THargreaves
commented
1 month ago Also worth noting that I noticed some inconsistency in whether state or step came first in the parameters for when transitioning/observing. Since the inclusion of step is the key thing that distinguishes initialisation from transitioning, I've put it first. Happy to flip them if people think state first is more sensible.
Solves #40.
This should also get rid of the possible method call errors.
The change also had the benefit of allowing the default simulate/log density definitions to be combined with the interface definitions.