I find myself using observables quite heavily to implement non-ODE parts of a model.
For example, I start with a base model with some state variables and parameters. I also have the parameter PPI (unused in any rate law but necessary part of the model) and the observable ASI = 1 - PPI / 100.
I stratified it as such:
model_country = stratify(
template_model = model,
key = 'country',
strata = ['US', 'CD', 'CN', 'NO', 'AU'],
structure = [],
directed = False,
cartesian_control = False,
params_to_stratify = ['DeltaS', 'PP', 'SP', 'g'], # all parameters used in the rate laws
concepts_to_stratify = ['GDP', 'R', 'OS', 'S'], # all state variables
param_renaming_uses_strata_names = True
)As of now, I cannot go straight to this model.json (representing the supply chain of cobalt) where I have the parameters and observables of the form:
PPI_c # stratified unused-in-rate-laws-but-used-in-observable parameters
ASI_c = 1 - PPI_c / 100 # stratified observable
(c = AU, CD, CN, NO, US).
What do you think of having either a observables_to_stratify key? I may be abusing the term "stratification" by applying its logic to things unrelated to the transitions and rate laws!
I find myself using observables quite heavily to implement non-ODE parts of a model.
For example, I start with a base model with some state variables and parameters. I also have the parameter
PPI(unused in any rate law but necessary part of the model) and the observableASI = 1 - PPI / 100.I stratified it as such:
As of now, I cannot go straight to this model.json (representing the supply chain of cobalt) where I have the parameters and observables of the form:
(
c = AU, CD, CN, NO, US).What do you think of having either a
observables_to_stratifykey? I may be abusing the term "stratification" by applying its logic to things unrelated to the transitions and rate laws!