Now samplers return a tuple which is out of the box compatible with s…

…urrogates and optimization methods

src/Sampling.jl

@@ -16,7 +16,7 @@ function sample(n,lb,ub,S::GridSample)
     else
         d = length(lb)
         x = [[rand(lb[j]:dx[j]:ub[j]) for j = 1:d] for i in 1:n]
-        return x
+        return Tuple.(x)
     end
 end
 
@@ -30,7 +30,7 @@ function sample(n,lb,ub,::UniformSample)
     else
         d = length(lb)
         x = [[rand(Uniform(lb[j],ub[j])) for j in 1:d] for i in 1:n]
-        return x
+        return Tuple.(x)
     end
 end
 
@@ -45,7 +45,7 @@ function sample(n,lb,ub,::SobolSample)
     if lb isa Number
         return [next!(s)[1] for i = 1:n]
     else
-        return [next!(s) for i = 1:n]
+        return Tuple.([next!(s) for i = 1:n])
     end
 end
 
@@ -66,7 +66,7 @@ function sample(n,lb,ub,::LatinHypercubeSample)
         @inbounds for c = 1:d
             x[:,c] = (ub[c]-lb[c])*x[:,c]/n .+ lb[c]
         end
-        return x
+        return Tuple.(x)
     end
 end
 
@@ -115,6 +115,6 @@ function sample(n,lb,ub,S::LowDiscrepancySample)
         @inbounds for c = 1:d
             x[:,c] = (ub[c]-lb[c])*x[:,c] .+ lb[c]
         end
-        return x
+        return Tuple.(x)
     end
 end

test/SVMSurrogate.jl

@@ -16,8 +16,7 @@ add_point!(my_svm_1D,[3.2,3.5],[7.4,8.0])
 obj_N = x -> x[1]^2*x[2]
 lb = [0.0,0.0]
 ub = [10.0,10.0]
-s = sample(100,lb,ub,UniformSample())
-x = Tuple.(s)
+x = sample(100,lb,ub,UniformSample())
 y = obj_N.(x)
 my_svm_ND = SVMSurrogate(x,y,lb,ub)
 val = my_svm_ND((5.0,1.2))

test/linearSurrogate.jl

@@ -13,8 +13,7 @@ add_point!(my_linear_surr_1D,[5.0,6.0],[8.3,9.7])
 #ND
 lb = [0.0,0.0]
 ub = [10.0,10.0]
-s = sample(5,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(5,lb,ub,SobolSample())
 y = [4.0,5.0,6.0,7.0,8.0]
 my_linear_ND = LinearSurrogate(x,y,lb,ub)
 add_point!(my_linear_ND,(10.0,11.0),9.0)

test/lobachesky.jl

@@ -30,8 +30,7 @@ lb = [0.0,0.0]
 ub = [8.0,8.0]
 alpha = 2.4
 n = 8
-s = sample(3200,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(3200,lb,ub,SobolSample())
 y = obj.(x)
 my_loba_ND = LobacheskySurrogate(x,y,alpha,n,lb,ub)
 

test/neuralSurrogate.jl

@@ -22,8 +22,7 @@ println(val)
 
 lb = [0.0,0.0]
 ub = [5.0,5.0]
-s = sample(5,lb,ub, SobolSample())
-x = Tuple.(s)
+x = sample(5,lb,ub, SobolSample())
 obj_ND_neural(x) = x[1]*x[2];
 y = obj_ND_neural.(x)
 model = Chain(Dense(2,1))

test/optimization.jl

@@ -47,8 +47,7 @@ my_rad_SRBFN = RadialBasis(x,y,bounds,z->norm(z),1)
 surrogate_optimize(objective_function_ND,SRBF(),lb,ub,my_rad_SRBFN,UniformSample())
 
 # Lobachesky
-s = sample(5,lb,ub,UniformSample())
-x = Tuple.(s)
+x = sample(5,lb,ub,UniformSample())
 y = objective_function_ND.(x)
 alpha = 2.0
 n = 4
@@ -58,8 +57,7 @@ surrogate_optimize(objective_function_ND,SRBF(),lb,ub,my_loba_ND,UniformSample()
 #Linear
 lb = [1.0,1.0]
 ub = [6.0,6.0]
-s = sample(500,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(500,lb,ub,SobolSample())
 objective_function_ND = z -> 3*norm(z)+1
 y = objective_function_ND.(x)
 my_linear_ND = LinearSurrogate(x,y,lb,ub)
@@ -68,8 +66,7 @@ surrogate_optimize(objective_function_ND,SRBF(),lb,ub,my_linear_ND,SobolSample()
 #SVM
 lb = [1.0,1.0]
 ub = [6.0,6.0]
-s = sample(5,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(5,lb,ub,SobolSample())
 objective_function_ND = z -> 3*norm(z)+1
 y = objective_function_ND.(x)
 my_SVM_ND = SVMSurrogate(x,y,lb,ub)
@@ -78,8 +75,7 @@ surrogate_optimize(objective_function_ND,SRBF(),lb,ub,my_SVM_ND,SobolSample(),ma
 #Neural
 lb = [1.0,1.0]
 ub = [6.0,6.0]
-s = sample(5,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(5,lb,ub,SobolSample())
 objective_function_ND = z -> 3*norm(z)+1
 y = objective_function_ND.(x)
 model = Chain(Dense(2,1))
@@ -92,8 +88,7 @@ surrogate_optimize(objective_function_ND,SRBF(),lb,ub,my_neural_ND_neural,SobolS
 #Random Forest
 lb = [1.0,1.0]
 ub = [6.0,6.0]
-s = sample(5,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(5,lb,ub,SobolSample())
 objective_function_ND = z -> 3*norm(z)+1
 y = objective_function_ND.(x)
 num_round = 2

test/random_forest.jl

@@ -15,8 +15,7 @@ add_point!(my_forest_1D,[7.0,8.0],obj_1D.([7.0,8.0]))
 #ND
 lb = [0.0,0.0]
 ub = [10.0,10.0]
-s = sample(5,lb,ub,SobolSample())
-x = Tuple.(s)
+x = sample(5,lb,ub,SobolSample())
 obj_ND = x -> x[1] * x[2]^2
 y = obj_ND.(x)
 my_forest_ND = RandomForestSurrogate(x,y,lb,ub,num_round)