rho_{\nu\mu}} - \color{red}{\partial_\nu(g^{\mu\nu}\Gamma^\rho_{\rho\mu})} + \color{green}{(\partial_\nu g^{\mu\nu})\Gamma^\rho_{\rho\mu}} + \color{green}{g^{\mu\nu}\Gamma^\rho_{\rho\lambda}\Gamma^\lambda_{\mu\nu}} - \color{blue}{g^{\mu\nu}\Gamma^\rho_{\mu\lambda}\Gamma^\lambda_{\rho\nu}} \\ = \color{red}{-\partial_\nu(g^{\nu\sigma}\Gamma^\rho_{\rho\sigma} - g^{\mu\sigma}\Gamma^\nu_{\mu\sigma})} + \color{green}{\Gamma^\rho_{\rho\mu}(\partial_\nu g^{\nu\mu}+g^{\lambda\nu}\Gamma^\mu_{\lambda\nu})} - \color{blue}{\Gamma^\rho_{\mu\nu}(\partial_\rho g^{\mu\nu}+g^{\mu\lambda}\Gamma^\nu_{\rho\lambda})}.$$ Applying the identities, $$\color{green}{\color{green}{\Gamma^\rho_{\rho\mu}(\partial_\nu g^{\nu\mu}+g^{\nu\lambda}\Gamma^\mu_{\nu\lambda})} = \Gamma^\rho_{\rho\mu}(-g^{\nu\sigma}g^{\mu\lambda}\partial_\nu g_{\sigma\lambda}+g^{\nu\sigma}\Gamma^\mu_{\nu\sigma}) = -\Gamma^\rho_{\rho\mu}g^{\mu\lambda}\Gamma^\nu_{\nu\lambda} = -g^{\mu\nu}\Gamma^\rho_{\rho\mu}\Gamma^\sigma_{\sigma\nu}} \\ \color{blue}{-\Gamma^\rho_{\mu\nu}(\partial_\rho g^{\mu\nu}+g^{\mu\lambda}\Gamma^\nu_{\rho\lambda}) = -\Gamma^\rho_{\mu\nu}(g^{\mu\lambda}\Gamma^\nu_{\rho\lambda} -g^{\mu\sigma}g^{\nu\lambda}\partial_\rho g_{\sigma\lambda}) = \Gamma^\rho_{\mu\nu}g^{\nu\lambda}\Gamma^\mu_{\rho\lambda} = g^{\mu\nu}\Gamma^\sigma_{\mu\rho}\Gamma^\rho_{\nu\sigma}}.$$ Putting it together, we finally get $$R = \color{red}{-\partial_\nu(g^{\nu\sigma}\Gamma^\rho_{\rho\sigma} - g^{\mu\sigma}\Gamma^\nu_{\mu\sigma})} - g^{\mu\nu}(\color{green}{\Gamma^\rho_{\rho\mu}\Gamma^\sigma_{\sigma\nu}}-\color{blue}{\Gamma^\sigma_{\mu\rho}\Gamma^\rho_{\nu\sigma}}).$$ Expanding this expression and applying the identities again, the standard formula is rederived.

The proofs of the identities are left as an exercise.

$\endgroup$
6
  • $\begingroup$ @Eletie I have narrowed down the issue. It appears to be the switching of the $\mu$ and $\nu$ in my second identity. My identity switches their order (which I have confirmed to be correct). So if I violate my identity and apply it to the green term without switching it, it produces the OP's expression. $\endgroup$ Commented Apr 29 at 18:58
  • $\begingroup$ Update: I have tested my formula for the unit 2-sphere and it produces the right Ricci scalar of 2. $\endgroup$ Commented Apr 29 at 22:14
  • 1
    $\begingroup$ I now agree with you - I believe the OP's equation has the correct indices only when a factor of the metric determinant is included in the derivative terms (making it a proper boundary term). Without that factor, your equations are what one would obtain $\endgroup$
    – Eletie
    Commented Apr 30 at 6:41
  • $\begingroup$ @Eletie Thanks for the confirmation. It is interesting that by adding the metric determinant as in 4.1.6 of 2401.12567, a pair of indices of one term gets swapped while all the indices on the other three terms remain the same. $\endgroup$ Commented Apr 30 at 10:46
  • $\begingroup$ I suppose it comes from $\partial_{\rho}\sqrt{-g} = \Gamma_{\rho \lambda}^{\lambda}$, but yes it's interesting how it all works out. $\endgroup$
    – Eletie
    Commented Apr 30 at 11:05
2
$\begingroup$

By performing a few integration by parts on the terms with derivatives of the connection, one can obtain this natural boundary term (total derivative). As a sketch, try using relations of the form $$ \partial ( g \Gamma) \propto \Gamma \partial g + g \partial \Gamma \ $$ with the original Ricci scalar definition, and you will obtain the definition above.

If you want more information, you can search for the 'Einstein' action or 'Gamma squared action', which refers to the quadratic connection term part. See also https://arxiv.org/abs/2103.15906, which uses this decomposition.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.