Bias data in the splatter plot

[Peter Peterson]

The splatter plot currently "randomly" selects events to plot from the event workspace attached to the MD workspace. Modify the vtkSplatterPlotFactory to sort the boxes by density (using refactored code from FindPeaksMD) and take events from the densest boxes until the total number of events is achieved.

[Peter Peterson]

I think Dennis has this well in hand.

[Dennis Mikkelson]

To allow more flexibility, in addition to sorting the list of boxes in order of decreasing normalized signal, a control will be added to allow the user to specify what percent of the boxes should be used starting with the box with the largest normalized signal. Finally, the default number of points to show will be increased to 150,000.

[Dennis Mikkelson]

RE 7091 bias data in splatterplot

Comment out a couple of un-used variables so it compiles cleanly. The variables will be used for testing other strategies.

refs #7091

Changeset: 3190544d5913da0b6b28017b061ccb1ff67c3a68

[Michael Reuter]

Refs #7091. Adding threshold parameter for top percentile.

Changeset: 9c528e4532de42bc9be262834a5bd1956c14f788

[Michael Reuter]

Refs #7091. Changing default value for number of points.

Changeset: 8be0de4572a19d3ee74a7f7701fc5589b10ae7a9

[Dennis Mikkelson]

RE 7091 bias data in splatterplot

Removed a couple of variables intended for an alternate biasing strategy that won't be used. Put cout statements in if (VERBOSE) structure that was originally present in the code. Tried an alternate biasing strategy that used points from the boxes above the threshold, based on the number of points in the box. For example, if the number of points requested was equal to 1/2 the total number of points in boxes above the threshold, then 1/2 of the points from each box would be used. This alternate strategy emphasized bright peaks much more than the weaker peaks and required many more points to see the full lattice. Also, it was more difficult to get a good view of scattering between the main Bragg peaks in TOPAZ 7424 or of the diffuse scattering in TOPAZ 7238, with the alternate strategy.

refs #7091

Changeset: c9039ec1ee3ece23e3e4c86e211f443dbca01a8a

[Dennis Mikkelson]

RE 7091 bias data in splatterplot

Changing the expected number of points, after updating the strategy for selecting points to plot in the splatter plot.

refs #7091

Changeset: 43f31c1e470b014d668ebfb37f58bd840435c771

[Michael Reuter]

Refs #7091. Change threshold parameter to double.

Changeset: 93cde903d1956e99e38eda44fc8a4fd4246906e1

[Dennis Mikkelson]

NOTES TO TESTER:

1. The branch name is feature/7091_bias_data_in_splatterplot.

2. Start MantidPlot and load a TOPAZ run, such as TOPAZ_3132_event.nxs, and convert it to an MD workspace. The quickest way to do this is to bring up the Interface: "SCD Event Data Reduction", and use the Browse button on the first form to select the event nexus file. Set the Max |Q| to Map to MD to 25, and turn on the Lorentz Correction. Press the Apply button to both load the file and convert it (properly) to an MD workspace.

3. When the MD workspace has been created, right click on the workspace in MantidPlot and choose "Show Vates Simple Interface".

4. Select the Scatter Plot Mode by clicking the button with the three colored dots near the top center of the Vates window. The view should change to a 3D view of events in reciprocal space and the reciprocal lattice should be immediately apparent.

5. Increasing or decreasing the Number of Points in the control at the left side of the Vates window and pressing Apply should adjust how many points are plotted, although with the default "Top Percentile" setting of 5%, increasing the Number of Points beyond the default of 150,000 has only a slight effect.

6. With the default Number of Points (150,000) changing the Top Percentile to 2% and pressing Apply should reduce the number of Bragg peaks visible in the display. Changing the Top Percentile to 25% should start to show the background counts near the Bragg peaks and at low |Q|.

[Peter Peterson]

This is fine. #7213 was generated for a performance issue.

[Stuart Campbell]

This ticket has been transferred to github issue ​7937