[BUG]: "Stratify" seems to make unexpected changes to model structure (overwrites a template)

[liunelson]

Before stratification:

After stratification:

This model was generated from LaTeX equations.
Model A (local, edited).json

Stratified model:
Model A (local, edited, strat).json

Note that the Sl(t) equation had a term with factor Sl(t) * v * (...) but it gets overwritten with Sh(t) * (...) after stratification:

model = stratify(
    template_model=model,
    key= "location",
    strata=['eu', 'af'],
    structure= [],
    directed=False,
    cartesian_control=False,
    modify_names=True,
    concepts_to_stratify=['Eh', 'P', 'I1', 'I2', 'H', 'Rh', 'Er', 'Ir', 'Rr', 'Sr', 'Sl', 'Sh'], #If none given, will stratify all concepts.
    concepts_to_preserve=None, #If none given, will stratify all concepts.
    params_to_stratify= None, #If none given, will stratify all parameters.
    params_to_preserve= None, #If none given, will stratify all parameters.
    param_renaming_uses_strata_names = True
)

[liunelson]

@nanglo123 @kkaris @bgyori Could you take a look when convenient? This is a step in the Hackathon Decision Maker Scenario, where Model A is stratified into a Europe-Africa model.

[liunelson]

I think Tenzin and I found the source of the problem: MIRA template_model_from_sympy_odes generates GroupedControlledConversion templates that do not have names so stratify creates bad stratified template names.

[nanglo123]

So this issue actually stems from the fact that our stratification method will assign template names if no template names exist. If there are multiple template names under the same strata then they will have the same name. Since Terarium relies on template naming to assign rate laws to their appropriate template, this is where the issue of mismtached rate laws derives from.

It could be a discussion topic on whether we want to remove template renaming altogether and have Terarium rely on other parts of a template to identify them or if we should change our renaming process.

I'm closing the issue now as the issue has been partially adressed as we now make sure to add a template name for all template types during ingestion if they have a name, removing lossiness for template names in the end-to-end proccesing of models in MIRA. Maybe this topic could be repurposed as a feature request as I don't think it's necessarily a "bug" if we have "bad" template names as it may not be the best approach to identify templates just by name or to name templates if they don't have one.