[Paraview] Comparing results computed with 2 different meshes

[Cory Quammen]

Evan,

The ordering of inputs to vtkProgrammableFilter is set up within
ParaView which is why you won't find any useful documentation about
this in vtkProgrammableFilter. It appears to me that ParaView sorts
the inputs by the memory address of the proxy for the VTK object
rather than using the selection order or the pipeline order. For all
intents and purposes, this is essentially a random ordering.

As a workaround, you could add another vtkProgrammableFilter between
each of your data sets and the main vtkProgrammableFilter. This
vtkProgrammableFilter could mark the data by adding a field data
consisting of a single string array with one entry, namely an
identifier for the data. Then in your main vtkProgrammableFilter, you
could query the field data on all the inputs to find the data set you
need.

Hope that helps,
Cory

Cory

On Tue, Feb 4, 2014 at 3:15 PM, Evan Kao <tossin at gmail.com> wrote:
>
>> No, that appears to not be the case in a quick test I ran. They appear
>> to be ordered in the order in which they appear in the pipeline
>> browser..
>
>
> Unfortunately, it doesn't seem like it follows that order either (in a quick
> test I did using Python Calculator).  Perhaps it's random?  I also tried at
> the documentation for vtkProgrammableFilter but couldn't find any answers
> (though I could've missed something).  For the Python Calculator, this might
> not be as big a deal since you can quickly check the output, but it seems
> more cumbersome to check in the Programmable Filter (at least for me - I
> don't have much experience with it or Python generally).  Is there an easy
> method to check with the Programmable Filter?
>
> - Evan
>
>
> On Mon, Feb 3, 2014 at 5:12 AM, Cory Quammen <cory.quammen at kitware.com>
> wrote:
>>
>> On Fri, Jan 31, 2014 at 1:04 PM, Evan Kao <tossin at gmail.com> wrote:
>> > Thanks a lot for the help.  I think Merge Blocks > Resample With Dataset
>> > >
>> > Python Calculator (or Append Attributes > Calculator) allows me to do
>> > what I
>> > originally asked for, although now I'm realizing the Python Calculator
>> > may
>> > not be enough to do the calculations I want.  I had some questions
>> > regarding
>> > the Python Calculator and the Programmable Filter:
>> >
>> > 1.  The data I'm looking at is from a transient simulation.  If I wanted
>> > to
>> > standardize a parameter for each time step by dividing it by the
>> > standard
>> > deviation of the population for that time point, am I understanding
>> > correctly that I would either have to calculate the standard deviation
>> > first
>> > and then manually type in the number into the Python Calculator or use
>> > the
>> > Python Programmable Filter?
>> >
>> > 2.  Additionally, I've been looking around and it seems the only way to
>> > perform a summing operation is to use the Programmable Filter.  Is that
>> > correct?
>>
>> For doing anything mildly complicated like summing data arrays, you're
>> probably better off with the Programmable Filter. You might
>> theoretically be able to use the Python Calculator, but you'll have
>> some long expressions to fit into that small Expression line.
>>
>> > 3.  Regarding both the Python Calculator or Programmable Filter, is
>> > there an
>> > easy way to tell which index corresponds to which input?  I thought
>> > perhaps
>> > it had to do with just order you click the inputs in, but I'm not so
>> > sure
>> > about that.
>>
>> No, that appears to not be the case in a quick test I ran. They appear
>> to be ordered in the order in which they appear in the pipeline
>> browser..
>>
>> > Thanks for your time,
>> > Evan
>> >
>> >
>> > On Thu, Jan 30, 2014 at 12:25 PM, Cory Quammen
>> > <cory.quammen at kitware.com>
>> > wrote:
>> >>
>> >> On Thu, Jan 30, 2014 at 2:31 PM, Evan Kao <tossin at gmail.com> wrote:
>> >> > Hi Cory,
>> >> >
>> >> > Thanks for the suggestion.  Resample With Dataset solves the problem
