[aspect-devel] Need help with fe/mapping_real_to_unit_q4_sphere_x

[Wolfgang Bangerth]

Seba,
I believe I need some help from you since you're more familiar with this 
code than anyone else I believe.

Take a look at the fe/mapping_real_to_unit_q4_sphere_x testcase. It dies 
with this exception:

  --------------------------------------------------------
An error occurred in line <1780> of file 
</home/people/wolf/p/deal.II/ultra450/deal.II/source/fe/mapping_q1.cc> 
in function
     dealii::Point<dim, double> dealii::MappingQ1<dim, 
spacedim>::transform_real_to_unit_cell_internal(const typename 
dealii::Triangulation<dim, spacedim>::cell_iterator&, const 
dealii::Point<spacedim>&, const dealii::Point<dim, double>&, 
dealii::MappingQ1<dim, spacedim>::InternalData&) const [with int dim = 
3, int spacedim = 3, typename dealii::Triangulation<dim, 
spacedim>::cell_iterator = dealii::TriaIterator<dealii::CellAccessor<3, 
3> >]
The violated condition was:
     loop<loop_limit
The name and call sequence of the exception was:
     ExcTransformationFailed()
Additional Information:
(none)
--------------------------------------------------------

You can get in there using the debugger by converting the AssertThrow 
into an Assert.

I believe that part of the problem is that the point we'd like to find 
the reference point for is (may be?) outside the Q1-mapped cell. In 
general it might be legitimate to say that this can't work then, except 
that as the fe/mapping_real_to_unit_q4_sphere test shows we call the Q1 
transform_real_to_unit_cell function to get an initial guess for the 
inverse Q4 mapping, and the point is definitely inside the Q4-mapped cell.

Do you have any suggestions on how this could be solved?

Best & many thanks in advance
  Wolfgang

-- 
------------------------------------------------------------------------
Wolfgang Bangerth               email:            bangerth at math.tamu.edu
                                 www: http://www.math.tamu.edu/~bangerth/