Re: Neuman condition for Poisson problem..

On Jun 22, 12:06 am, Fedor <malabar_caro...@xxxxxxxxx> wrote:
Hi all,

I m trying to solve the Poisson problem
\Delta u = f
on the unit square, with Neuman condition on the 4 boudaries:
du/dn = g
To do that, I am using finite difference. There are 2 major problems:

1/ of course, if u is a solution, then u+Cst is also a solution, so
the problem is not well-posed. To remedy that, I had first thought to
impose the value u(1,1)=0, or something like that ...

2/ but one needs a compability condition for the existence and
unicity:
\int_{square} f = \int_{boudary} g

That means that if one writes the matrix associated to the finite
difference scheme:
A * U = V
where V depends on f and g, the matrix A will not be invertible.. so
what are the usual ways to solve that problem ? Any goods references
on the net, books, or ideas ?

Thank you,
Fedor

It was my understanding that no math would be required for this ng.

.

Relevant Pages