>> >> > of
>> >> > matching up the Point IDs, but only passes data from the Input (which
>> >> > I
>> >> > guess makes sense given its purpose), so I think regardless of what I
>> >> > rename
>> >> > the arrays, the data will always be from only one of the meshes
>> >> > meaning
>> >> > I
>> >> > can't use Calculator to compare the data.
>> >>
>> >> How about using Append Attributes to combine the arrays of Resample
>> >> With Dataset and the Source?
>> >>
>> >> > I think my main obstacle is the data type.  Paraview reads in Ensight
>> >> > data
>> >> > as a Multi-Block data set, but neither Append Datasets nor the Python
>> >> > Calculator seem to work on that data type.  Is there a simple way to
>> >> > convert
>> >> > them into a format that can be read by those filters?
>> >>
>> >> You can use Merge Blocks to combine all the blocks in a Multi-Block
>> >> data set to an unstructured grid.
>> >>
>> >> Best,
>> >> Cory
>> >>
>> >> > Thanks,
>> >> > Evan
>> >> >
>> >> >
>> >> > On Thu, Jan 30, 2014 at 6:10 AM, Cory Quammen
>> >> > <cory.quammen at kitware.com>
>> >> > wrote:
>> >> >>
>> >> >> Evan,
>> >> >>
>> >> >> You might want to take a look at the Resample With Dataset filter.
>> >> >> One
>> >> >> of your meshes would be the Input and one would be the Source. What
>> >> >> this filter will do is sample the data values in the Input mesh at
>> >> >> the
>> >> >> locations of points in the Source. If I understand the setup of your
>> >> >> meshes, this should essentially take care of the problem where your
>> >> >> point IDs are not the same. The result will be a single mesh with
>> >> >> two
>> >> >> data arrays that you can then compare point by point.
>> >> >>
>> >> >> You might run into a problem if both meshes use the same name for
>> >> >> the
>> >> >> point array from which you want to get values (one will clobber the
>> >> >> other). You can take care of that by using the Calculator to rename
>> >> >> one of the mesh's arrays, then using the result of the calculator as
>> >> >> the Input or Source.
>> >> >>
>> >> >> Hope that helps,
>> >> >> Cory
>> >> >>
>> >> >> On Mon, Jan 27, 2014 at 4:18 PM, Evan Kao <tossin at gmail.com> wrote:
>> >> >> > Hello all,
>> >> >> >
>> >> >> > I'm trying to compare the results of 2 CFD simulations (which are
>> >> >> > imported
>> >> >> > into Paraview as Ensight data) point-by-point but I'm having
>> >> >> > trouble
>> >> >> > figuring out how to even combine the data into a single Pipeline
>> >> >> > object
>> >> >> > so
>> >> >> > that I can do some simple calculations with the Calculator Filter.
>> >> >> > I'd
>> >> >> > appreciate it if anyone could let me know if what I'm trying to do
>> >> >> > is
>> >> >> > natively possible in Paraview, and if not, provide some
>> >> >> > suggestions
>> >> >> > on
>> >> >> > how
>> >> >> > to approach it.
>> >> >> >
>> >> >> > What might complicate the process is that the results were
>> >> >> > obtained
>> >> >> > using
>> >> >> > two different meshes (in fact, it's a mesh comparison test).  The
>> >> >> > points
>> >> >> > I
>> >> >> > want to look at are the same for both meshes, but the point IDs
>> >> >> > are
>> >> >> > different.
>> >> >> >
>> >> >> > So what I'd like to do in summary is:
>> >> >> > 1) Is there a way to extract and sort the data in a way such that
>> >> >> > the
>> >> >> > point
>> >> >> > IDs are the same for both meshes?
>> >> >> >
>> >> >> > 2) Is there a way to convert/manipulate the data into a single
>> >> >> > Pipeline
>> >> >> > object for comparison?  Or to compare the results of different
>> >> >> > Pipeline
>> >> >> > objects?  I've tried various combinations of filters like
>> >> >> > AppendDatasets,
>> >> >> > GroupDatasets, and MergeBlocks with little success.
>> >> >> >
>> >> >> > Thanks,
>> >> >> > Evan
>> >> >> >
>> >> >> > _______________________________________________
>> >> >> > 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
>> >> >> >
>> >> >
>> >> >
>> >
>> >
>
>