[Paraview] confused with getting nan values in plot over line

[Moreland, Kenneth]

I'm not familier with the term P1-discontinuous.  Is that the same as not C0 continuous (or simply not continuous)?

If so, then yes that is a problem.  The VTK data sets do not directly represent discontinuous interpolated fields.  The only way to do it is how you describe.  Replicate the point so that two distinct field values can be placed at the same coordinates.  But, as you discovered, this means that the connectivity cannot be represented, which can foul up some operations like finding a cell.

I know of no simple away around the problem.  You would need a different cell locator that used a different slower but more reliable search.

-Ken

From: Ali Rostai <ali.rostai at gmail.com<mailto:ali.rostai at gmail.com>>
Date: Mon, 17 Oct 2011 09:22:45 -0700
To: Kenneth Moreland <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>
Cc: "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: Re: [Paraview] confused with getting nan values in plot over line

Hi and excuse me for late reply,

I tried the "Clean to grid" filter and it seems that problems has been solved. But I noticed that
the data was coming from a finite element method using P1-discontinuous Triangular elements.
And it looks for me the problem is somehow related to that.

But as a general question Can paraview show P1-discontinuous data? then for each node in mesh
we have several values. But I think vtk format accepts only one value for each noed.

Regards,
Ali


On Thu, Oct 6, 2011 at 10:34 AM, Moreland, Kenneth <kmorel at sandia.gov<mailto:kmorel at sandia.gov>> wrote:
For starters, the going to infinity thing in the plot is weird. There should just be gaps in the curve. You may have hit a bug.

As far as why you are getting nans in the first place I'm not sure. I can't really tell from your screenshots, but it looks your geometry might be a bunch of polygons on a plane. Is that the case? If so, make sure that all z coordinates are zero. If all the coordinates in pone dimension are exactly the same, then I would expect it to work. Otherwise the line will not be completely on it.

If that is not the case, another problem could be with your topology. The find cell method could fail if your mesh is not fully connected. For example, if you have repeated coincident points, then VTK won't be able to find neighbors and trace to a cell. Try running one of the clean filter on the data. I think that should fix the problem. I think there are some corner cases where even that might not work, but your mesh looks like it has good quality triangles or tetrahedra, so that seems unlikely unless you have degenerate cells or gaps I cannot see.

-Ken


From: ali rostai [mailto:ali.rostai at gmail.com<mailto:ali.rostai at gmail.com>]
Sent: Thursday, October 06, 2011 08:12 AM
To: Moreland, Kenneth; paraview at paraview.org<mailto:paraview at paraview.org> <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: Re: [Paraview] confused with getting nan values in plot over line

Hi,

I tried and tried to move the start and end of line inside the domain.
Actually I see two lines going to infinity (nan) in xyplot pane now! (screen shot attached)

Regards,
Ali

On Thu, Oct 6, 2011 at 5:08 AM, Moreland, Kenneth <kmorel at sandia.gov<mailto:kmorel at sandia.gov>> wrote:
Like I said previously, it looks like the line you define is right on the boarder of your volume. Try moving the line slightly more inside the volume.

-Ken

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20111017/39c2817b/attachment.htm>