[petsc-dev] Ghost values in sieve

[Chris Eldred]

Awesome- that is easy. How do I access that SF?

On Thu, Aug 23, 2012 at 12:54 PM, Matthew Knepley <knepley at gmail.com> wrote:
> On Thu, Aug 23, 2012 at 1:42 PM, Chris Eldred <chris.eldred at gmail.com>
> wrote:
>>
>> I need the adjacency relations discussed in my other post- the only
>> one that is not part of closure(p) U star(p) is: U
>> cone(support(edge)). Given an edge p, I need all of the edges that
>> cover the same cell as edge p.
>
>
> Okay, then for parallelism, I think you need nothing more than the SF we get
> from Jacobian preallocation.
>
>   Thanks,
>
>      Matt
>
>>
>> On Thu, Aug 23, 2012 at 12:17 PM, Matthew Knepley <knepley at gmail.com>
>> wrote:
>> > On Thu, Aug 23, 2012 at 12:26 PM, Chris Eldred <chris.eldred at gmail.com>
>> > wrote:
>> >>
>> >> Yes- I am implementing the TriSK scheme
>> >> (www.mmm.ucar.edu/people/skamarock/Ringler_et_al_JCP_2009.pdf) on
>> >> arbitrary Voronoi meshes. In order to do wind/flux reconstruction at
>> >> the cell edges, it needs to know about the edges of adjacent cells-
>> >> which are outside of closure(p) U star(p).
>> >
>> >
>> > Great! Stuff that cannot be done with that structured crap. However,
>> > from
>> > quickly looking at
>> > the paper, there is nothing beyond the neighbors, so we can reuse the
>> > code
>> > from
>> > Jacobian preallocation. If you could tell me exactly what adjacency you
>> > need, we might be
>> > able to do it even more simply.
>> >
>> >    Matt
>> >
>> >>
>> >> On Thu, Aug 23, 2012 at 11:16 AM, Jed Brown <jedbrown at mcs.anl.gov>
>> >> wrote:
>> >> > On Thu, Aug 23, 2012 at 12:12 PM, Chris Eldred
>> >> > <chris.eldred at gmail.com>
>> >> > wrote:
>> >> >>
>> >> >> Thanks- that helps a lot. If I need stencils that are larger than
>> >> >> closure(p) U star(p) (for a higher-order finite difference method,
>> >> >> for
>> >> >> example), I assume that I need to create my own PetscSF's that
>> >> >> describe which points need to be ghosted?
>> >> >
>> >> >
>> >> > Is this still a fully unstructured method? The Sieve formalism
>> >> > doesn't
>> >> > give
>> >> > you a very efficient way to do this for structured or semi-structured
>> >> > grids.
>> >> >
>> >> > Even so, if wider stencils are to be supported, I think it should be
>> >> > implemented within the library. Doing it outside with the current
>> >> > infrastructure is going to be quite a rabbit hole.
>> >> >
>> >> >>
>> >> >> Is there some documentation or example code that explains the theory
>> >> >> behind star forests?
>> >> >
>> >> >
>> >> > Docs for the basic operations:
>> >> >
>> >> > http://59A2.org/files/StarForest.pdf
>> >>
>> >>
>> >>
>> >> --
>> >> Chris Eldred
>> >> DOE Computational Science Graduate Fellow
>> >> Graduate Student, Atmospheric Science, Colorado State University
>> >> B.S. Applied Computational Physics, Carnegie Mellon University, 2009
>> >> chris.eldred at gmail.com
>> >
>> >
>> >
>> >
>> > --
>> > 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
>>
>>
>>
>> --
>> Chris Eldred
>> DOE Computational Science Graduate Fellow
>> Graduate Student, Atmospheric Science, Colorado State University
>> B.S. Applied Computational Physics, Carnegie Mellon University, 2009
>> chris.eldred at gmail.com
>
>
>
>
> --
> 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



-- 
Chris Eldred
DOE Computational Science Graduate Fellow
Graduate Student, Atmospheric Science, Colorado State University
B.S. Applied Computational Physics, Carnegie Mellon University, 2009
chris.eldred at gmail.com