On Wed 2008-09-10 12:11, Roy Stogner wrote:
> 
> Would it make sense to add the option to include global scalar
> parameters to the vectors of variables included in a System?  I'm
> thinking right now about a surface ablation rate that ends up coupled
> to every element in a system via a moving-mesh model, but I think John
> had to workaround our lack of the same feature when doing his arc
> length continuation solver.

It's also needed for LNKS optimization where the size is the number of
global controls.

> The trouble would be the dense rows and columns that each new scalar
> would add to the sparsity pattern.

In the applications I can think of, it's not useful to assemble this
contribution into the global vector since the problem is indefinite.
Better is to keep it as a separate block and employ the appropriate
Schur-complement preconditioners.  Of course you wrap the multi-block
matrix so that the solvers don't know the difference.  The dense rows
can be applied unassembled so that no single process ends up needing the
global state.  DMComposite has support for managing the redundant array,
but you have to do a small amount of work to apply the coupling
unassembled.

Jed

Attachment: pgpjtr402Zf0w.pgp
Description: PGP signature

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to