Describe the bug π
setp_oop / setsym_oop setters whose symbols span different MTKParameters portions
(e.g. Initial(v) in Initials + a plain parameter in Tunable) fail with
MethodError: no method matching Float64(::ForwardDiff.Dual{...})
as soon as the values are non-Float64 (ForwardDiff Duals being the common case).
Homogeneous single-portion setters promote correctly. This breaks
ForwardDiff.gradient over any solve-based loss that tunes an initial condition together
with a parameter β e.g. parameter estimation with mixed u0/p search spaces
Expected behavior
no error.
Minimal Reproducible Example π
using ModelingToolkit
using ModelingToolkit: D_nounits as D, t_nounits as t
using SymbolicIndexingInterface
using ForwardDiff: Dual
@parameters m=1.5 d=9.0
@variables s(t)=1.0 v(t)=1.0
eqs = [D(s) ~ v, m * D(v) ~ 1 - d * v]
sys = mtkcompile(System(eqs, t; name = :model))
prob = ODEProblem(sys, [], (0.0, 1.0))
xd = [Dual(1.1, 1.0), Dual(4.3, 1.0)]
setp_oop(prob, [Initial(sys.v)])(prob, xd[1:1]) # works, initials promoted to Dual
setp_oop(prob, [sys.d])(prob, xd[2:2]) # works, tunable promoted to Dual
setp_oop(prob, [Initial(sys.v), sys.d])(prob, xd) # MethodError
Error & Stacktrace β οΈ
ERROR: MethodError: no method matching Float64(::ForwardDiff.Dual{Nothing, Float64, 1})
The type `Float64` exists, but no method is defined for this combination of argument types when trying to construct it.
Closest candidates are:
(::Type{T})(::Real, ::RoundingMode) where T<:AbstractFloat
@ Base rounding.jl:265
(::Type{T})(::T) where T<:Number
@ Core boot.jl:965
Float64(::Int16)
@ Base float.jl:245
...
Stacktrace:
[1] convert(::Type{Float64}, x::ForwardDiff.Dual{Nothing, Float64, 1})
@ Base ./number.jl:7
[2] setindex!
@ ./array.jl:985 [inlined]
[3] set_parameter!(p::ModelingToolkitBase.MTKParameters{β¦}, val::ForwardDiff.Dual{β¦}, pidx::ModelingToolkitBase.ParameterIndex{β¦})
@ ModelingToolkitBase ~/.julia/packages/ModelingToolkitBase/tQDig/src/systems/parameter_buffer.jl:505
[4] setindex!(p::ModelingToolkitBase.MTKParameters{β¦}, val::ForwardDiff.Dual{β¦}, pind::ModelingToolkitBase.ParameterIndex{β¦})
@ ModelingToolkitBase ~/.julia/packages/ModelingToolkitBase/tQDig/src/systems/parameter_buffer.jl:1135
[5] macro expansion
@ ~/.julia/packages/ModelingToolkitBase/tQDig/src/systems/parameter_buffer.jl:919 [inlined]
[6] _remake_buffer
@ ~/.julia/packages/ModelingToolkitBase/tQDig/src/systems/parameter_buffer.jl:784 [inlined]
[7] remake_buffer
@ ~/.julia/packages/ModelingToolkitBase/tQDig/src/systems/parameter_buffer.jl:659 [inlined]
[8] (::SymbolicIndexingInterface.OOPSetter{β¦})(valp::SciMLBase.ODEProblem{β¦}, val::Vector{β¦})
@ SymbolicIndexingInterface ~/.julia/packages/SymbolicIndexingInterface/v22YL/src/value_provider_interface.jl:272
[9] top-level scope
@ ~/work/dev/upstream_bugs/remake_buffer_mixed_promotion/issue_draft.md:41
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 `~/work/dev/upstream_bugs/remake_buffer_mixed_promotion/Project.toml`
[f6369f11] ForwardDiff v1.4.3
