fix tests

Author: GiggleLiu

test/dragondrop.jl

@@ -14,11 +14,12 @@ using Random
     r1 = solve(GenericTensorNetwork(IndependentSet(graph, weights)), SingleConfigMax())[]
     J2 = vcat([Float64[J[i,j] for j=i+1:n] for i=1:n]...)
     # note the different sign convention of J
-    r2 = solve(GenericTensorNetwork(SpinGlass(complete_graph(n), -J2, H)), SingleConfigMax())[]
+    sg = SpinGlass(complete_graph(n), -J2, H)
+    r2 = solve(GenericTensorNetwork(sg), SingleConfigMax())[]
     @test r1.n - qubo.mis_overhead ≈ r2.n
     @test r1.n % 1 ≈ r2.n % 1
     c1 = map_config_back(qubo, r1.c.data)
-    @test spinglass_energy(complete_graph(n), c1; J=-J2, h=H) ≈ spinglass_energy(complete_graph(n), r2.c.data; J=-J2, h=H)
+    @test GenericTensorNetworks.energy(sg, 1 .- 2 .* Int.(c1)) ≈ GenericTensorNetworks.energy(sg, 1 .- 2 .* Int.(r2.c.data))
     #display(MappingGrid(UnitDiskMapping.CopyLine[], 0, qubo))
 end
 
@@ -56,7 +57,7 @@ end
         add_edge!(g2, (i-1)*n+j, (i2-1)*n+j2)
         push!(weights, J)
     end
-    r2 = solve(GenericTensorNetwork(SpinGlass(g2, -weights)), SingleConfigMax())[]
+    r2 = solve(GenericTensorNetwork(SpinGlass(g2, -weights, zeros(Int, n*n))), SingleConfigMax())[]
     @show r1, r2
 end
 
@@ -88,9 +89,10 @@ end
     for (i,j,h) in onsite
         hs[i+(j-1)*m] = h
     end
-    r2 = solve(GenericTensorNetwork(SpinGlass(g2, -Js, hs)), SingleConfigMax())[]
+    sg = SpinGlass(g2, -Js, hs)
+    r2 = solve(GenericTensorNetwork(sg), SingleConfigMax())[]
     @test r1.n - qubo.mis_overhead ≈ r2.n
     c1 = map_config_back(qubo, collect(Int,r1.c.data))
     c2 = collect(r2.c.data)
-    @test spinglass_energy(g2, c1; J=-Js, h=hs) ≈ spinglass_energy(g2, c2; J=-Js, h=hs)
-end
+    @test GenericTensorNetworks.energy(sg, 1 .- 2 .* Int.(c1)) ≈ GenericTensorNetworks.energy(sg, 1 .- 2 .* Int.(c2))
+end

test/reduceto.jl

@@ -26,14 +26,14 @@ end
         ]
         @info "Testing reduction from $(typeof(source)) to $(target_type)"
         # directly solve
-        best_source = ProblemReductions.findbest(source, ProblemReductions.BruteForce())
+        solver = GTNSolver(optimizer=TreeSA(ntrials=1))
+        best_source = ProblemReductions.findbest(source, source isa ConstraintSatisfactionProblem ? solver : ProblemReductions.BruteForce())
 
         # reduce and solve
         result = ProblemReductions.reduceto(target_type, source)
         target = ProblemReductions.target_problem(result)
         @test target isa target_type
-        #best_target = findbest(target, BruteForce())
-        best_target = GenericTensorNetworks.solve(GenericTensorNetwork(GenericTensorNetworks.IndependentSet(SimpleGraph(target.graph), collect(target.weights))), ConfigsMax())[].c.data
+        best_target = ProblemReductions.findbest(IndependentSet(SimpleGraph(target.graph), collect(target.weights)), solver)
 
         # extract the solution
         best_source_extracted_single = unique( ProblemReductions.extract_solution.(Ref(result), best_target) )