Fails to solve numeric problem, which is successfully generated

[B-LIE]

Question❓

I am working on components for a chemical reactor. The problem is possibly numerically tricky, as it involves compression of liquid with thermodynamics from Clapeyron.jl.

I have successfully compiled the symbolic model (sys_rx), and successfully created a numeric problem:

julia> prob_rx = ODEProblem(sys_rx, [], tspan)
ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
Initialization status: FULLY_DETERMINED
Non-trivial mass matrix: true
timespan: (0.0, 30.0)
u0: 8-element Vector{Float64}:
   298.15
   298.15
   298.15
   358.15
    -1.2078062302879227e10
  5969.36198084991
   199.9832285042051
 33891.033956759304

When I try to solve the model, I get an error message... Are you guys able to see where the problem lies? Note: pb_e is one of three instances of a PressureBoundaryR component I have created. It has a single port a, and mass fraction vector x (three elements).

julia> sol_rx = solve(prob_rx,Rodas5())
UndefVarError: `pb_e₊a₊x` not defined in `ModelingToolkitBase`
Suggestion: check for spelling errors or missing imports.

Stacktrace:
  [1] macro expansion
    @ C:\Users\Bernt\.julia\packages\Symbolics\29fkz\src\build_function.jl:143 [inlined]
  [2] macro expansion
    @ C:\Users\Bernt\.julia\packages\RuntimeGeneratedFunctions\9kbBw\src\RuntimeGeneratedFunctions.jl:200 [inlined]
  [3] macro expansion
    @ .\none:0 [inlined]
  [4] generated_callfunc(::RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc9f8d13b, 0x149ba748, 0xea26c03f, 0xcdd5fdda, 0x0aafbe4f), Nothing}, ::Tuple{Vector{Float64}, Vector{Vector{Float64}}}, ::Tuple{SciMLBase.NonlinearSolution{Float64, 1, Vector{Float64}, Vector{Float64}, @NamedTuple{p::Nothing}, Nothing, Nothing, Nothing, Nothing, Nothing}}, ::MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}})
    @ ModelingToolkitBase .\none:0
  [5] iteratively_build_sols(alg::SCCNonlinearSolve.SCCAlg{NonlinearSolveBase.NonlinearSolvePolyAlgorithm{Val{6}, Tuple{NonlinearSolveQuasiNewton.QuasiNewtonAlgorithm{Nothing, Missing, NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveQuasiNewton.GoodBroydenUpdateRule, NonlinearSolveQuasiNewton.NoChangeInStateReset{Nothing}, NonlinearSolveQuasiNewton.IdentityInitialization{Nothing, NonlinearSolveQuasiNewton.FullStructure}, Val{false}}, NonlinearSolveQuasiNewton.QuasiNewtonAlgorithm{Nothing, Missing, NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveQuasiNewton.KlementUpdateRule, NonlinearSolveQuasiNewton.IllConditionedJacobianReset, NonlinearSolveQuasiNewton.IdentityInitialization{Nothing, NonlinearSolveQuasiNewton.DiagonalStructure}, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, Missing, NonlinearSolveBase.NewtonDescent{Nothing}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, NonlinearSolveFirstOrder.GenericTrustRegionScheme{NonlinearSolveFirstOrder.RadiusUpdateSchemes.__Simple, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}}, NonlinearSolveBase.Dogleg{NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveBase.SteepestDescent{Nothing}}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, NonlinearSolveFirstOrder.GenericTrustRegionScheme{NonlinearSolveFirstOrder.RadiusUpdateSchemes.__Bastin, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}}, NonlinearSolveBase.Dogleg{NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveBase.SteepestDescent{Nothing}}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, NonlinearSolveFirstOrder.LevenbergMarquardtTrustRegion{Float64}, NonlinearSolveBase.GeodesicAcceleration{NonlinearSolveBase.DampedNewtonDescent{Nothing, Float64, NonlinearSolveFirstOrder.LevenbergMarquardtDampingFunction{Float64, Float64, Float64}}, Float64, Float64}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{true}}}}, Nothing}, probs::Tuple{NonlinearProblem{Vector{Float64}, true, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, NonlinearFunction{true, SciMLBase.FullSpecialize, ModelingToolkitBase.GeneratedFunctionWrapper{(2, 2, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x47db40c9, 