Consider a domain $$D= \{(x,y) : x>0, y>0 \}$$ Let $\textbf x = (x,y)$ and $\xi =(\xi_x , \xi_y)$. Find the Green's function, $G(\textbf x , \xi)$ such that $$\nabla ^2 G=\delta (\textbf x - \xi), \, \, \, \, \, \, \textbf x \in D$$ subject to $$\frac{\partial G}{\partial x}(0,y,\xi)=0, \, \, \, \, \, \text{for} \, \, \, \, y>0$$ and $$G(x,0,\xi)=0, \, \, \, \, \, \text{for} \, \, \, \, x>0$$

Using method of images:

We have the source $\xi=(\xi_{x},\xi_{y})$ and images sources:

$\xi_1=(-\xi_{x},\xi_{y})$, $\xi_2=(\xi_{x},-\xi_{y})$, $\xi_3=(-\xi_{x},-\xi_{y})$

So we have $\nabla ^2 G=\delta ( \textbf x - \xi )\pm \delta ( \textbf x - \xi_1 ) \pm \delta ( \textbf x - \xi_2 )\pm \delta ( \textbf x - \xi_3 )$

Can someone please clearly explain how you determine if the plus/minus signs are in fact plus or minus please.

I know that it is to do with the boundary conditions but don't understand how.


The correct extension is $$\nabla^2G = \delta(\mathbf{x}-\xi) + \delta(\mathbf{x}-\xi_1) - \delta(\mathbf{x}-\xi_2) - \delta(\mathbf{x}-\xi_3)$$ To see why, interpret $G$ as electric potential. Then, the delta functions correspond to point charges, and we have the conditions that (i) the potential is symmetric across $x=0$, and (ii) the potential is zero along $y=0$. The first condition implies that the sign on the delta function at $\xi_1$ should be the same as the sign on the delta function at $\xi$, while the second condition implies that the sign on delta function at $\xi_2$ should be opposite the sign at $\xi$. Then, since $\xi_3$ is the reflection of $\xi_2$ over $x=0$, the delta function there should have the same sign as $\xi_2$ (alternatively, since it is the reflection of $\xi_1$ over $y=0$, it should have the opposite sign from the delta function at $\xi_1$).

Solving for $G$ yields $$G = \ln\frac{1}{|\mathbf{x} - \xi|} + \ln\frac{1}{|\mathbf{x} - \xi_1|} - \ln\frac{1}{|\mathbf{x} - \xi_2|} - \ln\frac{1}{|\mathbf{x} - \xi_3|}$$ and we can verify that $G$ satisfies the required boundary conditions.

  • $\begingroup$ Does $\xi$ basically mean an arbitrary point anywhere on the xy plane? And does $\xi_1$ mean an arbitrary point anywhere on left quadrants? And $\xi_2$ etc...? $\endgroup$ – snowman May 7 '16 at 10:58
  • $\begingroup$ Is there an informal way to do this? I still don't get it... $\endgroup$ – snowman May 7 '16 at 10:59
  • 1
    $\begingroup$ Remember that the Green's function is defined such that the solution to $$\Delta u = f$$ is $u(x) = (G(x,\cdots) * f(\cdot))(x)$. In other words, the Greens function tells you how the differential equation responds to an impulse of one unit at the point $\xi$. The points $\xi_1$, $\xi_2$, and $\xi_3$ are obtained by reflecting $\xi$ over the boundary lines $x=0$ and $y=0$. $\endgroup$ – Strants May 7 '16 at 21:03
  • 1
    $\begingroup$ I'm also really not sure what you mean by an 'informal' way to find the Greens function. What do you mean by 'informal'? $\endgroup$ – Strants May 7 '16 at 21:06
  • 1
    $\begingroup$ (1): $\xi_x$ and $\xi_y$ are positive real numbers, but they are not fixed. Remember that for $$\Delta u = f$$ with the given boundary conditions, we have $$u(\mathbf{x}) = \iint_{x>0,y>0} G(\mathbf{x}, \xi) f(\xi)\;d\xi_x d\xi_y$$ (2): Sorry, I meant to write $\xi_2$. (3): Well, you've already noticed that if you reflect over a line with Neumann boundary conditions, you don't change the sign of the delta function. Why would that change when you reflect a second time? $\endgroup$ – Strants May 11 '16 at 5:55

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.