FYP Project Contributions

EAplayground.py

@@ -1,11 +1,8 @@
 # Evolutionary algorim
 from code_construction.code_construction import CodeConstructor
-import numpy as np
 from evolutionary_algorithm.ea import BivariateBicycleCodeEvolutionaryOptimization
-from pymoo.operators.crossover.sbx import SBX
 from pymoo.operators.mutation.pm import PM
 from pymoo.operators.repair.rounding import RoundingRepair
-from pymoo.operators.sampling.rnd import IntegerRandomSampling
 from pymoo.algorithms.soo.nonconvex.ga import GA
 from pymoo.operators.crossover.ux import UniformCrossover
 from pymoo.optimize import minimize
@@ -18,22 +15,27 @@
 # print(undetectable_error_rate.evaluate(p=0.01))
 from pymoo.core.sampling import Sampling
 import sys
+
+
 if len(sys.argv) == 3:
-        seed= int(sys.argv[1])
-        dataset_index = int(sys.argv[2])
-        lambda_ = 1
-elif len(sys.argv) >3:
-    seed= int(sys.argv[1])
+    seed = int(sys.argv[1])
+    dataset_index = int(sys.argv[2])
+    lambda_ = 1
+elif len(sys.argv) > 3:
+    seed = int(sys.argv[1])
     dataset_index = int(sys.argv[2])
     lambda_ = float(sys.argv[3])
 else:
     seed = 42
     dataset_index = 0
     lambda_ = 1
 
-l=12
-g=6
-print(f'(l,g)=({l},{g}), dataset_index = {dataset_index}, seed={seed}, lambda = {lambda_}')
+l = 12
+g = 6
+print(
+    f"(l,g)=({l},{g}), dataset_index = {dataset_index}, seed={seed}, lambda = {lambda_}"
+)
+
 
 class MySampling(Sampling):
     def __init__(self, init_samples):
@@ -42,38 +44,47 @@ def __init__(self, init_samples):
 
     def _do(self, problem, n_samples, **kwargs):
         return self.init_samples
-para_dict = {'l':l,'g':g}
-code_class = 'bb'
 
 
-code_constructor = CodeConstructor(method=code_class,para_dict = para_dict)
+para_dict = {"l": l, "g": g}
+code_class = "bb"
+
+
+code_constructor = CodeConstructor(method=code_class, para_dict=para_dict)
 # define objective function
-pp=0.05
-Obj_Func = ObjectiveFunction(code_constructor,lambda_=lambda_, pp=pp,decoder_param={'trail':10_000})
+pp = 0.05
+Obj_Func = ObjectiveFunction(
+    code_constructor, lambda_=lambda_, pp=pp, decoder_param={"trail": 10_000}
+)
 obj_func = Obj_Func.forward
-if l ==6 and g==3:
-    init_data_file = f"./data/BO_initial_points/BO_initial_points_{dataset_index}_{lambda_}_63.pkl"
+if l == 6 and g == 3:
+    init_data_file = (
+        f"./data/BO_initial_points/BO_initial_points_{dataset_index}_{lambda_}_63.pkl"
+    )
 else:
-    init_data_file = f"./data/BO_initial_points/BO_initial_points_{dataset_index}_{lambda_}.pkl"
+    init_data_file = (
+        f"./data/BO_initial_points/BO_initial_points_{dataset_index}_{lambda_}.pkl"
+    )
 # file with 63 suffix has (l,m)=(6,3). Otherwise (l,m)=(12,6)
 with open(init_data_file, "rb") as f:
     data = pickle.load(f)
-    X_init = data['X']
-    y_init = data['y']
-    pl_init = data['pl']
-problem = BivariateBicycleCodeEvolutionaryOptimization(l=l,m=g,obj_func=Obj_Func)        
+    X_init = data["X"]
+    y_init = data["y"]
+    pl_init = data["pl"]
+problem = BivariateBicycleCodeEvolutionaryOptimization(l=l, m=g, obj_func=Obj_Func)
 algorithm = GA(
     pop_size=20,
     sampling=MySampling(X_init),
     crossover=UniformCrossover(prob=1.0),
     mutation=PM(prob=1.0, eta=3.0, vtype=float, repair=RoundingRepair()),
-    eliminate_duplicates=True
+    eliminate_duplicates=True,
 )
 
-res = minimize(problem,
-            algorithm,
-            termination=('n_gen', 11),
-            seed=seed,
+res = minimize(
+    problem,
+    algorithm,
+    termination=("n_gen", 11),
+    seed=seed,
 )
 flat2 = [-v for row in problem.evaluation_history for v in row]
 
@@ -82,10 +93,12 @@ def _do(self, problem, n_samples, **kwargs):
 print(problem.best_parameters)
 print("Best solution found: \nX = %s\nF = %s" % (res.X, -res.F))
 
-with open(f'./data/BO_results/EA_{l}_{g}_{dataset_index}_{seed}_{lambda_}.pkl','wb') as f:
+with open(
+    f"./data/BO_results/EA_{l}_{g}_{dataset_index}_{seed}_{lambda_}.pkl", "wb"
+) as f:
     results = {
-        'best_x':problem.best_parameters,
-        'best_y':problem.best_result,
-        'evaluation_history':flat2,
+        "best_x": problem.best_parameters,
+        "best_y": problem.best_result,
+        "evaluation_history": flat2,
     }
     pickle.dump(results, f)