Merge pull request #2041 from JuliaOpt/bl/dual_objective_value

Add dual_objective_function and update doc links to MOI

docs/src/constraints.md

@@ -106,7 +106,7 @@ julia> @constraint(model, 2x + 1 <= 4x + 4)
 
 ## [Duality](@id constraint_duality)
 
-JuMP adopts the notion of [conic duality from MOI](http://www.juliaopt.org/MathOptInterface.jl/v0.8.1/apimanual.html#Duals-1).
+JuMP adopts the notion of [conic duality from MOI](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apimanual/#Duals-1).
 For linear programs, a feasible dual on a `>=` constraint is nonnegative and a
 feasible dual on a `<=` constraint is nonpositive. If the constraint is an
 equality constraint, it depends on which direction is binding.
@@ -350,13 +350,13 @@ julia> @constraint(model, A * x - b in MOI.Nonnegatives(2))
 
 In addition to the `Nonnegatives` set, MOI defines a number of
 other vector-valued sets such as `Nonpositives`. See the
-[MOI documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.8.1/apireference/#Sets-1)
+[MOI documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apireference/#Sets-1)
 for more information.
 
 Note also that for the first time we have used an explicit *function-in-set*
 description of the constraint. Read more about this representation for
 constraints in the
-[MOI documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.8.1/apimanual/#Constraints-by-function-set-pairs-1).
+[MOI documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apimanual/#Constraints-by-function-set-pairs-1).
 
 ## Constraints on a single variable
 
@@ -440,7 +440,7 @@ julia> @constraint(model, [t, u, x[1], x[2]] in RotatedSecondOrderCone())
 
 In addition to the second order cone and rotated second order cone,
 MOI defines a number of other conic sets such as the exponential
-and power cones. See the [MathOptInterface documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.8.1/apireference/#Sets-1)
+and power cones. See the [MathOptInterface documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apireference/#Sets-1)
 for more information.
 
 ## Constraints on a collection of variables

docs/src/extensions.md

@@ -22,7 +22,7 @@ Extending JuMP
 # TODO: create new bridge
 ```
 
-See the [bridge section in the MOI manual](http://www.juliaopt.org/MathOptInterface.jl/v0.8/apimanual/#Constraint-bridges-1).
+See the [bridge section in the MOI manual](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apimanual/#Automatic-reformulation-1).
 
 ```@docs
 add_bridge

docs/src/installation.md

@@ -23,7 +23,7 @@ is often more complex. We list below the currently available solvers.
 !!! note
     This list is open for new contributions. See also
     [Interacting with solvers](@ref) and the
-    [MathOptInterface docs](http://www.juliaopt.org/MathOptInterface.jl/v0.6.1/)
+    [MathOptInterface docs](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/)
     for more details on how JuMP interacts with solvers. Please get in touch
     with any questions about connecting new solvers with JuMP.
 

docs/src/nlp.md

@@ -333,7 +333,7 @@ may be expected to be within a factor of 5 of AMPL's.
 For some advanced use cases, one may want to directly query the derivatives of a
 JuMP model instead of handing the problem off to a solver.
 Internally, JuMP implements the `AbstractNLPEvaluator` interface from
-[MathOptInterface](http://www.juliaopt.org/MathOptInterface.jl/v0.6.1/apireference.html#NLP-evaluator-methods-1).
+[MathOptInterface](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apireference/#NLP-evaluator-methods-1).
 To obtain an NLP evaluator object from a JuMP model, use `JuMP.NLPEvaluator`.
 `JuMP.index` returns the `MOI.VariableIndex` corresponding to a JuMP variable.
 `MOI.VariableIndex` itself is a type-safe wrapper for `Int64` (stored in the

docs/src/objective.md

@@ -17,7 +17,9 @@ To query the objective function from a model, see [`objective_sense`](@ref),
 
 To query the optimal objective value or best known bound after a solve, see
 [`objective_value`](@ref) and [`objective_bound`](@ref). These two functions
-apply to nonlinear objectives also.
+apply to nonlinear objectives also. The optimal value of the
+[dual](@ref constraint_duality) objective can be obtained via
+[`dual_objective_value`](@ref).
 
 
 ## Reference
@@ -34,4 +36,5 @@ JuMP.objective_function_type
 
 JuMP.objective_bound
 JuMP.objective_value
+JuMP.dual_objective_value
 ```

docs/src/solutions.md

@@ -31,7 +31,7 @@ see the [Termination statuses](@ref) section below.
 
 Second, we can query the [`primal_status`](@ref) and the [`dual_status`](@ref),
 which will tell us what kind of results we have for our primal and dual
-solutions. This might be an optimal primal-dual pair, a primal solution without 
+solutions. This might be an optimal primal-dual pair, a primal solution without
 a corresponding dual solution, or a certificate of primal or dual infeasibility.
 For more information, see the [Solution statuses](@ref) section below.
 
@@ -70,15 +70,15 @@ MOI.ResultStatusCode
 ```
 
 Common status situations are described in the
-[MOI docs](http://www.juliaopt.org/MathOptInterface.jl/v0.8/apimanual/#Common-status-situations-1).
+[MOI docs](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/apimanual/#Common-status-situations-1).
 
