How to set up integral for solution to PDE using Green's function?

34 Views Asked by At

I've found Green's function, the problem is I've never done the integration part to find the solution for a 2-D problem or higher before, so I'd like a little guidance for setting up the integral. The integral should look like this:

$$\tag{1}u(\textbf{x}) = \int_a^b f(\textbf{x'}) G(\textbf{x},\textbf{x'}) d\textbf{x'}$$

I just don't know what that's supposed to look like when we expand from vector (or bold) notation. I know it should expand into a double integral like this:

$$\tag{2}u(x,y) = \int_0^H\int_0^L f(x',y')G(x,y,x',y') dx'dy'$$ I'm just hung up on the piecewise definition of Green's function complicating things. In my example, I had a piecewise definition along y, and it was continuous along x. To illustrate it had the form $G(x,y,x',y') =G(x,x')G(y,y') = G(x,x')\cases{G(y,y'^-) & y<y' \\ G(y,y'^+) & y>y'}$. This is my guess for what I think it should be:

$$\tag{3}\boxed{u(x,y) = \int_0^L\bigg(\int_0^y f(x',y')G(x,y,x',y'^-)dy'+\int_y^H f(x',y')G(x,y,x',y'^+)dy'\bigg)dx'}$$

Please let me know if this looks correct!

Below is my work for my specific case


If my guess if correct, then for my problem:

$$u_{xx} + u_{yy} = x(x-1)y(y-1); \ \ 0 \le x \le L, 0 \le y \le H \\ BCs \cases{u(0,y) = 0 \\ u(L,y) = 0 \\ u(x,0) = 0 \\ u(x,H) = 0}$$

where I used an eigenfunction expansion to set up my Green's function and I arrived to:

$$\tag{1}G(x,y,x',y') = \sum_{m=1}^\infty c_m G(x,x') \begin{cases} G(y,y'^-) & y<y' \\ G(y,y'^+) & y>y'\end{cases}$$ $$\tag{2}c_m = \frac{2}{m\pi \sinh(\frac{m\pi}{L} H)}\\ G(x,x') = \sin(\frac{m\pi}{L}x)\sin(\frac{m\pi}{L}x'), \ \ \ G(y,y'^-) = \sinh(\frac{m\pi}{L}(y'-H)) \sinh(\frac{m\pi}{L}y), \\ G(y,y'^+) = \sinh(\frac{m\pi}{L}y')\sinh(\frac{m\pi}{L}(y-H)) $$

$$\tag{3}\boxed{u(x,y) = \sum_{m=1}^\infty c_m \Bigg[\int_0^L\bigg(\int_0^y g(y')G(y,y'^-) + \int_y^H g(y')G(y,y'^+) dy'\bigg)h(x')G(x,x') dx'\Bigg]}$$

Where I split up $f(x,y) = g(x)h(y); g(x) = x(x-1),h(y) = y(y-1)$ to make the equation look a bit cleaner.


EDIT: I computed the results for this integral and the results were atrocious not to mention it was a highly verbose result that took a very long time to find all in addition to being incorrect and I do NOT want to go back and pick through where the mistake was made. Is this where I use Green's Theorem to sidestep the area integral?