Helmholtz theorem states that given a smooth vector field $\mathbf{H}$, there are a scalar field $\phi$ and a vector field $\mathbf{G}$ such that

$\mathbf{H}=\nabla \phi +\nabla \times \mathbf{G}$


$\nabla \mathbf{\cdot G}=0$

Is this decomposition unique? That is, given $\mathbf{H}$, are the fields $\phi$, $\mathbf{G}$ satisfying the above equations unique?

Edit: Unique, up to an additive constant.


  • 1
    $\begingroup$ Do you mean "unique up to an additive constant"? (A latex-suggestion, use \mathbf{G} $\mathbf{G}$ instead of "poor man's bold" \pmb{G} $\pmb{G}$ which looks ugly if you do something wrong like $\pmb{\cdot G}$) $\endgroup$ – Myself May 28 '11 at 17:29
  • $\begingroup$ @Myself: Yes, unique up to an additive constant. Edited my post to correct that. $\endgroup$ – becko May 28 '11 at 17:38
  • $\begingroup$ Look here en.wikipedia.org/wiki/Talk:Helmholtz_decomposition $\endgroup$ – user24703 Feb 9 '12 at 18:58
  • $\begingroup$ @Alexandr What's the problem? The Helmholtz/Hodge decompositions only make sense when you have a Riemannian metric anyway, and in the Helmholtz type statement, like most statements in classical vector calculus, the Euclidean metric is implicitly used to identify vectors and covectors. $\endgroup$ – Willie Wong Feb 9 '12 at 19:19
  • $\begingroup$ cross-post: physics.stackexchange.com/q/10522/3064 $\endgroup$ – becko May 30 '13 at 20:56

The decomposition is not unique without further conditions. You can add linear terms to $\phi$ and $\mathbf G$ that yield constant contributions to $\mathbf H$ that cancel:

$$ \begin{eqnarray} \phi &\to& \phi + z\;, \\ \nabla\phi &\to& \nabla\phi + \mathbf e_z\;, \\ \mathbf G &\to& \mathbf G + \frac{1}{2}(y\mathbf e_x-x\mathbf e_y)\;, \\ \nabla\times\mathbf G &\to& \nabla\times\mathbf G - \mathbf e_z\;. \end{eqnarray} $$

However, I think that if you impose suitable conditions that the fields decay at infinity, the decomposition should be unique.

  • $\begingroup$ Thanks. Yes, I realize now that boundary conditions are essential. After boundary conditions have been specified, the resolution is unique. $\endgroup$ – becko May 29 '11 at 0:07

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.