[Paraview] VTK parallel unstructured grid and halo elements

[Louie Cardone-Noott]

Hi Berk,

Thanks for your reply. Yes that does sound like what I need to do.
Shouldn't the nodes of elements with vtkGhostLevels>0 already be ignored in
the range calculation though?

Is there no way to do this currently then? I tried marking them with
vtkGhostPoints but it didn't help last time I tried.

Cheers,
Louie

On 25 February 2015 at 18:31, Berk Geveci <berk.geveci at kitware.com> wrote:

> Hi Louie,
>
> Ghost/halo nodes are considered valid (but duplicated) so they are taken
> into account when calculating ranges, statistics etc. What you need is to
> mark them as invalid, it sounds like? We are making slow progress towards
> having an infrastructure that supports this but currently it is only
> partially support for structured grids and AMR meshes and a subset of
> filters.
>
> Best,
> -berk
>
> On Wed, Feb 25, 2015 at 11:06 AM, Louie Cardone-Noott <
> louie.cardone-noott at linacre.ox.ac.uk> wrote:
>
>> Hi again list,
>>
>> Inspired by David Ortley's recent success (see "Smooth seams in SILO
>> files") I wonder if perhaps the way I'm doing it is correct, but Paraview
>> is incorrectly considering the values at hidden nodes when calculating data
>> ranges.
>>
>> Essentially my "option 3" (writing a vtkGhostLevels array) appears to
>> work as it should, and the "seams" are hidden in my parallel data. But, if
>> I then Extract Surface and look at the Data Arrays -> Data Ranges I see
>> values which can only be from non-surface nodes. I suspect they are
>> probably from nodes on the (hidden) interior surfaces. This is obviously a
>> bug.
>>
>> I could probably provide a minimal example if useful.
>>
>> Best wishes,
>> Louie
>>
>>
>> On 17 January 2015 at 16:26, Louie Cardone-Noott <
>> louie.cardone-noott at linacre.ox.ac.uk> wrote:
>>
>>> Hi Paraview list,
>>>
>>> I use a finite element code that generates VTK output, but I'm having
>>> trouble dealing with halo (ghost) nodes/elements correctly.
>>>
>>> We currently partition meshes node-wise, and each process writes all its
>>> owned nodes and one level of halo nodes to a chunk. Once all the chunks are
>>> loaded there is a "double layer" of internal boundaries between processes.
>>> I'd really like to be able to use the Extract Surface filter correctly but
>>> as you would expect it also finds all the internal surfaces. From searching
>>> I've tried:
>>>
>>> * Clean To Grid, but this removes both sides of the double layer leaving
>>> unsightly gaps.
>>> * Writing a vtkGhostPoints array (on nodes) but this seems not to be
>>> used by this filter (or Paraview at all?)
>>> * Writing a vtkGhostLevels array (on elements). This correctly hides all
>>> the ghost elements at the internal boundaries (great!) but still doesn't
>>> let me do everything I want to. For example, the extracted surface looks
>>> right, but the array limits still seem to be based on the max/min of the
>>> external AND internal surfaces, so I suspect the internal ones are just
>>> "hidden from view".
>>>
>>> What's the right way to deal with this?
>>>
>>> Thanks in advance.
>>>
>>> Louie
>>>
>>> _______________________________________________
>>> 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
>>>
>>> Search the list archives at: http://markmail.org/search/?q=ParaView
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://public.kitware.com/mailman/listinfo/paraview
>>>
>>>
>>
>> _______________________________________________
>> 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
>>
>> Search the list archives at: http://markmail.org/search/?q=ParaView
>>
>> Follow this link to subscribe/unsubscribe:
>> http://public.kitware.com/mailman/listinfo/paraview
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20150227/42a57084/attachment.html>