0xe298923d, 0xb41b6f8a, 0x7aefd986, 0x69b98c8b), Nothing}, RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:ˍ₋out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc6ae11a3, 0xbcc3220a, 0x19add288, 0x6a187cca, 0xb2e71098), Nothing}}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, @Kwargs{}, SciMLBase.StandardNonlinearProblem, Nothing, Nothing}, NonlinearProblem{Vector{Float64}, true, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, NonlinearFunction{true, SciMLBase.FullSpecialize, ModelingToolkitBase.GeneratedFunctionWrapper{(2, 2, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x697b509a, 0xd09feca7, 0xa845e4f0, 0x0a8725ae, 0x6bd5de47), Nothing}, RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:ˍ₋out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x65040671, 0x92ffaf79, 0x5c6df019, 0x81e065de, 0x987668b7), Nothing}}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, @Kwargs{}, SciMLBase.StandardNonlinearProblem, Nothing, Nothing}, NonlinearProblem{Vector{Float64}, true, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, NonlinearFunction{true, SciMLBase.FullSpecialize, ModelingToolkitBase.GeneratedFunctionWrapper{(2, 2, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xfd9f3524, 0xa2c748d2, 0x76d6c94e, 0x46380581, 0x3cb7586f), Nothing}, RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:ˍ₋out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x44200bc6, 0xd4d163b8, 0x0ab21b94, 0xc677f686, 0xba7df4d3), Nothing}}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, @Kwargs{}, SciMLBase.StandardNonlinearProblem, Nothing, Nothing}}, explicitfuns::Tuple{ModelingToolkit.CacheWriter{ModelingToolkitBase.GeneratedFunctionWrapper{(3, 3, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc6013997, 0xd53377e6, 0x0997f324, 0xe074eff3, 0x338bbd24), Nothing}, Nothing}}, ModelingToolkit.CacheWriter{ModelingToolkitBase.GeneratedFunctionWrapper{(3, 3, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc9f8d13b, 0x149ba748, 0xea26c03f, 0xcdd5fdda, 0x0aafbe4f), Nothing}, Nothing}}, ModelingToolkit.CacheWriter{ModelingToolkitBase.GeneratedFunctionWrapper{(3, 3, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x3809ac20, 0x4313bc5d, 0x5b071d07, 0xa511ee84, 0x356e452b), Nothing}, Nothing}}}, ::Val{3}; kwargs::@Kwargs{abstol::Float64, reltol::Float64})
    @ SCCNonlinearSolve C:\Users\Bernt\.julia\packages\SCCNonlinearSolve\xmKJS\src\SCCNonlinearSolve.jl:113
  [6] _scc_solve(prob::SCCNonlinearProblem{Vector{Float64}, false, Tuple{NonlinearProblem{Vector{Float64}, true, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, NonlinearFunction{true, SciMLBase.FullSpecialize, ModelingToolkitBase.GeneratedFunctionWrapper{(2, 2, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x47db40c9, 0xe298923d, 0xb41b6f8a, 0x7aefd986, 0x69b98c8b), Nothing}, RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:ˍ₋out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc6ae11a3, 0xbcc3220a, 0x19add288, 0x6a187cca, 0xb2e71098), Nothing}}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, @Kwargs{}, SciMLBase.StandardNonlinearProblem, Nothing, Nothing}, NonlinearProblem{Vector{Float64}, true, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, NonlinearFunction{true, SciMLBase.FullSpecialize, ModelingToolkitBase.GeneratedFunctionWrapper{(2, 2, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x697b509a, 0xd09feca7, 0xa845e4f0, 0x0a8725ae, 0x6bd5de47), Nothing}, RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:ˍ₋out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x65040671, 0x92ffaf79, 0x5c6df019, 0x81e065de, 0x987668b7), Nothing}}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, @Kwargs{}, SciMLBase.StandardNonlinearProblem, Nothing, Nothing}, NonlinearProblem{Vector{Float64}, true, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, NonlinearFunction{true, SciMLBase.FullSpecialize, ModelingToolkitBase.GeneratedFunctionWrapper{(2, 2, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xfd9f3524, 0xa2c748d2, 0x76d6c94e, 0x46380581, 