Passing Initials as parameters does not work consistently

[hersle]

It is encouraged to get/set initial ODE values with e.g.

prob.ps[Initial(x)] = 1.0

Therefore, for consistency, I really think it should also work to create an ODEProblem like this:

using ModelingToolkit, OrdinaryDiffEq
using ModelingToolkit: t_nounits as t, D_nounits as D
@variables x(t)
@named M = ODESystem([D(x) ~ 0], t)
Ms = structural_simplify(M)
prob = ODEProblem(Ms, [], (0.0, 1.0), [Initial(x) => 1.0])

However, this fails:

ERROR: Initialization incomplete. Not all of the state variables of the
DAE system can be determined by the initialization. Missing
variables:

Any[x(t)]

I think Initial(x) should be recognized as a parameter consistently across XProblem, remake, setp, getp, etc.

[ChrisRackauckas]

I think this all got fixed to just be x => 1.0 everywhere?

• feat: use guesses as temporary values for variables solved by initialization #3605
• fix: allow solving initialization separately #3599

Should have recently fixed the workflow hurdles around that.

[hersle]

I know that u0 = [x => 1.0] works. I only think it's an inconsistency that

• Initial(x) is a parameter, and
• p = [Initial(x) => 1.0] is not recognized.

Anyway, my real wish was a way to get the (initial) value of a variable, regardless of whether it is an unknown/observed/parameter. Your links made me realize that I can do that easily with getsym:

using ModelingToolkit, OrdinaryDiffEq, SymbolicIndexingInterface
using ModelingToolkit: t_nounits as t, D_nounits as D
@variables x(t)
@parameters p
@named M = ODESystem([D(x) ~ p], t)
Ms = structural_simplify(M)
prob = ODEProblem(Ms, [x => 1.0], (0.0, 1.0), [p => 2.0])
getsym(prob, x)(prob) # 1.0
getsym(prob, p)(prob) # 2.0

And I guess setsym should do the job if I need to set them instead. Thank you.