On Thu, Mar 10, 2016 at 9:02 AM, Roy Stogner <royst...@ices.utexas.edu> wrote:
> > On Thu, 10 Mar 2016, Paul T. Bauman wrote: > > On Thu, Mar 10, 2016 at 8:47 AM, Roy Stogner <royst...@ices.utexas.edu> >> wrote: >> >> On Thu, 10 Mar 2016, li....@siat.ac.cn wrote: >> >> > Is the normal direction from "fe_face->get_normals(); " a >> outside normal or inside normal? >> >> Outside normal, IIRC. >> >> >> Correct. >> > > We should probably document this somewhere. fe_abstract.h? Good idea. Probably in FEMap as well (since the FEAbstract call is a just a shim to the FEMap call). > One catch is that "outside" is determined by the local node ordering >> so that, if you're using an external mesh generator, you need to be >> sure that it orients them correctly so that outside is what you >> think it is. For example, Cubit has a "flip normal" function to do >> this. If you're using libMesh mesh generation, outside is what you >> think it is. >> > > I'm a bit surprised that that doesn't cause more problems as well. I > guess if JxW is negative *everywhere* then you just end up solving > -R(u)=0 instead of R(u)=0? > Arg, I wasn't paying attention. I was thinking about manifolds and the normal of the interior, not of the face. What I said doesn't apply to the face normals, those will alway be outward. I shouldn't answer email before I've finished my first cup of coffee. Sorry for any confusion. I'll drop in a documentation PR for get_normals() ------------------------------------------------------------------------------ Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785111&iu=/4140 _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
