[petsc-users] Boundary conditions using DMComplex

[Subramanya G]

Hi Matt,
I got what you were saying about the boundary conditions. But I am
unable to figure out which routines to use for selecting only a
section of the mesh while setting up a Petsc Section. The
DMComplexCreateSection does not seem to offer anyway of restricting
the mesh.
Thanks,


Subramanya G Sadasiva,

Graduate Research Assistant,
Hierarchical Design and Characterization Laboratory,
School of Mechanical Engineering,
Purdue University.

"The art of structure is where to put the holes"
Robert Le Ricolais, 1894-1977


On Fri, Sep 21, 2012 at 3:42 PM, Matthew Knepley <knepley at gmail.com> wrote:
> On Fri, Sep 21, 2012 at 2:20 PM, Subramanya G <subramanya.g at gmail.com>
> wrote:
>>
>> I have  a small question about using DMComplex.
>>  How does one keep track of external boundaries. I found no methods to
>> check if  a particular node/face belonged to a particular external
>> edge set.
>
>
> First you want to distinguish between
>
> a) Topological boundaries
>
> For this you use Labels, so you can mark parts of the mesh using
> DMCcomplexSetLabelValue()
>
> b) Algebraic Boundary Conditions
>
> These are specified in the PetscSection using PetscSectionSetConstraintDof()
> and
> PetscSectionSetConstraintIndices(). To construct these, you typically use
> the information
> in the mesh labels.
>
>>
>> Also, Is it possible to set up a problem over a part of the mesh
>> instead of the entire mesh?
>
>
> This is very easy. Just create a PetscSection that only has unknowns over
> part of the mesh.
>
>    Matt
>
>>
>> Thanks
>>
>>
>> Subramanya G Sadasiva,
>>
>> Graduate Research Assistant,
>> Hierarchical Design and Characterization Laboratory,
>> School of Mechanical Engineering,
>> Purdue University.
>>
>> "The art of structure is where to put the holes"
>> Robert Le Ricolais, 1894-1977
>
>
>
>
> --
> What most experimenters take for granted before they begin their experiments
> is infinitely more interesting than any results to which their experiments
> lead.
> -- Norbert Wiener