ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

Looks like a bug in FIAT in DiscontinuousVectorLagrange with q = 0.

This is entity_ids for q = 0:

{0: {0: {}, 1: {}, 2: {}, 3: {}}, 1: {0: {}, 1: {}, 2: {}, 3: {}, 4:
{}, 5: {}}, 2: {0: {}, 1: {}, 2: {}, 3: {}}, 3: {0: [0, 1, 2]}}

This is entity_ids for q = 1:

{0: {0: {}, 1: {}, 2: {}, 3: {}}, 1: {0: {}, 1: {}, 2: {}, 3: {}, 4:
{}, 5: {}}, 2: {0: {}, 1: {}, 2: {}, 3: {}}, 3: {0: [0, 1, 2, 3]}}

Shouldn't the last dictionary should be {0: [0]} for q = 0?
The number of nodes should be 0, not 3. Right?

/Anders

On Sat, Jun 04, 2005 at 11:01:20AM -0500, Anders Logg wrote:
> Yes, this seems to be a bug. I will take a look.
> 
> /Anders
> 
> On Sat, Jun 04, 2005 at 03:54:31PM +0200, Johan Jansson wrote:
> > Hi!
> > 
> > There seems to be a bug in the dofmap generation for discontinuous
> > Lagrange elements. What happens is that dofmap() seems to think there
> > are more dofs than there really are. Consider this form:
> > 
> > # Test form for updated elasticity
> > 
> > name = "ElasticityUpdatedSigma0"
> > element1 = FiniteElement("Vector Lagrange", "tetrahedron", 1)
> > element2 = FiniteElement("Discontinuous vector Lagrange", "tetrahedron", 0)
> > 
> > c1 = Constant() # Lame coefficient
> > c2 = Constant() # Lame coefficient
> > 
> > q = BasisFunction(element2)
> > v = Function(element1)
> > 
> > L = ((2 * c1 * v[i].dx(i) * q[0]) +
> >         (c2 * (v[i].dx(0) + v[0].dx(i))) * q[i]) * dx
> > 
> > The generated code looks like this:
> > 
> > ...
> > 
> >     inline unsigned int spacedim() const
> >     {
> >       return 3;
> >     }
> > 
> > ...
> > 
> >     void dofmap(int dofs[], const Cell& cell, const Mesh& mesh) const
> >     {
> >       dofs[0] = 3*cell.id() + 0;
> >       dofs[1] = 3*cell.id() + 1;
> >       dofs[2] = 3*cell.id() + 2;
> >       int offset = 3*mesh.noCells();
> >       dofs[3] = offset + 3*cell.id() + 0;
> >       dofs[4] = offset + 3*cell.id() + 1;
> >       dofs[5] = offset + 3*cell.id() + 2;
> >       offset = offset + 3*mesh.noCells();
> >       dofs[6] = offset + 3*cell.id() + 0;
> >       dofs[7] = offset + 3*cell.id() + 1;
> >       dofs[8] = offset + 3*cell.id() + 2;
> >     }
> > 
> > ...
> > 
> > For some reason dofmap() seems to think there are 9 dofs on the
> > element. If I just remove the extra lines, then it works ok.
> > 
> >   Johan
> > 
> > _______________________________________________
> > FFC-dev mailing list
> > FFC-dev@xxxxxxxxxx
> > http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
> > 
> 

-- 
Anders Logg
Research Assistant Professor
Toyota Technological Institute at Chicago
http://www.tti-c.org/logg/