[Libmesh-users] DirichletBoundary question From: David Knezevic - 2013-08-05 11:34 ```I'm passing a subclass of FunctionBase to a DirichletBoundary object, and the FunctionBase uses a MeshFunction from a separate (dim-1)-dimensional mesh to return a value. In this context, the lower-dimensional mesh "matches" the relevant boundary of the dim-dimensional mesh. I sometimes have some tolerance issues, so that I have to find the "nearest point" in the (dim-1)-dimensional mesh. At present, I just find the "nearest node" on the (dim-1)-dimensional mesh (rather than the nearest interior point or whatever), and it works fine. I gather that this is because (at least in my case) the DirichletBoundary only evaluates the FunctionBase at the nodes on the boundary of the dim-dimensional mesh. I'm wondering if this will be true in general, or if this is just due to the fact that I'm using nodal (i.e. LAGRANGE) basis functions in this case? Thanks! David ```
 Re: [Libmesh-users] DirichletBoundary question From: Roy Stogner - 2013-08-05 13:05 ```On Mon, 5 Aug 2013, David Knezevic wrote: > I sometimes have some tolerance issues, so that I have to find the > "nearest point" in the (dim-1)-dimensional mesh. At present, I just find > the "nearest node" on the (dim-1)-dimensional mesh (rather than the > nearest interior point or whatever), and it works fine. > > I gather that this is because (at least in my case) the > DirichletBoundary only evaluates the FunctionBase at the nodes on the > boundary of the dim-dimensional mesh. I'm wondering if this will be true > in general, or if this is just due to the fact that I'm using nodal > (i.e. LAGRANGE) basis functions in this case? On p==1 elements our projection is just nodal interpolation and the evaluation will only use nodal data. Even p==2 Lagrange elements may produce data requests from non-nodal quadrature points, though. --- Roy ```
 Re: [Libmesh-users] DirichletBoundary question From: David Knezevic - 2013-08-05 13:06 ```On 08/05/2013 03:04 PM, Roy Stogner wrote: > > On Mon, 5 Aug 2013, David Knezevic wrote: > >> I sometimes have some tolerance issues, so that I have to find the >> "nearest point" in the (dim-1)-dimensional mesh. At present, I just find >> the "nearest node" on the (dim-1)-dimensional mesh (rather than the >> nearest interior point or whatever), and it works fine. >> >> I gather that this is because (at least in my case) the >> DirichletBoundary only evaluates the FunctionBase at the nodes on the >> boundary of the dim-dimensional mesh. I'm wondering if this will be true >> in general, or if this is just due to the fact that I'm using nodal >> (i.e. LAGRANGE) basis functions in this case? > > On p==1 elements our projection is just nodal interpolation and the > evaluation will only use nodal data. Even p==2 Lagrange elements may > produce data requests from non-nodal quadrature points, though. Understood, thanks! ```