[Dune] Reading arbitrary polygonal meshes in Dune

[Christian Engwer]

On Thu, Feb 26, 2015 at 03:05:13PM +0100, Benedikt Oswald wrote:
> aha. But still, what about triangulating into triangles and assigning a
> certain tag to a group of elements
> that form the original polygon, then evaluating the algorthm based on
> triangles with that same tag ? Benedikt

It will make handling of DOFs nearly impossible. Often you have things
like one DOF per intersection/face (in the case of conforming polygons
every intersection corresponfs to a face), but now you have polygon
internal faces and resolving this is all but simple.

Christian

> Am 26/02/15 um 15:01 schrieb Christian Engwer:
> > Hi Benedikt,
> >
> > he says he wants to use the virtual element method and there the main
> > idea is to have a method which wokrs on arbitrary polygons.
> >
> > Christian
> >
> > On Thu, Feb 26, 2015 at 02:56:01PM +0100, Benedikt Oswald wrote:
> >> Hello Ganesh
> >>
> >> a simpler, relatively straighforward approach could be
> >> to read your polygons into gmsh and then have gmsh
> >> triangulate them, into triangles. The same approach
> >> would also work in 3D. Then, there is not need to
> >> write a new grid. Greetings, Benedikt
> >>
> >> Am 26/02/15 um 14:49 schrieb Christian Engwer:
> >>> Hi,
> >>>
> >>> currently no grid can handle such elements. Once upon a time there was
> >>> a simple implementation, which was able to handle arbitrary polygonal
> >>> elements in 2D with out refinement. But this doesn't work with the
> >>> current Dune anymore. On the other hand I believe it wouldn't be too
> >>> difficult to implement such a mesh, as we have introducent several
> >>> changes in the interface to allow for non-standard meshes (this was
> >>> all with the cornerpoint mesh in mind).
> >>>
> >>> Ciao
> >>> Christian
> >>>
> >>> On Thu, Feb 26, 2015 at 12:02:02PM +0000, Andreas Dedner wrote:
> >>>> Hi.
> >>>> The main problem is not the grid type or the factory but that you
> >>>> need an actual
> >>>> grid implementation that can handle these general meshes. ALU and
> >>>> SPGrid etc.
> >>>> can only handle cubes and simplex elements and that is the only
> >>>> geometry types
> >>>> they can understand in their GridFactory.
> >>>>
> >>>> So for something more general you need to provide a new grid
> >>>> implementation that
> >>>> allows for more general grid types. The geometry type of these
> >>>> elements would be
> >>>> "none" and probably you would also be implementing only part of the
> >>>> geometry interface.
> >>>> Something along those lines was done in the CornerPointGrid implementation.
> >>>> https://github.com/OPM/dune-cornerpoint
> >>>>
> >>>> Best
> >>>> Andreas
> >>>>
> >>>>
> >>>>
> >>>> On 26/02/15 11:38, Ganesh Diwan wrote:
> >>>>> Dear list,
> >>>>>
> >>>>> I am working on implementing virtual element computations inside
> >>>>> Dune. This will involve implementing the algorithm on arbitrary
> >>>>> shaped elements. Is there a way to read arbitrary polygonal (2d)
> >>>>> or polyhedral (3d) meshes in Dune? Currently implemented Grid
> >>>>> managers (ALUGrid, SPGrid etc.) identify either simplices or cube
> >>>>> types.
> >>>>>
> >>>>> Can the GridFactory class be used for this purpose? Grid-howto
> >>>>> document says when building a new gridtype Dune may have a problem
> >>>>> for unidentified element type. Say, I have an element with five
> >>>>> edges (a pentagon). In this case, is Dune likely to throw an
> >>>>> exception for unidentified element type?
> >>>>>
> >>>>> I wanted to get the Grid guru's thoughts before investing time in this.
> >>>>>
> >>>>> Thank you!
> >>>>> Ganesh
> >>>>>
> >>>>>
> >>>>> _______________________________________________
> >>>>> Dune mailing list
> >>>>> Dune at dune-project.org
> >>>>> http://lists.dune-project.org/mailman/listinfo/dune
> >>>> _______________________________________________
> >>>> Dune mailing list
> >>>> Dune at dune-project.org
> >>>> http://lists.dune-project.org/mailman/listinfo/dune
> >> -- 
> >> ------------------------------------------------------------------------------------------------
> >> Dr. sc. techn. Benedikt Oswald - first engineer - LSPR AG - phone - +41 43 366 90 74
> >> Technoparkstrasse 1, CH-8005 Zürich, benedikt.oswald at lspr.ch - labor vincit omnia improbus
> >> ------------------------------------------------------------------------------------------------
> >>
> >>
> >> _______________________________________________
> >> Dune mailing list
> >> Dune at dune-project.org
> >> http://lists.dune-project.org/mailman/listinfo/dune
> >>
> 
> -- 
> ------------------------------------------------------------------------------------------------
> Dr. sc. techn. Benedikt Oswald - first engineer - LSPR AG - phone - +41 43 366 90 74
> Technoparkstrasse 1, CH-8005 Zürich, benedikt.oswald at lspr.ch - labor vincit omnia improbus
> ------------------------------------------------------------------------------------------------
> 
> 

-- 
Prof. Dr. Christian Engwer 
Institut für Numerische und Angewandte Mathematik
Fachbereich Mathematik und Informatik der Universität Münster
Einsteinstrasse 62
48149 Münster

E-Mail	christian.engwer at uni-muenster.de
Telefon	+49 251 83-35067
FAX		+49 251 83-32729