Issue with 'gulf' when use_nls= true

[farhadrclass]

Hi,

I ran the following code and encountered an error when passing the T (Type), n (size), and use_nls parameters to the gulf() function. Here's the code and the resulting output:

julia> nlp = OptimizationProblems.ADNLPProblems.gulf()
ADNLPModel - Model with automatic differentiation backend ADModelBackend{
  ForwardDiffADGradient,
  ForwardDiffADHvprod,
  EmptyADbackend,
  EmptyADbackend,
  EmptyADbackend,
  SparseADHessian,
  EmptyADbackend,
}
  Problem name: gulf
   All variables: ████████████████████ 3      All constraints: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
            free: ████████████████████ 3                 free: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
           lower: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                lower: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
           upper: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                upper: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
         low/upp: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0              low/upp: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
           fixed: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                fixed: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
          infeas: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0               infeas: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
            nnzh: (  0.00% sparsity)   6               linear: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
                                                    nonlinear: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
                                                         nnzj: (------% sparsity)

  Counters:
             obj: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                 grad: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                 cons: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
        cons_lin: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0             cons_nln: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                 jcon: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
           jgrad: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                  jac: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0              jac_lin: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
         jac_nln: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                jprod: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0            jprod_lin: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
       jprod_nln: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0               jtprod: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0           jtprod_lin: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
      jtprod_nln: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                 hess: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                hprod: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0
           jhess: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0               jhprod: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0


julia> nlp = OptimizationProblems.ADNLPProblems.gulf(T=T,n=10, use_nls=true)
ERROR: UndefRefError: access to undefined reference
Stacktrace:
  [1] getindex
    @ .\essentials.jl:13 [inlined]
  [2] macro expansion
    @ .\reduce.jl:264 [inlined]
  [3] macro expansion
    @ .\simdloop.jl:77 [inlined]
  [4] mapreduce_impl(f::ADNLPModels.var"#8#18", op::typeof(+), A::Vector{…}, ifirst::Int64, ilast::Int64, blksize::Int64)
    @ Base .\reduce.jl:263
  [5] mapreduce_impl
    @ .\reduce.jl:277 [inlined]
  [6] _mapreduce(f::ADNLPModels.var"#8#18", op::typeof(+), ::IndexLinear, A::Vector{…})
    @ Base .\reduce.jl:447
  [7] _mapreduce_dim(f::Function, op::Function, ::Base._InitialValue, A::Vector{…}, ::Colon)
    @ Base .\reducedim.jl:365
  [8] mapreduce(f::Function, op::Function, A::Vector{SparseConnectivityTracer.HessianTracer{…}})
    @ Base .\reducedim.jl:357
  [9] (::ADNLPModels.var"#7#17")(x::Vector{SparseConnectivityTracer.HessianTracer{…}})
    @ ADNLPModels C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\ad.jl:236
 [10] (::ADNLPModels.var"#lagrangian#55"{…})(x::Vector{…})
    @ ADNLPModels C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\sparsity_pattern.jl:44
 [11] trace_function(::Type{…}, f::ADNLPModels.var"#lagrangian#55"{…}, x::Vector{…})
    @ SparseConnectivityTracer C:\Users\Farhad\.julia\packages\SparseConnectivityTracer\22dWp\src\trace_functions.jl:48
 [12] _hessian_sparsity(f::Function, x::Vector{Float64}, ::Type{SparseConnectivityTracer.HessianTracer{…}})
    @ SparseConnectivityTracer C:\Users\Farhad\.julia\packages\SparseConnectivityTracer\22dWp\src\trace_functions.jl:106
 [13] hessian_sparsity
    @ C:\Users\Farhad\.julia\packages\SparseConnectivityTracer\22dWp\src\adtypes_interface.jl:64 [inlined]
 [14] compute_hessian_sparsity(f::ADNLPModels.var"#7#17", nvar::Int64, c!::ADNLPModels.var"#9#19", ncon::Int64; detector::SparseConnectivityTracer.TracerSparsityDetector{…})
    @ ADNLPModels C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\sparsity_pattern.jl:53
 [15] ADNLPModels.SparseADHessian(nvar::Int64, f::Function, ncon::Int64, c!::ADNLPModels.var"#9#19"; x0::Vector{…}, coloring_algorithm::SparseMatrixColorings.GreedyColoringAlgorithm{…}, detector::SparseConnectivityTracer.TracerSparsityDetector{…}, kwargs::@Kwargs{…})
    @ ADNLPModels C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\sparse_hessian.jl:29
 [16] macro expansion
    @ C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\ad.jl:266 [inlined]
 [17] macro expansion
    @ .\timing.jl:395 [inlined]
 [18] ADModelNLSBackend(nvar::Int64, F!::OptimizationProblems.ADNLPProblems.var"#F!#784"{…}, nequ::Int64; backend::Symbol, matrix_free::Bool, show_time::Bool, gradient_backend::Type, hprod_backend::Type, hessian_backend::Type, hprod_residual_backend::Type, jprod_residual_backend::Type, jtprod_residual_backend::Type, jacobian_residual_backend::Type, hessian_residual_backend::Type, kwargs::@Kwargs{…})
    @ ADNLPModels C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\ad.jl:262
 [19] ADNLSModel!(F!::Function, x0::Vector{…}, nequ::Int64; linequ::Vector{…}, name::String, minimize::Bool, kwargs::@Kwargs{…})
    @ ADNLPModels C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\nls.jl:163
 [20] ADNLSModel!
    @ C:\Users\Farhad\.julia\packages\ADNLPModels\hHKR7\src\nls.jl:151 [inlined]
 [21] #gulf#782
    @ C:\Users\Farhad\.julia\packages\OptimizationProblems\G0vFO\src\ADNLPProblems\gulf.jl:46 [inlined]
 [22] gulf
    @ C:\Users\Farhad\.julia\packages\OptimizationProblems\G0vFO\src\ADNLPProblems\gulf.jl:28 [inlined]
 [23] gulf(; use_nls::Bool, kwargs::@Kwargs{T::DataType, n::Int64})
    @ OptimizationProblems.ADNLPProblems C:\Users\Farhad\.julia\packages\OptimizationProblems\G0vFO\src\ADNLPProblems\gulf.jl:5
 [24] top-level scope
    @ REPL[38]:1
Some type information was truncated. Use `show(err)` to see complete types.

