Use the Green's function for the half-plane to solve the problem $$\begin{cases} \Delta u(x_1,x_2) = 0 \ \ \text{in the half-plane} \ x_2 > 0\\ u(x_1,0) = g(x_1) \ \ \text{on the boundary} \ x_2 = 0 \end{cases}$$ where the boundary condition is $$g(x_1) = \begin{cases} \pi \ \ \text{if} \ |x_1| \leq 1\\ 0 \ \ \text{if} \ |x_1| > 1\\ \end{cases}$$ Verify that the solution is, in fact, harmonic, and satisfies the given boundary condition.\ \noindent Remark: In our theorems, we assumed that $g$ is continuous, which this one is not. What is happening at the two discontinuities?

Attempted (beginning) solution - The Green's function for the half-plane is given by $$G(x,x_0) = \frac{1}{2\pi}\log |x - x_0| - \frac{1}{2\pi}\log |x - x^{*}_0|$$ where $x = x(x,y)$, $x_0 = (x_0,y_0)$, and $x^{*}_0 = (x_0,-y_0)$.

Now I am pretty confused with what to do next here, going through my professors notes I can seem to find anything related to Green's function in the half-plane. I had to search for it online to find it which is what is presented above.

The only part that seems remotely related in my notes is on the Half-space where we found the solution to Poisson's problem in the upper half place for $2$ dimensions which is $$u(x_1,x_2) = -\frac{1}{4\pi}\int_{-\infty}^{\infty}\int_{0}^{\infty}\log \frac{(y_1 - y_2)^2 + (y_2 - x_2)^2}{(y_1 - x_1)^2 + (y_2 + x_2)^2}f(y_1,y_2)dy_2 dy_1 + \frac{1}{\pi}\int_{-\infty}^{\infty}\frac{x_2}{(y_1 - x_1)^2 + x_2^{2}}g(y_1)dy_1$$

My apologies ahead of time if this question is a bit messy. I just don't understand what I need to do here. Any suggestions are greatly appreciated.

Attempted solution (sort of) - The Green's function for the half-plane is given by $$G(x,x_0) = \frac{1}{2\pi}\log |x - x_0| - \frac{1}{2\pi}\log |x - x^{*}_0|$$ where $x = x(x,y)$, $x_0 = (x_0,y_0)$, and $x^{*}_0 = (x_0,-y_0)$. In our case we have $$G(x_1,x_2) = \frac{1}{2\pi}\log|x_1 - x_2| - \frac{1}{2\pi}\log|x_1 - \tilde{x_2}|$$ Thus $$\frac{\partial G}{\partial n} = \frac{1}{2\pi}\left(\frac{x_1 - \tilde{x_2}}{|\tilde{x_2} - x_1|^2}\right)\Bigg|_{\tilde{x_2} = 0} = \frac{1}{2\pi x_1}$$ Thus we have $$u(x_1,x_2) = \frac{1}{2\pi}\int_{-\infty}^{\infty}\frac{1}{x_1}g(x_1)dx_1$$ Not sure if this is correct, any suggestions or comments are appreciated. If it is correct, is it possible to evaluate further? How do I show that the solution is harmonic?

Attempted solution 3 - The Green's function for the half-plane is given by $$G(x,x_0) = \frac{1}{2\pi}\log |x - x_0| - \frac{1}{2\pi}\log |x - x^{*}_0|$$ where $x = x(x,y)$, $x_0 = (x_0,y_0)$, and $x^{*}_0 = (x_0,-y_0)$. In our case we have \begin{align*} G &= \frac{1}{2\pi}\log\left(\sqrt{(\tilde{x_1} - x_1)^2}\right) - \frac{1}{2\pi}\log\left(\sqrt{(\tilde{x_1}-x_1)^2 + (\tilde{x_2} + x_2)^2}\right)\\ &= -\frac{1}{4\pi}\left[\log\left((\tilde{x_1} - x_1)^2 + (\tilde{x_2} - x_2)^2\right) - \log\left((\tilde{x_1} - x_1)^2 + (\tilde{x_2} + x_2)^2\right)\right] \end{align*} Now,

