Revamp for Julia v0.6

.travis.yml

@@ -3,8 +3,7 @@ os:
   - linux
   - osx
 julia:
-  - 0.5
-  - nightly
+  - 0.6
 notifications:
   email: false
 sudo: false

REQUIRE

@@ -1,2 +1,2 @@
-julia 0.5
-MathProgBase 0.5 0.7
+julia 0.6
+MathProgBase 0.5 0.8

appveyor.yml

@@ -1,9 +1,7 @@
 environment:
   matrix:
-  - JULIA_URL: "https://julialang-s3.julialang.org/bin/winnt/x86/0.5/julia-0.5-latest-win32.exe"
-  - JULIA_URL: "https://julialang-s3.julialang.org/bin/winnt/x64/0.5/julia-0.5-latest-win64.exe"
-  - JULIA_URL: "https://julialangnightlies-s3.julialang.org/bin/winnt/x86/julia-latest-win32.exe"
-  - JULIA_URL: "https://julialangnightlies-s3.julialang.org/bin/winnt/x64/julia-latest-win64.exe"
+  - JULIA_URL: "https://julialang-s3.julialang.org/bin/winnt/x86/0.6/julia-0.6-latest-win32.exe"
+  - JULIA_URL: "https://julialang-s3.julialang.org/bin/winnt/x64/0.6/julia-0.6-latest-win64.exe"
 
 branches:
   only:

examples/jump_const_obj.jl

@@ -1,18 +1,18 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 # Example with no objective (#50)
 
-if !isdefined(:solver); solver = BonminNLSolver(); end
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-m = Model(solver=solver)
-@variable(m, 0 <= yp <= 1, Int)
-@variable(m, 0 <= l <= 1000.0)
-@variable(m, 0 <= f <= 1000.0)
-@NLconstraint(m, .087 * l >= f ^ 2)
-@constraint(m, l <= yp * 1000.0)
-@objective(m, Min, 5)
+@testset "example: jump_no_obj" begin
+    m = Model(solver=solver)
+    @variable(m, 0 <= yp <= 1, Int)
+    @variable(m, 0 <= l <= 1000.0)
+    @variable(m, 0 <= f <= 1000.0)
+    @NLconstraint(m, .087 * l >= f ^ 2)
+    @constraint(m, l <= yp * 1000.0)
+    @objective(m, Min, 5)
 
-context("example: jump_no_obj") do
-    @fact solve(m) --> :Optimal
-    @fact getobjectivevalue(m) --> 5
+    @test solve(m) == :Optimal
+    @test getobjectivevalue(m) == 5
 end

examples/jump_maxmin.jl

@@ -1,6 +1,7 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
+
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-if !isdefined(:solver); solver = IpoptNLSolver(); end
 # Note min and max not implemented in Couenne
 
 ## Solve test problem with simple min functions
@@ -11,13 +12,13 @@ if !isdefined(:solver); solver = IpoptNLSolver(); end
  #  The optimal objective value is 0.25.
  #      x = 0.5
 ##
-context("example: maxmin") do
+@testset "example: maxmin" begin
     m = Model(solver=solver)
     @variable(m, -0.5 <= x <= 0.5, start = 0.25)
     @NLobjective(m, Max, min(x^2, 0.3, x))
-    @fact solve(m) --> :Optimal
-    @fact getobjectivevalue(m) --> roughly(0.25, 1e-2)
-    @fact getvalue(x) --> roughly(0.5, 1e-2)
+    @test solve(m) == :Optimal
+    @test isapprox(getobjectivevalue(m), 0.25, atol=1e-2)
+    @test isapprox(getvalue(x), 0.5, atol=1e-2)
 end
 
 ## Solve test problem with simple max functions
@@ -28,11 +29,11 @@ end
  #  The optimal objective value is 0.
  #      x = 0.
 ##
-context("example: minmax") do
+@testset "example: minmax" begin
     m = Model(solver=solver)
     @variable(m, -1 <= x <= 1, start=-1)
     @NLobjective(m, Min, max(x^2, x, -1))
-    @fact solve(m) --> :Optimal
-    @fact getobjectivevalue(m) --> roughly(0, 1e-2)
-    @fact getvalue(x) --> roughly(0, 1e-2)
+    @test solve(m) == :Optimal
+    @test isapprox(getobjectivevalue(m), 0, atol=1e-2)
+    @test isapprox(getvalue(x), 0, atol=1e-2)
 end

examples/jump_minlp.jl

@@ -1,4 +1,4 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 ## Solve test problem 1 (Synthesis of processing system) in
  #  M. Duran & I.E. Grossmann, "An outer approximation algorithm for
@@ -24,24 +24,33 @@ using JuMP, FactCheck, AmplNLWriter
  #  The solution is (1.30098, 0, 1, 0, 1, 0).
  ##
 
