Quadratic Expression

Specific Domains Optimization (Mathematical)
,
1

Hi, I’m trying to create a quadratic expression using a variable attached in model m :

using JuMP
using Ipopt

m= Model(with_optimizer(Ipopt.Optimizer))

@variables m begin
   u[1:(time_lapses)*2,1:number_of_players]  
end

cf=@expression(m,[1:(time_lapses*2),1:number_of_players],0)

for i in 1:number_of_prosumers
   for j in 1:(time_lapses*2)
       cf[j,i]=u[j,i]*u[j,i]
   end
end

I get this error :

MethodError: Cannot `convert` an object of type GenericQuadExpr{Float64,VariableRef} to an object of type GenericAffExpr{Float64,VariableRef}
2

The @expression macro creates a container of JuMP.AffExpr (it will create a container of Integer in JuMP master, see https://github.com/JuliaOpt/JuMP.jl/pull/2070).
You can do

cf = Matrix{JuMP.QuadExpr}(undef, 2time_lapses, number_of_players)
for i in 1:2time_lapses
    for j in 1:number_of_players
        cf[i, j] = zero(JuMP.QuadExpr)
    end
end
for i in 1:number_of_prosumers
   for j in 1:(time_lapses*2)
       JuMP.add_to_expression(cf[j,i], u[j,i], u[j,i])
   end
end

but in JuMP master, you should be able to do

cf = @expression(m, [1:(time_lapses*2),1:number_of_players],
     sum(u[j,i]*u[j,i] for i in 1:number_of_prosumers, j in 1:(time_lapses*2)))
3

Thanks for your help! It worked