Gradient Calculation fails for DAE (Mass-Matrix ODEProblem) during backward pass initialization with ReverseDiffAdjoint

[AndreasHofmann217]

Describe the bug 🐞

Hi,
trying to train parameters within a singular mass-matrix ODEProblem fails in the backward pass with ReverseDiffAdjoint() during the initialization of the DAE (also reinitialization after an event). Providing the correct initial guess values for the NLS does mitigate the initialization problem. However, I argue that it is not typical to know all correct guess values. For reinitialization after an events the issue still remains.

Since the Julia Error points to initialize_dae.jl within OrdinaryDiffEqNonlinearSolve and a retcode-check that is not compatible with ReverseDiff.jl I was assuming that commenting this retcode-check would resolve the issue. Unfortunately it did not.

Minimal Reproducible Example 👇

using ComponentArrays
using LinearAlgebra
using OrdinaryDiffEq
using ReverseDiff
using SciMLSensitivity

# model where parameters shall be tuned
function trainable_pendulum!(du,u,p,t)
    y, vy, x, vx, f = u
    L = 1.0
    g = p.g
    m = p.m
    du[1] = vy
    du[2] = (y*f/ L + g*m) / m
    du[3] = L^2 - x^2 - y^2
    du[4] = -2*x*vx - 2*y*vy
    du[5] = (-2*x^2*f)/ (L*m) + (-2*((y*f)/L + g*m)*y) / m - 2*(vx^2) - 2*vy^2
end

# add some dummy time events here to emphasize issues with reinitialization of DAE
t_stops = [0.2,0.4]
condition(u,t,integrator) = t in t_stops
affect!(integrator) = () -> ()
callback = DiscreteCallback(condition,affect!)

# provide incorrect initialization guess value, since for complex models this is typically not known
inital_pendulum = [0.0,0.0,0.8,0.0,0.0] # valid initialization: [0.0,0.0,1.0,0.0,0.0]
mass_matrix = diagm([1.0,1.0,0.0,0.0,0.0])
pend_fun = ODEFunction(trainable_pendulum!;mass_matrix=mass_matrix)

# reference solution
ref_params = ComponentArray{Float64}(g = -9.81, m = 1.0)
ref_pend_prob = ODEProblem(pend_fun,inital_pendulum,(0.0,0.5),ref_params)
ref_pend_sol = solve(ref_pend_prob,QBDF(),abstol=1e-6,reltol=1e-6, callback=callback, tstops=t_stops, saveat=1e-3)

# training part
train_pend_fun = ODEFunction(trainable_pendulum!;mass_matrix = mass_matrix)
trainable_params = ComponentArray{Float64}(g = -5.0, m = 0.4)
train_pend_prob = ODEProblem(pend_fun,inital_pendulum,(0.0,0.5),trainable_params)

solvers = [Rodas5P(autodiff=false), QBDF(autodiff=false)] # to have  a single and a multistep solver

# simple mae loss
function loss(p,solver)

    prob_ = remake(train_pend_prob;p=p)
    sol_ = solve(prob_, solver,abstol=1e-6,reltol=1e-6, tstops=t_stops, callback=callback, saveat=1e-3, sensealg=ReverseDiffAdjoint())
    return sum(1/503*sum(abs.(view(Array(sol_),1:4,:) .- view(Array(ref_pend_sol),1:4,:)),dims=2))
end

for used_solver in solvers
    try
        println("\n\n\nSolver: $(split(string(used_solver),"(")[1])")
        _loss_ReverseDiffGradient(par) = loss(par,used_solver)
        grad = similar(trainable_params)
        println("loss: $(_loss_ReverseDiffGradient(trainable_params))")
        ReverseDiff.gradient!(grad,_loss_ReverseDiffGradient,trainable_params)

        println("gradient for solver: $(split(string(used_solver),"(")[1]) is $(grad)")    
    catch err
        println("$(err) - Failed to calculate gradient")
    end
end   

Error & Stacktrace ⚠️

┌ Warning: Potential performance improvement omitted. EnzymeVJP tried and failed in the automated AD choice algorithm. To show the stack trace, set SciMLSensitivity.STACKTRACE_WITH_VJPWARN[] = true. To turn off this printing, add `verbose = false` to the `solve` call.
└ @ SciMLSensitivity C:\Users\hofmaand\.julia\packages\SciMLSensitivity\AjuEJ\src\concrete_solve.jl:24

