From: Roy Stogner <roystgnr@ic...>  20100427 21:31:29

On Tue, 27 Apr 2010, Wharmby, Andrew W Civ USAF AFMC 711 HPW/RHDO wrote: > I’m in the process of learning libmesh and have been going through > the examples but can seem to figure out how to apply a source term > to a system. For instance, if I build a cube and want to apply a > gaussian heat source to one of its faces on its center, how is that > done? In other words, you want to apply a heat flux on a Neumann boundary? When you integrate the heat transfer PDE by parts, you'll be left with a boundary flux term. Our examples use Dirichlet boundary conditions, the test functions are zero there, and the Neumann terms disappear; however the code that integrates penalty Dirichlet boundary condition terms is very similar to what you'll need to integrate a weak Neumann boundary condition term. We ought to add a Neumann example; we seem to get this question once or twice a year. > In example 9 where does the .5*dt in the Fe and Ke build loops come > from? I want to make sure I’m translating my derivation correctly > into the code so I’m trying to follow example 9. Those fall out of the "1/2"s in CrankNicolson time integration.  Roy 