[vtk-developers] nonlinear cell types

[David C Thompson]

On Thu, 2008-01-10 at 06:02 -0700, Soeren Gebbert wrote:
> 2008/1/9, David Thompson <dcthomp at sandia.gov>:
>         Soeren Gebbert wrote:
>         > there are several patches pending in the vtk bug tracker related to
>         > the new cell types
>         > and some new cubic cells.
>         Hi Soeren,
>         
>         I've looked at the patches (bugs #5466, 5467, 5477) that fix the 
>         tessellation of the vtkBiQuadraticXXX cells already in VTK and have
>         permission from Berk to apply the fixes; I'll check them in soon.
> 
> That's really good the hear, thanks. If you have time please have a look at these bugs too: 
> 0005481, 0005478, 0005483, 0005482.
Done.

>         As for new cell types, I haven't seen any patches in the VTK bug tracker 
>         that add new types. Are there any issues beyond the 3 above? (Note: as I
>         understand things, Kitware doesn't want to add new cell types right now,
>         so I'm not volunteering to commit those patches if there are any. I 
>         would like to look at them, though.)
> 
> Those cubic cells are hidden in the feature request number 5485.
> There is a diff and a tar.gz archive with the sources. 
> An important fix in these sources includes a new method to validate cell derivations.
I'll take a look at pulling out the code to validate derivatives and
incorporate it into existing tests (it may take me a little while to get
to it).

Thanks again for the new cell types.

	David

>  
> I have implemented a central difference scheme to evaluate the interpolation derives.
> Here a comment from the diff c++ test code:
> 
> +  // To check if the derivation numbering and scale is correct (different
> +  // parametric spaces are used eg: [-1; 1] and [ 0; 1]) 
> +  // we check the derivations with numerical differentiation
> +  // BUT this algorithm needs the correct implementation of the cell interpolation functions. 
> +  //
> +  //                f(x + h, y, z) - f(x - h, y, z)
> +  //  f'(x,y,z)|x = -------------------------------
> +  //                              2h
> +  //                f(x, y + h, z) - f(x, y - h, z) 
> +  //  f'(x,y,z)|y = -------------------------------
> +  //                              2h
> +  //                f(x, y, z + h) - f(x, y, z - h)
> +  //  f'(x,y,z)|z = -------------------------------
> +  // 
>  
> I was able to detect with this test several errors in the derives computation in many cells.
> I can also provide the maxima script code which i have used for derives and interpolation function 
> calculation. 
> 
> I will try to answer any questions you have about the quadratic and cubic cells as well as the patched tests.
> 
> Best regards
> Soeren
> 
>             David
>         > I don't know if the vtk development team plans
>         > to put those cells in a new release, because there is a generic cell
>         > approach
>         > too which allows connect over a bridge many more cell types to vtk 
>         > (i.e. high non linear cells ...  im not sure if hermitian cell types
>         > will function too).
>         >
>         > But i will try to answer you any questions related to those new
>         > quadratic and cubic cells.
>         > I can also provide a small patch for the bridge classes to enable
>         > tessellation.
>         >
>         > Best regards
>         > Soeren
>         >
>         >
>         > 2008/1/9, Martin Baumann < mailsgetlost at web.de
>         > <mailto:mailsgetlost at web.de>>:
>         >
>         >     Hi,
>         >
>         >     I found the nonlinear cell types such as vtkBiQuadraticQuad that were 
>         >     developed by Soeren Gebbert and find them very useful. Now I wonder
>         >     when these will become part of a release of vtk. Will these be
>         >     part of a
>         >     vtk release at all?
>         >
>         >     Principal question: Who decides what will appear in a new release
>         >     and what won't?
>         >
>         >     Best regards,
>         >     M.B.
>         >     _______________________________________________
>         >     vtk-developers mailing list
>         >     vtk-developers at vtk.org <mailto:vtk-developers at vtk.org>
>         >     http://www.vtk.org/mailman/listinfo/vtk-developers
>         >
>         >
>         
>         
>