\begin{align*} \frac{\partial G}{\partial n}\Bigg|_{\tilde{x_2} = 0} &= -\frac{1}{4\pi}\left[\frac{2(\tilde{x_2} - x_2)}{(\tilde{x_1} - x_1)^2 + (\tilde{x_2} - x_2)^2} - \frac{2(\tilde{x_2} + x_2)}{(\tilde{x_1} - x_1)^2 + (\tilde{x_2} + x_2)^2} \right]\Bigg|_{\tilde{x_2} = 0}\\ &= \frac{x_2}{\pi((\tilde{x_1} - x_1)^2 + x_2^{2})} \end{align*} Thus we have $$u(x_1,x_2) = \frac{1}{\pi}\int_{-\infty}^{\infty}\frac{x_2}{(\tilde{x_1} - x_1)^2 + x_2^{2}}g(x_1)dx_1$$ We see that $\Delta u(x_1,x_2) = 0$ since $g(x_1)$ is a constant function and taking the derivative with respect to $x_1$ will equal $0$. Therefore the solution $u(x_1,x_2)$ is harmonic.

I am not sure if this is correct or not. Also, how do we show that this solution satisfies the boundary condition?



Using Green's Second Identity with $\nabla'^2 G(x_1,x_2|x_1',x_2')=-\delta(x_1-x_1')\delta(x_2-x_2')$ in distribution to write

$$\begin{align} u(x_1,x_2)&=\int_{x_2\ge 0}\left(G(x_1,x_2|x_1',x_2')\underbrace{\nabla'^2 u(x_1',x_2')}_{=0}-u(x_1',x_2')\underbrace{\nabla'^2 G(x_1,x_2|x_1',x_2')}_{=-\delta(x_1-x_1')\delta(x_2-x_2')}\right)\,dx_1'\,dx_2'\\\\ &=-\int_{-\infty}^\infty \left.\left(\underbrace{G(x_1,x_2|x_1',x_2')}_{=0\,\text{at}\,x_2'=0}\frac{\partial u(x_1',x_2')}{\partial x_2'}-\underbrace{u(x_1',x_2')}_{=g(x_1')\,\text{at}\,x_2'=0}\frac{\partial G(x_1,x_2|x_1',x_2')}{\partial x_2'}\right)\right|_{x_2'=0}\,dx_1'\\\\ &=\int_{-\infty}^\infty g(x_1')\left.\left(\frac{\partial G(x_1,x_2|x_1',x_2')}{\partial x_2'}\right)\right|_{x_2'=0}\,dx_1 \end{align}$$

  • $\begingroup$ I am not sure if I follow from my book: Partial Differential Equations by Walter A. Strauss, Green's second identity is given by $$\iiint_{D}(u\Delta v - v\Delta u)dx = \iint_{\partial D}\left(u\frac{\partial v}{\partial n} - v\frac{\partial u}{\partial n}\right)dS$$ $\endgroup$
    – justanewb
    Jul 19 '17 at 23:18
  • $\begingroup$ I probably just don't understand the formula or how to apply it to this problem. $\endgroup$
    – justanewb
    Jul 19 '17 at 23:19
  • $\begingroup$ You have the correct formula for $\mathbb{R}^3$ if $v=G$; for $\mathbb{R}^2$ reduce the 3-D integral over $D$ to a 2-D integral over the upper-half plane and the integral over $\partial D$ to a contour integral that is comprised of the real line and a semicircle of "infinite radius." The integral over the "infinite semicircle" is $0$ given the behavior of $G$. The left-hand side of the one you wrote becomes $u(x_1,x_2)$. The right-hand side becomes $- \int_{-\infty}^\infty g(x_1')\frac{\partial G}{\partial x_2'}\,dx_1'$ as expected. $\endgroup$
    – Mark Viola
    Jul 19 '17 at 23:32
  • $\begingroup$ My apologies, my intellectual inferiority does not comprehend. I should have mentioned the formula is for $\mathbb{R}^3$. But I just don't understand how we use it for this particular problem it just doesn't make sense to me on where I begin given what I have. $\endgroup$
    – justanewb
    Jul 19 '17 at 23:35
  • $\begingroup$ The solution I posted is all you need. It gives you the answer. Just use the form of $G$ you wrote in your OP, take its derivative with respect to $x_2'$, then after that let $x_2'=0$. You will get an integral representation for $u$ $\endgroup$
    – Mark Viola
    Jul 19 '17 at 23:39

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.