-if !isdefined(:solver); solver = BonminNLSolver(); end
+ # solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-m = Model(solver=solver)
-x_U = [2,2,1]
-@variable(m, x_U[i] >= x[i=1:3] >= 0)
-@variable(m, y[4:6], Bin)
 
-@NLobjective(m, Min, 10 + 10*x[1] - 7*x[3] + 5*y[4] + 6*y[5] + 8*y[6] - 18*log(x[2]+1) - 19.2*log(x[1]-x[2]+1))
-@NLconstraint(m, 0.8*log(x[2] + 1) + 0.96*log(x[1] - x[2] + 1) - 0.8*x[3] >= 0)
-@NLconstraint(m, log(x[2] + 1) + 1.2*log(x[1] - x[2] + 1) - x[3] - 2*y[6] >= -2)
-@NLconstraint(m, x[2] - x[1] <= 0)
-@NLconstraint(m, x[2] - 2*y[4] <= 0)
-@NLconstraint(m, x[1] - x[2] - 2*y[5] <= 0)
-@NLconstraint(m, y[4] + 2*y[5]*0.5 <= 1)
+@testset "example: jump_minlp" begin
+    m = Model(solver=solver)
+    x_U = [2,2,1]
+    @variable(m, x_U[i] >= x[i=1:3] >= 0)
+    @variable(m, y[i=4:6], Bin)
 
-context("example: jump_minlp") do
-    @fact solve(m) --> :Optimal
-    @fact getvalue(x)[:] --> roughly([1.30098, 0.0, 1.0], 1e-5)
-    @fact getvalue(y)[:] --> roughly([0.0, 1.0, 0.0], 1e-5)
-    @fact getobjectivevalue(m) --> roughly(6.00975, 1e-5)
+    @NLobjective(m, Min, 10 + 10*x[1] - 7*x[3] + 5*y[4] + 6*y[5] + 8*y[6] - 18*log(x[2]+1) - 19.2*log(x[1]-x[2]+1))
+    @NLconstraint(m, 0.8*log(x[2] + 1) + 0.96*log(x[1] - x[2] + 1) - 0.8*x[3] >= 0)
+    @NLconstraint(m, log(x[2] + 1) + 1.2*log(x[1] - x[2] + 1) - x[3] - 2*y[6] >= -2)
+    @NLconstraint(m, x[2] - x[1] <= 0)
+    @NLconstraint(m, x[2] - 2*y[4] <= 0)
+    @NLconstraint(m, x[1] - x[2] - 2*y[5] <= 0)
+    @NLconstraint(m, y[4] + 2*y[5]*0.5 <= 1)
+
+    @test solve(m) == :Optimal
+
+    if getsolvername(solver) == "ipopt"
+        # Ipopt solves the relaxation
+        @test isapprox(getvalue(x)[:], [1.14652, 0.546596, 1.0], atol=1e-5)
+        @test isapprox(getvalue(y)[:], [0.27330, 0.299959, 0.0], atol=1e-5)
+        @test isapprox(getobjectivevalue(m), 0.75928, atol=1e-5)
+    else
+        @test isapprox(getvalue(x)[:], [1.30098, 0.0, 1.0], atol=1e-5)
+        @test isapprox(getvalue(y)[:], [0.0, 1.0, 0.0], atol=1e-5)
+        @test isapprox(getobjectivevalue(m), 6.00975, atol=1e-5)
+    end
 end

examples/jump_nlexpr.jl

@@ -1,32 +1,32 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 # Example testing basic use of NLExpr with AmplNLWriter.jl
 
-if !isdefined(:solver); solver = IpoptNLSolver(); end
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-m = Model(solver=solver)
 
-n = 30
-l = -ones(n); l[1] = 0
-u = ones(n)
-@variable(m, l[i] <= x[i=1:n] <= u[i])
-@NLexpression(m, f1, x[1])
-@NLexpression(m, g, 1 + 9 * sum{x[j] ^ 2, j = 2:n} / (n - 1))
-@NLexpression(m, h, 1 - (f1 / g) ^ 2)
-@NLexpression(m, f2, g * h)
+@testset "example: jump_nlexpr" begin
+    m = Model(solver=solver)
 
-setvalue(x[1], 1)
-setvalue(x[2:n], zeros(n - 1))
-@NLobjective(m, :Min, f2)
+    n = 30
+    l = -ones(n); l[1] = 0
+    u = ones(n)
+    @variable(m, l[i] <= x[i=1:n] <= u[i])
+    @NLexpression(m, f1, x[1])
+    @NLexpression(m, g, 1 + 9 * sum(x[j] ^ 2 for j = 2:n) / (n - 1))
+    @NLexpression(m, h, 1 - (f1 / g) ^ 2)
+    @NLexpression(m, f2, g * h)
 
