“…[40,32], for example), the question of developing efficient iterative solvers for the solution of the resulting (linear) system of equations has been addressed only recently and only in the framework of the h-version of the DGFEM (h-DGFEM). For example, a wide class of domain decomposition methods for discontinuous Galerkin approximations of elliptic problems has been proposed and analyzed in [24,34,12,27,3,5,21]. We point out that, because of differences in the variational formulation associated with the underlying discontinuous polynomial spaces, the stiffness matrices arising from discontinuous Galerkin approximations possess different sparsity structures compared to those from conforming methods, and indeed, for a given mesh and polynomial degree distribution, the underlying matrix is typically larger in the DGFEM setting.…”