# NL Model Issue during forward diff

**URL:** <https://discourse.julialang.org/t/nl-model-issue-during-forward-diff/39888>\
**Category:** Optimization (Mathematical)\
**Tags:** jump\
**Created:** [May 21, 2020, 3:34pm UTC](https://discourse.julialang.org/t/nl-model-issue-during-forward-diff/39888 "2020-05-21T15:34:38Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Dominic\_Pazzula](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/dominic_pazzula/32/12783_2.png) [@Dominic\_Pazzula](https://discourse.julialang.org/u/Dominic_Pazzula)\
**Post date:** [May 21, 2020, 3:34pm UTC](https://discourse.julialang.org/t/nl-model-issue-during-forward-diff/39888/1 "2020-05-21T15:34:38Z")

</div>

I’m still new to Julia and JuMP, so this might be an easy one.

Setup: I have some complex business logic that takes in some planned changes in a DataFrame, and returns a DataFrame with projected output and error variables. I want to put this into an optimization.

```julia
model = Model(optimizer_with_attributes(Ipopt.Optimizer, "print_level"=>3))

@variable(model, max_val >= plan[1:nPotential] >= 0.)

function ssePlan(plan...)
    global _plan, p
    p = collect(plan)
    _plan = DataFrame(:ID => planTable.ID, :new_plan => p)
    _planAggTable = evalPlan(currentPlan,planTable=_plan)
    e1 = _planAggTable.E1
    e2 = _planAggTable.E2
    sse = e1'*e1 + e2'*e2
    return(sse)
end

_temp = ssePlan(planTable.new_plan...)

register(model,:ssePlanFn,nPotential,ssePlan,autodiff=true)

@NLobjective(model,Min,ssePlanFn(plan...))

optimize!(model)

```

`evalPlan()` contains the business logic.

I get the following error on the last line:

> **MethodError: no method matching Float64(::ForwardDiff.Dual{ForwardDiff.Tag{JuMP.var"#107#109"{typeof(ssePlan)},Float64},Float64,8})**
> 
> **Closest candidates are:  
> Float64(::Real, !Matched::RoundingMode) where T\<:AbstractFloat at rounding.jl:200  
> Float64(::T) where T\<:Number at boot.jl:718  
> Float64(!Matched::Int8) at float.jl:60  
> …**

The stack trace shows this occurring deep in the business logic.

I put the `global` statement into the objective function so I could see what was being passed in.

```julia
julia> _plan
16×2 DataFrame
│ Row │ ID │ commitment │
│ │ Int64 │ ForwardDiff.Dual{ForwardDiff.Tag{JuMP.var"#107#109"{typeof(ssePlan)},Float64},Float64,8} │
├─────┼────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ 1 │ -11100 │ Dual{ForwardDiff.Tag{JuMP.var"#107#109"{typeof(ssePlan)},Float64}}(0.0,1.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0) │
⋮
│ 15 │ 82 │ Dual{ForwardDiff.Tag{JuMP.var"#107#109"{typeof(ssePlan)},Float64}}(0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0) │
│ 16 │ 83 │ Dual{ForwardDiff.Tag{JuMP.var"#107#109"{typeof(ssePlan)},Float64}}(0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0) │

```

I’m confused. I thought the values being passed to the function through the model variables `plan[1:16]` would be single values.

What am I doing wrong here?

---

<div class="post-metadata">

**Author:** ![Dominic\_Pazzula](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/dominic_pazzula/32/12783_2.png) [@Dominic\_Pazzula](https://discourse.julialang.org/u/Dominic_Pazzula)\
**Post date:** [May 21, 2020, 5:49pm UTC](https://discourse.julialang.org/t/nl-model-issue-during-forward-diff/39888/2 "2020-05-21T17:49:35Z")

</div>

Solved it myself after some digging.

There was a `@byrow!` from DataFramesMeta. A new column was being explicitly created as Float64. Changed to Any

```julia
        @byrow! aggTable begin
            # @newcol cash::Array{Float64} <-- Old code
            @newcol cash::Array{Any}

```