- #1
PDE_warrior
- 1
- 0
First, my deepest apologies if I am asking a trivial question, or asking it in the wrong forum.
I am trying to solve a PDE, which I have already reduced to canonical form and simplified to the full extent of my abilities. The PDE is:
u_xy + a(x,y) u_x + b u_y = 0, with a(x,y)=2/(x+y) and b=-1/2.
The method of separation of variables does not work. I have tried posing u(x,y)=v(x,y) w(x,y) and choosing v in such a way as to simplify the resulting equation, to no avail.
Since I know there are two families of solutions involving each an arbitrary function of one variable, I tried setting u(x,y)=v(x,y,F(z(x,y))), with F arbitrary. This leads to two cases: z_x = 0 OR z_y = 0, as well as an ODE that must be satisfied to remove the dependency in F_z. I can solve that ODE for either case, but either way the leftover equation then can only be satisfied by setting F=0.
I am aiming for a solution of the form u(x,y) = anything involving x, y, up to two arbitrary functions and any number of integrations, so that I could potentially plug in the proper functions once I have decided exactly what the boundary conditions should be. In particular, any solution expressed using a function that solves the adjoint problem (as is found in Garabedian) is totally useless to me since the adjoint problem is just as hard to solve as the original one.
I am thankful for any and all help anyone may provide, even just suggestions or pointing me in the correct direction.
I am trying to solve a PDE, which I have already reduced to canonical form and simplified to the full extent of my abilities. The PDE is:
u_xy + a(x,y) u_x + b u_y = 0, with a(x,y)=2/(x+y) and b=-1/2.
The method of separation of variables does not work. I have tried posing u(x,y)=v(x,y) w(x,y) and choosing v in such a way as to simplify the resulting equation, to no avail.
Since I know there are two families of solutions involving each an arbitrary function of one variable, I tried setting u(x,y)=v(x,y,F(z(x,y))), with F arbitrary. This leads to two cases: z_x = 0 OR z_y = 0, as well as an ODE that must be satisfied to remove the dependency in F_z. I can solve that ODE for either case, but either way the leftover equation then can only be satisfied by setting F=0.
I am aiming for a solution of the form u(x,y) = anything involving x, y, up to two arbitrary functions and any number of integrations, so that I could potentially plug in the proper functions once I have decided exactly what the boundary conditions should be. In particular, any solution expressed using a function that solves the adjoint problem (as is found in Garabedian) is totally useless to me since the adjoint problem is just as hard to solve as the original one.
I am thankful for any and all help anyone may provide, even just suggestions or pointing me in the correct direction.
Last edited: