BOX3

src/ADNLPProblems/BOX3.jl

@@ -0,0 +1,11 @@
+export BOX3
+
+function BOX3(args...; n::Int = default_nvar, type::Val{T} = Val(Float64), m::Int = 2n, kwargs...) where {T}
+  m < 3 && @warn("BOX3: must have m ≥ 3")
+  m = max(m, 3)
+
+  x0 = T[0; 10; 20]
+  f(x) = one(T)/2*sum((exp(-one(T)/10*j*x[1]) - exp(-one(T)/10*j*x[2]) -x[3]*(exp(-one(T)/10*j) - exp(-T(j))))^2 for j = 1:m)
+
+  return ADNLPModels.ADNLPModel(f, x0, name = "BOX3"; kwargs...)
+end

src/Meta/BOX3.jl

@@ -0,0 +1,25 @@
+BOX3_meta = Dict(
+  :nvar => 3,
+  :variable_nvar => false,
+  :ncon => 0,
+  :variable_ncon => false,
+  :minimize => true,
+  :name => "BOX3",
+  :has_equalities_only => false,
+  :has_inequalities_only => false,
+  :has_bounds => false,
+  :has_fixed_variables => false,
+  :objtype => :other,
+  :contype => :unconstrained,
+  :best_known_lower_bound => -Inf,
+  :best_known_upper_bound => 662.8684162588787,
+  :is_feasible => true,
+  :defined_everywhere => missing,
+  :origin => :unknown,
+)
+get_BOX3_nvar(; n::Integer = default_nvar, kwargs...) = 3
+get_BOX3_ncon(; n::Integer = default_nvar, kwargs...) = 0
+get_BOX3_nlin(; n::Integer = default_nvar, kwargs...) = 0
+get_BOX3_nnln(; n::Integer = default_nvar, kwargs...) = 0
+get_BOX3_nequ(; n::Integer = default_nvar, kwargs...) = 0
+get_BOX3_nineq(; n::Integer = default_nvar, kwargs...) = 0

src/PureJuMP/BOX3.jl

@@ -0,0 +1,33 @@
+#   Source: problem 12 in
+#   J.J. More', B.S. Garbow and K.E. Hillstrom,
+#   "Testing Unconstrained Optimization Software",
+#   ACM Transactions on Mathematical Software, vol. 7(1), pp. 17-41, 1981.
+
+#   Source: Problem 12 in
+#   J.J. More', B.S. Garbow and K.E. Hillstrom,
+#   "Testing Unconstrained Optimization Software",
+#   ACM Transactions on Mathematical Software, vol. 7(1), pp. 17-41, 1981.
+
+#   See also Buckley#BOX663
+#   SIF input: Ph. Toint, Dec 1989.
+
+#   classification SUR2-AN-3-0
+export BOX3
+
+function BOX3(args...; n::Int = default_nvar, m::Int = 2n, kwargs...)
+  m < 3 && @warn("BOX3: must have m ≥ 3")
+  m = max(m, 3)
+
+  nlp = Model()
+
+  x0 = [0.0; 10.0; 20.0]
+  @variable(nlp, x[i=1:3], start = x0[i])
+
+  @NLobjective(
+  nlp,
+  Min,
+  0.5*sum((exp(-0.1*j*x[1]) - exp(-0.1*j*x[2]) -x[3]*(exp(-0.1*j) - exp(-j)))^2 for j = 1:m))
+
+  return nlp
+end
+