Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. It's 100% free.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Maxwell's equations on a pseudo-Riemannian manifold $(M,g_{ab})$ say,

$$d_a F_{bc} = \nabla_{[a}F_{bc]} = 0,$$

$$\nabla_a F^{ab} = J^b,$$

where $d_a$ is the exterior derivative, $\nabla_a$ is the covariant derivative compatible with $g_{ab}$ (i.e. $\nabla_a g_{bc}=0$), and $F_{ab}$ is antisymmetric ($F_{ab}=-F_{ba}$).

Let $J^a, F_{ab}$ and $\tilde{J}_a, \tilde{F}_{ab}$ be two solutions to Maxwell's equations in this form. Suppose that $J^a$ and $\tilde{J}^a$ are equivalent up to isometry, in that $\varphi_*J^a = \tilde{J}^a$ for some isometry $\varphi:M\rightarrow M$.

Question: Does it follow that $F_{ab}$ and $\tilde{F}_{ab}$ are equivalent up to isometry, in that $\varphi_*F_{ab} = \tilde{F}_{ab}$?

Wald's General Relativity (1984, Chapter 10 Problem 2) shows a sense in which the answer is yes in the source-free case ($J^a=\mathbf{0}$). I wonder what is known about the non-source-free case.

share|cite|improve this question
up vote 7 down vote accepted

No, you also need an initial condition for the field, and a boundary condition for the field. A plane wave solution to the Maxwell equations has the same 4-current as vacuum, after all.

share|cite|improve this answer
Just for clarification, what do you mean by "initial and boundary conditions" in this context? – soulphysics Aug 8 '14 at 15:16
@soulphysics: Let's say we solve this in terms of the vector potential $A_{a}(x,t)$. Then, to find a unique solution to Maxwell's equations, we need to know the value of $A_{a}(x,0)$ and ${\dot A}_{a}(x,0)$, as well as either a falloff condition for $A$ at infinity, or $A(0,t)$ on some surface. – Jerry Schirmer Aug 8 '14 at 15:32
It's not clear to me how fixing a vector potential helps the uniqueness problem. Since $F_{ab}$ is closed and antisymmetric, I can always (locally) write it as the exterior derivative of a vector potential, $F_{ab}=d_a A_b$. And of course there are many other equivalent vector potentials $\hat{A}_{a}=A_a+\nabla_a\chi$ for some scalar field $\chi$ (i.e. related by a gauge transformation). But all of them give rise to the same Maxwell-Faraday tensor $F_{ab}$. So, how does this provide a further constraint? – soulphysics Aug 8 '14 at 16:04
@soulphysics: because a plane wave has a different maxwell tensor from vacuum. The two solutions aren't related by a gauge difference. They are different physical states, but they both have $j^{\mu} =0$. – Jerry Schirmer Aug 8 '14 at 16:06
This is standard for the solution of a system of PDEs. You need an initial condition, and a set of boundary conditions. – Jerry Schirmer Aug 8 '14 at 16:07

Ok, here's my own (pedantic) argument that the answer is "No." Three background steps:

  1. Take $(M,g_{ab})$ and $(\tilde{M},\tilde{g}_{ab})$ to be two pseudo-Riemannian manifolds related by an isometry $\varphi:M\rightarrow\tilde{M}$. I will consistently use a tilde ($\tilde{}$) to refer to objects on the second manifold and no tilde to refer to objects on the first. Let $\varphi_*$ be the pushforward and $\varphi^*$ the pullback of $\varphi$.

  2. Let $F_{ab}$ and $J^a$ be any solution to Maxwell's equations on $(M,g_{ab})$ that is not source free, i.e., $\nabla_a F^{ab} = J^b \neq \mathbf{0}$.

  3. Let $\tilde{E}_{ab}$ and $\mathbf{0}$ be a non-trivial solution to Maxwell's equations on $(\tilde{M},\tilde{g}_{ab})$ that is source free, i.e. $\nabla_a E^{ab} = \mathbf{0}$ and $E_{ab}\neq\mathbf{0}$.

Now, we can state the counterexample: let $\tilde{F}_{ab}$ be the tensor field on $(\tilde{M},\tilde{g}_{ab})$ defined by,

$$\tilde{F}_{ab} := \varphi_*F_{ab} + \tilde{E}_{ab}.$$

It is antisymmetric and closed, since both $\varphi_*F_{ab}$ and $\tilde{E}_{ab}$ are, and so it provides a solution to Maxwell's equations. Let the 4-current associated with $\tilde{F}_{ab}$ be given by,

$$\tilde{J}^b := \tilde{\nabla}_a \tilde{F}^{ab}.$$

Then, $J^b$ and $\tilde{J}^b$ are related by an isometry:

$$\varphi_*J^b = \varphi_*\nabla_a F^{ab} = \nabla_a \varphi_*F^{ab} = \nabla_a(\tilde{F}^{ab} - \tilde{E}^{ab}) = \nabla_a\tilde{F}^{ab} = \tilde{J}^b,$$

where the second equality applies the fact that $\varphi$ is an isometry, the third the definition of $\tilde{F}_{ab}$, and the fourth the fact that $\tilde{E}_{ab}$ is source-free. And yet, $\varphi_*F_{ab}\neq \tilde{F}_{ab}$ by construction.

How curious.

share|cite|improve this answer
I'm probably missing something obvious, but where did you use the hypothesis $J^b \neq 0$? Or does this hold without that assumption? – Chris White Aug 8 '14 at 17:15
@ChrisWhite -- no, I think you're right. That assumption isn't necessary. – soulphysics Aug 8 '14 at 18:44
So then you're saying "no" to a situation where Wald said "yes"? I feel there's something subtle I'm missing still, perhaps in the particular way Wald actually formulates the question ($D_a E^a = D_a B^a = 0$ iirc), but I haven't fully thought this through. – Chris White Aug 8 '14 at 19:05
I believe Wald has more initial conditions than just a 4-current $J^a$, in specifying the value of $E^a$ and $B^a$ on a hypersurface (assuming then, also, that the spacetime is globally hyperbolic). – soulphysics Aug 8 '14 at 19:55

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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