Re: [Libmesh-users] Special Refinement

[<ti...@ce...>]

Dear Roy and John,

On Tue, 19 Jun 2007, Roy Stogner wrote:

> On Tue, 19 Jun 2007, John Peterson wrote:
>
>> I would expect it to be pretty difficult. We've always wanted to support
>> some form of anisotropic refinement in the library but never quite got
>> around to it.
>
> It's not that we never got around to it so much as we couldn't figure
> out how to implement it.  The whole concept of Elem::level() basically
> breaks down once you introduce topologically anisotropic refinement.
>
> You could do geometrically anisotropic refinement without changing the
> library, though; just refine as normal and move the new points
> created.

That's a wonderful idea!  Why didn't I think of this myself? 
Actually, geometrically anisotropic refinement is (nearly) all I 
currently need.

There is only one thing I would require further: For the Tet4 (and 
Tet10) elements, I would like the user to be able to influence the 
choice of the diagonal along which the middle octrahedron of the 
tetrahedron is divided into four tetrahedra.  I.e., I would like a 
function

Tet4::influence_diagonal_selection(Tet4::Diagonal);

(and the same for Tet10).  Would you accept such a function in your 
library?

Best Regards,

Tim