A characterization of harmonic \(L^r\)-vector fields in three dimensional exterior domains (Q2141388)

The Helmholtz-Weyl decomposition is a fundamental theorem in fluid mechanics. Roughly said, it describes a vector field in terms of its divergence-free and rotation-free components. Geometrically, such decomposition leads to the study of the de Rham-Hodge-Kodaira decomposition. This paper shows that, despite the lack of compactness of an exterior domain \(\Omega\), the dimension of the space of harmonic vector fields for \(\Omega\) is finite and is topologically invariant. For a domain \(D\subset\mathbb{R}^3\) with smooth boundary \(\partial D\), let \begin{align*} X_{\mathrm{har}}(D) &:=\left\{h \in C^{\infty}(\bar{D}) ; \operatorname{div} h=0, \operatorname{rot} h=0 \text{ in } D,\left.h \cdot v\right|_{\partial D}=0\right\}, \\ V_{\mathrm{har}}(D) &:=\left\{h \in C^{\infty}(\bar{D}) ; \operatorname{div} h=0, \operatorname{rot} h=0 \text{ in } D, h \times\left. v\right|_{\partial D}=0\right\}. \end{align*} The finiteness of \(\dim(V_{\mathrm{har}}(D))\) and \(\dim(X_{\mathrm{har}}(D))\) is crucial for applying the Fredholm alternative to determine the solvability of certain PDEs in fluid mechanics, which stems from the estimate \[ \|\nabla u\|_{L^{r}(D)} \leq C\left(\|\operatorname{rot} u\|_{L^{r}(D)}+\|\operatorname{div} u\|_{L^{r}(D)}+\|u\|_{L^{r}(D)}\right) \] for all \(u \in H^{1, r}(D)\) satisfying \(u \cdot v=0\) or \(u \times v=0\) on \(\partial D\) and Rellich's theorem when \(D\) is bounded. The paper, however, focuses on the case when \(\Omega\) is an exterior domain, in which case the embedding of \(H^{1, r}(\Omega)\) into \(L^r(\Omega)\) is not compact in general. For an exterior domain \(\Omega\), the authors consider the spaces \(X_{\mathrm{har}}^{r}(\Omega)\) and \(V_{\mathrm{har}}^{r}(\Omega)\) instead: \[ X_{\mathrm{har}}^{r}(\Omega):=\left\{h \in L^{r}(\Omega) ; \operatorname{div} h=0, \operatorname{rot} h=0 \text{ in } \Omega,\left.h \cdot \nu\right|_{\partial \Omega}=0\right\}, \] \[ V_{\mathrm{har}}^{r}(\Omega):=\left\{h \in L^{r}(\Omega) ; \operatorname{div} h=0, \operatorname{rot} h=0 \text{ in } \Omega, h \times\left.\nu\right|_{\partial \Omega}=0\right\}. \] Then they prove that \textbf{Theorem 2.1.} Let \(1<r<\infty\) and \(\Omega \subset \mathbb{R}^{3}\) be an exterior domain in \(\mathbb{R}^{3}\) with smooth boundary \(\partial \Omega\). Then the spaces \(X_{\mathrm{har}}^{r}(\Omega)\) and \(V_{\mathrm{har}}^{r}(\Omega)\) are finite dimensional. The proof is based on the estimate \[ \|\nabla u\|_{L^{r}(\Omega)} \leq C\left(\|\operatorname{rot} u\|_{L^{r}(\Omega)}+\|\operatorname{div} u\|_{L^{r}(\Omega)}+\|u\|_{L^{r}(\Omega\cap B)}\right) \] and the fact that if \(h\) is bounded in \(X_{\mathrm{har}}^{r}(\Omega)\) or in \(V_{\mathrm{har}}^{r}(\Omega)\) respectively, the norm \(\left\|h\right\|_{L^{r}(\Omega \backslash B)}\) will be arbitrarily small for uniformly large \(B\). It is also very interesting to explore the relationship between the dimensions of \(X_{\mathrm{har}}^{r}(\Omega)\) and \(V_{\mathrm{har}}^{r}(\Omega)\) and certain topological invariances of the domain \(\Omega\). Let \(L\) and \(N\) be the first and the second Betti number of \(\Omega\) respectively: \textbf{Theorem 2.2.} Let \(1<r<\infty\) and \(\Omega \subset \mathbb{R}^{3}\) be an exterior domain with smooth boundary \(\partial \Omega\) satisfying some assumptions. Then the following assertions hold. \begin{itemize} \item[(i)] \(\operatorname{dim} X_{\mathrm{har}}^{r}(\Omega)=N\). \item[(ii)] If \(3 / 2<r<\infty\), then \(\operatorname{dim} V_{\mathrm{har}}^{r}(\Omega)=L\). \item[(iii)] If \(1<r \leq 3 / 2\), then \(\operatorname{dim} V_{\mathrm{har}}^{r}(\Omega)=L-1\). \end{itemize} One notices that \(\operatorname{dim} V_{\mathrm{har}}^{r}(\Omega)\) depends on both topological information of \(\Omega\) and the Lebesgue's exponent \(r\), which is quite different from the situations of bounded domains. The proof relies on the existence of an appropriate extension of functions on certain exterior domains \(\Omega\) and the solvability of the weak Neumann problem of the Poisson equations.