┌ Warning: Potential performance improvement omitted. ReverseDiffVJP tried and failed in the automated AD choice algorithm. To show the stack trace, set SciMLSensitivity.STACKTRACE_WITH_VJPWARN[] = true. To turn off this printing, add `verbose = false` to the `solve` call.
└ @ SciMLSensitivity C:\Users\hofmaand\.julia\packages\SciMLSensitivity\AjuEJ\src\concrete_solve.jl:67

ERROR: type TrackedArray has no field retcode
Stacktrace:
  [1] getproperty(x::ReverseDiff.TrackedArray{Float64, Float64, 1, Vector{Float64}, Vector{Float64}}, f::Symbol)
    @ Base .\Base.jl:49
  [2] _initialize_dae!(integrator::OrdinaryDiffEqCore.ODEIntegrator{…}, prob::ODEProblem{…}, alg::BrownFullBasicInit{…}, isinplace::Val{…})
    @ OrdinaryDiffEqNonlinearSolve C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqNonlinearSolve\s1Hh9\src\initialize_dae.jl:442
  [3] _initialize_dae!
    @ C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\initialize_dae.jl:56 [inlined]
  [4] initialize_dae!
    @ C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\initialize_dae.jl:40 [inlined]
  [5] initialize_dae!(integrator::OrdinaryDiffEqCore.ODEIntegrator{…})
    @ OrdinaryDiffEqCore C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\initialize_dae.jl:40
  [6] __init(prob::ODEProblem{…}, alg::Rodas5P{…}, timeseries_init::Tuple{}, ts_init::Tuple{}, ks_init::Tuple{}, recompile::Type{…}; saveat::Float64, tstops::Vector{…}, d_discontinuities::Tuple{}, save_idxs::Nothing, save_everystep::Bool, save_on::Bool, save_start::Bool, save_end::Nothing, callback::DiscreteCallback{…}, dense::Bool, calck::Bool, dt::Float64, dtmin::Float64, dtmax::Float64, force_dtmin::Bool, adaptive::Bool, gamma::Rational{…}, abstol::Float64, reltol::Float64, qmin::Rational{…}, qmax::Int64, qsteady_min::Int64, qsteady_max::Rational{…}, beta1::Nothing, beta2::Nothing, qoldinit::Rational{…}, controller::Nothing, fullnormalize::Bool, failfactor::Int64, maxiters::Int64, internalnorm::typeof(DiffEqBase.ODE_DEFAULT_NORM), internalopnorm::typeof(opnorm), isoutofdomain::typeof(DiffEqBase.ODE_DEFAULT_ISOUTOFDOMAIN), unstable_check::typeof(DiffEqBase.ODE_DEFAULT_UNSTABLE_CHECK), verbose::Bool, timeseries_errors::Bool, dense_errors::Bool, advance_to_tstop::Bool, stop_at_next_tstop::Bool, initialize_save::Bool, progress::Bool, progress_steps::Int64, progress_name::String, progress_message::typeof(DiffEqBase.ODE_DEFAULT_PROG_MESSAGE), progress_id::Symbol, userdata::Nothing, allow_extrapolation::Bool, initialize_integrator::Bool, alias::ODEAliasSpecifier, initializealg::OrdinaryDiffEqCore.DefaultInit, kwargs::@Kwargs{})
    @ OrdinaryDiffEqCore C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\solve.jl:565
  [7] __init (repeats 5 times)
    @ C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\solve.jl:11 [inlined]
  [8] __solve(::ODEProblem{…}, ::Rodas5P{…}; kwargs::@Kwargs{…})
    @ OrdinaryDiffEqCore C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\solve.jl:6
  [9] __solve
    @ C:\Users\hofmaand\.julia\packages\OrdinaryDiffEqCore\CNpVk\src\solve.jl:1 [inlined]
 [10] #solve_call#35
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:635 [inlined]
 [11] solve_call
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:592 [inlined]
 [12] #solve_up#44
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:1142 [inlined]
 [13] solve_up
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:1120 [inlined]
 [14] solve(prob::ODEProblem{…}, args::Rodas5P{…}; sensealg::SensitivityADPassThrough, u0::Nothing, p::Nothing, wrap::Val{…}, kwargs::@Kwargs{…})
