Error required: check that units in AMR are internally consistent

[djinnome]

We need unit tests that check to make sure that consistent dimensions and units are being used among model variables and parameters, and between the data and the model variables and parameters

[SamWitty]

@djinnome , could you please complete the description for this issue or close it? Thanks!

[SamWitty]

This should be able to handle the wastewater example, where different state variables have different units. To handle this, use the units on parameters and rate laws to reconcile differences.

[djinnome]

I looked into the AMR, and there is no explicit annotation of dimensions or units. The closest thing is the DKG identifiers, and I don't think that has enough information to perform a dimensional analysis. I suppose we can assume that parameters are 1/day, that compartments are people and that fluxes are people/day, but without a set of clear use cases that would fail or pass, especially since the wastewater state variable has different units, I am not sure this is actionable.

[djinnome]

I did find one model that had units: https://github.com/DARPA-ASKEM/simulation-integration/blob/main/data/models/SEIRHD_base_model01_petrinet.json

So at least I could perform a dimensional analysis on the rate law to make sure it works out to person/day.

[djinnome]

Since we are already using sympy, I think https://docs.sympy.org/latest/modules/physics/units/index.html is probably the simplest way to perform dimensional analysis.

[djinnome]

Questions:

• Is person a dimension, a unit, or a quantity?

[SamWitty]

@djinnome , should this be re-opened now that you've found an AMR with units?

[djinnome]

Re-opening, since a path to success exists.

[djinnome]

Honestly, there isn't much to do for Epi models, but I think the dimensional analysis will be critical for managing the complexity of climate models in phase 2.