[petsc-users] snes ex12 question about DMPlexProjectFunctionLocal

[Geoffrey Irving]

On Thu, Nov 14, 2013 at 6:08 PM, Matthew Knepley <knepley at gmail.com> wrote:
> On Thu, Nov 14, 2013 at 7:38 PM, Geoffrey Irving <irving at naml.us> wrote:
>>
>> On Thu, Nov 14, 2013 at 5:35 PM, Matthew Knepley <knepley at gmail.com>
>> wrote:
>> > On Thu, Nov 14, 2013 at 7:32 PM, Geoffrey Irving <irving at naml.us> wrote:
>> >>
>> >> If this line in ex12 is part of Dirichlet boundary conditions, why is
>> >> it done unconditionally even for Neumann?  Is this a mistake that
>> >> fails to test correct Neumann conditions?
>> >>
>> >>     ierr = DMPlexProjectFunctionLocal(dm, user.fe, user.exactFuncs,
>> >> INSERT_BC_VALUES, userJ.u);CHKERRQ(ierr);
>> >
>> > No, its just idempotent in that case.
>>
>> That's what I meant by fail to test: if there was a bug related to
>> null spaces, it wouldn't show up in this example.
>
>
> I do not understand. If you have Neumann conditions, nothing happens, so
> it would check that case.

I may be misunderstanding something, but wouldn't nothing happen only
if the rest of the code is correct?

Geoffrey