I am trying to derive the covariant formulation of Maxwell's equations.

I understand that all four of Maxwell's equations can be written compactly as

$$\partial_{\mu}F^{\mu\nu} - j^{\mu} = 0 \;, \tag{1}$$


$$\partial_{[\mu}F_{\alpha\beta]} = 0\;. \tag{2}$$

However, the second of these equations is often expressed in terms of the dual $\tilde{F}^{\mu\nu}$ as $$\partial_{\mu}\tilde{F}^{\mu\nu} = 0\;.$$ I am trying to derive eqn. (2) from this one. Until now, I have

$$\partial_{\mu}\tilde{F}^{\mu\nu} = \epsilon^{\mu\nu\alpha\beta}\partial_{\mu}F_{\alpha\beta}.$$ (where $\epsilon^{\mu\nu\alpha\beta}$ is the 4-dimensional Levi-Civita symbol)

I don't know how to prove that the RHS is the same as eqn.(2). Any help is greatly appreciated!


Since the Levi-Civita symbol is antisymmetric, contraction with it extracts the antisymmetric part, $$\epsilon^{\mu\nu\alpha\beta} \partial_\mu F_{\alpha \beta} = \epsilon^{\mu\nu\alpha\beta} \partial_{[\mu} F_{\alpha \beta]} = 0.$$ However, the Levi-Civita symbol is also invertible, i.e. we can remove it by contracting with another, $$\epsilon_{\mu' \nu \alpha' \beta'} \epsilon^{\mu\nu\alpha\beta} \partial_{[\mu} F_{\alpha \beta]} = 3! \, \partial_{[\mu'} F_{\alpha' \beta']}.$$ We know the left-hand side is zero, so the right-hand side is zero, recovering the result.

  • $\begingroup$ Thanks! That makes sense, intuitively. However, is there any rigorous way of showing that the Levi-Civita "extracts" the antisymmetric part of the tensor? Is it simply a question of expanding out in terms of all the permutations and putting it back together appropriately? $\endgroup$ – Sankarshana Dec 19 '18 at 15:08
  • $\begingroup$ @Sankarshana Fix a pair of indices, say $T_{ab \ldots}$. You can always write the tensor as $T_{ab \ldots} = T_{(ab) \ldots} + T_{[ab] \ldots}$. Then it's easy to see contracting with the Levi-Civita kills the first term, because it is antisymmetric in any two indices. Since this holds for every pair of indices, the only surviving part is antisymmetric in all indices. $\endgroup$ – knzhou Dec 19 '18 at 15:17

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.