Convergence of natural adaptive least squares finite element methods (Q2407473)

This paper analyses the convergence of natural adaptive mesh-refining first-order div least squares finite element methods. The authors prove a \(Q\)-linear convergence of the associated adaptive mesh-refining strategy for a sufficiently fine initial mesh with some sufficiently large bulk parameter for piecewise constant right-hand sides in a Poisson model problem. The proof is based mainly on some modifications of known supercloseness results to non-convex polygonal domains plus the flux representation formula. For the sake of simplicity, the analysis is performed for the lowest-order case in two-dimensions.