0x3cb7586f), Nothing}, RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:ˍ₋out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x44200bc6, 0xd4d163b8, 0x0ab21b94, 0xc677f686, 0xba7df4d3), Nothing}}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, @Kwargs{}, SciMLBase.StandardNonlinearProblem, Nothing, Nothing}}, Tuple{ModelingToolkit.CacheWriter{ModelingToolkitBase.GeneratedFunctionWrapper{(3, 3, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc6013997, 0xd53377e6, 0x0997f324, 0xe074eff3, 0x338bbd24), Nothing}, Nothing}}, ModelingToolkit.CacheWriter{ModelingToolkitBase.GeneratedFunctionWrapper{(3, 3, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0xc9f8d13b, 0x149ba748, 0xea26c03f, 0xcdd5fdda, 0x0aafbe4f), Nothing}, Nothing}}, ModelingToolkit.CacheWriter{ModelingToolkitBase.GeneratedFunctionWrapper{(3, 3, true), RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(:out, :__mtk_arg_1, :___mtkparameters___), ModelingToolkitBase.var"#_RGF_ModTag", ModelingToolkitBase.var"#_RGF_ModTag", (0x3809ac20, 0x4313bc5d, 0x5b071d07, 0xa511ee84, 0x356e452b), Nothing}, Nothing}}}, NonlinearFunction{false, SciMLBase.FullSpecialize, Returns{Nothing}, LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED_NO_TIME), Nothing, System, Nothing, Nothing}, MTKParameters{Vector{Float64}, StaticArraysCore.SVector{0, Float64}, Tuple{}, Tuple{}, Tuple{}, Tuple{Vector{Float64}, Vector{Vector{Float64}}}}, Val{true}}, alg::SCCNonlinearSolve.SCCAlg{NonlinearSolveBase.NonlinearSolvePolyAlgorithm{Val{6}, Tuple{NonlinearSolveQuasiNewton.QuasiNewtonAlgorithm{Nothing, Missing, NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveQuasiNewton.GoodBroydenUpdateRule, NonlinearSolveQuasiNewton.NoChangeInStateReset{Nothing}, NonlinearSolveQuasiNewton.IdentityInitialization{Nothing, NonlinearSolveQuasiNewton.FullStructure}, Val{false}}, NonlinearSolveQuasiNewton.QuasiNewtonAlgorithm{Nothing, Missing, NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveQuasiNewton.KlementUpdateRule, NonlinearSolveQuasiNewton.IllConditionedJacobianReset, NonlinearSolveQuasiNewton.IdentityInitialization{Nothing, NonlinearSolveQuasiNewton.DiagonalStructure}, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, Missing, NonlinearSolveBase.NewtonDescent{Nothing}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, NonlinearSolveFirstOrder.GenericTrustRegionScheme{NonlinearSolveFirstOrder.RadiusUpdateSchemes.__Simple, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}}, NonlinearSolveBase.Dogleg{NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveBase.SteepestDescent{Nothing}}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, NonlinearSolveFirstOrder.GenericTrustRegionScheme{NonlinearSolveFirstOrder.RadiusUpdateSchemes.__Bastin, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}, Rational{Int64}}, NonlinearSolveBase.Dogleg{NonlinearSolveBase.NewtonDescent{Nothing}, NonlinearSolveBase.SteepestDescent{Nothing}}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{false}}, NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithm{Missing, NonlinearSolveFirstOrder.LevenbergMarquardtTrustRegion{Float64}, NonlinearSolveBase.GeodesicAcceleration{NonlinearSolveBase.DampedNewtonDescent{Nothing, Float64, NonlinearSolveFirstOrder.LevenbergMarquardtDampingFunction{Float64, Float64, Float64}}, Float64, Float64}, Nothing, ADTypes.AutoForwardDiff{1, ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}}, Nothing, Nothing, Val{true}}}}, Nothing}; kwargs::@Kwargs{abstol::Float64, reltol::Float64})
    @ SCCNonlinearSolve C:\Users\Bernt\.julia\packages\SCCNonlinearSolve\xmKJS\src\SCCNonlinearSolve.jl:153
  [7] _scc_solve
    @ C:\Users\Bernt\.julia\packages\SCCNonlinearSolve\xmKJS\src\SCCNonlinearSolve.jl:153 [inlined]
  [8] #scc_solve_up#6
    @ C:\Users\Bernt\.julia\packages\SCCNonlinearSolve\xmKJS\src\SCCNonlinearSolve.jl:62 [inlined]
  [9] scc_solve_up
    @ C:\Users\Bernt\.julia\packages\SCCNonlinearSolve\xmKJS\src\SCCNonlinearSolve.jl:58 [inlined]
 [10] #solve#5
    @ C:\Users\Bernt\.julia\packages\SCCNonlinearSolve\xmKJS\src\SCCNonlinearSolve.jl:52 [inlined]
 [11] solve

