Physics 7xx: Axially symmetric Green function

Remember that inside of an integration over axially symmetric charge distributions

    \[{1\over |\bm{r}-\bm{r'}|}=\sum_{n=0} {r_<^n\over r_>^{n+1}}P_n(\cos\theta)P_n(\cos\theta')\]

and the left-hand side is a Green function for \mathbb{R}^3

    \[\nabla^2 G(x,x')=-4\pi \delta^3(x-x')\]

Example 1
A sphere of radius a is permanently and uniformly polarized \bm{P}=P\bm{k}. Find the electric field at it’s center.\\
This is equivalent (in the sense that the fields are the same) to a sphere with the following surface and bulk charge densities

    \[\rho_{pol}=-\bm{\nabla}\cdot\bm{P}=0\quad\mbox{and}\quad \sigma_{pol}=\bm{P}\cdot\bm{r}=P \cos\theta\]

When we have a given charge density, we always use

    \[V(r)=\int{\rho(r')\over 4\pi\epsilon_0 |\bm{r}-\bm{r}'|} d^3 r, \qquad \int{\sigma(r')\over 4\pi\epsilon_0 |\bm{r}-\bm{r}'|} dA\]

to find V at a point, or

    \[ \bm{E}(r)=\int{\rho(r') (\bm{r}-\bm{r'})\over 4\pi\epsilon_0 |\bm{r}-\bm{r}'|^3} d^3 r, \qquad \int{\sigma(r') (\bm{r}-\bm{r'})\over 4\pi\epsilon_0 |\bm{r}-\bm{r}'|^3} dA\]

to find the electric field, so in this case

    \[\bm{E}(0)=-\frac{1}{4\pi\epsilon_0 a^3}\int_0^{2\pi} d\phi \int_0^{\pi} \sin\theta \, d\theta \,  a^2 \, \Big(a\bm{k}\cos\theta +a\bm{i}\sin\theta\cos\phi\]

    \[+a\bm{j}\sin\theta \sin\phi\Big)\Big(P \cos\theta\Big)=-\frac{P}{3\epsilon_0}\bm{k}\]

Example 2
Find the equivalent magnetization “charge” density \rho_M and equivalent magnetization “surface charge density” \sigma_m=\bm{M}\cdot\bm{n} that has the same magnetic field and scalar potential that a magnetized sphere of radius a, magnetization \bm{M}=M_0 \bm{k} would make.

Because there are no conduction currents and \bm{\nabla}\times\bm{M}=0 we can use a scalar potential \Omega, that will satisfy \nabla^2\Omega=0. Uniqueness says

    \[\Omega_{in}=Ar\cos\theta, \quad \Omega_{out}={B\over r^2}\cos\theta\]

Continuity of H_\theta tells us that Aa={B\over a^2}, and of B_r with \bm{B}=\mu_0(\bm{H}+\bm{M})=\mu_0(-\bm{\nabla}\Omega+\bm{M})

The field is equivalent to that made by a surface monopole density \sigma_M=\bm{M}\cdot\bm{n}=M \, \cos\theta=M \, P_1(\cos\theta).

    \[ \Omega={1\over 4\pi}\int_0^{2\pi} a^2 \, d\phi' \int_0^\pi \sin\theta' d\theta' \Big(M \, P_1(\cos\theta')\Big)\]

    \[\cdot \Big( \sum_{n=0}^\infty {a^n\over r^{n+1}} P_n(\cos\theta')P_n(\cos\theta)\Big)\]

    \[={Ma^3\over 3\pi}{P_1(\cos\theta)\over r^2}\]

the potential of a magnetic dipole m={4\pi a^3 M\over 3}.

Suppose that you have a region R bounded by a surface S and can construct (by images or variable separation) a Green function for R that vanishes of S. Suppose that you know the potential V(x') on all points x'\in S and want V(x) for x\in R, in which \nabla^2V=0.

    \[\int_R \bm{\nabla}'\cdot(V(x')\bm{\nabla}'G(x,x'))d^3x'=\oint_S V(x')\bm{\nabla}'G(x,x')\cdot{n}'d^2x'\]

    \[=\oint_S V(x'){\partial G\over \partial n'}(x-x') d^2x'\]

by the divergence theorem. This depends only on the potential on the surface. But

    \[\int_R \bm{\nabla}'\cdot(V(x')\bm{\nabla}'G(x,x'))d^3x'=\int_R \bm{\nabla}'V(x')\cdot\bm{\nabla}'G(x,x')d^3x'\]

    \[+\int_R V(x')\nabla'^2G(x,x'))d^3x'\]

Integrate the first term by parts, which removes the derivative from G, and then evaluates G on the surface S where it vanishes, leaving only

    \[-\int_R \nabla'^2V(x')'G(x,x'))d^3x'+\int_R V(x')\nabla'^2G(x,x'))d^3x'\]

    \[=0-4\pi \int_R V(x')\delta^2(x-x')d^3x'=-4\pi V(x)\]

and so we have the very useful result

    \[V(x)=-{1\over 4\pi}\oint_S V(x'){\partial G\over \partial n'}(x-x') d^2x'\]

relating V in R to V on S.

Home 2.0
error: Content is protected !!