Mixed boundary value problems on cylindrical domains

From MaRDI portal
(Redirected from Publication:513033)



Abstract: We study second-order divergence-form systems on half-infinite cylindrical domains with a bounded and possibly rough base, subject to homogeneous mixed boundary conditions on the lateral boundary and square integrable Dirichlet, Neumann, or regularity data on the cylinder base. Assuming that the coefficients A are close to coefficients A_0 that are independent of the unbounded direction with respect to the modified Carleson norm of Dahlberg, we prove a priori estimates and establish well-posedness if A_0 has a special structure. We obtain a complete characterization of weak solutions whose gradient either has an L^2-bounded non-tangential maximal function or satisfies a Lusin area bound. Our method relies on the first-order formalism of Axelsson, McIntosh, and the first author and the recent solution of Kato's conjecture for mixed boundary conditions due to Haller-Dintelmann, Tolksdorf, and the second author.


Let \(\Omega\) is a bounded set in \(\mathbb{R}^d\) (\(d\geq 2\)). The authors consider the elliptic system \[ -\sum_{i,j=0}^d\sum_{k=1}^m\partial_i(a_{i,j}^{l,k}(t,x)\partial_ju_k(t,x))=0, \quad l=1,\dots,m, \quad (t,x)\in \mathbb{R}_+\times \Omega, \] subjected to the mixed Dirichlet/Neumann lateral boundary conditions \[ u=0\text{ on }\mathbb{R}_+\times \mathcal{D},\quad \nu \cdot A\nabla_{t,x}u=0\text{ on }\mathbb{R}_+\times(\partial \Omega \setminus \mathcal{D}). \] Here, \(\nu\) is the outer normal vector to \(\mathbb{R}_+\times \Omega\), \(\mathcal{D}\) is a closed subset of \(\partial \Omega\) and \(A=A(t,x):= \big(a_{i,j}^{l,k}(t,x)\big)_{i,j=0,\dots,d}^{l,k=1,\dots,m}\) is the coefficient tensor (the coefficients being complex). The set \(\Omega\) is assumed to be bounded, open, connected, and \(d\)-regular (with respect to the Lebesgue measure), and the set \(\mathcal{D}\) is assumed either empty or \(d-1\)-regular (with respect to the Hausdorff measure). Finally, the set \(\partial \Omega \setminus \mathcal{D}\) is assumed to be locally Lipschitz. The coefficient tensor \(A\) is assumed bounded in \(\mathbb{R}_+\times \Omega\), satisfying an ellipticity/accretivity condition, and such that \(\|A-A_0\|_C<\infty\), for some \(t\)-independent tensor \(A_0\), where \(\|\cdot\|_C\) is a modified Carleson norm. The authors establish \textit{a priori} estimates for weak solutions to the above problem satisfying one of the three following inhomogeneous boundary conditions on \(\{0\}\times \Omega\): 1) \(u_{\mid t=0}=\varphi\in L^2(\Omega)^m\), 2) \({(A\nabla_{t,x}u)_{\bot}}_{\mid t=0}=\varphi\in L^2(\Omega)^m\), 3) \((A\nabla_xu)_{\mid t=0}=\varphi\in L^2(\Omega)^m\). The \textit{a priori} estimate is proved for weak solutions whose non-tangential maximal function has finite \(L^2\)-norm in the case of the boundary conditions \(2)\) and \(3)\), and for weak solutions satisfying a Lusin area bound in the case of the boundary condition \(1)\). In both cases, additional regularity results are established when \(A=A_0\). A well posedness result is also proved for \(t\)-independent coefficient tensors satisfying some additional condition and for \(t\)-dependent coefficient tensors sufficiently near (with respect to the above modified Carleson norm) to a \(t\)-independent coefficient tensor for which well posedness holds. The proof relies on the transformation of the above elliptic system with the mixed lateral boundary conditions into an equivalent non-autonomous evolution first order equation and on the Kato root estimate for operators with mixed boundary conditions.











This page was built for publication: Mixed boundary value problems on cylindrical domains

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q513033)