 ## Obtaining solutions
 
 Provided the primal status is not `MOI.NO_SOLUTION`, the primal solution can
 be obtained by calling [`value`](@ref). For the dual solution, the function
 is [`dual`](@ref). Calling [`has_values`](@ref) for the primal status and
 [`has_duals`](@ref) for the dual solution is an equivalent way to check whether
-the status is `MOI.NO_SOLUTION`. 
+the status is `MOI.NO_SOLUTION`.
 
 It is important to note that if [`has_values`](@ref) or [`has_duals`](@ref)
 return false, calls to [`value`](@ref) and [`dual`](@ref) might throw an error
@@ -95,10 +95,12 @@ The container type (e.g., scalar, vector, or matrix) of the returned solution
 
 The objective value of a solved problem can be obtained via
 [`objective_value`](@ref). The best known bound on the optimal objective
-value can be obtained via [`objective_bound`](@ref).
+value can be obtained via [`objective_bound`](@ref). If the solver supports it,
+the value of the dual objective can be obtained via
+[`dual_objective_value`](@ref).
 
 The following is a recommended workflow for solving a model and querying the
-solution: 
+solution:
 ```julia
 using JuMP
 model = Model()

docs/src/solvers.md

@@ -56,7 +56,7 @@ the optimizer:
   MOI but new ones can be defined and added to the `LazyBridgeOptimizer` used by
   JuMP.
 
-See the [MOI documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.8.1/)
+See the [MOI documentation](http://www.juliaopt.org/MathOptInterface.jl/v0.9.1/)
 for more details on these two MOI layers.
 
 To attach an optimizer to a JuMP model, JuMP needs to create a new empty
@@ -109,4 +109,4 @@ set_silent
 unset_silent
 
 set_parameter
-```
+```

src/objective.jl

@@ -24,6 +24,15 @@ Return the objective value after a call to `optimize!(model)`.
 """
 objective_value(model::Model)::Float64 = MOI.get(model, MOI.ObjectiveValue())
 
+"""
+    dual_objective_value(model::Model)
+
+Return the value of the objective of the dual problem after a call to
+`optimize!(model)`. Throws `MOI.UnsupportedAttribute{MOI.DualObjectiveValue}` if
+the solver does not support this attribute.
+"""
+dual_objective_value(model::Model)::Float64 = MOI.get(model, MOI.DualObjectiveValue())
+
 """
     objective_sense(model::Model)::MathOptInterface.OptimizationSense
 

test/generate_and_solve.jl

@@ -71,6 +71,7 @@ using JuMP
         @test  1.0 == @inferred JuMP.value(x + y)
         @test  1.0 == @inferred JuMP.value(c)
         @test -1.0 == @inferred JuMP.objective_value(m)
+        @test -1.0 == @inferred JuMP.dual_objective_value(m)
 
         @test MOI.FEASIBLE_POINT == @inferred JuMP.dual_status(m)
         @test -1.0 == @inferred JuMP.dual(c)
@@ -225,6 +226,7 @@ using JuMP
         @test 1.0 == @inferred JuMP.value(x)
         @test 0.0 == @inferred JuMP.value(y)
         @test -1.0 == @inferred JuMP.objective_value(m)
+        @test 5.0 == @inferred JuMP.dual_objective_value(m)
 
         @test MOI.FEASIBLE_POINT == @inferred JuMP.dual_status(m)
         @test -1.0 == @inferred JuMP.dual(c1)

test/model.jl

@@ -14,21 +14,12 @@ using JuMP
 
 using Test
 
-using MathOptInterface
-const MOI = MathOptInterface
-const MOIT = MOI.Test
-const MOIU = MOI.Utilities
-
 # Simple LP model not supporting Interval
-@MOIU.model(SimpleLPModel,
-            (),
-            (MOI.EqualTo, MOI.GreaterThan, MOI.LessThan),
-            (),
-            (),
-            (),
-            (MOI.ScalarAffineFunction,),
-            (),
-            ())
+MOIU.@model(
+    SimpleLPModel,
+    (), (MOI.EqualTo, MOI.GreaterThan, MOI.LessThan), (), (),
+    (), (MOI.ScalarAffineFunction,), (), ()
+)
 
 struct Optimizer
     a::Int
@@ -63,6 +54,7 @@ function test_model()
         c = @constraint(model, x ≤ 0)
         @objective(model, Max, x)
         @test_throws err JuMP.objective_value(model)
+        @test_throws err JuMP.dual_objective_value(model)
         @test_throws err JuMP.objective_bound(model)
         @test_throws err JuMP.value(x)
         @test_throws err JuMP.value(c)
@@ -164,8 +156,8 @@ function test_model()
 
     @testset "UniversalFallback" begin
         m = Model()
-        MOI.set(m, MOIT.UnknownModelAttribute(), 1)
-        @test MOI.get(m, MOIT.UnknownModelAttribute()) == 1
+        MOI.set(m, MOI.Test.UnknownModelAttribute(), 1)
+        @test MOI.get(m, MOI.Test.UnknownModelAttribute()) == 1
     end
 
     @testset "Bridges" begin

test/nonnegative_bridge.jl

@@ -10,8 +10,7 @@
 
 # This file contains an example bridge used for tests.
 
-using MathOptInterface
-const MOI = MathOptInterface
+using JuMP
 const MOIB = MOI.Bridges
 const MOIBC = MOI.Bridges.Constraint