[961ee093] ModelingToolkit v11.38.0
[2efcf032] SymbolicIndexingInterface v0.3.51
- Output of
using Pkg; Pkg.status(; mode = PKGMODE_MANIFEST)
Status `~/work/dev/upstream_bugs/remake_buffer_mixed_promotion/Manifest.toml`
[47edcb42] ADTypes v1.22.3
[14f7f29c] AMD v0.5.3
[6e696c72] AbstractPlutoDingetjes v1.4.0
[1520ce14] AbstractTrees v0.4.5
[7d9f7c33] Accessors v0.1.45
[79e6a3ab] Adapt v4.7.0
[ec485272] ArnoldiMethod v0.4.0
[4fba245c] ArrayInterface v7.28.1
[4c555306] ArrayLayouts v1.12.2
[aae01518] BandedMatrices v1.11.0
[e2ed5e7c] Bijections v0.2.2
[b2a6c25c] BinaryHeaps v1.0.4
[caf10ac8] BipartiteGraphs v0.1.10
[8e7c35d0] BlockArrays v1.10.0
[70df07ce] BracketingNonlinearSolve v1.12.4
β
[861a8166] Combinatorics v1.0.2
[38540f10] CommonSolve v0.2.12
[bbf7d656] CommonSubexpressions v0.3.1
[f70d9fcc] CommonWorldInvalidations v1.1.2
[34da2185] Compat v4.18.1
[b152e2b5] CompositeTypes v0.1.4
[a33af91c] CompositionsBase v0.1.2
[2569d6c7] ConcreteStructs v0.2.7
[187b0558] ConstructionBase v1.6.0
[a8cc5b0e] Crayons v4.2.0
[9a962f9c] DataAPI v1.16.0
[864edb3b] DataStructures v0.19.6
[e2d170a0] DataValueInterfaces v1.0.0
[2b5f629d] DiffEqBase v7.10.0
[459566f4] DiffEqCallbacks v4.19.0
[163ba53b] DiffResults v1.1.0
[b552c78f] DiffRules v1.16.0
[a0c0ee7d] DifferentiationInterface v0.7.20
[ffbed154] DocStringExtensions v0.9.5
[5b8099bc] DomainSets v0.8.1
[7c1d4256] DynamicPolynomials v0.6.6
[4e289a0a] EnumX v1.0.7
[f151be2c] EnzymeCore v0.8.21
[e2ba6199] ExprTools v0.1.11
[55351af7] ExproniconLite v0.10.14
[7034ab61] FastBroadcast v1.3.5
[9aa1b823] FastClosures v0.3.2
[a4df4552] FastPower v1.4.0
[1a297f60] FillArrays v1.17.0
[64ca27bc] FindFirstFunctions v3.2.1
[6a86dc24] FiniteDiff v2.32.0
[f6369f11] ForwardDiff v1.4.3
[a85aefff] FunctionMaps v0.1.2
[069b7b12] FunctionWrappers v1.1.3
[77dc65aa] FunctionWrappersWrappers v1.12.0
[46192b85] GPUArraysCore v0.2.0
[86223c79] Graphs v1.14.0
[3263718b] ImplicitDiscreteSolve v2.1.5
[d25df0c9] Inflate v0.1.5
[18e54dd8] IntegerMathUtils v0.1.4
[8197267c] IntervalSets v0.7.14
[3587e190] InverseFunctions v0.1.17
[92d709cd] IrrationalConstants v0.2.6
[82899510] IteratorInterfaceExtensions v1.0.0
[692b3bcd] JLLWrappers v1.8.0
[ae98c720] Jieko v0.2.1
[ccbc3e58] JumpProcesses v9.29.2
[ba0b0d4f] Krylov v0.10.8
[b964fa9f] LaTeXStrings v1.4.0
[87fe0de2] LineSearch v0.1.12
[7ed4a6bd] LinearSolve v5.4.0
[2ab3a3ac] LogExpFunctions v1.0.1
[e6f89c97] LoggingExtras v1.2.0
[1914dd2f] MacroTools v0.5.16
[bb5d69b7] MaybeInplace v0.1.7
[961ee093] ModelingToolkit v11.38.0
[7771a370] ModelingToolkitBase v1.59.0
[6bb917b9] ModelingToolkitTearing v1.20.4
[2e0e35c7] Moshi v0.3.12
[46d2c3a1] MuladdMacro v0.2.7
[102ac46a] MultivariatePolynomials v0.5.19
[d8a4904e] MutableArithmetics v1.8.0
[77ba4419] NaNMath v1.1.4
[be0214bd] NonlinearSolveBase v2.39.1
[5959db7a] NonlinearSolveFirstOrder v2.2.3
[6fe1bfb0] OffsetArrays v1.17.0
β
[bac558e1] OrderedCollections v1.8.2
[bbf590c4] OrdinaryDiffEqCore v4.12.0
[e409e4f3] PoissonRandom v0.4.13
[d236fae5] PreallocationTools v1.4.0
