fix: construct initializeprob if initial value is symbolic

Author: AayushSabharwal

src/systems/problem_utils.jl

@@ -433,9 +433,14 @@ function process_SciMLProblem(
         solvablepars = [p
                         for p in parameters(sys)
                         if is_parameter_solvable(p, pmap, defs, guesses)]
+        has_dependent_unknowns = any(unknowns(sys)) do sym
+            val = get(op, sym, nothing)
+            val === nothing && return false
+            return symbolic_type(val) != NotSymbolic() || is_array_of_symbolics(val)
+        end
         if build_initializeprob &&
            (((implicit_dae || has_observed_u0s || !isempty(missing_unknowns) ||
-              !isempty(solvablepars)) &&
+              !isempty(solvablepars) || has_dependent_unknowns) &&
              get_tearing_state(sys) !== nothing) ||
             !isempty(initialization_equations(sys))) && t !== nothing
             initializeprob = ModelingToolkit.InitializationProblem(

test/initializationsystem.jl

@@ -844,3 +844,19 @@ end
     isys = ModelingToolkit.generate_initializesystem(sys)
     @test isequal(defaults(isys)[y], 2x + 1)
 end
+
+@testset "Create initializeprob when unknown has dependent value" begin
+    @variables x(t) y(t)
+    @mtkbuild sys = ODESystem([D(x) ~ x, D(y) ~ t * y], t; defaults = [x => 2y])
+    prob = ODEProblem(sys, [y => 1.0], (0.0, 1.0))
+    @test prob.f.initializeprob !== nothing
+    integ = init(prob)
+    @test integ[x] ≈ 2.0
+
+    @variables x(t)[1:2] y(t)
+    @mtkbuild sys = ODESystem([D(x) ~ x, D(y) ~ t], t; defaults = [x => [y, 3.0]])
+    prob = ODEProblem(sys, [y => 1.0], (0.0, 1.0))
+    @test prob.f.initializeprob !== nothing
+    integ = init(prob)
+    @test integ[x] ≈ [1.0, 3.0]
+end