From: Roy S. <roy...@ic...> - 2006-03-29 18:48:24
|
On Tue, 28 Mar 2006, Roy Stogner wrote: > On that note, we probably ought to put a global tag on the CVS head > before I do my next big commit. I don't think there's anything that > should break any code which doesn't do p refinement, but it's a big > change that touches a lot of files, and just in case people do start > running into bugs it might be prudent to make reverting the whole > changeset easy. This morning's CVS head has been tagged, and my p refinement code is committed. If anyone's feeling adventurous enough to cvs update, let me know (preferably with a stack trace) if you encounter any bugs. The commit log: Initial p refinement support - uniform p refinement is working; adaptive p and hp refinement has yet to be well tested. Major changes include: DofMap now examines element p_level when assigning and indexing degrees of freedom. MeshRefinement now tests for level one p_level conformity when asked, and can do uniform p refinement and coarsening. Quadrature rules and FE objects now examine Elem::p_level() when they are able to. Library internals have been changed to pass total polynomial order (= base polynomial order + element p_level) to functions that cannot examine Elem::p_level() themselves. Major limitations include: System::project_vector does not yet work with p refinement. The option to turn off projection of solution vectors has been added as a temporary workaround. Non-hierarchical element types cannot support adaptive p refinement, and will be limited to uniform p refinement for the forseeable future. --- Roy |