I read multiple sources about raising and lowering indices and Einstein summary notation but am having problems doing it.

Here is an example. I am trying to derive Maxwell's inhomogeneous equations in the Minkowski vacuum \begin{equation} \partial_\mu F^{\mu\nu} = 0 \end{equation} using the alternative form in curved space: \begin{equation} \frac{1}{\sqrt{-g}}\partial_\mu\Bigg[(\sqrt{-g\,}g^{\mu\alpha}g^{\nu\beta}F_{\alpha\beta}\Bigg] = 0 \end{equation} Substituting $g^{\phi\theta} = \eta^{\phi\theta} = \text{diag} (+,-,-,-)$: \begin{equation} \partial_\mu F_{\alpha\beta} = 0 \end{equation}

What is the proper way to convert this to $\partial_\mu F^{\mu\nu} = 0$ or $\partial_\alpha F^{\alpha\beta} = 0$ or $\partial_\lambda F^{\lambda\kappa} = 0$? I keep getting a mess.


I am at a level of trying to make sure I am doing the mechanics properly. I think in the back of my mind I knew I could contract indices to get your alternative form. Here is my messy approach: \begin{equation} \eta_{\mu\phi} \: \partial^\phi F_\alpha \, ^\rho \: \eta_{\beta\rho} = 0 \end{equation} \begin{equation} \eta_{\mu\phi} \: \eta^{\phi\sigma} \: \partial_\sigma F_\alpha \, ^\rho \: \eta_{\beta\rho} = 0 \end{equation} \begin{equation} \eta_{\mu\phi} \: \eta^{\phi\sigma} \: \partial_\sigma \: \eta_{\lambda\alpha} \: F^{\lambda\rho} \: \eta_{\beta\rho} = 0 \end{equation} Substituting $\eta^{\phi\theta} = \text{diag} (+,-,-,-)$: \begin{equation} \partial_\sigma F^{\lambda\rho} = 0 \end{equation} Then I would keep raising and lowering and substituting until I get the preferred form, if ever. I assume I am missing the proper approach.

  • $\begingroup$ How did you mange to lose the sums over $\mu$, $\alpha$ and $\beta$? Do you understand how $A_\mu \delta^\mu_\nu=A_\nu$? $\endgroup$
    – mike stone
    Nov 7, 2020 at 13:11

1 Answer 1


You are overthinking this.

Multiplication by a $g$ tensor is what's usually meant by 'raising' or 'lowering' indices. Depending on how you structure your axioms, it is possible to take $g^{\mu\alpha}F_{\alpha\beta} = F^\alpha_\beta$ as a definition of what $F$ with an upper index means.

Generally, $F$ is defined as a covariant (bottom indices) tensor. This has various nice properties, chief among which being that it also makes $F$ a 2-form, which is a notion that is invariant under coordinate changes. The presence of a metric makes it possible to convert covariant objects to contravariant ones, so the top line could be equivalently written as

$$\frac{1}{\sqrt{-g}} \partial_\mu \left[\sqrt{-g} F^{\mu\nu} \right]$$ The only reason it is not written like that is to make the metric tensor dependence explicit, while leaving $F_{\mu\nu}$ in its 'natural' state.


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.