DifferentialEquations.jl invalidates OrdinaryDiffEq.jl precompile caches for lu-factorization

[ChrisRackauckas]

using OrdinaryDiffEq, SnoopCompile

function lorenz(du,u,p,t)
 du[1] = 10.0(u[2]-u[1])
 du[2] = u[1]*(28.0-u[3]) - u[2]
 du[3] = u[1]*u[2] - (8/3)*u[3]
end

u0 = [1.0;0.0;0.0]
tspan = (0.0,100.0)
prob = ODEProblem(lorenz,u0,tspan)
alg = Rodas5()
tinf = @snoopi_deep solve(prob,alg)

InferenceTimingNode: 1.142492/2.968307 on Core.Compiler.Timings.ROOT() with 9 direct children

# New session

using DifferentialEquations, SnoopCompile

function lorenz(du,u,p,t)
 du[1] = 10.0(u[2]-u[1])
 du[2] = u[1]*(28.0-u[3]) - u[2]
 du[3] = u[1]*u[2] - (8/3)*u[3]
end

u0 = [1.0;0.0;0.0]
tspan = (0.0,100.0)
prob = ODEProblem(lorenz,u0,tspan)
alg = Rodas5()
tinf = @snoopi_deep solve(prob,alg)

InferenceTimingNode: 1.535779/13.754596 on Core.Compiler.Timings.ROOT() with 7 direct children

And better inferred version:

using DifferentialEquations, SnoopCompile

function lorenz(du,u,p,t)
 du[1] = 10.0(u[2]-u[1])
 du[2] = u[1]*(28.0-u[3]) - u[2]
 du[3] = u[1]*u[2] - (8/3)*u[3]
end

u0 = [1.0;0.0;0.0]
tspan = (0.0,100.0)
prob = ODEProblem(lorenz,u0,tspan)
alg = Rodas5(chunk_size = Val{3}(), linsolve = DiffEqBase.LUFactorize())
tinf = @snoopi_deep solve(prob,alg)

[ChrisRackauckas]

What's up?

[chriselrod]

Compiling Julia to use max_methods=1:

julia> using DifferentialEquations, SnoopCompile

julia> function lorenz(du,u,p,t)
        du[1] = 10.0(u[2]-u[1])
        du[2] = u[1]*(28.0-u[3]) - u[2]
        du[3] = u[1]*u[2] - (8/3)*u[3]
       end
lorenz (generic function with 1 method)

julia> u0 = [1.0;0.0;0.0]
3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> tspan = (0.0,100.0)
(0.0, 100.0)

julia> prob = ODEProblem(lorenz,u0,tspan)
ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
timespan: (0.0, 100.0)
u0: 3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> alg = Rodas5()
Rodas5{0, true, DefaultLinSolve, Val{:forward}}(DefaultLinSolve(nothing, nothing, nothing))

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 1.474002/3.069819 on Core.Compiler.Timings.ROOT() with 9 direct children

Julia with max_methods=3, the default:

julia> using DifferentialEquations, SnoopCompile

julia> function lorenz(du,u,p,t)
        du[1] = 10.0(u[2]-u[1])
        du[2] = u[1]*(28.0-u[3]) - u[2]
        du[3] = u[1]*u[2] - (8/3)*u[3]
       end
lorenz (generic function with 1 method)

julia> u0 = [1.0;0.0;0.0]
3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> tspan = (0.0,100.0)
(0.0, 100.0)

julia> prob = ODEProblem(lorenz,u0,tspan)
ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
timespan: (0.0, 100.0)
u0: 3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> alg = Rodas5()
Rodas5{0, true, DefaultLinSolve, Val{:forward}}(DefaultLinSolve(nothing, nothing, nothing))

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 4.389920/25.203741 on Core.Compiler.Timings.ROOT() with 7 direct children

using DiffEqFlux seems to still cause a regression compared to using DifferentialEquations:

julia> using DifferentialEquations, SnoopCompile, DiffEqFlux

julia> function lorenz(du,u,p,t)
        du[1] = 10.0(u[2]-u[1])
        du[2] = u[1]*(28.0-u[3]) - u[2]
        du[3] = u[1]*u[2] - (8/3)*u[3]
       end
lorenz (generic function with 1 method)

julia> u0 = [1.0;0.0;0.0]
3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> tspan = (0.0,100.0)
(0.0, 100.0)

julia> prob = ODEProblem(lorenz,u0,tspan)
ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
timespan: (0.0, 100.0)
u0: 3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> alg = Rodas5()
Rodas5{0, true, DefaultLinSolve, Val{:forward}}(DefaultLinSolve(nothing, nothing, nothing))

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 1.507790/3.427428 on Core.Compiler.Timings.ROOT() with 9 direct children

But this is still substantially better than max_methods=3.