    @ DiffEqBase C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:1057
 [15] (::SciMLSensitivity.var"#reversediff_adjoint_forwardpass#402"{…})(_u0::ReverseDiff.TrackedArray{…}, _p::ReverseDiff.TrackedArray{…})
    @ SciMLSensitivity C:\Users\hofmaand\.julia\packages\SciMLSensitivity\AjuEJ\src\concrete_solve.jl:1507
 [16] ReverseDiff.GradientTape(f::Function, input::Tuple{…}, cfg::ReverseDiff.GradientConfig{…})
    @ ReverseDiff C:\Users\hofmaand\.julia\packages\ReverseDiff\p1MzG\src\api\tape.jl:207
 [17] GradientTape
    @ C:\Users\hofmaand\.julia\packages\ReverseDiff\p1MzG\src\api\tape.jl:204 [inlined]
 [18] _concrete_solve_adjoint(::ODEProblem{…}, ::Rodas5P{…}, ::ReverseDiffAdjoint, ::Vector{…}, ::ComponentVector{…}, ::SciMLBase.ReverseDiffOriginator; kwargs::@Kwargs{…})
    @ SciMLSensitivity C:\Users\hofmaand\.julia\packages\SciMLSensitivity\AjuEJ\src\concrete_solve.jl:1519
 [19] _solve_adjoint(prob::ODEProblem{…}, sensealg::ReverseDiffAdjoint, u0::Vector{…}, p::ComponentVector{…}, originator::SciMLBase.ReverseDiffOriginator, args::Rodas5P{…}; merge_callbacks::Bool, kwargs::@Kwargs{…})
    @ DiffEqBase C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:1598
 [20] _solve_adjoint
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:1571 [inlined]
 [21] (::DiffEqBaseReverseDiffExt.var"##solve_up#230#23"{…})(prob::ODEProblem{…}, sensealg::ReverseDiffAdjoint, u0::Vector{…}, p::ReverseDiff.TrackedArray{…}, args::Rodas5P{…}; kwargs::@Kwargs{…})
    @ DiffEqBaseReverseDiffExt C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\ext\DiffEqBaseReverseDiffExt.jl:180
 [22] ##solve_up#230#23
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\ext\DiffEqBaseReverseDiffExt.jl:179 [inlined]
 [23] track(::typeof(DiffEqBase.solve_up), prob::ODEProblem{…}, sensealg::ReverseDiffAdjoint, u0::Vector{…}, p::ReverseDiff.TrackedArray{…}, args::Rodas5P{…}; kwargs::@Kwargs{…})
    @ DiffEqBaseReverseDiffExt C:\Users\hofmaand\.julia\packages\ReverseDiff\p1MzG\src\macros.jl:195
 [24] track
    @ C:\Users\hofmaand\.julia\packages\ReverseDiff\p1MzG\src\macros.jl:190 [inlined]
 [25] #solve_up#14
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\ext\DiffEqBaseReverseDiffExt.jl:113 [inlined]
 [26] solve_up
    @ C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\ext\DiffEqBaseReverseDiffExt.jl:108 [inlined]
 [27] solve(prob::ODEProblem{…}, args::Rodas5P{…}; sensealg::ReverseDiffAdjoint, u0::Nothing, p::Nothing, wrap::Val{…}, kwargs::@Kwargs{…})
    @ DiffEqBase C:\Users\hofmaand\.julia\packages\DiffEqBase\HyX7M\src\solve.jl:1057
 [28] loss(p::ReverseDiff.TrackedArray{…}, solver::Rodas5P{…})
    @ Main c:\Users\hofmaand\Desktop\TrainableDAE\mwe.jl:47
 [29] _loss_ReverseDiffGradient(par::ReverseDiff.TrackedArray{Float64, Float64, 1, ComponentVector{…}, ComponentVector{…}})
    @ Main c:\Users\hofmaand\Desktop\TrainableDAE\mwe.jl:54
 [30] ReverseDiff.GradientTape(f::typeof(_loss_ReverseDiffGradient), input::ComponentVector{…}, cfg::ReverseDiff.GradientConfig{…})
    @ ReverseDiff C:\Users\hofmaand\.julia\packages\ReverseDiff\p1MzG\src\api\tape.jl:199
 [31] gradient!(result::ComponentVector{…}, f::Function, input::ComponentVector{…}, cfg::ReverseDiff.GradientConfig{…})
    @ ReverseDiff C:\Users\hofmaand\.julia\packages\ReverseDiff\p1MzG\src\api\gradients.jl:41
 [32] top-level scope
    @ c:\Users\hofmaand\Desktop\TrainableDAE\mwe.jl:57
Some type information was truncated. Use `show(err)` to see complete types.

