= Constant-drift massive-Laplacian Dirichlet problem
Let $U\subset\mathbb R^d$ be a bounded open set, $A\in\mathbb R^d$ a constant <vector>, and $f\in L^2(U)$. On the real <Sobolev space> $H_0^1(U)$, consider the bounded <bilinear form>
$$
B(u,v)=\int_U\bigl(\nabla u\cdot\nabla v+uv+uA\cdot\nabla v\bigr).
$$
Its <weak solution> solves
$$
-\Delta u+u-A\cdot\nabla u=f
$$
with zero <Dirichlet boundary condition>. The minus sign follows from <integration by parts>: $\int uA\cdot\nabla v=-\int(A\cdot\nabla u)v$ because $A$ is constant. To represent the plus-drift equation, instead use the form with $-uA\cdot\nabla v$.
No smallness condition on $\|A\|$ is needed. For a smooth compactly supported $u$,
$$
\int_UuA\cdot\nabla u=\tfrac12\int_UA\cdot\nabla(u^2)=0.
$$
Density extends the identity to $H_0^1(U)$, since the drift pairing is continuous in the $H^1$ <norm>. Therefore $B(u,u)=\|u\|_{H^1}^2$ for every constant $A$. This is a <coercive bilinear form>, so the <Lax-Milgram theorem> gives a unique <weak solution> for arbitrary constant drift. The drift changes the equation but contributes no energy to this homogeneous-Dirichlet estimate.
Back to article page