[chriselrod]

It may lead to a small performance regression, so where this comes from is something we should look into within the packages.
Julia 1.6, max_methods = 1:

julia> using DifferentialEquations, SnoopCompile

julia> function lorenz(du,u,p,t)
        du[1] = 10.0(u[2]-u[1])
        du[2] = u[1]*(28.0-u[3]) - u[2]
        du[3] = u[1]*u[2] - (8/3)*u[3]
       end
lorenz (generic function with 1 method)

julia> u0 = [1.0;0.0;0.0]
3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> tspan = (0.0,100.0)
(0.0, 100.0)

julia> prob = ODEProblem(lorenz,u0,tspan)
ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
timespan: (0.0, 100.0)
u0: 3-element Vector{Float64}:
 1.0
 0.0
 0.0

julia> alg = Rodas5()
Rodas5{0, true, DefaultLinSolve, Val{:forward}}(DefaultLinSolve(nothing, nothing, nothing))

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 1.540056/3.249808 on Core.Compiler.Timings.ROOT() with 9 direct children

julia> using BenchmarkTools; @benchmark solve($prob,$alg)
BenchmarkTools.Trial: 1518 samples with 1 evaluation.
 Range (min … max):  3.153 ms …   9.154 ms  ┊ GC (min … max): 0.00% … 62.66%
 Time  (median):     3.222 ms               ┊ GC (median):    0.00%
 Time  (mean ± σ):   3.290 ms ± 577.525 μs  ┊ GC (mean ± σ):  1.75% ±  6.29%

  █
  █▅▄▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂ ▂
  3.15 ms         Histogram: frequency by time        8.67 ms <

 Memory estimate: 954.20 KiB, allocs estimate: 10579.

Julia 1.6, max_methods=3:

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 4.364899/22.365325 on Core.Compiler.Timings.ROOT() with 8 direct children

julia> using BenchmarkTools; @benchmark solve($prob,$alg)
BenchmarkTools.Trial: 1547 samples with 1 evaluation.
 Range (min … max):  3.106 ms …   9.695 ms  ┊ GC (min … max): 0.00% … 63.91%
 Time  (median):     3.149 ms               ┊ GC (median):    0.00%
 Time  (mean ± σ):   3.229 ms ± 594.692 μs  ┊ GC (mean ± σ):  1.68% ±  6.02%

      ▅▆▆██▃▁▁▁
  ▂▃▆▇██████████▆▅▅▄▃▃▃▂▂▂▂▂▂▂▁▂▂▁▁▁▂▂▂▂▃▄▅▅▄▅▄▃▃▃▃▃▂▂▂▂▂▂▁▂▂ ▃
  3.11 ms         Histogram: frequency by time        3.38 ms <

 Memory estimate: 903.67 KiB, allocs estimate: 8962.

Julia 1.7, max_methods=1:

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 1.579549/3.173668 on Core.Compiler.Timings.ROOT() with 8 direct children

julia> using BenchmarkTools; @benchmark solve($prob,$alg)
BenchmarkTools.Trial: 1569 samples with 1 evaluation.
 Range (min … max):  3.087 ms …  10.379 ms  ┊ GC (min … max): 0.00% … 68.20%
 Time  (median):     3.117 ms               ┊ GC (median):    0.00%
 Time  (mean ± σ):   3.185 ms ± 611.371 μs  ┊ GC (mean ± σ):  1.71% ±  6.06%

    ██▄
  ▃█████▆▅▄▄▃▃▂▂▂▂▃▃▃▃▃▂▂▂▂▁▁▁▂▁▁▁▁▁▁▁▁▁▁▁▁▂▁▁▁▁▁▁▂▁▁▁▁▁▁▁▁▁▂ ▃
  3.09 ms         Histogram: frequency by time        3.54 ms <

 Memory estimate: 717.47 KiB, allocs estimate: 10572.

