Finite element approximation of the pure Neumann problem using the iterative penalty method (Q884596)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Finite element approximation of the pure Neumann problem using the iterative penalty method |
scientific article |
Statements
Finite element approximation of the pure Neumann problem using the iterative penalty method (English)
0 references
6 June 2007
0 references
The author considers the finite element solution of the pure Neumann problem \(-\Delta u=f\) in \(\Omega\) with \(\partial u/\partial n=0\) on \(\partial\Omega\), where \(\Omega\subset {\mathbb R}^d\), \(d=1,2,3\), is a bounded open domain with boundary \(\partial\Omega\). The penalty method is used to solve the following equivalent variational equation for \(u \in H^1(\Omega)\): \(a(u_\rho,v)+\rho\int_{\partial \Omega}u_\rho v\,ds=(f,v)\; \forall v\in H^1(\Omega)\). This variational equation is solved by means of an approximate solution, \(u_{\rho,h}\in V_h\), of the equation \(a(u_{\rho,h},v_h)+\rho\int_{\partial \Omega}u_{\rho,h} v_h\,ds=(f,v_h)\; \forall v_h\in V_h\). In this paper, only the influence of \(\rho\) is considered. It is observed that the condition number depends heavily on \(\rho\) for fixed \(h\). The following iterative scheme based on the penalty method: \(a(u_{\rho,h}^0,v_h)+\rho\int_{\partial \Omega}u_{\rho,h}^0 v_h\,ds=(f,v_h)\; \forall v_h\in V_h\), and for \(n=1,2,\ldots\), \(a(u_{\rho,h}^n,v_h)+\rho\int_{\partial \Omega}u_{\rho,h}^n v_h\,ds= (u_{\rho,h}^{n-1},v_h)\; \forall v_h\in V_h\), is designed to avoid unstable computation by taking \(\rho\) not too small since the error estimate is shown to be \(O(\rho^{n+(1/2)}+h)\).
0 references
Neumann problem
0 references
penalty method
0 references
iterative scheme
0 references
condition number
0 references
Poisson equation
0 references
finite element
0 references
error estimate
0 references
0 references