Environment (please complete the following information):

• Output of using Pkg; Pkg.status()

Status `C:\Users\hofmaand\Desktop\TrainableDAE\Project.toml`
  [b0b7db55] ComponentArrays v0.15.25
  [1dea7af3] OrdinaryDiffEq v6.93.0
  [37e2e3b7] ReverseDiff v1.15.3
⌃ [1ed8b502] SciMLSensitivity v7.75.0
Info Packages marked with ⌃ have new versions available and may be upgradable.

• Output of using Pkg; Pkg.status(; mode = PKGMODE_MANIFEST)

Status `C:\Users\hofmaand\Desktop\TrainableDAE\Manifest.toml`
  [47edcb42] ADTypes v1.14.0
  [621f4979] AbstractFFTs v1.5.0
  [7d9f7c33] Accessors v0.1.42
  [79e6a3ab] Adapt v4.3.0
  [66dad0bd] AliasTables v1.1.3
  [ec485272] ArnoldiMethod v0.4.0
  [4fba245c] ArrayInterface v7.18.0
  [4c555306] ArrayLayouts v1.11.1
  [a9b6321e] Atomix v1.1.1
  [62783981] BitTwiddlingConvenienceFunctions v0.1.6
  [70df07ce] BracketingNonlinearSolve v1.1.1
  [fa961155] CEnum v0.5.0
  [2a0fbf3d] CPUSummary v0.2.6
  [7057c7e9] Cassette v0.3.14
  [082447d4] ChainRules v1.72.3
  [d360d2e6] ChainRulesCore v1.25.1
  [fb6a15b2] CloseOpenIntervals v0.1.13
  [38540f10] CommonSolve v0.2.4
  [bbf7d656] CommonSubexpressions v0.3.1
  [f70d9fcc] CommonWorldInvalidations v1.0.0
  [34da2185] Compat v4.16.0
  [b0b7db55] ComponentArrays v0.15.25
  [a33af91c] CompositionsBase v0.1.2
  [2569d6c7] ConcreteStructs v0.2.3
  [187b0558] ConstructionBase v1.5.8
  [adafc99b] CpuId v0.3.1
  [9a962f9c] DataAPI v1.16.0
  [864edb3b] DataStructures v0.18.22
  [e2d170a0] DataValueInterfaces v1.0.0
  [2b5f629d] DiffEqBase v6.167.0
  [459566f4] DiffEqCallbacks v4.3.0
  [77a26b50] DiffEqNoiseProcess v5.24.1
  [163ba53b] DiffResults v1.1.0
  [b552c78f] DiffRules v1.15.1
  [a0c0ee7d] DifferentiationInterface v0.6.48
  [31c24e10] Distributions v0.25.118
  [ffbed154] DocStringExtensions v0.9.3
  [4e289a0a] EnumX v1.0.4
  [7da242da] Enzyme v0.13.34
  [f151be2c] EnzymeCore v0.8.8
  [d4d017d3] ExponentialUtilities v1.27.0
  [e2ba6199] ExprTools v0.1.10
  [55351af7] ExproniconLite v0.10.14
  [7034ab61] FastBroadcast v0.3.5
  [9aa1b823] FastClosures v0.3.2
  [442a2c76] FastGaussQuadrature v1.0.2
  [a4df4552] FastPower v1.1.1
  [1a297f60] FillArrays v1.13.0
  [6a86dc24] FiniteDiff v2.27.0
  [f6369f11] ForwardDiff v0.10.38
  [f62d2435] FunctionProperties v0.1.2
  [069b7b12] FunctionWrappers v1.1.3
  [77dc65aa] FunctionWrappersWrappers v0.1.3
  [d9f16b24] Functors v0.5.2
  [0c68f7d7] GPUArrays v11.2.2
  [46192b85] GPUArraysCore v0.2.0
⌃ [61eb1bfa] GPUCompiler v1.2.0
  [c145ed77] GenericSchur v0.5.4
  [86223c79] Graphs v1.12.0
  [076d061b] HashArrayMappedTries v0.2.0
  [34004b35] HypergeometricFunctions v0.3.28
  [7869d1d1] IRTools v0.4.14
  [615f187c] IfElse v0.1.1
  [d25df0c9] Inflate v0.1.5
  [3587e190] InverseFunctions v0.1.17
  [92d709cd] IrrationalConstants v0.2.4
  [82899510] IteratorInterfaceExtensions v1.0.0
  [692b3bcd] JLLWrappers v1.7.0
  [ae98c720] Jieko v0.2.1
  [63c18a36] KernelAbstractions v0.9.34
  [ba0b0d4f] Krylov v0.9.10
  [929cbde3] LLVM v9.2.0
  [10f19ff3] LayoutPointers v0.1.17
