[vtkusers] Painting finite element meshes with information defined at quadrature points, not at nodal locations

[Jeff Amelang]

Hello,

I sincerely appreciate any time you take to answer.  In finite element
calculations, you often end up with quantities (such as stresses) defined
at integration points, not at the nodes.  I would like to, without any
further interpolation or processing, hand 1) such an array of values at
integration points and 2) a mesh to vtk and have it paint the cells by
interpolating within each cell.  As an example, consider the attached
picture (hopefully I can attach things here).  It contains two adjacent
triangles, each with values of a field defined at 3 interior quadrature
points.  I would provide the values of the field at the integration points
within the cells and vtk would paint each cell by interpolating within that
cell.  So, even though the values of the field within the left triangle are
around 4 and 5, the left triangle would be painted with a value of around 3
on the left point and a value of around 6 on the right edge.  The right
triangle would be painted with a value of around 4 on the left edge and a
value of around 1 on the right edge.  The colors would not match on either
side of the adjoining edge (just as if you were painting each triangle by a
different CellData scalar, except it wouldn't be constant in each cell).
 That really won't make sense if the attachment doesn't go through.

Please do not refer me to this:
http://www.vtk.org/Wiki/VTK/VTK_integration_point_support or this:
http://www.vtk.org/Wiki/File:VTK-Quadrature-Point-Design-Doc.pdf without
giving me some sort of guidance on how the classes in those documents can
be used to accomplish what I'm describing.  To my reading, the pdf is
describing a way to get vtk to interpolate from fields defined at nodes to
integration points.  That is not what I am looking for.  Nor do I need to
interpolate from integration points to nodal values.

I have fields defined at integration points and need to paint the mesh by
those fields interpolated within each cell.  If the mesh were a 2d
triangular mesh with N triangles, for a 3-point quadrature rule I would be
providing 3N field values defined at the 3 quadrature points within each
cell, the positions of the nodes of the mesh, and the connectivities to
create the cells.  That's it.

Please feel free to ask clarifying questions.  Again, I sincerely
appreciate any time you take to answer.  Does vtk have support for this?
jeff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20130110/d83ea046/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: VTKIntegrationPointsQuestion.png
Type: image/png
Size: 33562 bytes
Desc: not available
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20130110/d83ea046/attachment.png>