Re: [Paraview] Using Probe Filter To Get The Average Value Around A Point

[Moreland, Kenneth]

As far as I know, there is no way to get the probe filter to get all points 
within a radius as you describe. You may instead consider the Point Data to 
Cell Data Filter. That will for each cell take the values of all attached 
points and average them.

The point volume interpolator filter will create a grid (by default 100^3) and, 
in gaussian kernel mode, will “splat” a gaussian function onto the grid from 
every point. Another way to think of it is that a 3D gaussian function is 
convolved with a 3D function comprising an impulse function at every point in 
the mesh scaled by the field value (and then sampled on the grid).

-Ken

Sent from my iPad

> On Feb 6, 2018, at 8:21 PM, Jeremias Gonzalez  
> wrote:
> 
> Thank you very much for your explanations and suggestions. Responses 
> interspersed below.
> 
>> On 2/5/2018 4:57 PM, Moreland, Kenneth wrote:
>> Jeremias,
>> When you set a radius and number of points in the probe filter, then the 
>> filter will randomly sample the volume within the defined sphere the number 
>> of times requested. The resulting values are the field values at those 
>> randomly sampled locations. >
>> An easy way to get an average of your samples is to run the result of the 
>> probe filter through the descriptive statistics filter. Look at the 
>> "Statistical Model" table and it will report the mean value for each field. 
>> (Note that if you are using ParaView 5.4 there is a bug, #17627, that shows 
>> the Statistical Model table wrong by default. You have to also change the 
>> Composite Data Set Index parameter in the Display part of the properties 
>> panel to select only the Derived Statistics block.)
>> A couple of caveats to this approach. First, because the sampling is random, 
>> don't expect the exact same answer every time you run it. Second, if one of 
>> the samples happens to lie outside of the mesh, that sample will be filled 
>> with 0's for all fields. That will throw off the average value.
> 
> Is there a probe setting that will simply grab all the points living in the 
> original mesh within the radius of the sphere I choose?
> 
>> That said, another approach you might want to take is to first filter the 
>> data in a way that blurs out the noise first. One way you can do that is to 
>> run the Point Volume Interpolator filter. Change the Kernel to something 
>> like Gaussian (the default Voronoi filter will not do the averaging that you 
>> want). Set the radius appropriately. You can then probe the resulting data 
>> set with a single value (radius 0) and immediate see the "averaged" result.
>> -Ken
> 
> I don't seem to be finding any information on what exactly the Gaussian 
> kernel does with the data, so how close is it to the plain averaging I would 
> like it to be doing?
___
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:
https://paraview.org/mailman/listinfo/paraview

Re: [Paraview] Using Probe Filter To Get The Average Value Around A Point

[Jeremias Gonzalez]

Thank you very much for your explanations and suggestions. Responses 
interspersed below.


On 2/5/2018 4:57 PM, Moreland, Kenneth wrote:

Jeremias,

When you set a radius and number of points in the probe filter, then the filter 
will randomly sample the volume within the defined sphere the number of times 
requested. The resulting values are the field values at those randomly sampled 
locations. >
An easy way to get an average of your samples is to run the result of the probe filter 
through the descriptive statistics filter. Look at the "Statistical Model" 
table and it will report the mean value for each field. (Note that if you are using 
ParaView 5.4 there is a bug, #17627, that shows the Statistical Model table wrong by 
default. You have to also change the Composite Data Set Index parameter in the Display 
part of the properties panel to select only the Derived Statistics block.)

A couple of caveats to this approach. First, because the sampling is random, 
don't expect the exact same answer every time you run it. Second, if one of the 
samples happens to lie outside of the mesh, that sample will be filled with 0's 
for all fields. That will throw off the average value.


Is there a probe setting that will simply grab all the points living in 
the original mesh within the radius of the sphere I choose?




That said, another approach you might want to take is to first filter the data in a way 
that blurs out the noise first. One way you can do that is to run the Point Volume 
Interpolator filter. Change the Kernel to something like Gaussian (the default Voronoi 
filter will not do the averaging that you want). Set the radius appropriately. You can 
then probe the resulting data set with a single value (radius 0) and immediate see the 
"averaged" result.

-Ken


I don't seem to be finding any information on what exactly the Gaussian 
kernel does with the data, so how close is it to the plain averaging I 
would like it to be doing?

___
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:
https://paraview.org/mailman/listinfo/paraview

Re: [Paraview] Using Probe Filter To Get The Average Value Around A Point

[Moreland, Kenneth]

Jeremias,

When you set a radius and number of points in the probe filter, then the filter 
will randomly sample the volume within the defined sphere the number of times 
requested. The resulting values are the field values at those randomly sampled 
locations.

An easy way to get an average of your samples is to run the result of the probe 
filter through the descriptive statistics filter. Look at the "Statistical 
Model" table and it will report the mean value for each field. (Note that if 
you are using ParaView 5.4 there is a bug, #17627, that shows the Statistical 
Model table wrong by default. You have to also change the Composite Data Set 
Index parameter in the Display part of the properties panel to select only the 
Derived Statistics block.)

A couple of caveats to this approach. First, because the sampling is random, 
don't expect the exact same answer every time you run it. Second, if one of the 
samples happens to lie outside of the mesh, that sample will be filled with 0's 
for all fields. That will throw off the average value.

That said, another approach you might want to take is to first filter the data 
in a way that blurs out the noise first. One way you can do that is to run the 
Point Volume Interpolator filter. Change the Kernel to something like Gaussian 
(the default Voronoi filter will not do the averaging that you want). Set the 
radius appropriately. You can then probe the resulting data set with a single 
value (radius 0) and immediate see the "averaged" result.

-Ken


On 2/5/18, 5:27 PM, "ParaView on behalf of Jeremias Gonzalez" 
 wrote:

Hi, I'm trying to find a way to get the average value around a point in 
a mesh that I know to be noisy due to its coarseness. Currently, I am 
unable to understand determine the exact nature of the radius and number 
of point parameters from the documentation ( 

https://www.paraview.org/ParaView/Doc/Nightly/www/py-doc/paraview.simple.ProbeLocation.html
 
), but I am guessing from some third party posts that the radius enables 
one to find a point nearby to a desired point in a given region, and the 
number of points expands the amount captured. The problem I have past 
that, if those are correct understandings, is what to do with the probe 
once I have it. Looking at the resulting spreadsheet from using the 
probe location with a given radius and number of points each labelled 
from 0 to 99, for example, it seems that I may have to use another loop, 
after I introduce and use the probe, with code like

my_running_total=0

for y in range(0, 99):
my_running_total += 
mycalcprobepoint.GetPointData(y).GetArray('Result').GetValue(0)

my_running_total /= 100

that will take that batch of points collected by the probe and average 
all the values I want. Is this the correct interpretation, and a valid 
way to carry out this objective?
___
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:
https://paraview.org/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:
https://paraview.org/mailman/listinfo/paraview