Least-squares finite element method and preconditioned conjugate gradient solution

From MaRDI portal





This work provides a receipt to construct least-squares finite element approximations for first-order systems of differential equations. Through the analysis of an elementary example the basic ideas of the method are clearly reviewed and the authors' modifications introduced. The general treatment of first-order systems are then examined in depth, showing the similarity between the authors' least-squares method and the standard Galerkin finite element approximation. Special preconditioners are derived regarding the iterative solution of the finite element equations by preconditioned conjugate gradients. An error analysis of the method is shown and error estimates derived which are confirmed by an experimental numerical analysis of a selected model problem. Bounds for the condition number of the finite element matrix of the first-order least-squares problem are evaluated. Finally, the optimality condition of a weighted least-squares functional is discussed and also confirmed by numerical experiments. Therefore, this work is a comprehensive study on least-squares finite elements, mainly on their mathematical aspects. However, no comparisons with standard Galerkin finite element methods regarding computer effectiveness and accuracy were included, and the reliability of this approach remains to be demonstrated.




Cited in
(21)








This page was built for publication: Least-squares finite element method and preconditioned conjugate gradient solution

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3771549)