On Thu, Oct 27, 2022 at 1:20 PM Yann Jobic <[email protected]> wrote:
There is this factor also in front of the rectangle, which should 1/4 for a unit rectangle of dim 2, but it's also 0.5.
Hi Yann, The reference element in Petsc is [-1, -1] to [1, 1]. This matches FEniCS/Firedrake and also makes it easier to use orthogonal polynomials. Thanks, Matt
Thanks, Yann
Le 10/27/2022 à 6:23 PM, Mark Adams a écrit :
Area of a unit triangle is (1x1)/2 I would guess.
On Thu, Oct 27, 2022 at 11:32 AM Yann Jobic <[email protected] <mailto:[email protected]>> wrote:
Hello,
I'm trying to understand how to use the lowlevel Petsc Finit Element Framework. I've got few questions about it.
1) I'm checking the determinant of the Jacobian transformation from real space to the parametric one. The source code is in :
https://petsc.org/main/src/dm/impls/plex/plexgeometry.c.html#DMPlexComputeTr... < https://petsc.org/main/src/dm/impls/plex/plexgeometry.c.html#DMPlexComputeTr...
I don't quite understand the 0.5 factor in front of it. Geometrically, it should be the ratio of the area/volume of the reel element over the referenced one. It's not the case here no ? It seems that, for other elements, this factor still applies. What this factor represent ?
2) I only checked triangle elements here. The ordering of the test functions and its derivatives is different it seems, with also the
0.5
factor. Am i mistaken ?
I could not find any informations about it in the user-doc. Maybe
it's
written somewhere else ?
Many Thanks,
Best regards,
Yann
-- 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 https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>