search

baggepinnen / ControlSystemIdentification.jl

System Identification toolbox, compatible with ControlSystems.jl
https://baggepinnen.github.io/ControlSystemIdentification.jl/dev
MIT License
132 stars 13 forks source link

Errors in using input_nonlinearity with newpem #156

Closed anchal-physics closed 4 weeks ago

anchal-physics commented 1 month ago

Hi

I have simple SISO system which is slighlty non-linear and I wanted to see if I can fit a Hammerstein-Wiener form non-linear system to it using newpem method. Unfortunately, it is giving me intractable errors. I think the error originates in functions of other libraries but maybe you can shed some light on how to properly define input_nonlinearity function?

I can't show the data here but it is a failry simple SISO and following runs without errors:

newpem(id_data, 2)
Output (Click to expand): Iter Function value Gradient norm 0 1.303796e-01 1.252132e+01 * time: 9.584426879882812e-5 (sys = ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, Matrix{Float64}, Matrix{Float64}, Matrix{Float64}} A = 0.952510834582067 -0.37171223660279223 0.0012212299487706242 0.9884628684643999 B = -0.03076361699634558 -0.0029576156945034823 C = 0.10795214583161879 -1.4966601166648088 D = 0.0 Sample Time: 0.001 (seconds) Discrete-time state-space model, x0 = [0.11360903072925647, 0.009137177133436751], res = * Status: success * Candidate solution Final objective value: 9.228208e-02 * Found with Algorithm: BFGS * Convergence measures |x - x'| = 1.56e-09 ≰ 0.0e+00 |x - x'|/|x'| = 1.65e-10 ≰ 0.0e+00 |f(x) - f(x')| = 0.00e+00 ≤ 1.0e-16 |f(x) - f(x')|/|f(x')| = 0.00e+00 ≤ 0.0e+00 |g(x)| = 5.81e-09 ≰ 1.0e-12 * Work counters Seconds run: 0 (vs limit 100) Iterations: 45 f(x) calls: 79 ∇f(x) calls: 46 , nlp = Float64[])

Then, if I try to give it a simple quadratic non-linearity, first I got this 

function inp_nl(u::AbstractArray{T}, p::AbstractArray{U}) where {T,U}
    u .= u .+ p[1] .* u .^ 2
end
newpem(id_data, 2; input_nonlinearity=inp_nl, nlp=Float64[0.00001])

Output (Click to expand):

