Let $u\in H^1(\Omega)$ with $\nabla\times u=0$ in $\Omega\subset\mathbb{R}^3$ (open bounded domain), $u\times n=0$ on $\partial\Omega$ (where $n$ is a a normal vector to $\partial\Omega$), $\operatorname{div}(u)=0$ in $\Omega$ and $u\cdot n=0$ on $\partial\Omega$.

Prove that $u=0$.



2 Answers 2


For $u\in H^1(\Omega)$, you can write : \begin{equation} ||u||_{H^1(\Omega)}\leq C \{ ||u||_{L^2(\Omega)}+ ||div\ u||_{L^2(\Omega)} + ||curl\ u||_{L^2(\Omega)} + ||u\cdot n||_{H^{\frac{1}{2}}(\partial\Omega)}\} \end{equation} or, \begin{equation} ||u||_{H^1(\Omega)}\leq C \{ ||u||_{L^2(\Omega)}+ ||div\ u||_{L^2(\Omega)} + ||curl\ u||_{L^2(\Omega)} + ||u\times n||_{H^{\frac{1}{2}}(\partial\Omega)}\}. \end{equation} Now if $u\cdot n =0$ on $\partial\Omega$, then you have stronger estimate to write \begin{equation} ||u||_{H^1(\Omega)}\leq C \{ ||div\ u||_{L^2(\Omega)} + ||curl\ u||_{L^2(\Omega)}\} \end{equation} or, if $u\times n=0$ on $\partial\Omega$ then also \begin{equation} ||u||_{H^1(\Omega)}\leq C \{ ||div\ u||_{L^2(\Omega)} + ||curl\ u||_{L^2(\Omega)}\}. \end{equation}

Now, if $div\ u=curl\ u =0$ in $\Omega$ then it is straightforward to conclude $u = 0$ in $\Omega$.

Remark: The above statement holds for $u\in W^{1,p}(\Omega)$ but not necessarily for $u\in L^p(\Omega)$.

Source: (1) "On the Stokes equations with the Navier type boundary conditions": Cherif Amrouche and Nour Seloula.


I could not comment in the above answer by tuhin, but strangely it assumes $u=0$ to prove $u=0$!!!!

This result is actually far from trivial. It is not necessarily true if you only assume $\Omega$ to be just open and bounded. However, if you assume $\partial\Omega$ to be $C^{2},$ the result is true.

See Theorem 5.21 in the book "The Pullback Equation for Differential Forms" by Csato, Dacorogna, Kneuss, Progress in Nonlinear Differential Equations and their Applications, 83. Birkhäuser/Springer, New York, 2012.

The estimate, simplified to the case of the question is $$ \lVert u \rVert_{H^1(\Omega)} \leq C \left( \lVert \operatorname*{curl}u \rVert_{L^2(\Omega)} + \lVert \operatorname*{div}u \rVert_{L^2(\Omega)} + \lVert n \times u \rVert_{H^{1/2}(\partial\Omega)} + \lVert n \cdot u \rVert_{L^1(\partial\Omega)} \right).$$

The estimate itself is a corollary of Theorem 2 in Bolik's paper--- Bolik, Jürgen, "H. Weyls Boundary Value Problems for Differential Forms", Differential Integral Equations 14 (2001), no. 8, 937–952.


You must log in to answer this question.

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