RE: [Libmesh-users] 1D Meshes

[KIRK, B. (JSC-E. (NASA) <ben...@na...>]

Since this has come up a few times recently it is probably worth =
revisiting
the 1D mesh support in the library.  I will be taking a week or so off =
soon
and will try to play around with a 1D mesh, seeing where the problems =
lie.
My plan is to start with ex4, since its goal is to be "dimension
independent."  Then a simple hack to it to test adaptivity...

Most of the development has focused on hybrid mesh support in 2 and 3
dimensions, so the infrastructure to support 1D is certainly there.  =
There
are just certain places in the code where we have not done a full
specialization for 1D.  Implementing this will not be a big deal.

-Ben




=20



-----Original Message-----
From: lib...@li...
[mailto:lib...@li...] On Behalf Of Kai =
Ludwig
Sent: Wednesday, October 27, 2004 11:20 AM
To: pet...@cf...
Cc: lib...@li...
Subject: Re: [Libmesh-users] 1D Meshes


> First let me say that 1D meshes are almost completely unsupported in=20
> libMesh.  Usually we would use matlab for smaller problems, but this=20
> is one aspect of the library I hope to improve.  The change you have=20
> posted below simply skips the process of finding neighbors, which=20
> could indeed cause trouble later in the calculation.  Beyond this=20
> fact, there are probably several other places in the library which =
are=20
> inconsistent with respect to 1D meshes, so use with extreme caution!

yes - it took me five minutes to figure out that the trick below does =
not
work (e.g. while refining a mesh). Sure, matlab is fine and I've used =
it to
test a 1D adaptive multilevel FEM algorithm. However, we're currently
developing general OO simulation software that should support 1D, and =
in
(near - who nows) future 2D and 3D electrochemical models. In the next =
days,
I've to decide what (adaptive) FEM-library (libmesh, dealII,
getfem++ ... ) is suitable for our purpose, in particular at
longer sight. In my opinion, it would be a big advantage to employ only =
a
single library for all dimensions.

Altogether, libmesh seems to be the most fetching of all, except the =
fact
that, unfortunately, 1D is not supported.

Kai

>
> -John
>
> Kai Ludwig writes:
> > This August Rafa Rodriguez Galvan posted that solution
> > for the support of 1D meshes:
> >
> > > I have solved (temprorally) the problem with 1D meshes by =
changing=20
> > > MeshBase::find_neighbours():
> > > ---
> > >   if (_dim =3D=3D 1)
> > >      //!!! error();
> > >      return;
> > > ---
> >
> > I've seen that in the new release libmesh-0.4.3_rc2,
> > still, 1D meshes are not supported.
> > Does Rafaels sloution work properly, without getting in trouble=20
> > somewhere later in development? Is there a clean solution coming=20
> > soon?
> >
> > Best regards,
> > Kai


--=20
http://echempp.sourceforge.net

Kai Ludwig
Institut f=FCr Organische Chemie
Auf der Morgenstelle 18
72076 T=FCbingen
Tel.: 07071/29-73049
Mail: kai...@un...




-------------------------------------------------------
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE LinuxWorld =
Reader's
Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_idU88&alloc_id=12065&op=3Dclick
_______________________________________________
Libmesh-users mailing list
Lib...@li...
https://lists.sourceforge.net/lists/listinfo/libmesh-users