On Mar 15, 2007, at 7:03 PM, Charless Fowlkes wrote:
I am new to fipy but quite enjoying it thus far.
That's great to hear. What sort of applications are you working on?
I am running into
one problem with visualization. I would like to display the results
from solving a diffusion eqn on a 2D surface embedded in 3D, e.g.
a sphere.
We actually did a similar problem awhile back.
I was trying to get the mesh to display with MayaviViewer
but immediately run into the error below.
I assume numerix.sum used to have an index argument but no longer?
Yes, sorry. It looks like this was changed to "axis" to accommodate
NumPy:
https://www.matforge.org/fipy/changeset/1982#file98
This has been fixed in svn.
Beyond that, it looks like _getStructure is going to check the mesh
dimension and won't create the appropriate polygons for my surface
mesh.
Instead it will see 3 dimensions and start looking for tetrahedra?
Any suggestions? Am I looking at the wrong viewer module all
together?
You have the right viewer, but the wrong mesh. We're actually rather
surprised that GmshImporter2D() was willing to import a mesh with
three spatial dimensions. I've posted an explanation of how we went
about it at https://www.matforge.org/fipy/wiki/SurfaceOfSphere
You'll need to get the very latest FiPy code from svn in order to use
this. Your question highlighted a number of bugs in our MayaviViewer,
which evidently don't test enough.
You can either add comments to the wiki or ask questions here, as you
see fit.
--
Jonathan E. Guyer, PhD
Metallurgy Division
National Institute of Standards and Technology
<http://www.metallurgy.nist.gov/>
