(De)normalization needs to be done on every cdf- or sf-related API call at the moment. It's much simpler to handle it like the pmf vector and just do it once in new.
This breaks multinomial because it uses some of the Categorical helper functions in its implementation. This PR depends on another one resolving this, e.g. #287.
(De)normalization needs to be done on every cdf- or sf-related API call at the moment. It's much simpler to handle it like the pmf vector and just do it once in
new.This breaks multinomial because it uses some of the
Categoricalhelper functions in its implementation. This PR depends on another one resolving this, e.g. #287.