> The tool I work on here at work can compare any two arbitrary > solutions to each other... even with completely non-nested grids. The > user gets to decide what kind of crime they want to commit though. > Often (if the meshes are really dissimilar) we'll use an overkill > rendezvous mesh to transfer both of the original solutions to first... > then compare them on that grid. (Note: To do this in libMesh use two > MeshFunctions and 3 meshes) > > My favorite way though... is just to lay down a super fine quadrature > rule and sample both grids at those quadrature points. Sure... on one > of the meshes you'll be integrating non-smooth functions with Gauss > quadrature... but the results are usually pretty good. > > I do think that some of the capabilities in this code at my current > job are going to be put into libMesh over the next year. libMesh > already has a lot of the infrastructure for it... it wouldn't be too > hard to make it a little more generic.
Take a look at src/apps/grid2grid.cc. It is a little dated and tuned for a specific code, but it is an 70% solution that could guide the implementation. This was implemented specifically for Bill to do mesh refinement studies as part of his thesis. Specifically, he wanted to demonstrate ||u - u_fine|| where the meshes spanned the same space and were not nested. We essentially followed Derek's suggested approach... We used an overkill quadrature rule on the fine mesh. -Ben ------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
