Changing Plot Defaults

This example demonstrates how to change the default plot settings for the matrix plot. This includes changing the plot type on the diagonals, changing the plot type for the off-diagonals.

from tsunami_ip_utils.viz.viz import matrix_plot
from tsunami_ip_utils.viz.plot_utils import generate_plot_objects_array_from_contributions
from tsunami_ip_utils.integral_indices import get_uncertainty_contributions
from paths import EXAMPLES

application_files = [ EXAMPLES / 'data' / 'example_sdfs' / 'MCT' / f'MIX-COMP-THERM-001-00{i}.sdf' for i in range(1,4) ]
experiment_files = application_files

contributions_nuclide, contributions_nuclide_reaction = get_uncertainty_contributions(application_files, experiment_files)

labels = {
    'applications': [ application_file.name for application_file in application_files ],
    'experiments': [ experiment_file.name for experiment_file in experiment_files ],
}

Changing the Diagonal Plot Type

On the diagonals, where the application and experiment are the same, the plot is a pie chart of contributions to the nuclear-data indcued uncertainty, as shown in Visualizing Contributions. However, other types of plots, such as static bar plots can be used instead. This is done by specifying the diagonal_type in the tsunami_ip_utils.viz.plot_utils.generate_plot_objects_array_from_contributions() function. Note that any plot type available in tsunami_ip_utils.viz.viz.contribution_plot() can be used.

Static Pie Chart Diagonals

For a static pie chart, the diagonal_type should be set to 'pie'

plot_objects_array_nuclide = generate_plot_objects_array_from_contributions(
    contributions_nuclide,
    '(Δk/k)^2',
    diagonal_type='pie'
)
fig = matrix_plot(plot_objects_array_nuclide, plot_type='interactive', labels=labels)
fig.show()

- Zoom Out + Zoom In

Static Bar Chart Diagonals

For a static pie chart, the diagonal_type should be set to 'bar'

plot_objects_array_nuclide = generate_plot_objects_array_from_contributions(
    contributions_nuclide,
    '(Δk/k)^2',
    diagonal_type='bar'
)
fig = matrix_plot(plot_objects_array_nuclide, plot_type='interactive', labels=labels)
fig.show()

- Zoom Out + Zoom In

If a plot type for which a variant with an interactive legend exists (e.g. 'interactive_pie') is used, the interactive legend can be turned off by setting the interactive_contribution_legend parameter to False. These plots can also be saved to an image

plot_objects_array_nuclide = generate_plot_objects_array_from_contributions(
    contributions_nuclide,
    '(Δk/k)^2',
    diagonal_type='interactive_pie', # This is the default, but it is shown here for clarity
    interactive_contribution_legend=False
)
fig = matrix_plot(plot_objects_array_nuclide, plot_type='interactive', labels=labels)
fig.show()

- Zoom Out + Zoom In

Changing the Off-Diagonal Plot Type

The off-diagonal plots can also be changed to a different type. By default, the off-diagonal plots are interactive scatter plots with an interactive legend, but they can just as easily be changed to a static matplotlib scatter plots

plot_objects_array_nuclide = generate_plot_objects_array_from_contributions(
    contributions_nuclide,
    '$(\\Delta k/k)^2$', # Since all plots are matplotlib plots, pylatex formatting can be used to make the labels look nice
    off_diagonal_type='scatter',
    diagonal_type='bar',
)
fig = matrix_plot(plot_objects_array_nuclide, plot_type='interactive', labels=labels)
fig.to_image( EXAMPLES / '_static' / 'matrix_bar_diagonal.png')

# sphinx_gallery_thumbnail_path = '../../examples/_static/matrix_bar_diagonal.png'

- Zoom Out + Zoom In