This is a sketch implementation of a multi-sector model.
It provides the option to pass a dictionary prior args for xprior and a dictionary of sensitivities Hx (both dictionary need to have the same set of keys to use as labels).
Any sectors in the xprior dictionary with prior params set to None will be held constant.
The tests run without modification when this model is substituted for inferpymc. I added an extra test to show that if there are three sectors, where two of the sectors are sampled from and one is held constant, then there are x traces for each of the sampled variables (but not for the variable that is held constant).
Note: there isn't any data processing here to run real inversions with multiple sectors. This is just mocked to set up the test.
This is a sketch implementation of a multi-sector model.
It provides the option to pass a dictionary prior args for
xpriorand a dictionary of sensitivitiesHx(both dictionary need to have the same set of keys to use as labels).Any sectors in the
xpriordictionary with prior params set toNonewill be held constant.The tests run without modification when this model is substituted for
inferpymc. I added an extra test to show that if there are three sectors, where two of the sectors are sampled from and one is held constant, then there are x traces for each of the sampled variables (but not for the variable that is held constant).Note: there isn't any data processing here to run real inversions with multiple sectors. This is just mocked to set up the test.