  [5078a376] LazyArrays v2.6.1
  [87fe0de2] LineSearch v0.1.4
  [d3d80556] LineSearches v7.3.0
  [7ed4a6bd] LinearSolve v3.7.0
  [2ab3a3ac] LogExpFunctions v0.3.29
  [1914dd2f] MacroTools v0.5.15
  [d125e4d3] ManualMemory v0.1.8
  [bb5d69b7] MaybeInplace v0.1.4
  [e1d29d7a] Missings v1.2.0
  [2e0e35c7] Moshi v0.3.5
  [46d2c3a1] MuladdMacro v0.2.4
  [d41bc354] NLSolversBase v7.9.0
  [872c559c] NNlib v0.9.29
  [77ba4419] NaNMath v1.1.2
  [8913a72c] NonlinearSolve v4.5.0
  [be0214bd] NonlinearSolveBase v1.5.1
  [5959db7a] NonlinearSolveFirstOrder v1.3.0
  [9a2c21bd] NonlinearSolveQuasiNewton v1.2.0
  [26075421] NonlinearSolveSpectralMethods v1.1.0
  [d8793406] ObjectFile v0.4.4
  [429524aa] Optim v1.11.0
  [3bd65402] Optimisers v0.4.5
  [bac558e1] OrderedCollections v1.8.0
  [1dea7af3] OrdinaryDiffEq v6.93.0
  [89bda076] OrdinaryDiffEqAdamsBashforthMoulton v1.2.0
  [6ad6398a] OrdinaryDiffEqBDF v1.3.0
  [bbf590c4] OrdinaryDiffEqCore v1.20.0
  [50262376] OrdinaryDiffEqDefault v1.3.0
  [4302a76b] OrdinaryDiffEqDifferentiation v1.4.0
  [9286f039] OrdinaryDiffEqExplicitRK v1.1.0
  [e0540318] OrdinaryDiffEqExponentialRK v1.4.0
  [becaefa8] OrdinaryDiffEqExtrapolation v1.5.0
  [5960d6e9] OrdinaryDiffEqFIRK v1.9.0
  [101fe9f7] OrdinaryDiffEqFeagin v1.1.0
  [d3585ca7] OrdinaryDiffEqFunctionMap v1.1.1
  [d28bc4f8] OrdinaryDiffEqHighOrderRK v1.1.0
  [9f002381] OrdinaryDiffEqIMEXMultistep v1.3.0
  [521117fe] OrdinaryDiffEqLinear v1.1.0
  [1344f307] OrdinaryDiffEqLowOrderRK v1.2.0
  [b0944070] OrdinaryDiffEqLowStorageRK v1.3.0
  [127b3ac7] OrdinaryDiffEqNonlinearSolve v1.5.0
  [c9986a66] OrdinaryDiffEqNordsieck v1.1.0
  [5dd0a6cf] OrdinaryDiffEqPDIRK v1.3.0
  [5b33eab2] OrdinaryDiffEqPRK v1.1.0
  [04162be5] OrdinaryDiffEqQPRK v1.1.0
  [af6ede74] OrdinaryDiffEqRKN v1.1.0
  [43230ef6] OrdinaryDiffEqRosenbrock v1.8.0
  [2d112036] OrdinaryDiffEqSDIRK v1.3.0
  [669c94d9] OrdinaryDiffEqSSPRK v1.2.1
  [e3e12d00] OrdinaryDiffEqStabilizedIRK v1.3.0
  [358294b1] OrdinaryDiffEqStabilizedRK v1.1.0
  [fa646aed] OrdinaryDiffEqSymplecticRK v1.3.0
  [b1df2697] OrdinaryDiffEqTsit5 v1.1.0
  [79d7bb75] OrdinaryDiffEqVerner v1.1.1
  [90014a1f] PDMats v0.11.32
  [65ce6f38] PackageExtensionCompat v1.0.2
  [d96e819e] Parameters v0.12.3
  [e409e4f3] PoissonRandom v0.4.4
  [f517fe37] Polyester v0.7.16
  [1d0040c9] PolyesterWeave v0.2.2
  [85a6dd25] PositiveFactorizations v0.2.4
  [d236fae5] PreallocationTools v0.4.25
  [aea7be01] PrecompileTools v1.2.1
  [21216c6a] Preferences v1.4.3
  [43287f4e] PtrArrays v1.3.0
  [1fd47b50] QuadGK v2.11.2
  [74087812] Random123 v1.7.0
  [e6cf234a] RandomNumbers v1.6.0
  [c1ae055f] RealDot v0.1.0
  [3cdcf5f2] RecipesBase v1.3.4
  [731186ca] RecursiveArrayTools v3.31.1
  [189a3867] Reexport v1.2.2
  [ae029012] Requires v1.3.1
  [ae5879a3] ResettableStacks v1.1.1
  [37e2e3b7] ReverseDiff v1.15.3
  [79098fc4] Rmath v0.8.0
  [7e49a35a] RuntimeGeneratedFunctions v0.5.13
  [94e857df] SIMDTypes v0.1.0
  [0bca4576] SciMLBase v2.79.0
  [19f34311] SciMLJacobianOperators v0.1.1
