A trapping principle and convergence result for finite element approximate solutions of steady reaction/diffusion systems (Q2480893)

The author considers for \({\mathbf u}=(u_1,\ldots,u_m)^t\) the nonlinear steady state system \[ \nabla \cdot p_k(\vec{x},{\mathbf u}) \nabla u_k = f_k(\vec{x},{\mathbf u}), \quad k=1,\dots,m, \] in a convex polyhedral domain \(\Omega \subset \mathbb{R}^{N}\) subject to mixed Dirichlet-Neumann boundary conditions \[ {\mathbf u-\hat{u}}=0 \text{ on } \Sigma_D, \quad \frac{\partial {\mathbf u}}{\partial \nu}=0 \text{ on } \Sigma_N, \] where \(\Sigma_D\) is a nonempty relatively open portion of the boundary \(\partial \Omega\) and \(\Sigma_N=\partial \Omega \backslash \Sigma_D\). The function \({\mathbf f}=(f_1,\ldots,f_m)^t\) is assumed to be outward pointing on the trapping region boundary. For the quasi-uniform simplicial decomposition of \(\Omega\) the non-obtuseness (rather than acuteness) condition is required. It states that the inward normals, associated with an arbitrary pair of element's faces, determine an angle with nonpositive cosine. As a consequence, all off-diagonal elements of the stiffness matrix are nonpositive. Also the stiffness matrix is supposed to be irreducible. In the main theorem the existence of a weak solution \({\mathbf u}\) of the given problem is proved via the weak convergence in \((H^1(\Omega))^m\) of a subsequence of finite element approximations. This subsequence is also strongly convergent in \((L^2(\Omega))^m\). The uniqueness of solutions is not assumed neither proved.