scientific article; zbMATH DE number 3883555

From MaRDI portal
Publication:3217548





The author uses the classical results of the finite element theory in order to obtain optimal error estimates for a boundary penalty method applied to Poisson equations with non homogeneous Dirichlet boundary conditions. The variational formulation of the boundary value problem studied in this paper is: \[ (1)\quad To\quad find\quad u,\quad such\quad that\quad (u-g)\in H^ 1_ 0(\Omega)\int_{\Omega}\nabla u\cdot \nabla vdx=\int_{\Omega}fvdx\quad \forall v\in H^ 1_ 0(\Omega) \] where g is a given function in \(H^ 1(\Omega)\), and f in \(L^ 2(\Omega)\). This problem is approximated by using a boundary penalty method which consists in solving the following discrete version of (1): \[ (2)\quad To\quad find\quad u_ h\quad in\quad V_ h\int_{\Omega}\nabla u_ h\cdot \nabla v_ hdx+h^{-\sigma}\int_{\partial \Omega}(u_ h-g)v_ hds=\int_{\Omega}fv_ hdx\quad \forall v_ h\in V_ h \] where \(V_ h\) is the standard finite element approximation of \(H^ 1(\Omega)\) by piecewise linear functions, h is the step size of the triangulation, and \(\sigma\) is the penalty parameter. The main theorem of this paper shows that, if u(solution of (1)) belongs to \(H^ 2(\Omega)\), then there exists a constant C, independent on h, such that: \((3)\quad \| u-u_ h\|_{H^ 1(\Omega)}\leq Ch^{3/4}\| u\|_{H^ 2(\Omega)}\) where \(u_ h\) is the (unique) solution of (2) with \(\sigma =3/2\) (this value of \(\sigma\) appears in the proof of (3) as the optimal choice of the penalty boundary parameter. For other values of \(\sigma\) one obtains a smaller power of h in (3)). Finally, the author proves that the estimate (3) can be improved assuming that u belongs to \(H^ 3(\Omega)\). In this case, the optimal choice of the penalty parameter is \(\sigma =1\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3217548)