[vtkusers] Errors in vtkCellDerivatives for certain cell types

Berk Geveci berk.geveci at kitware.com
Mon Oct 2 16:49:28 EDT 2000


	vtkCellDerivatives is now (hopefully) fixed and works with all 
	(compatible) cell types. Wedge, trianglestrip, pixel, polygon,
	quad, line and polyline had to be fixed. There is an example
	in the contrib/examplesTcl directory called TestCellDerivs.tcl
	Please report any problems that you find.

	Berk

Richard Strelitz wrote:
> 
> Colleagues,
>   I have been testing vtkCellDerivatives, one of the contributed modules.
> I have modified the tcl example CellCenters.tcl which creates examples
> of all cell types by adding code to compute the cellDerivative.  This is
> what I've found:
> 
>   works for:  voxel, hexahedron, wedge, pyramid, tetrahedron triangle,
> trianglestrip
> 
>   memory read error  for Pixel and Polygon
> 
>   incorrect answer for Quad (always = 0,0,0)
> 
>   deranged answer for Line, polyline  (for 2-D in plane data, 3rd component
> is equivalent to NaN-
>    needs a warning, although cell derivative for polyline is probably best
> left undefined.
> 
>   provides answer though it perhaps shouldn't for vertex, polyvertex.
> 
>  The main problems are in PIXEL, POLYGON, and, most maddeningly, QUAD.
> The issues raised here precede the previously reported alleged error in
> QUAD::derivatives,
> if you compare the computations of Derivatives in polygon and hexahedron,
> the error in QUAD
> becomes more likely to exist.
> 
> Thank you,
>   Richard Strelitz,
>   Advanced Computing Group, LANL
>




More information about the vtkusers mailing list