Calling the submodel() function of a (Bayesian) HMM would not only return a sub-model but indeed overwrite the HMM instance directly. This makes sense when used within _estimate(), but not if a user wants to have an actual submodel. The PR submodels and returns a copy of self in these cases.
Calling the
submodel()function of a (Bayesian) HMM would not only return a sub-model but indeed overwrite the HMM instance directly. This makes sense when used within_estimate(), but not if a user wants to have an actual submodel. The PR submodels and returns a copy ofselfin these cases.