[aea7be01] PrecompileTools v1.3.4
[21216c6a] Preferences v1.5.2
[08abe8d2] PrettyTables v3.4.4
[27ebfcd6] Primes v0.5.7
[0c0d3e7f] PureKLU v1.2.0
[988b38a3] ReadOnlyArrays v0.2.0
[795d4caa] ReadOnlyDicts v1.0.1
[3cdcf5f2] RecipesBase v1.3.4
[731186ca] RecursiveArrayTools v4.3.5
[189a3867] Reexport v1.2.2
[9fe22ead] RespecializeParams v1.2.0
[7e49a35a] RuntimeGeneratedFunctions v0.5.24
[9dfe8606] SCCNonlinearSolve v1.14.0
[0bca4576] SciMLBase v3.41.0
[19f34311] SciMLJacobianOperators v0.1.16
[a6db7da4] SciMLLogging v2.0.3
[c0aeaf25] SciMLOperators v1.25.2
[431bcebd] SciMLPublic v1.2.4
[53ae85a6] SciMLStructures v1.10.4
[efcf1570] Setfield v1.1.2
[727e6d20] SimpleNonlinearSolve v2.13.1
[699a6c99] SimpleTraits v0.9.6
[a57abbd0] SparseColumnPivotedQR v2.1.5
[276daf66] SpecialFunctions v2.8.1
[0c0c59c1] StarAlgebras v0.3.0
[64909d44] StateSelection v1.11.0
[90137ffa] StaticArrays v1.9.18
[1e83bf80] StaticArraysCore v1.4.4
[10745b16] Statistics v1.11.1
[892a3eda] StringManipulation v0.4.7
[2efcf032] SymbolicIndexingInterface v0.3.51
[19f23fe9] SymbolicLimits v1.1.4
[d1185830] SymbolicUtils v4.44.0
[0c5d862f] Symbolics v7.35.0
[3783bdb8] TableTraits v1.0.1
[bd369af6] Tables v1.13.0
[ed4db957] TaskLocalValues v0.1.3
[8ea1fca8] TermInterface v2.0.0
[a759f4b9] TimerOutputs v1.1.0
[781d530d] TruncatedStacktraces v1.4.0
[3a884ed6] UnPack v1.0.2
[d30d5f5c] WeakCacheSets v0.1.0
[1d5cc7b8] IntelOpenMP_jll v2025.2.0+0
[856f044c] MKL_jll v2025.2.0+0
[efe28fd5] OpenSpecFun_jll v0.5.6+0
[1317d2d5] oneTBB_jll v2022.3.0+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 v1.7.0
[7b1f6079] FileWatching v1.11.0
[9fa8497b] Future v1.11.0
[b77e0a4c] InteractiveUtils v1.11.0
[ac6e5ff7] JuliaSyntaxHighlighting v1.12.0
[4af54fe1] LazyArtifacts v1.11.0
[b27032c2] LibCURL v0.6.4
[76f85450] LibGit2 v1.11.0
[8f399da3] Libdl v1.11.0
[37e2e46d] LinearAlgebra v1.12.0
[56ddb016] Logging v1.11.0
[d6f4376e] Markdown v1.11.0
[ca575930] NetworkOptions v1.3.0
[44cfe95a] Pkg v1.12.1
[de0858da] Printf v1.11.0
[3fa0cd96] REPL v1.11.0
[9a3f8284] Random v1.11.0
[ea8e919c] SHA v0.7.0
[9e88b42a] Serialization v1.11.0
[6462fe0b] Sockets v1.11.0
[2f01184e] SparseArrays v1.12.0
[f489334b] StyledStrings v1.11.0
[fa267f1f] TOML v1.0.3
[a4e569a6] Tar v1.10.0
[8dfed614] Test v1.11.0
[cf7118a7] UUIDs v1.11.0
[4ec0a83e] Unicode v1.11.0
[e66e0078] CompilerSupportLibraries_jll v1.3.0+1
[deac9b47] LibCURL_jll v8.15.0+0
[e37daf67] LibGit2_jll v1.9.0+0
[29816b5a] LibSSH2_jll v1.11.3+1
[14a3606d] MozillaCACerts_jll v2025.11.4
[4536629a] OpenBLAS_jll v0.3.29+0
[05823500] OpenLibm_jll v0.8.7+0
[458c3c95] OpenSSL_jll v3.5.4+0
[bea87d4a] SuiteSparse_jll v7.8.3+2
[83775a58] Zlib_jll v1.3.1+2
[8e850b90] libblastrampoline_jll v5.15.0+0
[8e850ede] nghttp2_jll v1.64.0+1
[3f19e933] p7zip_jll v17.7.0+0
Info Packages marked with β
have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
Julia Version 1.12.6
Commit 15346901f00 (2026-04-09 19:20 UTC)
Build Info:
Official https://julialang.org release
Platform Info:
OS: Linux (x86_64-linux-gnu)
