A new equilibrated residual method improving accuracy and efficiency of flux-free error estimates
This paper presents a new methodology to compute guaranteed upper bounds for the energy norm of the error in the context of linear finite element approximations of the reaction–diffusion equation. The new approach revisits the ideas in Parés et al. (2009) [6, 4], with the goal of substantially reduc...
| Autores: | , |
|---|---|
| Formato: | artículo |
| Fecha de publicación: | 2017 |
| País: | España |
| Recursos: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/99967 |
| Acesso em linha: | https://hdl.handle.net/2117/99967 https://dx.doi.org/10.1016/j.cma.2016.10.010 |
| Access Level: | acceso abierto |
| Palavra-chave: | Numerical analysis Exact/guaranteed/strict bounds Fully computable a posteriori error estimation Adaptivity Reaction–diffusion equation Flux-free Equilibrated boundary tractions Anàlisi numèrica Àrees temàtiques de la UPC::Matemàtiques i estadística::Anàlisi numèrica::Mètodes en elements finits |
| Resumo: | This paper presents a new methodology to compute guaranteed upper bounds for the energy norm of the error in the context of linear finite element approximations of the reaction–diffusion equation. The new approach revisits the ideas in Parés et al. (2009) [6, 4], with the goal of substantially reducing the computational cost of the flux-free method while retaining the good quality of the bounds. The new methodology provides also a technique to compute equilibrated boundary tractions improving the quality of standard equilibration strategies. The zeroth-order equilibration conditions are imposed using an alternative less restrictive form of the first-order equilibration conditions, along with a new efficient minimization criterion. This new equilibration strategy provides much more accurate upper bounds for the energy and requires only doubling the dimension of the local linear systems of equations to be solved. |
|---|