[AayushSabharwal]

What version of ModelingToolkit is this on? It seems like a codegen bug and might be related to #4345 but I'm not entirely sure.

[B-LIE]

I'm on ModelingToolkit v11.13.0.

Another questions... I have used ifelse(scalar>0, array, array) but that does not work. Seems like the test (scalar>0) has to have the same shape as array. I have got around it by a re-write as a comprehension -- which is fine, but takes more space.

Could I define an array of the same shape as array, filled with ones, and do ifelse(scalar*unitarray > 0, array, array) (or .> 0)?

I can provide you with the case I have, but it is relatively extensive... and I don't know how I can reduce it to a simple example.

[AayushSabharwal]

Another questions... I have used ifelse(scalar>0, array, array) but that does not work. Seems like the test (scalar>0) has to have the same shape as array. I have got around it by a re-write as a comprehension -- which is fine, but takes more space.

Yeah, this is tracked in JuliaSymbolics/Symbolics.jl#1771. It's a bit overdue. I'll try and get to it today, and ifelse(cond, arr1, arr2) should work if arr1 and arr2 have the same shape.

[AayushSabharwal]

I can provide you with the case I have, but it is relatively extensive... and I don't know how I can reduce it to a simple example.

Is this for the codegen error, or ifelse? If it's for the codegen error, please do provide whatever example you have. The bug is serious enough that I'd like to fix it ASAP.

[B-LIE]

It is for the codegen thing. The ifelse thing is a minor thing; more a convenience.

I normally use Jupyter notebooks within VSCode (not IJulia).This allows me to add comments in a more natural way than pure .jl files, but I guess you developers are not so fond of Jupyter notebook... Is it ok that I attach a stripped down version of the notebook (i.e., removing other code that is not relevant)?

[AayushSabharwal]

If there's a way to share it that isn't an .ipynb file I'd really appreciate it 😅 I don't have the tools installed to open such files. A quick google says there should be an option to export the notebook as a Julia file. If that's possible, it would be great.

