Re: [Getfem-users] getfem::new_interpolated_fem hanging

[Yves Renard]

Dear Andriy,

I tried your program on 32bits and 64bits processor and .. it does work on 
both cases. What version of Getfem++ do you use ?

Yves.


On Wednesday 10 October 2007 16:54, Andriy Andreykiv wrote:
> Dear Yves,
>
>       I have prepared a simple program where the problem manifests itself.
> Please unzip the attached archive with a program.
> Compile it with "make all". But before, please check the MAKEFILE if it
> properly links all the libraries on your system.
>
> Run the program with         ./level_set_elliptic ls.param
>
> It will hang exactly at the creation of the interpolated fem.
>
> The program does the following: it calculates an elliptic problem in a unit
> square mesh with Dirichlet BC's, imposed in a weak sence with Lagrange
> multiplier, as in fictitious domain method. The Dirichlet BC's are imposed
> on the boundary of another embedded mesh, imported into the program
> (embedded_mesh.gf - produced by another getfem program).
>
> In a parameter file ls.param you can set a TRANSLATION parameter, that will
> move the embedded mesh. In its current position of the embedded mesh, the
> program will hang during creation of the interpolated_fem. If you move the
> embedded mesh say by 0.2, the program will not hang. Please see the .vtk
> files in order to see what the program actually does.
>
> Regarding the triangular meshes, I did run them, but didn't have any
> problems there, although no guaranty that I was just lucky and missed "the
> bad" spot.
>
> But there is another strange thing with the triagular mesh. When I create a
> unit mesh in GETFEM with triangular elements and then postprocess it in
> MayaVI, the colors are really odd, it's like the orientation of the
> elements is wrong (please see the attached JPG). If I create the triangular
> mesh in GMSH, the resulting colors in MayaVI are OK, but the colored and
> blacked sides are switched, as compared to quad elements.
>
> Thank you very much in advance,
>                                                               Andriy
>
> On Wednesday 10 October 2007 14:00:22 you wrote:
> > On Tuesday 09 October 2007 13:24, Andriy Andreykiv wrote:
> > > Good day,
> > >
> > >   I have  a program, that uses getfem::new_interpolated_fem. The call to
> > > this functions occurs many times in the program. The program, actually,
> > > simulates a movement of conductor in a electrostatic field and I use a
> > > call to this function to enforce Dirichlet conditions with largrange
> > > multiplier along the boundary of the conductor. So, for some meshes
> > > (not all and not always) when a conductor reaches a certain position
> > > the call to
> > >
> > > getfem::pfem ifem=getfem::new_interpolated_fem(mf_electric,mim_line);
> > >
> > > hangs. Hangs for a very long time (for at least a night, while it
> > > shouldn't take even a second for other positions of the same
> > > conductor). It doesn't produce an error and uses all the computing
> > > power, like it entered an infinite loop.
> > >
> > > in the above case, for instance,
> > > mf_electric is FEM_QK(2,1), built on the mesh with GT_QK(2,1), using
> > > getfem::regular_unit_mesh (with noise ON) and  mim_line is
> > > IM_GAUSS1D(10).
> > >
> > > Do you, gentlemen, see any reason why this function would hang?
> >
> > No, it seems to be a bug !
> > Could you try to make the smallest program on which the problem appears
> > in such a way I can try to fix this ?
> >
> > Does the problem appear for triangular meshes ?
> >
> > Yves.

-- 

  Yves Renard (address@hidden)       tel : (33) 04.72.43.87.08
  Pole de Mathematiques,                       fax : (33) 04.72.43.85.29
  Institut Camille Jordan - CNRS UMR 5208
  INSA de Lyon, Universite de Lyon
  20, rue Albert Einstein
  69621 Villeurbanne Cedex, FRANCE
  http://math.univ-lyon1.fr/~renard

---------