JuMP 如何将优化问题嵌套进另一个优化中

using JuMP

import Ipopt

function ss1(z)

mss1 = Model(Ipopt.Optimizer)

set_silent(mss1)

@variable(

    mss1,

    x[1:7]

    #[2.6,0.7,17,7.3,7.3,2.9,5] <= x[1:7] <= [3.6,0.8,28,8.3,8.3,3.9,5.5]

)

@expression(

    mss1,

    J1,

    sum((x-z).^2)

)

@objective(

    mss1,

    Min,

    sum((x-z).^2)

)

@NLconstraints(

    mss1,

    begin

        27*x[1]^(-1)*x[2]^(-2)*x[3]^(-1)-1 <= 0

        397.5*x[1]^(-1)*x[2]^(-2)*x[3]^(-2)-1 <= 0

        x[2]*x[3]-40 <= 0

        x[1]/x[2]-12 <= 0

        5-x[1]/x[2] <= 0

        

    end

)

@constraints(

    mss1,

    begin

        [2.6,0.7,17,7.3,7.3,2.9,5] .<= x[1:7] 

        x[1:7] .<= [3.6,0.8,28,8.3,8.3,3.9,5.5]

    end

)

optimize!(mss1)

return value(J1)

end

function ss2(z)

mss2 = Model(Ipopt.Optimizer)

set_silent(mss2)

@variable(

    mss2,

    x[1:7]

)

@expression(

    mss2,

    J2,

    sum((x-z).^2)

)

@objective(

    mss2,

    Min,

    sum((x-z).^2)

)

@NLconstraints(

    mss2,

    begin

        27*x[1]^(-1)*x[2]^(-2)*x[3]^(-1)-1 <= 0

        97.5*x[1]^(-1)*x[2]^(-2)*x[3]^(-2)-1 <= 0

        1.93*x[2]^(-1)*x[3]^(-1)*x[4]^3*x[6]^(-4)-1 <= 0

        ((745*x[4]/(x[2]*x[3]))^2+1.69*1e7)^(1/2)/(0.1*x[6]^3)-1100 <= 0

        x[2]*x[3]-40 <= 0

        x[1]/x[2]-12 <= 0

        5-x[1]/x[2] <= 0

        (1.5*x[6]+1.9)*x[4]^(-1)-1 <= 0

    end

)

@constraints(

    mss2,

    begin

        [2.6,0.7,17,7.3,7.3,2.9,5] .<= x[1:7] 

        x[1:7] .<= [3.6,0.8,28,8.3,8.3,3.9,5.5]

    end

)

optimize!(mss2)

return value(J2)

end

function ss3(z)

mss3 = Model(Ipopt.Optimizer)

set_silent(mss3)

@variable(

    mss3,

    x[1:7]

)

@expression(

    mss3,

    J3,

    sum((x-z).^2)

)

@objective(

    mss3,

    Min,

    sum((x-z).^2)

)

@NLconstraints(

    mss3,

    begin

        27*x[1]^(-1)*x[2]^(-2)*x[3]^(-1)-1 <= 0

        397.5*x[1]^(-1)*x[2]^(-2)*x[3]^(-2)-1 <= 0

        1.93*x[2]^(-1)*x[3]^(-1)*x[5]^3*x[7]^(-4)-1 <= 0

        ((745*x[4]/(x[2]*x[3]))^2+1.69*1e7)^(1/2)/(0.1*x[6]^3)-1100 <= 0

        ((745*x[5]/(x[2]*x[3]))^2+1.575*1e8)^(1/2)/(0.1*x[7]^3)-850 <= 0

        x[2]*x[3]-40 <= 0

        x[1]/x[2]-12 <= 0

        5-x[1]/x[2] <= 0

        (1.1*x[7]+1.9)*x[5]^(-1)-1 <= 0

    end

)

    @constraints(

        mss3,

        begin

            [2.6,0.7,17,7.3,7.3,2.9,5] .<= x[1:7] 

        x[1:7] .<= [3.6,0.8,28,8.3,8.3,3.9,5.5]

        end

    )

optimize!(mss3)

return value(J3)

end

function opt()

msys = Model(Ipopt.Optimizer)

set_silent(msys)

@variable(

    msys,

   x[1:7]

)

@expressions(

    msys,

    begin

        J1, ss1(x)^2

        J2, ss2(x)^2

        J3, ss3(x)^2

    end

)

@NLobjective(

    msys,

    Min,

    0.7854*x[1]*x[2]^2*(3.333*x[3]^2+14.933*x[3]-43.0934)-1.508*x[1]*(x[6]^2+x[7]^2)+7.447*(x[6]^3+x[7]^3)+0.7854*(x[4]*x[6]^2+x[5]*x[7]^2)

)

@NLconstraints(

    msys,

    begin

        J1 <= 1e-5

        J2 <= 1e-5

        J3 <= 1e-5

     

    end

)

@constraints(

    msys,

    begin

        [2.6,0.7,17,7.3,7.3,2.9,5] .<= x[1:7] 

        x[1:7] .<= [3.6,0.8,28,8.3,8.3,3.9,5.5]

    end

)

    optimize!(sys)

return value(x)

end

opt()
我想将ss1 ss2 ss3 三个优化结果作为约束嵌套进opt这个优化问题中
min obj_opt
s.t. J1<=1E-5
J2<=1E-5
J3<=1E-5


但是会这样报错 具体应该怎么办呢

1、我印象里面好像Ipopt不支持隐函数?可以看看user-defined-functions那块。
2、你的子函数能顺利运行吗?
3、或者你看能不能把问题转化一下,一步优化到位