[B-LIE]

He he. I'll try to move it over to a pure .jo file. Sent from Outlook for Android<https://aka.ms/AAb9ysg>

…

________________________________ From: Aayush Sabharwal ***@***.***> Sent: Friday, March 6, 2026 10:11:10 AM To: SciML/ModelingToolkit.jl ***@***.***> Cc: Bernt Lie ***@***.***>; Author ***@***.***> Subject: Re: [SciML/ModelingToolkit.jl] Fails to solve numeric problem, which is successfully generated (Issue #4377) [https://avatars.githubusercontent.com/u/23384717?s=20&v=4]AayushSabharwal left a comment (SciML/ModelingToolkit.jl#4377)<#4377 (comment)> If there's a way to share it that isn't an .ipynb file I'd really appreciate it 😅 I don't have the tools installed to open such files. A quick google says there should be an option to export the notebook as a Julia file. If that's possible, it would be great. — Reply to this email directly, view it on GitHub<#4377 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AM5VM7355RTXQHFRMWHZV234PKI25AVCNFSM6AAAAACWIRHJOSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DAMJQGUYTSMJZGM>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[B-LIE]

OK, please find the attached file (Reactor_bug.txt). It is nearly 600 lines... with comments. I have tried to structure the code to make it readable... The error appears on the last line; all prior to that should run fine. I used the latest versions of all packages I import as of yesterday.

NOTE: I had to change the file type from *.jl to *.txt to be able to upload it... Just change it back to *.jl.

Reactor_bug.txt

[AayushSabharwal]

Awesome, thank you! I'll investigate soon and let you know what I find.

[B-LIE]

I should say... the problem is pretty problematic numerically, I would think. So perhaps the simulation crashes when it runs. But now it never even gets to the point of running.

[Essentially, it is a rigid tank filled with liquid, where a chemical reaction takes place. Pressure in tank should be around 20 bar. Liquid is pressured into the tank through a valve, thus compressing the liquid slightly, leading to a an increase in pressure in the tank, and more liquid pressured out to through a valve against a counter pressure.

The numeric difficulty lies in the following: with just a few grams of liquid too little in the tank, the fluid flashes and becomes liquid + vapor, with a pressure of perhaps 0.3 bar. If the amount of liquid is too high (just a few grams), the pressure in the tank can go to hundreds or thousands of bar pressure.]

[AayushSabharwal]

I've identified what is wrong. The explicitfun!s in the SCCNonlinearProblem used for initialization are wrong. I think something is wrong with the process that identifies the subset of observed equations to use.

In the meantime, it seems your initialization is small enough that the SCC form is not strictly necessary. Passing use_scc = false to ODEProblem will allow working around this issue until it is resolved.

[B-LIE]

OK, the code runs now. I have discovered some bugs in my code, but that is more related to the physics of the problem. I will update you with a correct model when I have figured out the problems.

[wenflo]

I'm facing the same error, on a maybe similar problem. I'm working on a package with components for modeling systems for (waste)water treatment (BioChemicalTreatment, https://gitlab.com/datinfo/BioChemicalTreatment.jl), which I'm working on updating to MTKv11.
On MTKv10 everything worked fine, but now it seems to produce an error.

As @B-LIE I'm able to generate the problem:

julia> prob = ODEProblem(sys_simp, [], (0, 5))
ODEProblem with uType Vector{Float64} and tType Int64. In-place: true
Initialization status: FULLY_DETERMINED
Non-trivial mass matrix: false
timespan: (0, 5)
u0: 2-element Vector{Float64}:
 1.0
 0.5

And, when solving, I get a very similar error:

