added test

Author: sshin23

src/nlp.jl

@@ -502,7 +502,6 @@ function _constraint!(c, f, pars)
     c.nconaug += nitr
     c.nnzj += nitr * f.o1step
     c.nnzh += nitr * f.o2step
-
     c.con = ConstraintAug(c.con, f, convert_array(pars, c.backend), oa)
 end
 
@@ -690,7 +689,7 @@ coord(itr, i, ::Tuple{}) = ()
 @inbounds @inline offset0(a::C, i) where {C<:ConstraintAug} = offset0(a.f, a.itr, i)
 @inbounds @inline offset0(f::F, itr, i) where {P<:Pair,F<:SIMDFunction{P}} =
     f.o0 + f.f.first(itr[i], nothing)
-@inbounds @inline offset0(f::F, itr, i) where {P<:Pair,F<:SIMDFunction{P}} = f.o0 + idxx(coord(itr, i, f.f.first), Base.size(itr))
+@inbounds @inline offset0(f::F, itr, i) where {T<:Tuple,P<:Pair{T},F<:SIMDFunction{P}} = f.o0 + idxx(coord(itr, i, f.f.first), Base.size(itr))
 
 for (thing, val) in [(:solution, 1), (:multipliers_L, 0), (:multipliers_U, 2)]
     @eval begin

test/NLPTest/NLPTest.jl

@@ -123,24 +123,24 @@ end
 
 function runtests()
     @testset "NLP test" begin
-        for (name, args) in NLP_TEST_ARGUMENTS
-            @testset "$name $args" begin
+        for backend in BACKENDS
+            @testset "$backend" begin
+                for (name, args) in NLP_TEST_ARGUMENTS
+                    @testset "$name $args" begin
 
-                exa_model = getfield(@__MODULE__, Symbol("_exa_$(name)_model"))
-                jump_model = getfield(@__MODULE__, Symbol("_jump_$(name)_model"))
+                        exa_model = getfield(@__MODULE__, Symbol("_exa_$(name)_model"))
+                        jump_model = getfield(@__MODULE__, Symbol("_jump_$(name)_model"))
 
-                m, vars0, cons0 = exa_model(nothing, args)
-                m0 = WrapperNLPModel(m)
+                        m, vars0, cons0 = exa_model(nothing, args)
+                        m0 = WrapperNLPModel(m)
 
-                m, vars2, cons2 = jump_model(nothing, args)
-                m2 = MathOptNLPModel(m)
+                        m, vars2, cons2 = jump_model(nothing, args)
+                        m2 = MathOptNLPModel(m)
+                        
+                        set_optimizer(m, MadNLP.Optimizer)
+                        set_optimizer_attribute(m, "print_level", MadNLP.ERROR)
+                        optimize!(m)
 
-                set_optimizer(m, MadNLP.Optimizer)
-                set_optimizer_attribute(m, "print_level", MadNLP.ERROR)
-                optimize!(m)
-
-                for backend in BACKENDS
-                    @testset "$backend" begin
 
                         m, vars1, cons1 = exa_model(backend, args)
                         m1 = WrapperNLPModel(m)
@@ -164,11 +164,21 @@ function runtests()
                                 end
                             end
                         end
-
                         result1 = madnlp(m1; print_level = MadNLP.ERROR)
                         test_api(result1, vars1, cons1, vars2, cons2)
                     end
                 end
+
+                m, vars0, cons0 = exa_luksan_vlcek_model(nothing, args; M = 2)
+                m3 = WrapperNLPModel(m)
+
+                m, vars2, cons2 = jump_luksan_vlcek_model(nothing, args; M = 2)
+                m4 = MathOptNLPModel(m)
+
+                @testset "Multi-column constraints" begin
+                    test_nlp(m3, m4; full = false)
+                end
+
             end
         end
     end

test/NLPTest/luksan.jl

@@ -1,47 +1,51 @@
-function luksan_vlcek_obj(x, i)
-    return 100 * (x[i-1]^2 - x[i])^2 + (x[i-1] - 1)^2
+function luksan_vlcek_obj(x, i, j)
+    return 100 * (x[i-1, j]^2 - x[i, j])^2 + (x[i-1, j] - 1)^2
 end
 
