[Paraview] Computation of gradients fails

[Moreland, Kenneth]

After taking a look at your mesh, I see that the problem is that you have some degenerate tetrahedra.  That is, there are tetrahedra where all 4 vertices are in a plane.  In this case the tetrahedron is self-intersecting and the gradient is ill-defined (over constrained I believe), which is why the linear solver is failing to compute it.  You can find the degenerate tetrahedra by running the Mesh Quality filter to compute the volume of the tetrahedra and look for any resulting "Quality" that is not strictly greater than 0.

So, you can do one of two things.  The first is to ignore the errors.  I don't think that any of the degenerate cells are part of the surface (although I can't be sure), and since the end game is to compute values on the surface it might not effect anything.

The second thing you can do is to fix the mesh.  The best thing to do is to go back to whatever created the original mesh, fix it, and rerun the simulation.  There is a reasonable chance that these degenerate tetrahedra also had a negative impact on the simulation results.  If that's not practical, you can use the volume computed by the Mesh Quality filter and the Threshold filter to remove any tetrahedra that don't have a strictly positive volume.  That will mess up the connectivity of the mesh though.

-Ken


On 9/22/10 9:30 AM, "Martin Vymazal" <martin.vymazal at vki.ac.be> wrote:

Hello Ken,

  I computed the x-velocity component using a calculator filter. Then
I applied the 'Gradient of Unstructured Data Set' filter to the output
of the previous filter. This is the place where the computation fails.
I sent you the data in a private email.

Best regards,

  Martin Vymazal


Quoting "Moreland, Kenneth" <kmorel at sandia.gov>:

> More information please.
>
> How are you extracting the x-component of the velocity?  What filter
> are you running to compute the gradient?  Would it be possible to
> post some example data that exhibits the problem so that we can
> replicate it?
>
> -Ken
>
>
> On 9/22/10 7:12 AM, "Martin Vymazal" <martin.vymazal at vki.ac.be> wrote:
>
> Dear ParaView developers,
>
>   I'm trying to compute gradient of my x-component of velocity on a
> grid of tetrahedra which has 55244 DOF. ParaView fails to compute the
> gradients with the following error message:
>
> Generic Warning: In
> /home/kitware/Dashboard/MyTests/ParaView-3-8/ParaView-3.8/ParaView/VTK/Common/vtkMath.cxx,
> line
> 1338
> Unable to factor linear system
>
> The mesh does not have degenerate/ugly elements, but it has some sharp
> edges (in the inlet/outlet area). Is this a problem? If not, is there
> a way of avoiding the error?
>
> Thank you very much.
>
>   Martin Vymazal
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://www.paraview.org/mailman/listinfo/paraview
>
>
>
>
>    ****      Kenneth Moreland
>     ***      Sandia National Laboratories
> ***********
> *** *** ***  email: kmorel at sandia.gov
> **  ***  **  phone: (505) 844-8919
>     ***      web:   http://www.cs.unm.edu/~kmorel
>
>



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.






   ****      Kenneth Moreland
    ***      Sandia National Laboratories
***********
*** *** ***  email: kmorel at sandia.gov
**  ***  **  phone: (505) 844-8919
    ***      web:   http://www.cs.unm.edu/~kmorel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20100922/5bd6b4b9/attachment.htm>