1
$\begingroup$

What's the solution to Laplace's equation $\nabla^2V=0 $ in the annulus with centre 0, inner radius 1, and outer radius 2, with boundary conditions $V=0$ on the inner boundary and $V=1$ on the outer boundary?

By separating variables, I got the general solution $V(r,\theta)=\Sigma[(C_n r^n + D_n r^{-n})(A_n \cos n\theta + B_n \sin n\theta)]$. The inner boundary condition gives $D_n = -C_n \forall n$. The outer boundary condition gives $\Sigma[C_n(2^n - 2^{-n})(A_n \cos n\theta + B_n \sin n\theta)] = 1.$ But this gives a Fourier series with zero constant term being identically equal to 1. By uniqueness of Fourier series, isn't this a contradiction?

Or is there a better method than using separation of variables, which gives a different general form for the solution?

Many thanks for any help with this!

$\endgroup$
  • 1
    $\begingroup$ Just a guess: have you tried a radially symmetric solution? $\endgroup$ – Siminore Apr 24 '12 at 9:55
0
$\begingroup$

Look for a solution $V=V(r)=V(|x|)$. Then $\Delta V=0$ can be rewritten as $$V''(r)+\frac{2}{r}V'(r)=0$$ in the interval $(1,2)$. Integrating this equation you find $V(r)=C_1+\frac{C_2}{r}$ for suitable constants $C_1$ and $C_2$. The boundary conditions imply $C_1=2$ and $C_2=-2$. Hence $V(x)=2-\frac{2}{|x|}$ solves your equation in the annulus.

Edit: as pointed out in the comments, this treated the case in dimension $3$. In dimension $2$, radially symmetric harmonic functions have the form $r \mapsto C_1+C_2 \log r$, but the conclusion is similar: just match the boundary conditions and find $C_1$ and $C_2$.

$\endgroup$
  • $\begingroup$ Your formula for $\Delta V$ applies to ${\mathbb R}^3$. In the plane there is no factor of $2$. $\endgroup$ – Christian Blatter Apr 24 '12 at 12:53
  • $\begingroup$ Sorry, when I read $\nabla^2$ I immediately think of the physical space. $\endgroup$ – Siminore Apr 24 '12 at 13:01
  • $\begingroup$ Your $V(r)$ does not solve the corrected equation. $\endgroup$ – Christian Blatter Apr 24 '12 at 13:26
2
$\begingroup$

I don't know what you did exactly in your computation process. At any rate the "eigenvalue" $n=0$ is a double eigenvalue, and besides the constant solution there is an additional solution, namely $u(r,\theta):=\log r$. This should give you enough manoeuvring space to satisfy the boundary conditions with a rotationally symmetric solution.

$\endgroup$
1
$\begingroup$

Functions of the form $A + B \ln(r)$ are harmonic, and you can use your boundary conditions to solve for $A$ and $B$. You obtain the solution ${\displaystyle {\ln(r) \over \ln(2)}}$.

$\endgroup$

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.