Can create `NonlinearProblem` with missing variables

[TorkelE]

using ModelingToolkit
using ModelingToolkit: t_nounits as t, D_nounits as D

@parameters p d
@variables X(t)
eqs = [0 ~ p - d*X]
@mtkbuild sys = NonlinearSystem(eqs)

u0 = []
ps = [p => 1.0, d => 1.0] # Value for `d` is missing

prob = NonlinearProblem(sys, u0, ps) # This passes without problem.
prob[X] # Is 1.0

[ChrisRackauckas]

I don't get it, aren't you setting d to 1.0 so it's not missing?

[TorkelE]

Actually, that might make sense. Previously you always had to provide guesses through u0 for everything, but now since structural_simplify does some solving before solve X gets eliminated and this behaviour makes sense.