CPU: 32 Γ Intel(R) Core(TM) i9-14900K
WORD_SIZE: 64
LLVM: libLLVM-18.1.7 (ORCJIT, alderlake)
GC: Built with stock GC
Threads: 32 default, 1 interactive, 32 GC (on 32 virtual cores)
Environment:
JULIA_EDITOR = code
JULIA_VSCODE_REPL = 1
Additional context
Root cause
The @generated fast path of _remake_buffer
(ModelingToolkitBase v1.59.0, src/systems/parameter_buffer.jl:784, comment
"For type-inference when using SII.setp_oop") decides per-portion buffer promotion
from the type parameters alone:
# parameter_buffer.jl:826-834
idxT <: ParameterIndex{SciMLStructures.Tunable} || continue
idxT <: ParameterIndex{SciMLStructures.Initials} || continue
with idxT = eltype(idxs). A mixed index vector (built by SII as
parameter_index.((indp,), sym), parameter_indexing.jl:834-837; the
BatchedInterface route in batched_interface.jl:351 splices an equivalent vector)
has the abstract typejoin eltype ParameterIndex{P, Int64} where P, which is a subtype
of none of the per-portion checks β so no buffer is promoted, and the generated
setindex! loop (parameter_buffer.jl:919) hits convert(Float64, ::Dual).
The existing fallback to the dynamic path only inspects the values:
# parameter_buffer.jl:791
if vals <: AbstractArray && !isconcretetype(eltype(vals))
eltype(vals) = Dual{...} is concrete, so the fallback doesn't trigger. The dynamic
__remake_buffer (Any buffers + narrowing) would have handled this case correctly.
Describe the bug π
setp_oop/setsym_oopsetters whose symbols span differentMTKParametersportions(e.g.
Initial(v)inInitials+ a plain parameter inTunable) fail withas soon as the values are non-
Float64(ForwardDiff Duals being the common case).Homogeneous single-portion setters promote correctly. This breaks
ForwardDiff.gradientover any solve-based loss that tunes an initial condition togetherwith a parameter β e.g. parameter estimation with mixed u0/p search spaces
Expected behavior
no error.
Minimal Reproducible Example π
Error & Stacktraceβ οΈ
Environment (please complete the following information):
using Pkg; Pkg.status()using Pkg; Pkg.status(; mode = PKGMODE_MANIFEST)versioninfo()Additional context
Root cause
The
@generatedfast path of_remake_buffer(ModelingToolkitBase v1.59.0,
src/systems/parameter_buffer.jl:784, comment"For type-inference when using
SII.setp_oop") decides per-portion buffer promotionfrom the type parameters alone:
with
idxT = eltype(idxs). A mixed index vector (built by SII asparameter_index.((indp,), sym),parameter_indexing.jl:834-837; theBatchedInterfaceroute inbatched_interface.jl:351splices an equivalent vector)has the abstract typejoin eltype
ParameterIndex{P, Int64} where P, which is a subtypeof none of the per-portion checks β so no buffer is promoted, and the generated
setindex!loop (parameter_buffer.jl:919) hitsconvert(Float64, ::Dual).The existing fallback to the dynamic path only inspects the values:
eltype(vals) = Dual{...}is concrete, so the fallback doesn't trigger. The dynamic__remake_buffer(Any buffers + narrowing) would have handled this case correctly.