An ill-posed moving boundary problem for doubly-connected domains (Q1097419): Difference between revisions

Given a doubly-connected domain \(\Omega\) in \({\mathbb{R}}^ N \)we look for a family \(\Omega\) (t), for t in some interval \(-\epsilon <t<\epsilon\), of doubly-connected domains such that \(\Omega (0)=\Omega\) and such that \(\partial \Omega (t)\) moves with the velocity \(-\nabla u_{\Omega (t)}\), where \(u_{\Omega (t)}\) denotes the harmonic function in \(\Omega\) (t) with boundary values zero and one on the outer and inner components of \(\partial \Omega (t)\) respectively. This problem, which may arise e.g. in electro-chemistry, is ill-posed for the inner component of \(\partial \Omega (t)\) as t increases and for the outer component as t decreases. In order for a solution to exist in some interval \(-\epsilon <t<\epsilon\) it is necessary that \(\partial \Omega\) is real analytic and we prove that, conversely, this condition also is sufficient at least for a solution in a certain weak sense to exist (for some \(\epsilon >0)\). The weak solution is constructed by first solving a Cauchy problem for the Laplacian operator with data on \(\partial \Omega\) (this step takes care of the ill- posedness) and then applying an elliptic variational inequality.