On Thu, 29 Apr 2010 15:46:54 +0200, Lorenzo Botti <[email protected]> 
wrote:
> Ok, I'm starting to understand.
> So the command line options won't consider the cases when you have more than
> one dof per node or (as in my dG code) per element.

It *only* handles the case where you have one kind of node, and *all*
dofs are at that node.

> Even if I use --node_major_dofs I'm getting something like
> u0,u1,u2,u3,v0,v1,v2,v3,w0,w1,w2,w3,p0,p1,p2,p3 in each element for a first
> order approximation in 3D.

Roy, what's going on here?  I thought you said that this option would do
nodal blocking (aka field interlacing).

> I'm trying something like
> --node_major_dofs -pc_type fieldsplit -pc_fieldsplit_block_size 16
> -pc_fieldsplit_0_fields 0,1,2,3,4,5,6,7,8,9,10,11 -pc_fieldsplit_1_fields
> 12,13,14,15 -pc_fieldsplit_type schur -pc_fieldsplit_schur_precondition self
> -fieldsplit_1_ksp_monitor
>
> Does it make sense?

Yes.

> I still need better preconditioning than nothing in my schur complement!

Yeah, these usually come from semi-implicit solvers or approximate
commutators (which are very closely related).

Jed

------------------------------------------------------------------------------
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to