  [c0aeaf25] SciMLOperators v0.3.13
⌃ [1ed8b502] SciMLSensitivity v7.75.0
  [53ae85a6] SciMLStructures v1.7.0
  [7e506255] ScopedValues v1.3.0
  [6c6a2e73] Scratch v1.2.1
  [efcf1570] Setfield v1.1.2
  [727e6d20] SimpleNonlinearSolve v2.2.0
  [699a6c99] SimpleTraits v0.9.4
  [ce78b400] SimpleUnPack v1.1.0
  [a2af1166] SortingAlgorithms v1.2.1
  [47a9eef4] SparseDiffTools v2.24.0
  [dc90abb0] SparseInverseSubset v0.1.2
  [0a514795] SparseMatrixColorings v0.4.14
  [276daf66] SpecialFunctions v2.5.0
  [aedffcd0] Static v1.2.0
  [0d7ed370] StaticArrayInterface v1.8.0
  [90137ffa] StaticArrays v1.9.13
  [1e83bf80] StaticArraysCore v1.4.3
  [10745b16] Statistics v1.11.1
  [82ae8749] StatsAPI v1.7.0
  [2913bbd2] StatsBase v0.34.4
  [4c63d2b9] StatsFuns v1.3.2
  [7792a7ef] StrideArraysCore v0.5.7
  [09ab397b] StructArrays v0.7.0
  [53d494c1] StructIO v0.3.1
  [2efcf032] SymbolicIndexingInterface v0.3.38
  [3783bdb8] TableTraits v1.0.1
  [bd369af6] Tables v1.12.0
  [8290d209] ThreadingUtilities v0.5.2
  [a759f4b9] TimerOutputs v0.5.28
  [9f7883ad] Tracker v0.2.37
  [781d530d] TruncatedStacktraces v1.4.0
  [3a884ed6] UnPack v1.0.2
  [013be700] UnsafeAtomics v0.3.0
  [19fa3120] VertexSafeGraphs v0.2.0
  [e88e6eb3] Zygote v0.7.4
  [700de1a5] ZygoteRules v0.2.7
  [7cc45869] Enzyme_jll v0.0.173+0
  [1d5cc7b8] IntelOpenMP_jll v2025.0.4+0
  [dad2f222] LLVMExtra_jll v0.0.35+0
  [856f044c] MKL_jll v2025.0.1+1
  [efe28fd5] OpenSpecFun_jll v0.5.6+0
  [f50d1b31] Rmath_jll v0.5.1+0
  [1317d2d5] oneTBB_jll v2022.0.0+0
  [0dad84c5] ArgTools v1.1.2
  [56f22d72] Artifacts v1.11.0
  [2a0f44e3] Base64 v1.11.0
  [ade2ca70] Dates v1.11.0
  [8ba89e20] Distributed v1.11.0
  [f43a241f] Downloads v1.6.0
  [7b1f6079] FileWatching v1.11.0
  [9fa8497b] Future v1.11.0
  [b77e0a4c] InteractiveUtils v1.11.0
  [4af54fe1] LazyArtifacts v1.11.0
  [b27032c2] LibCURL v0.6.4
  [76f85450] LibGit2 v1.11.0
  [8f399da3] Libdl v1.11.0
  [37e2e46d] LinearAlgebra v1.11.0
  [56ddb016] Logging v1.11.0
  [d6f4376e] Markdown v1.11.0
  [a63ad114] Mmap v1.11.0
  [ca575930] NetworkOptions v1.2.0
  [44cfe95a] Pkg v1.11.0
  [de0858da] Printf v1.11.0
  [9a3f8284] Random v1.11.0
  [ea8e919c] SHA v0.7.0
  [9e88b42a] Serialization v1.11.0
  [1a1011a3] SharedArrays v1.11.0
  [6462fe0b] Sockets v1.11.0
  [2f01184e] SparseArrays v1.11.0
  [4607b0f0] SuiteSparse
  [fa267f1f] TOML v1.0.3
  [a4e569a6] Tar v1.10.0
  [cf7118a7] UUIDs v1.11.0
  [4ec0a83e] Unicode v1.11.0
  [e66e0078] CompilerSupportLibraries_jll v1.1.1+0
  [deac9b47] LibCURL_jll v8.6.0+0
  [e37daf67] LibGit2_jll v1.7.2+0
  [29816b5a] LibSSH2_jll v1.11.0+1
  [c8ffd9c3] MbedTLS_jll v2.28.6+0
  [14a3606d] MozillaCACerts_jll v2023.12.12
  [4536629a] OpenBLAS_jll v0.3.27+1
  [05823500] OpenLibm_jll v0.8.1+4
  [bea87d4a] SuiteSparse_jll v7.7.0+0
  [83775a58] Zlib_jll v1.2.13+1
  [8e850b90] libblastrampoline_jll v5.11.0+0
  [8e850ede] nghttp2_jll v1.59.0+0
  [3f19e933] p7zip_jll v17.4.0+2
Info Packages marked with ⌃ have new versions available and may be upgradable.

