On Fri, Sep 14, 2012 at 4:11 PM, Paul T. Bauman <[email protected]> wrote:
> For the test problem, no. For the coupled magnetostatics problem I'm > working on that will use the Nedelec element capability - yes. But I'm not > far enough along yet to start down that road (still > debugging/characterizing application). But it's in terms of potentials > anyway, so I'm dealing with Laplace coupled to curl curl = f (coupled to > other things), so the analogy to the test problem is not really good anyway. > > But maybe I'm misunderstanding where you're going? > Fast/multilevel solvers for Maxwell are somewhat involved, either requiring you to use a very specific discretization or to assemble some auxiliary operators so that the MG cycle can run in a nicer auxiliary space. If it's going to be a bottleneck and you care about it a lot, you'll want to set up plumbing for these auxiliary operators and we'll need to prioritize an interface to make that as clean as possible. If it's not in a particularly crucial place, don't worry about it yet. ------------------------------------------------------------------------------ Got visibility? Most devs has no idea what their production app looks like. Find out how fast your code is with AppDynamics Lite. http://ad.doubleclick.net/clk;262219671;13503038;y? http://info.appdynamics.com/FreeJavaPerformanceDownload.html _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
