[Paraview] trying to achieve odd symmetry reflection where axis is inside original domain

Thu Nov 5 22:41:01 EST 2015

On Mon, Nov 2, 2015 at 11:30 AM, Noah Reddell <reddell at uw.edu> wrote:

> Hi Cory,
>
>    Thanks for your interest.  I can answer your two questions pretty
> compactly.
>
> *> Could you describe the glitches? I couldn't understand them from just
> the images you attached.*
> The scalar variable EM_field_n_c4 in the referenced .VTS file is the
> analytic function: *tanh(-x/4.082)*.
> The resulting odd symmetry reflection about x=0 should achieve the same
> function.  The function is smooth.
> I suspect the glitches come from the two nodes left of x=0 in my original
> data.  I speculate that the Reflect filter is reflecting those as well.
> Actually, I'm pretty sure I observed as much yesterday when not using the
> option to copy the original data.  This means some kind of funny overlap
> occurs between the before and after reflection data.
>

Ah, thanks for clarifying. Now I understand your grid layout better.

*> what do you expect to see at x=0 if your original data doesn't have
> anything at x = 0?*
> My preference would be to have no new nodes created along x=0.  I have
> nodes at x= { 0-d2,  0-d1,  0+d1, 0+d2, 0+d3, 0+d4, ... }.
> One way to look at the problem would be to reflect nodes {0+d3,0+d4,...}
> about x=0 because the the existing two nodes {0-d2, 0-d1} are already
> correct.
> Another way to look it it is to discard existing nodes {0-d2, 0-d1} and
> reflect what remains about x=0.
> In both cases, reflect with possible negative scaling!
>

Yes, I would try trimming the extent to get rid of 0-d2 and 0-d1, then
reflect the rest. This *should* solve the glitch you see, but you will
still have the center missing from the result.

> I don't care if a new node is created at x=0 which would be the average
> value of nodes {0-d1, 0+d1}.  But Paraview linear interpolation already
> does this for presentation purposes and it would not be strictly correct
> for the underlying polynomial element.
>

Your best bet is probably a programmable filter to get the control you
want. You should be able to take care of the trimming away of {0-d2, 0-d1},
do the reflection with odd symmetry, then fill in the center by adding
cells to the output.

- Cory

> Best Regards,
>
> Noah Reddell
>
> On Mon, Nov 2, 2015 at 7:42 AM, Cory Quammen <cory.quammen at kitware.com>
> wrote:
>
>> Hi Noah,
>>
>> See answers and questions for you inlined below.
>>
>>
>>> 1)  My original data extends beyond the x=0 symmetry plane by two data
>>> points.  And I have no data point on x=0 itself.  This is because in the
>>> simulation, there is a third-order polynomial element centered on x=0.
>>>
>>
>> What would you like to see on x=0?
>>
>>
>>> 2)  My data is represented as Structured Curvilinear Grid (.VTS files)
>>> though in actuality, it is rectilinear.
>>>
>>
>> Shouldn't be a problem.
>>
>>
>>> 3)  Some data arrays have odd symmetry, such that f(x<0) = -f(x>0)
>>>
>>
>> The Reflect filter doesn't support this, but you obviously know that.
>>
>>
>>> Here’s what I’ve tried.
>>> A) Use the Reflect filter, reflect about x=0.  Works quite well for even
>>> symmetry data, but the odd-symmetry data is misrepresented with opposite
>>> sign for (x<0).  Also, there are glitches with the doubled-up mesh points
>>> about x=0.  That becomes obvious when I tried to add a calculator to
>>> correct the odd symmetry based on (x<0) boolean expression.
>>>
>>
>> Could you describe the glitches? I couldn't understand them from just the
>> images you attached.
>>
>>
>>> B) Manually select data points with (x>0); Extract Selection; Reflect
>>> that extraction; Group original and new selection.  Problem:  I end up with
>>> a band of missing data for the cell centered on x=0.
>>>
>>
>> Again, what do you expect to see at x=0 if your original data doesn't
>> have anything at x = 0?
>>
>> Thanks,
>> Cory
>>
>>
>>> Ideally, I want to end up with a consistent full-domain representation
>>> such that I can perform stream line traces across the x=0 mid-plane.
>>>
>>> I’d be very grateful if anyone could point out a working solution.
>>> Here’s one of my data files and some screenshots of my failed attempts.  (
>>> The scalar data item EM_field_n_c4 is a good odd-symmetry example. )
>>>
>>> VTS data file 800 kB
>>> <https://dl.dropboxusercontent.com/u/54478577/supporting%20forum%20posts/paraview_odd_symmetry_help/harris_current_sheet_equilibreum_2.vts>
>>>
>>> Existing Curvilinear Mesh representation
>>> <https://dl.dropboxusercontent.com/u/54478577/supporting%20forum%20posts/paraview_odd_symmetry_help/By.jpg>
>>>
>>> ProblemA demonstration of missing band about x=0
>>> <https://dl.dropboxusercontent.com/u/54478577/supporting%20forum%20posts/paraview_odd_symmetry_help/missingband.jpg>
>>>
>>> ProblemB demonstration of glitch at centerline
>>> <https://dl.dropboxusercontent.com/u/54478577/supporting%20forum%20posts/paraview_odd_symmetry_help/glitch_at_centerline.jpg>
>>>
>>>
>>> Best Regards,
>>>
>>> Noah Reddell
>>> University of Washington
>>>
>>>
>>> _______________________________________________
>>> 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
>>>
>>>
>>
>>
>> --
>> Cory Quammen
>> R&D Engineer
>> Kitware, Inc.
>>
>
>

-- 
Cory Quammen
R&D Engineer
Kitware, Inc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20151105/1266c68d/attachment.html>