Does a chain rule for the covariant derivative exist so that we can evaluate an expression like $$\nabla_c\sqrt{t_{ab}}?$$

where $t_{ab}$ are tensor components?

More generally, how do we take the covariant derivative of a function of tensor components $f(t_{ab})$, i.e. $$\nabla_c f(t_{ab})?$$

  • 1
    $\begingroup$ $\sqrt{t_{ab}}$ are not components of a tensor so your expression is physically meaningless. $\endgroup$ – G. Smith Mar 19 at 17:57
  • $\begingroup$ @G.Smith why exactly is it not a tensor $\endgroup$ – Shashaank Mar 19 at 19:10
  • $\begingroup$ @Shashaank It doesn’t satisfy the transformation rule for a tensor with two covariant indices. $\endgroup$ – G. Smith Mar 19 at 20:34
  • $\begingroup$ @G.Smith I'm thinking of taking the covariant derivative $\nabla_c \sqrt{g}$ where $g$ is the metric determinant. How can I use some sort of chain rule to take care of the square root? $\endgroup$ – TaeNyFan Mar 20 at 11:15
  • $\begingroup$ Yes. The chain rule works for covariant derivatives. As far as I know, it works for any kind of derivative. $\endgroup$ – G. Smith Mar 20 at 21:14

There is a chain rule if the object is a tensor formed by tensor products and contractions, for instance $$\nabla_c (A^{bde}B_{bdf})=B_{bdf}\nabla_c A^{bde}+A^{bde}\nabla_c B_{bdf}.$$ Something like $f(t_{ab})$ is in general not a tensor so the action of the connection coefficient part of the covariant derivative is undefined on it.


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.