3
$\begingroup$

In his PDE book, Evans demonstrates that for elliptic PDEs with Dirichlet boundary condition, the boundary term can be eliminated: Evans PDE

I am now wondering if this also works with Neumann boundary conditions. More precisely:

Let $\Omega$ be open and bounded with $C^1$ boundary. Let $L$ be in divergence form $$Lu=-\sum_{i,j=1}^n(a_{ij}(x)u_{x_i})_{x_j}+\sum_{i=1}^nb_i(x)u_{x_i}+c(x)u$$ and uniformly elliptic with $a_{ij},b_i,c\in L^{\infty}(\Omega)$, $f\in L^2(\Omega)$, $\psi\in L^2(\partial\Omega)$ and $h\in L^{\infty}(\partial\Omega)$ with $h\geq 0$.

Consider now the generalized boundary value problem \begin{cases} Lu = f & \text{in } \Omega \newline Bu - hu=\psi & \text{on } \partial \Omega \end{cases} for $Bu=-\sum_{i,j=1}^na_{ij}(x)u_{x_i}\cdot\nu_j(x)$ with $x\in\partial\Omega$.

Under what conditions on $\psi \in L^2(\partial \Omega)$ can the solvability (with solutions $u \in H^2(\Omega)$, such that both equations hold in the $L^2$ sense) be reduced to an equivalent problem of the form \begin{cases} Lv = g & \text{in } \Omega \newline Bv - hv=0 & \text{on } \partial \Omega? \end{cases}

Is it sufficient to find a solution $w$ to the homogeneous problem and then argue with $v=u-w$?

$\endgroup$

1 Answer 1

2
$\begingroup$

Yes, this works, it is a general feature of linear equations (not restricted to pdes).

First find $w$ that satisfies the boundary condition $$ Bw-hw = \psi, $$ and has sufficient regularity, then solve $$ Lv = f - Lw, \ Bv-hv=0 $$ for $v$.

Note that this decomposition approach was necessary to get a weak formulation for the Dirichlet problem, while for Neumann boundary conditions this is not necessary.

$\endgroup$
3
  • $\begingroup$ Thank you for your answer! So is it correct that I do not need to impose any additional conditions on $\psi$? $\endgroup$
    – sina1357
    Commented Jul 4 at 11:39
  • 1
    $\begingroup$ $\psi\in L^2(\partial\Omega)$ yields $w\in H^1$, and then $v\in H^1$. $\endgroup$
    – daw
    Commented Jul 4 at 11:40
  • $\begingroup$ I have one last question: Did I understand correctly that for example $w\in H^2(\Omega)$ (and therefore $v\in H^2(\Omega)$) follows from $\psi\in H^1(\partial\Omega)$? Thank you! $\endgroup$
    – sina1357
    Commented Jul 4 at 11:55

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .