Problem createing StructuralIdentifiability ODE from certain system

[TorkelE]

I wish to create a StructuralIdentifaibiltiy ODE from a ModelingToolkit model. However, it fails if the measured quantities are a combination of known parameters and measured quantities:

# Fetch packages.
using ModelingToolkit, StructuralIdentifiability

# Create model
@parameters p1 p2
@variables t x1(t) x2(t) y(t)
D = Differential(t)
eqs = [D(x1) ~ p1* x2 - x1,
       D(x2) ~ p2 - p1* x2]
@named osys = ODESystem(eqs)

# Create StructuralIdentifiability ODE.
known_parameter = [p1]
measured_quantities = [y ~ x1]
known_stuff = [Num.(measured_quantities);]
si_ode = preprocess_ode(osys, known_stuff)

# This, providing either only known parameters or equations work:
si_ode = preprocess_ode(osys, known_parameter)
si_ode = preprocess_ode(osys, measured_quantities)

Is there a way to create a system with this kind of information?

[pogudingleb]

One can have two output variables y1(t), y2(t) and measured_quantities = [y1 ~ x1, y2 ~ p1]. Would this be appropriate for you?

Thanks again for working on this connection to Catalyst! If you think it would be easier for you, we could have a call about this at some point.

[TorkelE]

Yeah, that seems like a way to do it.

Actually, would you be available for a call tomorrow (or Friday, but tomorrow, Thursday, I am available all week). I have gotten a decent stretch on the implementation, but it would be good to go through everything with you, so that I do not miss any features.

[pogudingleb]

Another option, of course, is to have preprocess_ode(osys, [x1, p1]), that is, you are not obliged to give names to measured_quantities (I should have made it more explicit in the docs...).

For the call, I should be available on Friday but I will know the exact details tomorrow.

[TorkelE]

That sounds good, I will hold off on doing more until we are able to meet. On Friday I am available before 10 am, and after 4 pm (US EST time).

[pogudingleb]

Followed-up by email about the meeting