Extension and representation of divergence-free vector fields on bounded domains (Q5929467): Difference between revisions

The following problems are considered: 1) to extend a divergence-free vector field \(u\in H^{s,p} (\Omega)\) on a bounded connected domain \(\Omega\subset \mathbb{R}^n\), with \(\partial\Omega =\Gamma_0 \cup\cdots \cup \Gamma_b\), \(\Gamma_j= \partial O_j\), to a divergence-free vector field \(\widetilde u\) also smooth as \(u\) on a neighbourhood of \(\overline\Omega\), and 2) to produce an antisymmetric 2-tensor field \(\nu\in H^{s,p} (\Omega)\) such that \(u=\text{Div} \nu+ \sum\lambda_j g_j\), \(g_j(x)= {x-y_j \over|x-y_j|^n}\), \(y_j\in O_j\), \(x\in \mathbb{R}^n\setminus \{y_j\}\), \(\lambda_j \in\mathbb{R}\). After Problem 2 solved it is possible to extend \(\nu\) to \(\widetilde\nu\) and set \(\widetilde u=\text{Div} \widetilde\nu+ \sum\lambda_jg_j\). In \S 2 one treats \(u\in LP(\Omega)\), \(1<p< \infty\); in \S 3 one has \(u\in H^{s,p} (\Omega)\), \(s\geq 0\), \(1<p <\infty\), and in \S 4 the results of \S 3 are extended to the situation where \(\partial\Omega\) is of class \(C^{k,\sigma}\), \(k\in Z^+\), \(\sigma\in(0,1]\), \(k+\sigma\geq 1\). Section 5 discusses the extensions of class \(C^{k+1,\sigma}\). See also [\textit{M. Mitrea} and \textit{M. Taylor}, Commun. Partial Differ. Equations 25, 1487-1536 (2000; Zbl 1157.35345)].