-function luksan_vlcek_con(x, i)
-    return 3x[i+1]^3 + 2 * x[i+2] - 5 + sin(x[i+1] - x[i+2])sin(x[i+1] + x[i+2]) + 4x[i+1] -
-           x[i]exp(x[i] - x[i+1]) - 3
+function luksan_vlcek_con1(x, i, j)
+    return 3x[i+1, j]^3 + 2 * x[i+2, j] - 5
+end
+function luksan_vlcek_con2(x, i, j)
+    return sin(x[i+1, j] - x[i+2, j])sin(x[i+1, j] + x[i+2, j]) + 4x[i+1, j] -
+        x[i, j]exp(x[i, j] - x[i+1, j]) - 3
 end
 
 function luksan_vlcek_x0(i)
     return mod(i, 2) == 1 ? -1.2 : 1.0
 end
 
-function _exa_luksan_vlcek_model(backend, N)
+function _exa_luksan_vlcek_model(backend, N; M = 1)
 
     c = ExaCore(backend = backend)
-    x = variable(c, N; start = (luksan_vlcek_x0(i) for i = 1:N))
-    s = constraint(c, luksan_vlcek_con(x, i) for i = 1:N-2)
-    objective(c, luksan_vlcek_obj(x, i) for i = 2:N)
+    x = variable(c, N, M; start = [luksan_vlcek_x0(i) for i = 1:N, j=1:M])
+    s = constraint(c, luksan_vlcek_con1(x, i, j) for i = 1:N-2, j=1:M)
+    constraint!(c, s, (i,j) => luksan_vlcek_con1(x, i, j) for i = 1:N-2, j=1:M)
+    objective(c, luksan_vlcek_obj(x, i, j) for i = 2:N, j=1:M)
 
     return ExaModel(c; prod = true), (x,), (s,)
 end
 
-function exa_luksan_vlcek_model(backend, N)
-    m, vars, cons = _exa_luksan_vlcek_model(backend, N)
+function exa_luksan_vlcek_model(backend, N; M = 1)
+    m, vars, cons = _exa_luksan_vlcek_model(backend, N;M =  M)
     return m
 end
 
-function _jump_luksan_vlcek_model(backend, N)
+function _jump_luksan_vlcek_model(backend, N; M = 1)
     jm = JuMP.Model()
 
-    JuMP.@variable(jm, x[i = 1:N], start = mod(i, 2) == 1 ? -1.2 : 1.0)
+    JuMP.@variable(jm, x[i = 1:N, j=1:M], start = mod(i, 2) == 1 ? -1.2 : 1.0)
     JuMP.@NLconstraint(
         jm,
-        s[i = 1:N-2],
-        3x[i+1]^3 + 2x[i+2] - 5 + sin(x[i+1] - x[i+2])sin(x[i+1] + x[i+2]) + 4x[i+1] -
-        x[i]exp(x[i] - x[i+1]) - 3 == 0.0
+        s[i = 1:N-2, j=1:M],
+        3x[i+1,j]^3 + 2x[i+2,j] - 5 + sin(x[i+1,j] - x[i+2,j])sin(x[i+1,j] + x[i+2,j]) + 4x[i+1,j] -
+        x[i,j]exp(x[i,j] - x[i+1,j]) - 3 == 0.0
     )
-    JuMP.@NLobjective(jm, Min, sum(100(x[i-1]^2 - x[i])^2 + (x[i-1] - 1)^2 for i = 2:N))
+    JuMP.@NLobjective(jm, Min, sum(100(x[i-1,j]^2 - x[i,j])^2 + (x[i-1,j] - 1)^2 for i = 2:N, j=1:M))
 
     return jm, (x,), (s,)
 end
 
-function jump_luksan_vlcek_model(backend, N)
-    jm, vars, cons = _jump_luksan_vlcek_model(backend, N)
+function jump_luksan_vlcek_model(backend, N; M = 1)
+    jm, vars, cons = _jump_luksan_vlcek_model(backend, N; M = M)
     return MathOptNLPModel(jm)
 end