Error returning solution of a problem with DualExponentialCone using JuMP

[guilhermebodin]

Hello I am trying to run this problem using the JuMP interface

using MathOptInterface, JuMP, COSMO
model = Model(with_optimizer(COSMO.Optimizer))
@variable(model, u)
@variable(model, v)
@variable(model, w)
@constraint(model, [u; v; w] in MathOptInterface.DualExponentialCone())
@constraint(model, u == 1)
@constraint(model, v == 2)
@objective(model, Max, u + v + w)
optimize!(model)

apparently it solves but breaks writing the answer somewhere

julia> optimize!(model)
------------------------------------------------------------------
          COSMO v0.5.0 - A Quadratic Objective Conic Solver
                         Michael Garstka
                University of Oxford, 2017 - 2019
------------------------------------------------------------------

Problem:  x ∈ R^{3},
          constraints: A ∈ R^{5x3} (5 nnz),
          matrix size to factor: 8x8 (18 nnz)
Sets:     DualExponentialCone of dim: 3
          ZeroSet of dim: 2
Settings: ϵ_abs = 1.0e-04, ϵ_rel = 1.0e-04,
          ϵ_prim_inf = 1.0e-06, ϵ_dual_inf = 1.0e-04,
          ρ = 0.1, σ = 1.0e-6, α = 1.6,
          max_iter = 2500,
          scaling iter = 10 (on),
          check termination every 40 iter,
          check infeasibility every 40 iter,
          KKT system solver: QDLDL
Setup Time: 2156.02ms

Iter:   Objective:      Primal Res:     Dual Res:       Rho:
40      -6.4299e+02     9.9900e-01      1.0000e+00      1.0000e-01
ERROR: MethodError: no method matching in_cone(::Array{Float64,1}, ::Type{COSMO.ExponentialCone{Float64}}, ::Float64)
Closest candidates are:
  in_cone(::AbstractArray{T,1}, ::COSMO.ExponentialCone{T}, ::T) where T at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:507
  in_cone(::AbstractArray{T,1}, ::COSMO.PowerCone{T}, ::T) where T at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:624
  in_cone(::AbstractArray{T<:Real,1}, ::Union{DualExponentialCone, DualPowerCone}, ::Real) where T<:Real at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:678
Stacktrace:
 [1] in_dual(::Array{Float64,1}, ::COSMO.DualExponentialCone{Float64}, ::Float64) at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:679
 [2] support_function(::SubArray{Float64,1,Array{Float64,1},Tuple{UnitRange{Int64}},true}, ::COSMO.DualExponentialCone{Float64}, ::Float64) at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:809
 [3] (::getfield(COSMO, Symbol("##21#22")){Float64})(::Tuple{SubArray{Float64,1,Array{Float64,1},Tuple{UnitRange{Int64}},true},COSMO.DualExponentialCone{Float64}}) at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:775
 [4] mapfoldl_impl(::getfield(COSMO, Symbol("##21#22")){Float64}, ::typeof(Base.add_sum), ::NamedTuple{(:init,),Tuple{Float64}}, ::Base.Iterators.Zip2{Array{SubArray{Float64,1,Array{Float64,1},Tuple{UnitRange{Int64}},true},1},Array{COSMO.AbstractConvexSet,1}}, ::Tuple{Int64,Int64}) at ./reduce.jl:43
 [5] mapfoldl_impl(::Function, ::Function, ::NamedTuple{(),Tuple{}}, ::Base.Iterators.Zip2{Array{SubArray{Float64,1,Array{Float64,1},Tuple{UnitRange{Int64}},true},1},Array{COSMO.AbstractConvexSet,1}}) at ./reduce.jl:59
 [6] #mapfoldl#170 at ./reduce.jl:70 [inlined]
 [7] mapfoldl at ./reduce.jl:70 [inlined]
 [8] #mapreduce#174 at ./reduce.jl:203 [inlined]
 [9] mapreduce at ./reduce.jl:203 [inlined]
 [10] sum at ./reduce.jl:397 [inlined]
 [11] support_function at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/convexset.jl:775 [inlined]
 [12] is_primal_infeasible(::Array{Float64,1}, ::COSMO.Workspace{Float64}) at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/infeasibility.jl:14
 [13] optimize!(::COSMO.Workspace{Float64}) at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/solver.jl:135
 [14] optimize!(::Optimizer) at /home/guilhermebodin/.julia/packages/COSMO/Paq9I/src/MOIWrapper.jl:152
 [15] optimize!(::MathOptInterface.Utilities.CachingOptimizer{Optimizer,MathOptInterface.Utilities.UniversalFallback{MathOptInterface.Utilities.Model{Float64}}}) at /home/guilhermebodin/.julia/packages/MathOptInterface/WXD0G/src/Utilities/cachingoptimizer.jl:189
 [16] optimize!(::MathOptInterface.Bridges.LazyBridgeOptimizer{MathOptInterface.Utilities.CachingOptimizer{Optimizer,MathOptInterface.Utilities.UniversalFallback{MathOptInterface.Utilities.Model{Float64}}}}) at /home/guilhermebodin/.julia/packages/MathOptInterface/WXD0G/src/Bridges/bridge_optimizer.jl:184
 [17] optimize!(::MathOptInterface.Utilities.CachingOptimizer{MathOptInterface.AbstractOptimizer,MathOptInterface.Utilities.UniversalFallback{MathOptInterface.Utilities.Model{Float64}}}) at /home/guilhermebodin/.julia/packages/MathOptInterface/WXD0G/src/Utilities/cachingoptimizer.jl:189
 [18] #optimize!#78(::Bool, ::Bool, ::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::Function, ::Model, ::Nothing) at /home/guilhermebodin/.julia/packages/JuMP/iGamg/src/optimizer_interface.jl:141
 [19] optimize! at /home/guilhermebodin/.julia/packages/JuMP/iGamg/src/optimizer_interface.jl:111 [inlined] (repeats 2 times)
 [20] top-level scope at none:0