Julia 1.7, max_methods=3:

julia> tinf = @snoopi_deep solve(prob,alg)
InferenceTimingNode: 4.394989/21.127358 on Core.Compiler.Timings.ROOT() with 6 direct children

julia> using BenchmarkTools; @benchmark solve($prob,$alg)
BenchmarkTools.Trial: 1618 samples with 1 evaluation.
 Range (min … max):  2.985 ms …  11.468 ms  ┊ GC (min … max): 0.00% … 72.16%
 Time  (median):     3.022 ms               ┊ GC (median):    0.00%
 Time  (mean ± σ):   3.088 ms ± 662.641 μs  ┊ GC (mean ± σ):  1.74% ±  5.86%

        ▅█▇▇▅▅▄▁
  ▂▂▃▄▆█████████▇▆▆▅▅▄▃▄▃▃▃▃▃▃▂▂▂▂▂▂▂▂▂▂▂▂▂▃▃▃▃▃▂▃▃▁▂▂▂▂▂▂▂▂▂ ▃
  2.98 ms         Histogram: frequency by time        3.19 ms <

 Memory estimate: 666.94 KiB, allocs estimate: 8955.

[ChrisRackauckas]

The AD tooling gets a weird type thing right now that I saw in SciML/OrdinaryDiffEq.jl#1533 . I would re-benchmark after fixing that kind of stuff. It's either that or the instability of the factorization code.

[ChrisRackauckas]

SciML/OrdinaryDiffEq.jl#1535 + JuliaDiff/SparseDiffTools.jl#167 makes it so that there are only 2 itrigs left:

julia> itrigs = inference_triggers(tinf)
2-element Vector{InferenceTrigger}:
 Inference triggered to call setindex!(::Vector{Tuple{Float64, Float64, Float64}}, ::Tuple{Bool, Bool, Bool}, ::Int64) from generate_chunked_partials (C:\Users\accou\.julia\dev\SparseDiffTools\src\differentiation\compute_jacobian_ad.jl:75) with specialization SparseDiffTools.generate_chunked_partials(::Vector{Float64}, ::UnitRange{Int64}, ::Val{3})
 Inference triggered to call unsafe_copyto!(::Matrix{Float64}, ::Int64, ::Matrix{Float64}, ::Int64, ::Int64) from _copyto_impl! (.\array.jl:331) inlined into OrdinaryDiffEq.calc_W!(::Matrix{Float64}, ::OrdinaryDiffEq.ODEIntegrator{Rodas5{3, true, DefaultLinSolve, Val{:forward}}, true, Vector{Float64}, Nothing, Float64, SciMLBase.NullParameters, Float64, Float64, Float64, Float64, Vector{Vector{Float64}}, ODESolution{Float64, 2, Vector{Vector{Float64}}, Nothing, Nothing, Vector{Float64}, Vector{Vector{Vector{Float64}}}, ODEProblem{Vector{Float64}, Tuple{Float64, Float64}, true, SciMLBase.NullParameters, ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}}, SciMLBase.StandardODEProblem}, Rodas5{3, true, DefaultLinSolve, Val{:forward}}, OrdinaryDiffEq.InterpolationData{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Vector{Float64}}, Vector{Float64}, Vector{Vector{Vector{Float64}}}, OrdinaryDiffEq.Rosenbrock5Cache{Vector{Float64}, Vector{Float64}, Vector{Float64}, Matrix{Float64}, Matrix{Float64}, OrdinaryDiffEq.Rodas5Tableau{Float64, Float64}, SciMLBase.TimeGradientWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Float64}, SciMLBase.NullParameters}, SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, DefaultLinSolve, SparseDiffTools.ForwardColorJacCache{Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, Float64}, Float64, 3}}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, Float64}, Float64, 3}}, Vector{Float64}, Vector{Vector{Tuple{Float64, Float64, Float64}}}, UnitRange{Int64}, Nothing}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.TimeGradientWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Float64}, SciMLBase.NullParameters}, Float64}, Float64, 1}}}}, DiffEqBase.DEStats}, ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, OrdinaryDiffEq.Rosenbrock5Cache{Vector{Float64}, Vector{Float64}, Vector{Float64}, Matrix{Float64}, Matrix{Float64}, OrdinaryDiffEq.Rodas5Tableau{Float64, Float64}, SciMLBase.TimeGradientWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Float64}, SciMLBase.NullParameters}, SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, DefaultLinSolve, SparseDiffTools.ForwardColorJacCache{Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, Float64}, Float64, 3}}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, Float64}, Float64, 3}}, Vector{Float64}, Vector{Vector{Tuple{Float64, Float64, Float64}}}, UnitRange{Int64}, Nothing}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.TimeGradientWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Float64}, SciMLBase.NullParameters}, Float64}, Float64, 1}}}, OrdinaryDiffEq.DEOptions{Float64, Float64, Float64, Float64, PIController{Rational{Int64}}, typeof(DiffEqBase.ODE_DEFAULT_NORM), typeof(LinearAlgebra.opnorm), Nothing, CallbackSet{Tuple{}, Tuple{}}, typeof(DiffEqBase.ODE_DEFAULT_ISOUTOFDOMAIN), typeof(DiffEqBase.ODE_DEFAULT_PROG_MESSAGE), typeof(DiffEqBase.ODE_DEFAULT_UNSTABLE_CHECK), DataStructures.BinaryHeap{Float64, DataStructures.FasterForward}, DataStructures.BinaryHeap{Float64, DataStructures.FasterForward}, Nothing, Nothing, Int64, Tuple{}, Tuple{}, Tuple{}}, Vector{Float64}, Float64, Nothing, OrdinaryDiffEq.DefaultInit}, ::Nothing, ::OrdinaryDiffEq.Rosenbrock5Cache{Vector{Float64}, Vector{Float64}, Vector{Float64}, Matrix{Float64}, Matrix{Float64}, OrdinaryDiffEq.Rodas5Tableau{Float64, Float64}, SciMLBase.TimeGradientWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Float64}, SciMLBase.NullParameters}, SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, DefaultLinSolve, SparseDiffTools.ForwardColorJacCache{Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, Float64}, Float64, 3}}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Float64, SciMLBase.NullParameters}, Float64}, Float64, 3}}, Vector{Float64}, Vector{Vector{Tuple{Float64, Float64, Float64}}}, UnitRange{Int64}, Nothing}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{SciMLBase.TimeGradientWrapper{ODEFunction{true, typeof(lorenz), LinearAlgebra.UniformScaling{Bool}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing}, Vector{Float64}, SciMLBase.NullParameters}, Float64}, Float64, 1}}}, ::Float64, ::Bool, ::Bool) (C:\Users\accou\.julia\dev\OrdinaryDiffEq\src\derivative_utils.jl:503)