• Output of versioninfo()

Julia Version 1.11.4
Commit 8561cc3d68 (2025-03-10 11:36 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 16 × 11th Gen Intel(R) Core(TM) i7-11850H @ 2.50GHz
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, tigerlake)
Threads: 8 default, 0 interactive, 4 GC (on 16 virtual cores)
Environment:
  JULIA_NUM_THREADS = 8
  JULIA_EDITOR = code

Additional context

[ChrisRackauckas]

This is arguably more of a bug in ReverseDiff.jl. The problem is that NonlinearSolution <: AbstractNoTimeSolution <: AbstractArray. ReverseDiff.jl has a behavior such that for any type that is a subtype of AbstractArray, it just returns an Array from the operation that builds it instead of the other type. So https://github.com/SciML/OrdinaryDiffEq.jl/blob/v6.95.1/lib/OrdinaryDiffEqNonlinearSolve/src/initialize_dae.jl#L434 this solve call returns an Array instead of a NonlinearSolution when you run a ReverseDiff. This means that https://github.com/SciML/OrdinaryDiffEq.jl/blob/v6.95.1/lib/OrdinaryDiffEqNonlinearSolve/src/initialize_dae.jl#L442 it doesn't have a retcode here. We could just make this do hasfield(nlsol, :retcode) && nlsol.retcode != ReturnCode.Success, but that feels a bit odd to turn off correctness checking here, simply because ReverseDiff returns the wrong type?

This is generally an issue with ReverseDiff, which is why we're doing more and more with Enzyme these days. Hopefully EnzymeAdjoint will just handle this fine.