-context("example: jump_nlexpr") do
-    @fact solve(m) --> :Optimal
-    @fact getvalue(x[1]) --> roughly(1.0, 1e-5)
-    @fact getvalue(x[2:end]) --> roughly(zeros(n - 1), 1e-5)
-    @fact getvalue(f1) --> roughly(1.0, 1e-5)
-    @fact getvalue(f2) --> roughly(0.0, 1e-5)
-    @fact getvalue(g) --> roughly(1.0, 1e-5)
-    @fact getvalue(h) --> roughly(0.0, 1e-5)
-    @fact getobjectivevalue(m) --> roughly(0.0, 1e-5)
-end
+    setvalue(x[1], 1)
+    setvalue(x[2:n], zeros(n - 1))
+    @NLobjective(m, :Min, f2)
 
+    @test solve(m) == :Optimal
+    @test isapprox(getvalue(x[1]), 1.0, atol=1e-5)
+    @test isapprox(getvalue(x[2:end]), zeros(n - 1), atol=1e-5)
+    @test isapprox(getvalue(f1), 1.0, atol=1e-5)
+    @test isapprox(getvalue(f2), 0.0, atol=1e-5)
+    @test isapprox(getvalue(g), 1.0, atol=1e-5)
+    @test isapprox(getvalue(h), 0.0, atol=1e-5)
+    @test isapprox(getobjectivevalue(m), 0.0, atol=1e-5)
+end

examples/jump_nltrig.jl

@@ -1,4 +1,4 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 ## Solve test problem with sind and cosd functions
  #
@@ -8,26 +8,23 @@ using JuMP, FactCheck, AmplNLWriter
  #  The optimal objective value is 0
  ##
 
-# Allow resolving the model from multiple starts after NLP changes in JuMP 0.12
-EnableNLPResolve()
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-if !isdefined(:solver); solver = CouenneNLSolver(); end
+@testset "example: jump_nltrig" begin
+    m = Model(solver=solver)
+    @variable(m, x[1:2])
 
-m = Model(solver=solver)
-@variable(m, x[1:2])
+    @NLobjective(m, Min, (7 - (3*cosd(x[1]) + 5*cosd(x[2])))^2 + (0 - (3*sind(x[1]) + 5*sind(x[2])))^2)
 
-@NLobjective(m, Min, (7 - (3*cosd(x[1]) + 5*cosd(x[2])))^2 + (0 - (3*sind(x[1]) + 5*sind(x[2])))^2)
-
-context("example: jump_nltrig") do
     setvalue(x[1], 30)
     setvalue(x[2], -50)
-    @fact solve(m) --> :Optimal
-    @fact getvalue(x)[:] --> roughly([38.21321, -21.78678], 1e-5)
-    @fact getobjectivevalue(m) --> roughly(0.0, 1e-5)
+    @test solve(m) == :Optimal
+    @test isapprox(getvalue(x)[:], [38.21321, -21.78678], atol=1e-5)
+    @test isapprox(getobjectivevalue(m), 0.0, atol=1e-5)
     # Now try from the other side
     setvalue(x[1], -30)
     setvalue(x[2], 50)
-    @fact solve(m) --> :Optimal
-    @fact getvalue(x)[:] --> roughly([-38.21321, 21.78678], 1e-5)
-    @fact getobjectivevalue(m) --> roughly(0.0, 1e-5)
+    @test solve(m) == :Optimal
+    @test isapprox(getvalue(x)[:], [-38.21321, 21.78678], atol=1e-5)
+    @test isapprox(getobjectivevalue(m), 0.0, atol=1e-5)
 end

examples/jump_no_obj.jl

@@ -1,17 +1,17 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 # Example with no objective (#50)
 
-if !isdefined(:solver); solver = BonminNLSolver(); end
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-m = Model(solver=solver)
-@variable(m, 0 <= yp <= 1, Int)
-@variable(m, 0 <= l <= 1000.0)
-@variable(m, 0 <= f <= 1000.0)
-@NLconstraint(m, .087 * l >= f ^ 2)
-@constraint(m, l <= yp * 1000.0)
+@testset "example: jump_no_obj" begin
+    m = Model(solver=solver)
+    @variable(m, 0 <= yp <= 1, Int)
+    @variable(m, 0 <= l <= 1000.0)
+    @variable(m, 0 <= f <= 1000.0)
+    @NLconstraint(m, .087 * l >= f ^ 2)
+    @constraint(m, l <= yp * 1000.0)
 
