[Gmsh] Is Non-uniform p-refinement planned?

Aleksejs Fomins aleksejs.fomins at lspr.ch
Wed Oct 15 08:59:43 CEST 2014


Attempt No2, lets see how many I will need until someone hears me...

On 10/12/2014 03:38 PM, Aleksejs Fomins wrote:
> Dear GMSH,
> 
> Our FEM code has benefited from GMSH meshing capabilities for several
> years now, and this year we started using its curvilinear capabilities,
> which brings our code to a totally new level of performance.
> 
> However, we dream of yet another piece of functionality, which could
> save us a lot of memory and processor time, and that is the non-uniform
> p-refinement.
> 
> At the moment, when selecting curvilinear interpolation order, GMSH uses
> that interpolation order for all elements. Furthermore, it frequently
> makes the boundary between elements of the same geometric entity curved,
> with the strategy that I am not able to immediately comprehend.
> 
> Could someone from the developers please comment if the following
> functionality could become available in foreseeable future:
> 
> 1) Make all boundaries between elements of the same geometrical entity
> straight-sided, thus only preserving curvature on the boundaries between
> geometrical entities, and rendering all other elements as linear elements.
> 
> 2) Further, give user some control on how the internal geometry of a
> single geometrical entity is interpolated, in terms of some strategy
> parameters specified by the user. I was made aware there is research
> showing that, depending on the structure of PDE of interest, a certain
> internal curvature can improve convergence of the problem.
> 
> Regards,
> Aleksejs Fomins
> 
> 
> 
> _______________________________________________
> gmsh mailing list
> gmsh at geuz.org
> http://www.geuz.org/mailman/listinfo/gmsh
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 551 bytes
Desc: OpenPGP digital signature
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20141015/3ff2db82/attachment.asc>