[Insight-developers] FEM BoundaryCondition : Is it functional ?

[brian avants]

i think this would depend on whether the test problem has that type of
BC (though i'm not sure what type is used within that function).

some BC's affect the stiffness matrix and i believe the FIXME/BUG
referred to the need to check whether this was implemented correctly.

how does the vnl_svd look?  and the eigendecomposition?

brian




On Tue, Sep 13, 2011 at 4:50 PM, Luis Ibanez <luis.ibanez at kitware.com> wrote:
> Brian, Jim, Vincent,
>
> Julie and Chuck are making great progress
> on evaluating the Numerical Libraries in ITK.
>
>
> They have found an strange behavior in the
> FEM Testing related to the use of Boundary
> Conditions.
>
> In particular in:
>
> ITK/Modules/Numerics/FEM/include/
>              itkFEMSolverCrankNicolson.hxx
>
> in line 226-227:
>
>  /* step over all types of BCs */
>  this->ApplyBC();  // BUG  -- are BCs applied appropriately to the problem?
>
>
> The surprising thing is that when the
> line above is commented out, the tests
> still pass.
>
> We are wondering is the call to ApplyBC()
> is needed at all, or whether the tests are
> not performing a strict enough evaluation
> of the output.
>
> Could you please comment on this issue ?
>
>
>   Thanks
>
>
>        Luis
>