> There are higher order elements, just not higher order Lagrange > elements; try the HIERARCHIC basis instead if you want C0 higher order > polynomials. You can even do adaptive p refinement with them. > (probably hp, too, but I wouldn't guarantee that's working)
Thanks for the quick reply. I did think about Hierarchic basis but I'm currently working on a few diffusion problems and Lagrange basis seems to work quite well. I will try to switch my implementation to use Hierarchics instead and see if there is an increase/decrease in performance/accuracy. Thanks for the suggestion. I am currently looking for adaptive p or h but hp would be future work. So I do not care about hp adaptivity for now. And hopefully, I will be able to figure out the visualization bugs for higher order solution with VTK since that is the format I use most frequently. > In 2D, nth order HIERARCHICs should work on every second order > geometric element. In 3D, they work on hexes only at the moment. Well, that should be enough for me now to get started. If I add something, I'll send a patch. > If the impression you're getting is that we've each written exactly > enough higher order stuff for our own apps to work, you'd be right. I understand that all the developers are doing your PhDs and it is more important to finish your research before adding all the features you can imagine, into a library, that selected few people use everyday. I am using LibMesh for my research too and just wanted to get an idea on what I would need to add/change, if I need higher than second order elements. Thanks for all the suggestions thus far Roy. I will email you if I have more questions regarding this topic. Cheers, Vijay On Tue, Feb 24, 2009 at 1:31 PM, Roy Stogner <royst...@ices.utexas.edu> wrote: > > On Tue, 24 Feb 2009, Vijay S. Mahadevan wrote: > >> Are there any higher than second order Finite Elements in Libmesh ? I >> obviously do not see them in the class docs but just wondering if >> there was some trick to apply say a 5th order Lagrange basis to a >> QUAD4 elem. Then would Libmesh automatically create the extra dofs >> needed to make this unisolvent ?! > > There are higher order elements, just not higher order Lagrange > elements; try the HIERARCHIC basis instead if you want C0 higher order > polynomials. You can even do adaptive p refinement with them. > (probably hp, too, but I wouldn't guarantee that's working) > > There is a limitation on geometric element compatibility, though: you > need to have a node on every face or edge which corresponds to the > center of support of a basis function. So to use your example, a > QUAD4 won't work for most higher order elements (cubic HERMITEs being > the only exception) because there's no nodes to store edge degrees of > freedom. > > There's also a limitation on output: our .xda/.xdr formats will save > all your higher order data to machine precision, but at least the > visualization formats I use are first-order-only; I'm not sure we can > even plot quadratics except by turning them into 2^d linears first. > >> If this is gibberish and libmesh does not have higher than >> second-order elements, do let me know. > > In 2D, nth order HIERARCHICs should work on every second order > geometric element. In 3D, they work on hexes only at the moment. > >> If you have suggestions to implement higher order elements, I would >> be glad to hear them also. > > Adding support for HIERARCHIC tets and pyramids would first require > adding the geometric elements to support them: a PYRAMID5 doesn't have > the edge or face nodes necessary to support even second order > elements, and a TET10 doesn't have the face nodes. We need to round > things out with a PYRAMID14, PYRAMID19, and TET14 one of these days. > > If the impression you're getting is that we've each written exactly > enough higher order stuff for our own apps to work, you'd be right. > None of the missing bits I've mentioned would be hard to add to > libMesh if you need them; they're just tedious enough that nobody's > done so yet. > --- > Roy > ------------------------------------------------------------------------------ Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H _______________________________________________ Libmesh-devel mailing list Libmesh-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-devel
