2009/9/12 Roy Stogner <royst...@ices.utexas.edu> > > > On Sat, 12 Sep 2009, Ted Kord wrote: > > Whereas, this IS the correct thing to do: >> >> Fe(i) += phi_face[i][qp] * g; >> > > No, it isn't. > > The difference is there's no multiplication by JxW_face[qp] because >> applying >> the Neumann B.C does not require integration. >> > > Yes, it does. You've covered up a bug, not found it, I'm afraid. > This may work in 1-D (where JxW_face is supposed to simply be 1 on > your single "integration point"), but it will give inaccurate results > in 2D/3D. > > Thanks all. I'm very excited and motivated. >> > > Sorry if I've squelched that any, but I didn't want to see your code > break later when you move to higher dimensions. > --- > Roy >
It's good you did that 'cause having gone over the code again, it turns out that even though Fe(i) += phi_face[i][qp] * g; works fine (because JxW_face is supposed to simply be 1 on the single "integration point" like you pointed out), the reason I was getting incorrect results before was 'cause I was using JxW[qp] instead of JxW_face[qp] like you mentioned. So, it all ends well, motivation's still there and I've learnt quite a bit. Thanks. Ted ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
