Daniel,
Here's the bit of code (below) I'm using to sort the cell vertices so
hexahedrons can be plotted in mayavi2. I don't think this is very general,
and it might not work properly if the hexahedrons are far from
rectilinear. I'm not using the fipy viewer, and instead create vtk files
to be viewed externally.
The rectilinear grid I created that seems to be working was also extruded,
which is why I assumed the extrusion routine was O.K.
I'm a bit under the gun to get this working and turn in some results.
I'll look into making a 3-D grid with gmsh, once I clear this project off
my plate.
Andy
cellVerts=mesh._getCellVertexIDs().data
vertCoords=mesh.getVertexCoords()
#order=[0,1,2,3,4,5]
order=[0,1,3,2,4,5,7,6]
cellVerts2=array(cellVerts[:])
#for i in range(8):
# cellVerts2[i,:]=cellVerts[order[i],:]
cellVerts2=transpose(cellVerts2)
vertCoords=transpose(vertCoords)
dtype=[('x',float),('y',float),('z',float),('id',float)]
###FiPy mixes up cell verts...need to sort them to be able to plot
###sensibly in mayavi
for i in range(mesh.getNumberOfCells()):
IDs=array([[j] for j in cellVerts2[i]])
cell=[vertCoords[j] for j in cellVerts2[i]]
cell=hstack((cell,IDs))
cell=array([tuple(cell[j]) for j in range(8)],dtype=dtype)
cell.sort(order=['z','x','y'])
cellVerts2[i]=array([cell[j][3] for j in order])
VisGrid=VtkData(
UnstructuredGrid(vertCoords,hexahedron=cellVerts2)
,CellData(Scalars(head,'head'))
)
VisGrid.tofile('head_data','ascii')
Andrew Reeve
Associate Prof.
Dept. of Earth Sciences
University of Maine
207-581-2353
On Thu, 5 Feb 2009, Daniel Wheeler wrote:
On Thu, Feb 5, 2009 at 9:05 AM, A.S.Reeve <[email protected]> wrote:
I re-meshed my problem using a rectilinear grid, used the same script, and
everything works fine now. Now that I've worked through this, I recall
having this issue before. It seems like the gmsh generated grids only work
for homogeneous or nearly homogeneous problem domains. If this is true,
perhaps there should be some sort of a warning issued when importing gmsh
grids?
Hi Andrew, Hope all is well. I'd wager on the extrude function rather
than gmsh import as being the source of the problem. I wrote that when
you were here that time and I haven't looked at it since. One thing to
do is to create a 3D mesh in gmsh and use that without using the
extrude function. Have you already tried that? I'm going to look into
it, but anything you have will be helpful.
I also recall there being problems plotting hexahedrons in mayavi. Is this
true, or am I imagining this? If its true, I could send you the bit of code
I'm using to plot up a rectilinear hexahedral grid for inclusion in FiPy.
Yeah, we had problems with cell vertex ordering when we wrote the
mayavi viewer. I'm not sure we ever sorted that one out so any code
you have that deals with that issue would be very helpful. Thanks!
--
Daniel Wheeler