[ChrisRackauckas]

With some more fixes, we got to:

julia> itrigs = inference_triggers(tinf)
2-element Vector{InferenceTrigger}:
 Inference triggered to call setindex!(::Vector{Tuple{Float64, Float64, Float64}}, ::Tuple{Bool, Bool, Bool}, ::Int64) from generate_chunked_partials (C:\Users\accou\.julia\dev\SparseDiffTools\src\differentiation\compute_jacobian_ad.jl:75) with specialization SparseDiffTools.generate_chunked_partials(::Vector{Float64}, ::UnitRange{Int64}, ::Val{3})
 Inference triggered to call OrdinaryDiffEq.jacobian2W!(::Matrix{Float64}, ::LinearAlgebra.UniformScaling{Bool}, ::Float64, ::Matrix{Float64}, ::Bool) called from toplevel

The first inference trigger should be fixed by JuliaDiff/SparseDiffTools.jl#168, which will need some help. The second looks like SnoopCompile is bugging. @timholy what could cause that? It's definitely not from the top level 😅

[ChrisRackauckas]

The first is definitely fixed by JuliaDiff/SparseDiffTools.jl#169

julia> itrigs = inference_triggers(tinf)
1-element Vector{InferenceTrigger}:
 Inference triggered to call OrdinaryDiffEq.jacobian2W!(::Matrix{Float64}, ::LinearAlgebra.UniformScaling{Bool}, ::Float64, ::Matrix{Float64}, ::Bool) called from toplevel

One inference trigger left, but it's a buggy one 😅 .

[ChrisRackauckas]

Solved by removing one line from Symbolics.jl JuliaSymbolics/Symbolics.jl#471