[farhadrclass]

I don't get an error for PureJump

julia> nlp = OptimizationProblems.PureJuMP.gulf()
A JuMP Model
├ solver: none
├ objective_sense: MIN_SENSE
│ └ objective_function_type: JuMP.NonlinearExpr
├ num_variables: 3
├ num_constraints: 0
└ Names registered in the model
  └ :x

julia> nlp = OptimizationProblems.PureJuMP.gulf(T=Float32, n=10, use_nls=true)
A JuMP Model
├ solver: none
├ objective_sense: MIN_SENSE
│ └ objective_function_type: JuMP.NonlinearExpr
├ num_variables: 3
├ num_constraints: 0
└ Names registered in the model
  └ :x

[farhadrclass]

so it fails for all:

  :gradient_backend => ForwardDiffADGradient,
  :gradient_backend => ReverseDiffADGradient,
  :gradient_backend => GenericForwardDiffADGradient,
  :gradient_backend => EnzymeReverseADGradient,
  :gradient_backend => ZygoteADGradient,

[farhadrclass]

I think the issue is here:

 function F!(r, x; m = m)
    for i = 1:n
      r[i] =
        exp(-abs((25 + (-50 * log(i * one(T) / 100))^(2 // 3)) * m * i * x[2])^x[3] / x[1]) -
        i // 100
    end
    return r

I think we should use for i = 1:m ?

[dpo]

Yes, please open a pr.

[dpo]

@farhadrclass