[petsc-users] request related to DMPlexComputeCellGeometryFVM

[Matthew Knepley]

On Thu, Jan 16, 2014 at 3:04 PM, Dharmendar Reddy
<dharmareddy84 at gmail.com>wrote:

> On Thu, Jan 16, 2014 at 2:54 PM, Matthew Knepley <knepley at gmail.com>
> wrote:
> > On Thu, Jan 16, 2014 at 2:36 PM, Dharmendar Reddy <
> dharmareddy84 at gmail.com>
> > wrote:
> >>
> >> Hello,
> >>          I have a request related to cell geometry for FVM. I am
> >> working on implementing a solver for coupled electron hole transport
> >> problem in a semiconductor.
> >>
> >> For the descretization (see the link [1] below), i want to use the FVM
> >> which uses voronoi box around each node as the control volume. Current
> >> implementation of the ComputeCellGeometry, provide the centroid of the
> >> cell.
> >>
> >> Can i request for functionality similar to ComputeCellGeometry, which
> >> on call returns, the orthocenter and the area perpendiclar to each
> >> edge for flux computation ?
> >
> >
> > 1) I had to lookup orthocenter and I got this:
> > http://www.mathopenref.com/triangleorthocenter.html
> >      which only defines it in 2D, and says it can be outside the
> triangle.
> > This does not sound promising.
> >      If you explain the calculation to me, I can help you implement it.
> >
>     I have Fortran implementation, i have not fully tested it but has
> the core required formulas. I willtry to clean it and send it to you.
>
> > 2) I do give back the edge area if you pass in the edge. Have you looked
> at
> > TS ex12? It shows me
> >     using this routine.
> >
>
> I will take a look at this one.
>
> > I will take a look at the discretization and see if I can understand it
> (I
> > see they say there is a problem with
> > obtuse elements).
> >
>
> Yes there will be a problem, if the element is obtuse which results in
> the orthocenter being outside the element.  The flux computation step
> will be incorrect in that scenario. My other post on fixing non
> Delaunay elements is related to this issue.


Then I would at least consider the FVM we have implemented in TS ex11. It
does not have the sensitive mesh dependence, is 2nd order TVD, and works
in both 2D and 3D. You can see a discussion of the code in my slides from
Paris
on the website.

   Matt


>
> >    Matt
> >
> >>
> >> Or any suggested references to look for geometry related routines.
> >>
> >> [1] http://www.iue.tuwien.ac.at/phd/triebl/node30.html
> >>
> >>
> >> 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/20140116/647c3b77/attachment-0001.html>