-context("example: jump_no_obj") do
-    @fact solve(m) --> :Optimal
-    @fact getobjectivevalue(m) --> 0
+    @test solve(m) == :Optimal
+    @test getobjectivevalue(m) == 0
 end

examples/jump_nonlinearbinary.jl

@@ -1,4 +1,4 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 ## Solve test problem with non-linear binary variables
  #
@@ -8,20 +8,27 @@ using JuMP, FactCheck, AmplNLWriter
  #  The solution is (0, 0).
  ##
 
-if !isdefined(:solver); solver = BonminNLSolver(); end
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-m = Model(solver=solver)
-@variable(m, x[1:2], Bin)
+@testset "example: jump_nonlinearbinary" begin
+    m = Model(solver=solver)
+    @variable(m, x[1:2], Bin)
 
-# Set some non-binary bounds on x1 and x2. These should be ignored.
-# The optimal solution if x is Int is (1, 2) which is allowed by these bounds
-setupperbound(x[1], 2)
-setupperbound(x[2], 2)
+    # Set some non-binary bounds on x1 and x2. These should be ignored.
+    # The optimal solution if x is Int is (1, 2) which is allowed by these bounds
+    setupperbound(x[1], 2)
+    setupperbound(x[2], 2)
 
-@NLobjective(m, Min, 100*(x[2] - (0.5 + x[1])^2)^2 + (1 - x[1])^2)
+    @NLobjective(m, Min, 100*(x[2] - (0.5 + x[1])^2)^2 + (1 - x[1])^2)
 
-context("example: jump_nonlinearbinary") do
-    @fact solve(m) --> :Optimal
-    @fact getvalue(x)[:] --> [0.0, 0.0]
-    @fact getobjectivevalue(m) --> 7.25
+    @test solve(m) == :Optimal
+
+    if getsolvername(solver) == "ipopt"
+        # Ipopt solves the relaxation
+        @test isapprox(getvalue(x), [0.501245, 1.0], atol=1e-6)
+        @test isapprox(getobjectivevalue(m), 0.249377, atol=1e-6)
+    else
+        @test getvalue(x)[:] == [0.0, 0.0]
+        @test getobjectivevalue(m) == 7.25
+    end
 end

examples/jump_pruning.jl

@@ -1,4 +1,4 @@
-using JuMP, FactCheck, AmplNLWriter
+using JuMP, Base.Test, AmplNLWriter
 
 ## Solve test problem with lots of expressions to prune
  #
@@ -16,23 +16,24 @@ using JuMP, FactCheck, AmplNLWriter
  #  The optimal objective value is 400, solutions can vary.
  ##
 
-if !isdefined(:solver); solver = IpoptNLSolver(); end
-m = Model(solver=solver)
+# solver = AmplNLSolver(Ipopt.amplexe, ["print_level=0"])
 
-@variable(m, x[1:2] >= 0)
+@testset "example: jump_pruning" begin
+    m = Model(solver=solver)
 
-@NLobjective(m, Max, x[1]^2 * x[2]^2)
-@NLconstraint(m, x[1] * x[2] <= 20)
-@NLconstraint(m, x[1] + x[2] <= 40)
-@NLconstraint(m, x[1] * x[2] + x[1] + x[2] <= 60)
-@NLconstraint(m, x[1] + x[1] * x[2] + x[2] <= 60)
-@NLconstraint(m, x[1] * x[2] + x[1] + x[2] <= 60)
-@NLconstraint(m, x[1] * x[2] - x[1] - x[2] <= 0)
-@NLconstraint(m, x[2] - x[1] * x[2] + x[1] <= 60)
-@NLconstraint(m, x[2] - x[1] + x[1] * x[2] <= 0)
-@NLconstraint(m, 0 <= 1.0)
+    @variable(m, x[1:2] >= 0)
 
-context("example: jump_pruning") do
-    @fact solve(m) --> :Optimal
-    @fact getobjectivevalue(m) --> roughly(400, 1e-2)
+    @NLobjective(m, Max, x[1]^2 * x[2]^2)
+    @NLconstraint(m, x[1] * x[2] <= 20)
+    @NLconstraint(m, x[1] + x[2] <= 40)
+    @NLconstraint(m, x[1] * x[2] + x[1] + x[2] <= 60)
+    @NLconstraint(m, x[1] + x[1] * x[2] + x[2] <= 60)
+    @NLconstraint(m, x[1] * x[2] + x[1] + x[2] <= 60)
+    @NLconstraint(m, x[1] * x[2] - x[1] - x[2] <= 0)
+    @NLconstraint(m, x[2] - x[1] * x[2] + x[1] <= 60)
+    @NLconstraint(m, x[2] - x[1] + x[1] * x[2] <= 0)
+    @NLconstraint(m, 0 <= 1.0)
+
+    @test solve(m) == :Optimal
+    @test isapprox(getobjectivevalue(m), 400, atol=1e-2)
 end