On Fri, 1 Aug 2008, Benjamin Kirk wrote:

>> The hard (or at least tedious) part may be fixing our I/O classes to
>> write out and read in solutions with per-subdomain variables.  I'm not
>> familiar with the nitty-gritty details of our output formats, but I
>> wouldn't be surprised if they didn't all support such a thing.
>
> I'm thinking for most cases you would extend the DOFs to cover the entire
> domain (for tecplot I/O, for example) and write every variable everywhere,
> getting 0 on the domains where they are not defined.

That'll work for the visualization formats, with a little ugliness.
On continuous elements, instead of seeing a sharp boundary for
variables at subdomain edges you'll end up seeing the variables drop
to zero gradually in the bounding elements from the neighboring
domain.

For our xda/xdr formats, we'll need to decide how to add the "which
variables are defined where" information to the header; there's no
fudging that will work there.  But on the other hand there's no need
to stress too much about compatibility with other software, just as
long as we maintain backwards compatibility with ourselves when
reading old files.

> I'm with Roy in thinking the DofMap part of this is not too hard, and we
> should probably just do it and libmesh_experimental() the API.  If there is
> something at least we can start discussion about how to improve it.
>
> But I'd like to dump libmesh-0.6.3 first, since this could end up being a
> pretty big changset, and if nothing else it would be nice to freeze what we
> have currently, which has been stable for a while now.

Sounds good.  I'd like to hear back from the users who are seeing
sparse matrix assembly slowdowns with 0.6.2 before we release
anything, though; if that's not already working correctly in SVN, we
should fix it before making a new release.
---
Roy

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

Reply via email to