I'm trying to understand, as a self learner, the covariant formulation of Electromagnetism. In particular I've been stuck for a while on the Bianchi identity. As I've come to understand, when we introduce the field strength tensor $F_{\mu\nu}$ we can write the Maxwell's equation in an explicitly covariant form using observable quantities as the aforementioned tensor (because of its Gauge invariance). Maxwell's equation thus reduce to one single 4-tensor equation (Gauss's + Maxwell-Ampere's laws), plus an identity (Gauss's law for magnetism and Faraday-Maxwell's law): I don't understand how to derive the identity in terms of the field strength tensor. This is my reasoning up until now.

I will use latin indices that range from 1 to 3 and greek ones that range from 0 to 3. We can say that:

$$ F_{ij} = B_{ij} = \varepsilon_{ijk} B^k \Longrightarrow \varepsilon^{ijk} F_{ij}= \varepsilon_{ijk}\varepsilon^{ijk} B^k \Longrightarrow B^k = \frac{1}{3!} \varepsilon^{ijk} F_{ij}$$


$$ F_{0i} = -\frac{E_i}{c} \Longrightarrow E_i = -cF_{0i} $$

Let us consider a flat space-time, then Gauss's law for magnetism can be written as:

$$ \nabla \cdot \vec{B} = \partial_k B^k = 0 \Longrightarrow \partial_k \frac{1}{3!} \varepsilon^{ijk} F_{ij} = 0 \Longrightarrow \varepsilon^{ijk} \partial_k F_{ij} = 0 $$

and Faraday-Maxwell's law can be written as:

$$ \nabla \times \vec{E} + \frac{\partial \vec{B}}{\partial t} = \vec{0} \Longrightarrow \varepsilon^{kij} \partial_i E_j + c \partial_0 B^k = 0 \Longrightarrow \varepsilon^{kij} \partial_i \left( -c F_{0j} \right) + c \partial_0 \frac{1}{3!} \varepsilon^{ijk} F_{ij} = 0\Longrightarrow $$

$$ - \varepsilon^{kij} \partial_i F_{0j} + \frac{1}{3!} \varepsilon^{ijk} \partial_0 F_{ij} = 0$$

Now, if I'm not mistaken, $\varepsilon^{0 \beta \gamma \delta} = \varepsilon^{0ijk} = \varepsilon^{ijk}$, therefore I can sobsitute the three dimensional Levi-Civita symbol with its four dimensional variant:

$$ \varepsilon^{0ijk} \partial_k F_{ij} = 0 $$

$$ -\varepsilon^{0kij} \partial_i F_{0j} + \frac{1}{3!} \varepsilon^{0ijk} \partial_0 F_{ij} = 0$$

From here I don't know how to continue. The latter of the two equations really seems to involve some kind of Levi-Civita symbol, because of the swapping of the $0$ index between the 4-gradient component and the field tensor component, but the $\frac{1}{3!}$ is in the way. Moreover, even if I simplified this equation I wouldn't know how to unify the two to obtain the final identity in terms of the field strength tensor:

$$ \epsilon^{\alpha \beta \gamma \delta} \partial_{\beta} F_{\gamma \delta} = 0 $$

Thank you in advance for any advice!


1 Answer 1


I shall note that in my answer I consider $c=1$. If that is confusing, please let me know. The Gauss Law reads $\partial_i B^{i}=0$, which implies $$\partial_1 F_{23} + \partial_{2} F_{31} + \partial_{3} F_{12}=0.$$ Faraday's Law tells us that $\nabla \times \vec{E} + \partial_{t} \vec {B}=0$. With indices, we can rewrite this as $$ \varepsilon_{ijk} \partial_{j} E^{k} + \partial_{0} B^{i}=0.$$ From the above we can read off $$\partial_2 E^3 - \partial_3 E^2 + \partial_0 B^1=0 \iff \partial_2 F_{03}+ \partial_{3} F_{20} + \partial_{0} F_{32}=0;$$ $$\partial_3 E^1 - \partial_1 E^3 + \partial_0 B^2=0 \iff \partial_3 F_{01} + \partial_{1} F_{30} + \partial_{0} F_{13}=0;$$ $$\partial_{1} E^{2} - \partial_{2} E^{1} + \partial_{0} B^3=0 \iff \partial_1 F_{02} +\partial_{2} F_{10} + \partial_{0} F_{21}=0.$$ Based on the above, altogether, for $F$, we have: $$ \partial_{\alpha} F_{\beta \gamma} + \partial_{\beta} F_{\gamma \alpha} + \partial_{\gamma} F_{\alpha \beta}=0.$$ Multiplying this equation by $\frac{1}{3}$, it holds that: $$\frac{1}{3} \left(\partial_{\alpha} F_{\beta \gamma} + \partial_{\beta} F_{\gamma \alpha} + \partial_{\gamma} F_{\alpha \beta} \right)=0.$$ Using the symmetry properties of the Faraday tensor: $$\frac{1}{2 \cdot 3} \left(\partial_{\alpha} (F_{\beta \gamma}- F_{\gamma \beta})+\partial_{\beta}(F_{\gamma \alpha} - F_{\alpha \gamma})+ \partial_{\gamma}(F_{\alpha \beta} - F_{\beta \alpha})\right)=0.$$ Simplifying and regrouping gives: $$\frac{1}{3!}\left(\partial_{\alpha} F_{\beta \gamma} + \partial_{\beta} F_{\gamma \alpha}+ \partial_{\gamma} F_{\alpha \beta} - \partial_{\beta} F_{\alpha \gamma} - \partial_{\alpha} F_{\gamma \beta} - \partial_{\gamma} F_{\beta \alpha} \right)=0.$$ Now contracting the equation with the Levi-Civita tensor gives the result. For the relation with what you have done can be seen here

  • $\begingroup$ Thank you very much for your answer: it was very helpful! No problem in using $c=1$ $\endgroup$
    – Luke__
    Commented Sep 22, 2023 at 18:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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