Frontend adaption for independent variables

[jong42]

For each variable, if fields['independent'] == True, the following behaviour has to be guaranteed in the frontend:

• The variable always has to be set as independent (blue in the Specification panel)

• No marginal plots must be shown for the variable

• No density plots must be shown for the variable

[jong42]

I changed the backend so that density queries on a model with independent variables always return np.NaN. Now, when- in the frontend - the option to show marginal plots is selected for an independent variable, no distribution is shown anymore. The same should happen for contour plots, but here occurs a bug: When the option to show a contour plot is selected, the whole visualization stops, without showing an error.

To reproduce the problem:

• Switch to branch 'add_independent _variables

• Create a model with independent variables, e.g. by running the script models_core/tests/create_PyMC3_testmodels.py

• Display the model in Lumen