Access to subsystems and variables of these systems

[ValentinKaisermayer]

If I use the RC example from MTK:

rc_model = let
    @named resistor = Resistor(R=1)
    @named capacitor = Capacitor(C=1)
    @named source = ConstantVoltage(V=1)
    @named ground = Ground()

    rc_eqs = [
            connect(source.p, resistor.p)
            connect(resistor.n, capacitor.p)
            connect(capacitor.n, source.n)
            connect(capacitor.n, ground.g)
            ]

    @named rc_model = ODESystem(rc_eqs, t)
    compose(rc_model, [resistor, capacitor, source, ground])
end

I can access the subsystems directly:

julia> propertynames(rc_model)
18-element Vector{Symbol}:
 :resistor
 :capacitor
 :source
 :ground
 :source₊p₊i
 :resistor₊p₊i
 :source₊p₊v
 :resistor₊p₊v
 :capacitor₊p₊i
 :resistor₊n₊i
 :resistor₊n₊v
 :capacitor₊p₊v
 :capacitor₊n₊i
 :source₊n₊i
 :capacitor₊n₊v
 :source₊n₊v
 :ground₊g₊i
 :ground₊g₊v

However, if I apply rc_model_simp = structural_simplify(rc_model) I can no longer access the subsystems (:resistor, :capacitor, :source, :ground):

julia> propertynames(rc_model_simp)
23-element Vector{Symbol}:
 :capacitor₊v
 :resistor₊i
 :resistor₊R
 :capacitor₊C
 :source₊V
 :capacitor₊i
 :source₊i
 :source₊p₊i
 :source₊n₊i
 :capacitor₊p₊v
 :ground₊g₊i
 :capacitor₊n₊i
 :resistor₊n₊v
 :source₊n₊v
 :capacitor₊n₊v
 :ground₊g₊v
 :source₊v
 :resistor₊n₊i
 :resistor₊v
 :resistor₊p₊v
 :resistor₊p₊i
 :capacitor₊p₊i
 :source₊p₊v

So the structure changed I guess.

How can I access a specific variable?
In the documentation this is always done via a direct reference to the subsystems like:

u0 = [
      capacitor.v => 0.0
     ]

or

sol[resistor.v]

However, apart from some test scripts I will not always have those around.
I find e.g. rc_model_simp.capacitor₊v odd because of the unicode characters.

So what is the indented way of accessing a variable?

[ChrisRackauckas]

sol[resistor.v] works.

[ValentinKaisermayer]

Maybe I didn't formulate my post correctly, but my question was how do I get the variable if I do not have the reference to resistor and only the simplified model anymore? Thx

[ChrisRackauckas]

@unpack rc_model_simp = resistor
sol[resistor.v]

should be fine?

[ValentinKaisermayer]

@ChrisRackauckas Sry, in advance I think this should have been a discourse post and not an issue.
So what I think happens, is that rc_model still has all the model structure (in it's field systems) and so I'm able to access for instance rc_model.capacitor.p.i.
However, after the structural_simplify this structure ist lost and the system no longer has it's structure (the field systems ist empty). All variables are now flattened out via their namespace i.e. capacitor₊p₊i, which has to be understood as one variable name. Am I correct?

At least for me, a section in the docs about name-spacing in general (what e.g.@unpack, @nonamespace does) would be helpful. For instance the structural_simplify section is maybe too short.

It is not always clear why some things work:
sol[@nonamespace rc_model_simp.capacitor₊v] or sol[capacitor.v]
and others not:
sol[rc_model.capacitor.v]

[ChrisRackauckas]

sol[rc_model.capacitor.v] this doesn't work because that never works, even without structural simplify, because rc_model is not a variable.

[ChrisRackauckas]

@YingboMa could we just extend getproperty of a system for that to work?

[ValentinKaisermayer]

But it evaluates to a variable, which might not exist, if I interpret the name-spacing correctly:

julia> rc_model.capacitor.v
rc_model₊capacitor₊v(t)