Re: [Libmesh-devel] Periodic BCs patch

[John P. <pet...@cf...>]

On Mon, Aug 20, 2012 at 5:11 PM, John Peterson
<pet...@cf...> wrote:
> On Mon, Aug 20, 2012 at 4:46 PM, John Peterson
> <pet...@cf...> wrote:
>> On Mon, Aug 20, 2012 at 4:30 PM, Roy Stogner <roy...@ic...> wrote:
>>>
>>> This is against the SVN head.  It seems to work fine on the
>>> all-too-few tests I've run, but does it pass the INL test suite too?
>>
>> I'll give it a try now.
>
> Actually, this looks a lot better!  I think it's working (see attached
> screenshots).
>
> Gotta run now, I'll run it through the MOOSE test suite tomorrow...

So there are small differences in several of the tests which I am
investigating (surprisingly, the wedge test is now passing!) but I
think there may still be a problem with subdomain-restricted variables
and periodic BCs.

I think Cody has discussed this with you previously but those messages
did not make it to the list, so in summary for this test:

.) block 1 is the four elements on the left
.) block 2 is the four elements on the right
.) variable u (not shown) is restricted to block 1
.) diffusion is solved on both blocks
.) v is periodic from top to bottom along the entire mesh
.) u is periodic from top to bottom on block 1 only
.) Dirichlet boundaries on left and right

As you can see in 'patch.png', the values of v in the nodes between
the two subdomains are being set to zero as of your latest patch.  The
u values are not affected by the patch.

-- 
John