parser.add_argument("--situation", "-s", required=True, help="Situation")
args = parser.parse_args()
csv_dir = './../csv_files/latent_vectors/' + args.situation + "/"
if os.path.exists(csv_dir + args.output_dir) is False:
print("対象のファイルが見つかりません。")
exit()
original_data_path = csv_dir + args.old_dir + "/latent_vectors.csv"
parents_index_path = csv_dir + args.output_dir + "/parents_index.csv"
children_index_path = csv_dir + args.output_dir + "/children_index.csv"
children_path = csv_dir + args.output_dir + "/children.csv"
original_data = functions.read_csv(original_data_path)
parents_index = functions.read_csv(parents_index_path)
parents_index = functions.transform_to_int(parents_index)
parents_index = parents_index[0]
children_index = functions.read_csv(children_index_path)
children_index = functions.transform_to_int(children_index)
children_index = children_index[0]
children = functions.read_csv(children_path)
children = functions.transform_to_float(children)
new_data = functions.replace(original_data, children, parents_index,
children_index)
# 新しい潜在ベクトル群をcsvファイルに書き込む
functions.write_csv(csv_dir + args.output_dir + "/latent_vectors.csv",
new_data)
write_filename = 'MGGimprove/children'
# 実行回数
num_execute = int(5000/300)
# 潜在空間の次元数
num_dimentions = 100
# 局所解ファイル
solutions_file = 'pre_experiment/mock_solutions_100'
# 評価結果のファイル
result_file = 'MGGimprove/evaluation_result'
# 局所解ファイルの読み込み
solutions_data = functions.read_csv(solutions_file)
del solutions_data[0]
solutions_data = functions.transform_to_float(solutions_data)
# 局所解とバイアスに分ける
solutions, bias = functions.divide_solutions_bias(solutions_data)
# 評価値の結果のリスト
evaluations_result = []
for num_experiment in range(1, 101):
print(num_experiment)
# 対象のデータの読み込み
data = functions.read_csv(read_filename)
del data[0]
data = functions.transform_to_float(data)
# before_index = functions.get_result(data, functions.get_evaluations_list(data, solutions, bias), num_experiment, functions.get_best_solution_index(bias), solutions)[2]
# 次の世代の作成
parser.add_argument("--output_dir", "-o", required=True, help="Directory to output generated csv files to")
parser.add_argument("--situation", "-s", required=True, help="Situation")
parser.add_argument("--num_parent", "-p", default=4, type=int, help="Number of parents")
parser.add_argument("--num_children", "-c", default=16, type=int, help="Number of children")
args = parser.parse_args()
csv_dir = './../csv_files/latent_vectors/' + args.situation + "/"
if os.path.exists(csv_dir + args.output_dir) is True:
exit()
if os.path.exists(csv_dir + args.output_dir) is False:
os.mkdir(csv_dir + args.output_dir)
input_path = csv_dir + args.input_dir + '/latent_vectors.csv'
latent_vectors = functions.read_csv(input_path)
latent_vectors = functions.transform_to_float(latent_vectors)
latent_vectors = np.array(latent_vectors)
num_vectors = len(latent_vectors)
parents_index = functions.make_random_list(num_vectors, args.num_parent)
parent_vectors = np.zeros((args.num_parent, len(latent_vectors[0])))
for index in range(len(parents_index)):
parent_vectors[index] = np.copy(latent_vectors[parents_index[index]])
children = functions.make_children(parent_vectors, args.num_children)
# 親に使ったベクトルのインデックスのcsvファイルを作成
functions.write_csv_for_vector(csv_dir + args.output_dir + "/parents_index.csv", parents_index)
# 親に使ったベクトルのcsvファイルを作成
functions.write_csv(csv_dir + args.output_dir + "/parents.csv", parent_vectors)