julia> sol = solve(prob)
ERROR: UndefVarError: `reactor₊rates₊X_B` not defined in `ModelingToolkitBase`
Suggestion: check for spelling errors or missing imports.
Stacktrace:
  [1] macro expansion
    @ ~/.julia/packages/Symbolics/eUt8E/src/build_function.jl:381 [inlined]
  [2] macro expansion
    @ ~/.julia/packages/RuntimeGeneratedFunctions/9kbBw/src/RuntimeGeneratedFunctions.jl:200 [inlined]
  [3] macro expansion
    @ ./none:0 [inlined]
  [4] generated_callfunc(::RuntimeGeneratedFunctions.RuntimeGeneratedFunction{…}, ::Vector{…}, ::MTKParameters{…}, ::Int64)
    @ ModelingToolkitBase ./none:0
  [5] (::ModelingToolkitBase.ReconstructInitializeprob{…})(srcvalp::SymbolicIndexingInterface.ProblemState{…}, dstvalp::SCCNonlinearProblem{…})
    @ ModelingToolkitBase ~/.julia/packages/ModelingToolkitBase/uIKoY/src/systems/problem_utils.jl:912
  [6] remake_initialization_data(sys::System, odefn::ODEFunction{…}, u0::Vector{…}, t0::Int64, p::MTKParameters{…}, newu0::Vector{…}, newp::MTKParameters{…})
    @ ModelingToolkitBase ~/.julia/packages/ModelingToolkitBase/uIKoY/src/systems/nonlinear/initializesystem.jl:569
  [7] remake(prob::ODEProblem{…}; f::Missing, u0::Vector{…}, tspan::Missing, p::MTKParameters{…}, kwargs::Missing, interpret_symbolicmap::Bool, build_initializeprob::Type, use_defaults::Bool, lazy_initialization::Nothing, _kwargs::@Kwargs{})
    @ SciMLBase ~/.julia/packages/SciMLBase/RJROr/src/remake.jl:333
  [8] get_updated_symbolic_problem(sys::System, prob::ODEProblem{…}; u0::Vector{…}, p::MTKParameters{…}, kw::@Kwargs{})
    @ ModelingToolkitBase ~/.julia/packages/ModelingToolkitBase/uIKoY/src/systems/nonlinear/initializesystem.jl:786
  [9] get_updated_symbolic_problem
    @ ~/.julia/packages/ModelingToolkitBase/uIKoY/src/systems/nonlinear/initializesystem.jl:757 [inlined]
 [10] get_concrete_problem(prob::ODEProblem{…}, isadapt::Bool; alg::Nothing, kwargs::@Kwargs{…})
    @ DiffEqBase ~/.julia/packages/DiffEqBase/XLzeg/src/solve.jl:667
 [11] solve_up(::ODEProblem{…}, ::Nothing, ::Vector{…}, ::MTKParameters{…}; originator::SciMLBase.ChainRulesOriginator, kwargs::@Kwargs{})
    @ DiffEqBase ~/.julia/packages/DiffEqBase/XLzeg/src/solve.jl:603
 [12] solve(::ODEProblem{…}; sensealg::Nothing, u0::Nothing, p::Nothing, wrap::Val{…}, kwargs::@Kwargs{})
    @ DiffEqBase ~/.julia/packages/DiffEqBase/XLzeg/src/solve.jl:575

However, in my stacktrace, SCCNonlinearProblem does not appear and use_scc = false doesn't help.

My equations are rather simple, but the structure that is simplified is rather complex (to account for other uses with more complex models etc). The equations effectively boil down to the following after simplification:

julia> full_equations(sys_simp)
2-element Vector{Equation}:
 Differential(t, 1)(reactor₊states₊X_B(t)) ~ -disinfection₊kd*reactor₊states₊S_O3(t)*reactor₊states₊X_B(t)
 Differential(t, 1)(reactor₊states₊S_O3(t)) ~ -disinfection₊kO3*reactor₊states₊S_O3(t)

Thus my question (and why I thought to post here as well): Is this the same error? Or is it something else, maybe something in my code that worked with v10, but with v11 not anymore?

I just tested with the most recently released version of MTK (v11.17.0).

If it helps, I can provide a small example of this as well (easiest using my (registered) package, or alternatively I could try to copy everything in one file). I didn't manage to get a proper MWE using only MTK so far.

[AayushSabharwal]

Hi @wenflo. Thanks for reporting the issue. Your error looks slightly different than that of @B-LIE. If it's possible, I'd greatly appreciate a reproducer.