On Sun, 6 Jul 2014, Miguel Angel Salazar de Troya wrote: > I was wondering if the Kelly Error estimator can be used for problems with > different material properties. That would change the value of the flux > depending on the region and this is what the estimator uses in its > calculations. I've seen that the implemention in libmesh only uses the > gradient of the field, not including any material property. I guess my > question would be if there are other error estimators more appropriate for > heterogeneous materials.
Using the flux jump ought to be a decent estimator for low-order FEM in heterogenous media, but as you guessed the implementation in libMesh is not just fine. I'd expect the effectivity to be effectively divided by the ratio of largest to smallest material property values in the domain. All the tricky code has been factored into JumpErrorEstimator, though. Creating a new subclass shouldn't be too hard. --- Roy ------------------------------------------------------------------------------ Open source business process management suite built on Java and Eclipse Turn processes into business applications with Bonita BPM Community Edition Quickly connect people, data, and systems into organized workflows Winner of BOSSIE, CODIE, OW2 and Gartner awards http://p.sf.net/sfu/Bonitasoft _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
