How to sum a three-dimensional data along a certain coordinate axis to obtain a two-dimensional data

[ofsea]

Hello everyone!

I would like to know if visit can achieve the following requirements: I have a three-dimensional scalar data, and I hope to sum the scalar values of each point on the axis along the z-axis to obtain a two-dimensional path integral data.

In theory, similar functions can be achieved using python or other programs, so I want to know if visit can implement it conveniently.

Thanks for your help.

[markcmiller86]

I think you want to use a Data Binning operation.

1. Got to Plots->Pseudocolor
2. Select Operators->Data Binning->2D-><mesh your variable is defined on>
3. The above steps create a plot/operator combination in the plot list
4. Open the Data Binning operator attributes (you may need to expand the plot item in the plot list to get at it).
5. Select X and Number of bins (>= number of mesh nodes in X) for Dimension 1 and Y and Number of bins (>= number of mesh nodes in Y) for Dimension 2 of the operation
6. Select Average for Reduction Operator
7. Select Mesh for Output on
8. Draw.

[markcmiller86]

I found that what I got was still three-dimensional data, which did not meet my needs.

Sorry, change Output on to Bins

[markcmiller86]

Or slice the 3D mesh you get anywhere across z axis

[ofsea]

Thank you ! Actually , when I change the option to "Bins" , the result is much better ! My results are shown as follows. But I can't understand the principle why "data binning" can do this (Integrate along the z direction). So I hope you can recommend some more detailed documents to let me learn.

Thank you again for your kind help!

[markcmiller86]

Ok great. Regarding the theoretical basis for these mappings, I think @hankchilds dissertation is the best formal reference for this.

Over the years, we've opted to simplify the naming of these mappings. In Hank's dissertation, they are called equivalence class functions (ECFs). Later on, in VisIt, we called them derived data functions (ddfs) (which they are still called in our python CLI interface) and we eventually changed that to data binning (which they are called in our GUI).

But, the whole idea starts by treating any input mesh with a bunch of fields defined on it as a sort of point cloud of data where every node (or zone) is a point in a high dimensional, feature space, the dimensions of that space being all the physics variables computed on the mesh.

[ofsea]

Thank you for your kindness! I will read the dissertation .You have solved my problem!