Closed stevencarlislewalker closed 2 years ago
Not an issue. Initially I wondered (unnecessarily in hindsight) if we needed to treat time variation on start_date
specially, but it turns out that this is not required to match the R engine. If this issue really was a bug, it might have meant that we do not need this feature: #136.
When 'do_make_state == TRUE', there are two ways to supply parameter values for the linearized model that is used to compute the initial state: one using 'update_lin_params' and one on the start date in 'params_timevar'. The user needs to be warned if they specify both, and the 'params_timevar' takes precedence.
sp
immediately after it is created while still preserving thesp_orig
so thatrel_orig
still works later in the simulation loopdo_make_state
is on then also add an item tolinearized_params
so that the correct parameters are used in state construction as well