Re: [Getfem-users] Number of dofs and number of points

[Yves Renard]

Le lundi 9 juillet 2007 17:46, Giovani a écrit :
> Yves,
>
> Thank you very much! I'll try the solution!

Ok, I made the modifications in the node storage of Getfem such that
the identification of dof is more robust now. The problem you pointed out 
cannot occurs now ...
The problem was in the algorithm to find fastly if a node is near an already 
existing one or not.
Anyway, than you for the bug report.

A lot of files have been modified. You can download the corrected version on 
http://svn.gna.org/viewcvs/getfem/. (if you want ready to use tar.gz version, 
tell me).

Now, concerning your example of program with an interpolation, the archive 
linked to your email was empty. May be due to the use of the mailing list. 
Can you send it to me directly ?
Indeed your mesh you send me previously was a little bit special (with half of 
elements misssing, compared to structured mesh for a cube). So if you 
interpolate on such a mesh, ther is half a chance that the node does not 
correspond to an element ...

Yves.

>
> There's another issue happening which I'd like to call your attention
> to. When trying to interpolate data from a very compact mesh into a
> coarse mesh, the interpolation function does not seen to be working
> properly. Perhaps the solution to this issue is also to reduce the
> error, as you pointed out. But since I do not know getfem's inner
> mechanisms, I fell that it would be worth to call your attention to this
> as well, in case it's necessary to update other places in the code.
>
> Annexed to this email I'm sending a small program with 2 meshes, which
> tries to interpolate data from the compact one into the coarse one. The
> compact mesh is very similar to the one I sent you on the other mail,
> however it's more "organized"; that other one had some deformations
> applied to it which triggered the error in the dof calculation. In this
> case, getfem can count the mesh number of dofs correctly, however the
> interpolation fails for most of the dofs of the target mesh.
>
>
> Giovani


-------------------------------------------------------------------------
  Yves Renard (address@hidden)       tel : (33) 04.72.43.80.11
  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
-------------------------------------------------------------------------