Sounds good. -- Anders
On Fri, Mar 28, 2014 at 03:15:13PM +0100, Martin Sandve Alnæs wrote: > +1 > > Martin > > > On 28 March 2014 15:00, Garth N. Wells <[email protected]> wrote: > > Following some discussions on getting non-affine maps working with FEniCS, > I’d like to propose adding the function > > ufc::finite_element::tabulate_dof_coordinates(double* coordinates); > > This function will tabulate the dof coordinates on the reference element. > The purpose is to allow us to match the coordinates of ‘higher order > nodes’ > that come from a mesh generator (for example) to the UFC dofs. Coordinate > matching is necessary because UFC does not specify anything as to how dofs > are ordered locally for a cell. > > Related to this, it would make sense to move/change the function > > ufc::dofmap::tabulate_coordinates(double** dof_coordinates, const > double* vertex_coordinates) const = 0; > > to the ufc::finite_element class: > > ufc::finite_element::tabulate_coordinates(double* dof_coordinates, > const double* vertex_coordinates) const = 0; > > and to remove the function > > std::size_t ufc::dofmap::geometric_dimension() const = 0; > > This would make a ufc::dofmap purely topological, and would be consistent > with the proposed addition of > ufc::finite_element::tabulate_dof_coordinates. > > Garth > _______________________________________________ > fenics mailing list > [email protected] > http://fenicsproject.org/mailman/listinfo/fenics > > > _______________________________________________ > fenics mailing list > [email protected] > http://fenicsproject.org/mailman/listinfo/fenics _______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
