fenics team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Mon, Nov 21, 2005 at 05:27:26PM -0500, tomtzigt wrote:

> Understood, I was planning to do the adaptation code so that FEniCS stays
> clean. I see that CUBIT uses EXODUS-II as its mesh format, don't know what
> Simmetrix uses, either way, I can use all the direction of someone that has
> written a translator before. If that means starting with something you have
> laying around, power to me.
> 
> The other issues I like to understand up front is what could go wrong with
> the actual simulation. CUBIT will produce a hex-mesh (that is their
> strength, right?) and Simmetrix will give me a tet-mesh. What type of
> problems can I anticipate when using these meshes on a Navier-Stokes PDE?
> Remember, I am learning as I go here.

FIAT, FFC and DOLFIN only do tetrahedra and triangles (and lines in
the case of FIAT).

/Anders

> Theo
> 
> -----Original Message-----
> From: Matthew Knepley [mailto:knepley@xxxxxxxxxxx] 
> Sent: Monday, November 21, 2005 5:16 PM
> To: tomtzigt
> Cc: 'Discussion of FEniCS development'
> Subject: Re: [fenics-dev] Mesh to FEniCS
> 
> "tomtzigt" <tomtzigt@xxxxxxxxxxxxxxxx> writes:
> 
> > :o="urn:schemas-microsoft-com:office:office"
> xmlns:w="urn:schemas-microsoft-com:office:word"
> xmlns="http://www.w3.org/TR/REC-html40";>
> >
> > I need some help. I am looking at two different initial meshers: CUBIT and
> a fancy industrial mesher from Simmetrix. The latter is completely
> automatic, but gives you a tet-mesh. The test case I want to build is some
> funky
> > geometry in a flow field and then letting Navier-Stokes loose on it.
> Before I spent time translating their mesh descriptions to the FEniCS mesh,
> the question I have is, what could possibly go wrong with these two meshers
> that I
> > should know about before investing time into the adaptation. :p>
> 
>   Never heard of Symmetrix, but both the meshers are not open, so we can't
> really make changes.
> I am not sure what answer you are looking for, however I have written code
> to take a simple
> cell <--> vertices description to a Sieve-type mesh already (C++) which is
> in PETSc. Is this
> what you want or something fancier?
> 
>      Matt

-- 
Anders Logg
Research Assistant Professor
Toyota Technological Institute at Chicago
http://www.tti-c.org/logg/