0
$\begingroup$

Given a vector potential describing an infinitely thin line of flux,

$$\vec{A} = \frac{\Phi}{2\pi r} \vec{e}_\varphi,$$

How can I calculate the curl so that the magnetic field is given by

$$\vec{B} = \delta(x) \delta(y) \Phi \vec{e}_z,$$

or similar expression? The expression for the curl (in cylindrical coordinates) comes down to this:

$$\nabla \times \vec{A} = \vec{e}_z \frac{1}{r} \frac{\partial}{\partial r} (r A_\varphi).$$

I am aware there is a singularity at $r=0$ which will probably give me the answer. I just don't see how. I'm asking here, because well, a mathematician might get a fit caused by my liberal use of Dirac delta's above. Any help is appreciated.

$\endgroup$
  • 1
    $\begingroup$ Use the 2-dimensional Green's function for the Laplacian to find an alternative representation of the 2-dimensional delta function. $\endgroup$ – Ryan Unger Feb 9 '15 at 16:49
1
$\begingroup$

The explanation of my comment: Recall $$\Delta\log r=2\pi\delta^2(x,y)$$ On the right we have $$\frac{1}{r}\frac{\partial}{\partial r}\left(r\frac{\partial\log r}{\partial r}\right)$$ Your potential can be written as $$\vec A=\frac{\Phi}{2\pi}\frac{\partial\log r}{\partial r}\vec e_\varphi$$ The curl is then $$\vec\nabla\times\vec A=\frac{\Phi}{2\pi}\frac{1}{r}\left(r\frac{\partial\log r}{\partial r}\right)\vec e_z=\frac{\Phi}{2\pi}\Delta\log r\,\vec e_z=\delta^2(x,y)\Phi\vec e_z$$ As was to be shown.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Right, and where does the naive approach (i.e. filling in the curl directly) actually go wrong? I'm still unable to pinpoint exactly which step is illegal/wrong and how to perform that step generically in a mathematically correct way. Thanks. $\endgroup$ – rubenvb Feb 10 '15 at 15:21
  • $\begingroup$ @rubenvb: Filling in the curl directly? $\endgroup$ – Ryan Unger Feb 10 '15 at 16:37
  • $\begingroup$ $\nabla \times \boldsymbol A = \frac{1}{r} \partial_r r \frac{\Phi}{2\pi r} = \frac{\Phi}{2\pi r} \partial_r 1 = 0$. Or something like that. $\endgroup$ – rubenvb Feb 10 '15 at 19:40
  • $\begingroup$ Of course that is only the $\boldsymbol e_z$ part, the rest is identically zero. $\endgroup$ – rubenvb Feb 10 '15 at 19:47
  • $\begingroup$ @rubenvb: The fraction $\frac{1}{r}\partial_r 1$ is not well defined at $r=0$. It is the "peak" of the delta function, which is clear from the identity $\Delta\log r=2\pi\delta^2(x,y)$. $\endgroup$ – Ryan Unger Feb 10 '15 at 20:36
0
$\begingroup$

You have to take into account that your vector potential is defined only for $r >0$. For $r=0$, the vector potential actually vanishes, $\vec{A}(0) = \vec{0}$. You can show this by considering a thick solenoid and letting its width go to zero. Therefore, the correct expression for the vector potential of a thin solenoid is $$\vec{A} = \theta(r) \frac{\Phi}{2\pi r} \vec{e}_\varphi.$$ Here, $\theta(\cdot)$ is the Heaviside step function. This will give you a correct magnetic field which is proportional to a delta function.

| cite | improve this answer | |
$\endgroup$

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.