May question is about possible approaches to solve the following system $$ \begin{array}{rcl} \nabla{n}&=&n\,\mathbf{E},\\ \nabla\cdot\mathbf{E}&=&1-n, \end{array} $$ in general with some boundary conditions (BC). In particular, I'm interested to solve it in the polar coordinates $$ \begin{array}{rcl} \partial_rn&=&nE_r,\\ \partial_{\varphi}n&=&rnE_{\varphi},\\ \partial_r{(rE_r)}+\partial_{\varphi}{(E_{\varphi})}&=&r(1-n), \end{array} $$ plus some BC.

Any suggestions?

Update 1

According to request of @nicoguarolet let's say that boundary conditions are as follows $$ \begin{array}{c} \left.n(r,\varphi)\right|_{r=r_L}=n_0\approx1,~ \left.E_{\varphi}\equiv E_r(r,\varphi)\right|_{r=r_L}=E_0\approx0,\\ \left.\partial_rE(r,\varphi)\right|_{r=r_R}=0,\\ \end{array} $$ where $[r,\,\varphi]\in[r_L,\,r_R]\times[\varphi_L,\,\varphi_R].$ Hope, that I haven't missed somewhat.

  • $\begingroup$ Boundary conditions are important for the solution of PDE, can you add them? $\endgroup$
    – nicoguaro
    Commented Apr 30, 2020 at 19:24
  • $\begingroup$ @nicoguaro Original post was updated. $\endgroup$ Commented Apr 30, 2020 at 19:39

1 Answer 1


That's not a valid system of PDEs. In particular, the equation $$ \nabla n = n\mathbf E $$ implies that you can find a function $n$ whose gradient equals a given vector field (n\mathbf E), but that's not possible for arbitrary vector fields. To see this, apply the curl to both sides: $$ \nabla \times \nabla u = \nabla\times (n\mathbf E) $$ and remember that $\nabla \times \nabla = 0$. In other words, the equation will only have a solution if $$ \nabla\times (n\mathbf E) = 0, $$ but your other equation does not guarantee that that is the case.


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.