MethodError: no method matching Float64(::ForwardDiff.Dual{ForwardDiff.Tag{... ┌ Error: Optimization failed, call `newpem(...; safe=true) to exit gracefully returning the initial estimate. └ @ ControlSystemIdentification /Users/gupta/.julia/packages/ControlSystemIdentification/La3Ei/src/pem.jl:267 { "name": "MethodError", "message": "MethodError: no method matching Float64(::ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}) Closest candidates are: (::Type{T})(::Real, !Matched::RoundingMode) where T<:AbstractFloat @ Base rounding.jl:207 (::Type{T})(::T) where T<:Number @ Core boot.jl:792 Float64(!Matched::IrrationalConstants.Logtwo) @ IrrationalConstants ~/.julia/packages/IrrationalConstants/vp5v4/src/macro.jl:112 ... ", "stack": "MethodError: no method matching Float64(::ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}) Closest candidates are: (::Type{T})(::Real, !Matched::RoundingMode) where T<:AbstractFloat @ Base rounding.jl:207 (::Type{T})(::T) where T<:Number @ Core boot.jl:792 Float64(!Matched::IrrationalConstants.Logtwo) @ IrrationalConstants ~/.julia/packages/IrrationalConstants/vp5v4/src/macro.jl:112 ... Stacktrace: [1] convert(::Type{Float64}, x::ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}) @ Base ./number.jl:7 [2] setindex!(A::Vector{Float64}, x::ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}, i1::Int64) @ Base ./array.jl:1021 [3] setindex! @ ~/.julia/juliaup/julia-1.10.4+0.aarch64.apple.darwin14/share/julia/stdlib/v1.10/LinearAlgebra/src/adjtrans.jl:330 [inlined] [4] _setindex! @ ./abstractarray.jl:1419 [inlined] [5] setindex! @ ./abstractarray.jl:1396 [inlined] [6] macro expansion @ ./broadcast.jl:1004 [inlined] [7] macro expansion @ ./simdloop.jl:77 [inlined] [8] copyto! @ ./broadcast.jl:1003 [inlined] [9] copyto! @ ./broadcast.jl:956 [inlined] [10] materialize! @ ./broadcast.jl:914 [inlined] [11] materialize! @ ./broadcast.jl:911 [inlined] [12] inp_nl(u::LinearAlgebra.Transpose{Float64, Vector{Float64}}, p::Vector{ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}}) @ Main ~/Git/ProjectTorreyPines/Controllers/Julia/density_control/jl_notebook_cell_df34fa98e69747e1a8f8a730347b8e2f_Y226sZmlsZQ==.jl:2 [13] (::ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64})(p::Vector{ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}}) @ ControlSystemIdentification ~/.julia/packages/ControlSystemIdentification/La3Ei/src/pem.jl:219 [14] vector_mode_dual_eval! @ ~/.julia/packages/ForwardDiff/PcZ48/src/apiutils.jl:24 [inlined] [15] vector_mode_gradient!(result::DiffResults.MutableDiffResult{1, Float64, Tuple{Vector{Float64}}}, f::ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, x::Vector{Float64}, cfg::ForwardDiff.GradientConfig{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12, Vector{ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}}}) @ ForwardDiff ~/.julia/packages/ForwardDiff/PcZ48/src/gradient.jl:96 [16] gradient! @ ~/.julia/packages/ForwardDiff/PcZ48/src/gradient.jl:37 [inlined] [17] gradient! @ ~/.julia/packages/ForwardDiff/PcZ48/src/gradient.jl:35 [inlined] [18] (::NLSolversBase.var\"#14#18\"{Float64, ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, ForwardDiff.GradientConfig{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12, Vector{ForwardDiff.Dual{ForwardDiff.Tag{ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64}, Float64}, Float64, 12}}}})(out::Vector{Float64}, x::Vector{Float64}) @ NLSolversBase ~/.julia/packages/NLSolversBase/kavn7/src/objective_types/oncedifferentiable.jl:70 [19] value_gradient!!(obj::NLSolversBase.OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, x::Vector{Float64}) @ NLSolversBase ~/.julia/packages/NLSolversBase/kavn7/src/interface.jl:82 [20] initial_state(method::Optim.BFGS{LineSearches.InitialStatic{Float64}, LineSearches.BackTracking{Float64, Int64}, Nothing, Nothing, Optim.Flat}, options::Optim.Options{Float64, Nothing}, d::NLSolversBase.OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, initial_x::Vector{Float64}) @ Optim ~/.julia/packages/Optim/ZhuZN/src/multivariate/solvers/first_order/bfgs.jl:94 [21] optimize @ ~/.julia/packages/Optim/ZhuZN/src/multivariate/optimize/optimize.jl:36 [inlined] [22] #optimize#91 @ ~/.julia/packages/Optim/ZhuZN/src/multivariate/optimize/interface.jl:143 [inlined] [23] optimize @ ~/.julia/packages/Optim/ZhuZN/src/multivariate/optimize/interface.jl:139 [inlined] [24] newpem(d::ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, nx::Int64; zeroD::Bool, focus::Symbol, h::Int64, stable::Bool, output_nonlinearity::Nothing, input_nonlinearity::typeof(inp_nl), nlp::Vector{Float64}, sys0::ControlSystemIdentification.N4SIDStateSpace{StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Matrix{Float64}, Matrix{Float64}, Matrix{Float64}, Vector{Float64}, LinearAlgebra.SVD{Float64, Float64, Matrix{Float64}, Vector{Float64}}, Float64}, metric::typeof(abs2), regularizer::ControlSystemIdentification.var\"#42#46\", optimizer::Optim.BFGS{LineSearches.InitialStatic{Float64}, LineSearches.BackTracking{Float64, Int64}, Nothing, Nothing, Optim.Flat}, store_trace::Bool, show_trace::Bool, show_every::Int64, iterations::Int64, allow_f_increases::Bool, time_limit::Int64, x_tol::Int64, f_abstol::Float64, g_tol::Float64, f_calls_limit::Int64, g_calls_limit::Int64, safe::Bool) @ ControlSystemIdentification ~/.julia/packages/ControlSystemIdentification/La3Ei/src/pem.jl:253 [25] top-level scope @ ~/Git/ProjectTorreyPines/Controllers/Julia/density_control/jl_notebook_cell_df34fa98e69747e1a8f8a730347b8e2f_Y226sZmlsZQ==.jl:4" }

I read in your documentation that optimizer = Optim.NelderMead() can give better results, so I tried it. It gives another kind of error which is weird. Somewhere inside when lsim is called, state vector is reduced in length by 1.

newpem(id_data, 2; input_nonlinearity=inp_nl, nlp=Float64[0.00001], optimizer = Optim.NelderMead())

