[petsc-users] related to DMPLExvecsetclsoure

[Matthew Knepley]

On Tue, Feb 4, 2014 at 11:09 PM, Dharmendar Reddy
<dharmareddy84 at gmail.com>wrote:

> On Mon, Feb 3, 2014 at 5:49 PM, Matthew Knepley <knepley at gmail.com> wrote:
> > On Mon, Feb 3, 2014 at 5:29 PM, Dharmendar Reddy <
> dharmareddy84 at gmail.com>
> > wrote:
> >>
> >> Hello,
> >>         I am trying to understand the layout of the values array
> >> argument of the DMPLexVecSetClosure.
> >>
> >> Consider a triangular element with two fields at each node, say n, p
> >> vertices : v1,v2,v3
> >> If the value of fields in the local vector :  n1,p1, n2, p2, n3, p3
> >>
> >> If the equation set is: F(n,p) = 0 and G(n,p) = 0
> >>
> >> if the values of F and G evaluated at v1,v2, v3 are F1,F2,F3 and
> >> G1,G2G3 respectively.
> >>
> >> Then the elemVec that goes into DMPLexVecSetClosure is:
> >>
> >> elemVec = { F1,F2,F3,G1G2,G3} is that correct ?
> >>
> >> Now, i am looking for some functionality which will let me access the
> >> following mapping information, (local dof id)?
> >>
> >> what is the Local Dof Id , i.e, the position in elemVec given the
> >> field id, component id and basis id (or node id for the case of P1
> >> Lagrange ?)
> >>
> >>
> >> The usage will be like this:
> >
> >
> > If, as in this case, the fields have the same element, its just normal
> > array ordering:
> >
> >   (f, b, c) = (f*numBasis + b)*numComponents + c
>
> I used this approach in my current code which has same  number of
> fields in all regions of the mesh.
> I pre calculate and store the local dof ids in an array.
>
> But now i am trying do a case where number of fields is not same in
> all regions. I am a bit confused about the interface case.
>
> Consider for example of interface region
>  1----------2
>  |   c1      |
>  |             |
>  3---------4
>  |   c2      |
>  |             |
>  5---------6
> the variables defined at nodes is shown below:
> 1: (phi),  2:(phi), [3,4,5,6]: (phi,n,p)
>
> I think, c1 will have dof layout as :dofArray:
> (phi3,phi4,phi2,phi1,n3,n4,p3,p4) assuming anti clockwise ordering of
> nodes in ref cell. I can get dofArray using vecget closure on c1.
>

Yes, that looks right.

   Matt


> >
> > In general, all field values are stacked together, and ordered by
> > basis vec and then component.
> >
> >     Matt
> >
> >>
> >> do fid=1,numField
> >>    numBasis = getNumBasis(fid)
> >>    numComponent = getNumComponent(fid)
> >>    do bid=1,numBasis
> >>       do cid =1,numCompoent
> >>        localId = getLocalDofId( fid, bid, cid)
> >>         elemResVec(localId) = residual value
> >>      end do
> >>    end do
> >> enddo
> >>
> >> call
> >>
> DMPlexVecSetClosure(dm,section,localResVec,cellId,elemResVec,ADD_VALUES,ierr)
> >>
> >> Thanks
> >> Reddy
> >
> >
> >
> >
> > --
> > 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
>



-- 
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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140205/80726c1d/attachment.html>