I am running with this Project.toml

(Teste) pkg> st
    Status `~/Documents/Github/Teste/Project.toml`
  [1e616198] COSMO v0.5.0
  [4076af6c] JuMP v0.20.0
  [b8f27783] MathOptInterface v0.9.2
  [c946c3f1] SCS v0.6.1

[migarstka]

Thanks @guilhermebodin .
I fixed the bug in ed25b4b. If you want to try it, you can checkout master.

With this fix, I get:

------------------------------------------------------------------
          COSMO v0.5.0 - A Quadratic Objective Conic Solver
                         Michael Garstka
                University of Oxford, 2017 - 2019
------------------------------------------------------------------

Problem:  x ∈ R^{3},
          constraints: A ∈ R^{5x3} (5 nnz),
          matrix size to factor: 8x8 (18 nnz)
Sets:     DualExponentialCone of dim: 3
          ZeroSet of dim: 2
Settings: ϵ_abs = 1.0e-04, ϵ_rel = 1.0e-04,
          ϵ_prim_inf = 1.0e-06, ϵ_dual_inf = 1.0e-04,
          ρ = 0.1, σ = 1.0e-6, α = 1.6,
          max_iter = 2500,
          scaling iter = 10 (on),
          check termination every 40 iter,
          check infeasibility every 40 iter,
          KKT system solver: QDLDL
Setup Time: 0.27ms

Iter:	Objective:	Primal Res:	Dual Res:	Rho:
40	-6.4299e+02	9.9900e-01	1.0000e+00	1.0000e-01

------------------------------------------------------------------
>>> Results
Status: Primal_infeasible
Iterations: 40
Optimal objective: -642.9926
Runtime: 0.027s (27.28ms)

which is correct as you constrain u > 0 and the dual exponential cone is:

K_Dexp = { (u, v, w) : -u exp( v / u ) <= - exp(1) w, u < 0 }

[guilhermebodin]

Thanks!