Output (Click to expand):

x0 must have length 2: got length 1 ┌ Error: Optimization failed, call `newpem(...; safe=true) to exit gracefully returning the initial estimate. └ @ ControlSystemIdentification /Users/gupta/.julia/packages/ControlSystemIdentification/La3Ei/src/pem.jl:267 { "name": "ErrorException", "message": "x0 must have length 2: got length 1", "stack": "x0 must have length 2: got length 1 Stacktrace: [1] error(s::String) @ Base ./error.jl:35 [2] lsim(sys::StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, u::Matrix{Float64}, t::StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}; x0::Vector{Float64}, method::Symbol) @ ControlSystemsBase ~/.julia/packages/ControlSystemsBase/kvrre/src/timeresp.jl:182 [3] lsim @ ~/.julia/packages/ControlSystemsBase/kvrre/src/timeresp.jl:176 [inlined] [4] #lsim#238 @ ~/.julia/packages/ControlSystemsBase/kvrre/src/timeresp.jl:235 [inlined] [5] lsim @ ~/.julia/packages/ControlSystemsBase/kvrre/src/timeresp.jl:223 [inlined] [6] lsim(sys::StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, d::ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}; x0::Vector{Float64}) @ ControlSystemIdentification ~/.julia/packages/ControlSystemIdentification/La3Ei/src/ControlSystemIdentification.jl:238 [7] (::ControlSystemIdentification.var\"#predloss#43\"{Bool, Int64, Nothing, typeof(inp_nl), ControlSystemIdentification.PredictionStateSpace{ControlSystemsBase.Discrete{Float64}, StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, Matrix{Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Nothing}, typeof(abs2), ControlSystemIdentification.var\"#42#46\", ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, Int64, Matrix{Float64}, ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, Matrix{Float64}, Float64}, Bool, Matrix{Float64}, Int64, Int64, Int64})(p::Vector{Float64}) @ ControlSystemIdentification ~/.julia/packages/ControlSystemIdentification/La3Ei/src/pem.jl:223 [8] value!!(obj::NLSolversBase.NonDifferentiable{Float64, Vector{Float64}}, x::Vector{Float64}) @ NLSolversBase ~/.julia/packages/NLSolversBase/kavn7/src/interface.jl:9 [9] initial_state(method::Optim.NelderMead{Optim.AffineSimplexer, Optim.AdaptiveParameters}, options::Optim.Options{Float64, Nothing}, d::NLSolversBase.NonDifferentiable{Float64, Vector{Float64}}, initial_x::Vector{Float64}) @ Optim ~/.julia/packages/Optim/ZhuZN/src/multivariate/solvers/zeroth_order/nelder_mead.jl:171 [10] optimize @ ~/.julia/packages/Optim/ZhuZN/src/multivariate/optimize/optimize.jl:36 [inlined] [11] #optimize#91 @ ~/.julia/packages/Optim/ZhuZN/src/multivariate/optimize/interface.jl:143 [inlined] [12] optimize @ ~/.julia/packages/Optim/ZhuZN/src/multivariate/optimize/interface.jl:139 [inlined] [13] newpem(d::ControlSystemIdentification.InputOutputData{LinearAlgebra.Transpose{Float64, Vector{Float64}}, LinearAlgebra.Transpose{Float64, Vector{Float64}}, Float64}, nx::Int64; zeroD::Bool, focus::Symbol, h::Int64, stable::Bool, output_nonlinearity::Nothing, input_nonlinearity::typeof(inp_nl), nlp::Vector{Float64}, sys0::ControlSystemIdentification.N4SIDStateSpace{StateSpace{ControlSystemsBase.Discrete{Float64}, Float64}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, LinearAlgebra.Hermitian{Float64, Matrix{Float64}}, Matrix{Float64}, Matrix{Float64}, Matrix{Float64}, Vector{Float64}, LinearAlgebra.SVD{Float64, Float64, Matrix{Float64}, Vector{Float64}}, Float64}, metric::typeof(abs2), regularizer::ControlSystemIdentification.var\"#42#46\", optimizer::Optim.NelderMead{Optim.AffineSimplexer, Optim.AdaptiveParameters}, store_trace::Bool, show_trace::Bool, show_every::Int64, iterations::Int64, allow_f_increases::Bool, time_limit::Int64, x_tol::Int64, f_abstol::Float64, g_tol::Float64, f_calls_limit::Int64, g_calls_limit::Int64, safe::Bool) @ ControlSystemIdentification ~/.julia/packages/ControlSystemIdentification/La3Ei/src/pem.jl:253 [14] top-level scope @ ~/Git/ProjectTorreyPines/Controllers/Julia/density_control/jl_notebook_cell_df34fa98e69747e1a8f8a730347b8e2f_Y245sZmlsZQ==.jl:1" }

Please let me know if you know why this is happening?

baggepinnen commented 4 weeks ago

Hello :wave:

The first error looks like u have been of type Float64 and p of type Dual, causing an error while writing into u. That's an issue I need to fix in the package.

This PR should fix this issue

The second issue is harder for me to figure out without being able to reproduce the error.

baggepinnen commented 4 weeks ago

I think I found the second issue as well, #157 should solve that too