Lagrangian formulation of domain decomposition methods: a unified theory (Q2504446)

The authors present an elegant and unique interpretation with augmented Lagrangian of some domain decomposition methods. The continuous and discrete analysis are performed for the Laplace equation. The possible limitations arising in the case of general mesh partitioning with cross-points are indicated. The condensed interface problem issue from these methods is clearly analyzed. Domain decomposition methods based on one Lagrange multiplier are efficient for solving ill-conditioned problems in a parallel environment. Several variants of these methods have been developed in the last decade. Recent extensions of this method to two Lagrange multipliers and mixed type interface conditions have been analyzed in the literature. A suitable choice of these mixed interface conditions ensures the well-posedness of the local sub-problems in each subdomain. An interpretation framework of some of these domain decomposition methods with one and two Lagrange multipliers and with Neumann and mixed interface conditions are presented and analyzed here. Also, the parallel implementation of these methods with a message passing environment is analyzed.