A big result from vector calculus is the Helmholtz Decomposition: for any vector-valued function $\mathbf{F} : \mathbb{R}^{3} \to \mathbb{R}^3$ that is well-behaved enough, we can always decompose it as follows: $$ \mathbf{F}(\mathbf{r}) = - \boldsymbol{\nabla} \Phi(\mathbf{r}) + \boldsymbol{\nabla} \times \mathbf{C}(\mathbf{r}) $$

There always exist functions $\Phi : \mathbb{R}^3 \to \mathbb{R}$ and $\mathbf{C} : \mathbb{R}^3 \to \mathbb{R}^3$ for any choice of such $\mathbf{F}$.

$\ $

My question is, does there exist a similar result for $4$-vectors? I'm reading about gauge fields in QFT....is there some way to decompose any gauge field $A_{\mu}(x) = A_{\mu}(x_0,x_1,x_2,x_3)$ into a sum similar to the above?

I'm thinking something along the lines of: $A_{\mu}(x) = \partial_{\mu} \lambda(x) + \mathrm{something}$

I can't think of what the curl term would look like here.

  • 9
    $\begingroup$ You have Hodge's theorem, $f=\mathrm df_1+\mathrm d^\star f_2+f_3$, where $f_3$ is harmonic. $\endgroup$ Nov 19, 2017 at 21:05
  • 1
    $\begingroup$ @AccidentalFourierTransform: Now, please translate into physicist-speak ;) What would the second term look like? Something like $\epsilon_\mu{}^{\alpha\beta\gamma} \partial_{[\alpha}B_{\beta\gamma]}$? $\endgroup$
    – Christoph
    Nov 19, 2017 at 21:52
  • 3
    $\begingroup$ @Christoph For many of us, forms are now physicist speak too :) $\endgroup$
    – JamalS
    Nov 19, 2017 at 23:17
  • $\begingroup$ @Christoph, it is customary to use the metric tensor and leave all 4 indices of the Levi-Civita pseudotensor on the same position, i.e. either all „up” or all ”down”. $\endgroup$
    – DanielC
    Nov 19, 2017 at 23:39
  • 2
    $\begingroup$ According to chapter 2, page 21 of Prof Wheeler's classical field theory notes, Hodges theorem becomes: $V_\mu=S_\mu + I_\mu + V^0_{\mu} \;\text{where} \;\ \partial^{\mu}S_{\mu} = 0, \;\partial_{\mu}I_{\nu}-\partial_{\nu}I_{\mu}=0, \;\Box V^0_{\mu}=0$ Perhaps someone more competent than myself could show this as the accepted answer? $\endgroup$ Feb 14, 2018 at 1:04

1 Answer 1


The Hodge decomposition states that any $1$-form $\omega$ can be written as $$\omega = d\alpha+\ast\, d\ast\beta +\gamma$$ where $\alpha$ is a function, $\beta$ is a $2$-form and $\gamma$ is a harmonic $1$-form. Here $d$ is the exterior derivative and $\ast$ is the Hodge dual.

In coordinates, this means that we can write a covector $A_\mu$ as $$A_\mu = P_\mu + Q_\mu + R_\mu$$ The corresponding conditions are as follows. $P$ corresponds to $d\alpha$ and in Minkowski spacetime $$P = d\alpha\,\,\, \mathrm{for\,\,some}\, \alpha \iff dP=0$$ In coordinates this means that $\partial_\nu P_\mu -\partial_\mu P_\nu=0$. Similarly $Q$ satisfies $d\ast Q =0$, which in coordinates translates as $\partial^\mu Q_\mu=0$. Finally $R$ must be harmonic, so $\partial^\nu \partial_\nu R_\mu = 0$.

This question on Math Overflow may also be of interest.

  • $\begingroup$ How does one retreive the 3D fundamental theorem of vector calculus from this? $\endgroup$ Sep 2, 2018 at 2:54
  • $\begingroup$ @Greg.Paul Pardon me, what's the 3D fundamental theorem of calculus? $\endgroup$
    – John Donne
    Sep 2, 2018 at 8:05
  • $\begingroup$ Sorry, I mean the result I quoted in the question - the Helmholtz decomposition. This says that $F_{\mu} = \partial_{\mu} \Phi + \epsilon_{\mu\nu\sigma} \partial_{\nu} C_{\sigma}$ for 3D functions $F$ for some $\Phi$, $C$. Is there a way to see this specific result from the Hodge decomposition? $\endgroup$ Sep 5, 2018 at 13:48
  • 1
    $\begingroup$ @Greg.Paul Roughly speaking yes: take your 3D vector field $F$ and apply the Hodge decomposition. You get a gradient plus a curl plus a harmonic term. Then if $F$ vanishes sufficiently fast at infinity (faster than $1/r$) you know that the harmonic function goes to zero faster than $1/r$, so it must be identically zero (this is a standard result in analysis). Hence you're left with gradient + curl. The details are more complicated however; see the question I linked in the answer. This paper may also be of help: arxiv.org/abs/1404.3679 $\endgroup$
    – John Donne
    Sep 5, 2018 at 16:05

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.