def test_initialize_vertices():
initialize = {"vertices": 2}
opt = RandomSearchOptimizer(search_space, initialize=initialize)
opt.search(objective_function, n_iter=2)
assert abs(opt.best_score) - 10000 < 0.001
def test_verbosity_4():
opt = RandomSearchOptimizer(search_space)
opt.search(
objective_function,
n_iter=100,
verbosity=["progress_bar", "print_results"],
)
def test_attributes_iter_times_1():
c_time = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
diff_time = time.time() - c_time
assert np.array(opt.iter_times).sum() < diff_time
def test_max_time_1():
c_time1 = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=1000000, max_time=1)
diff_time1 = time.time() - c_time1
assert 0.3 < diff_time1 < 2
def test_attributes_results_5():
opt = RandomSearchOptimizer(search_space,
initialize={"warm_start": [{
"x1": 10
}]})
opt.search(objective_function, n_iter=1)
assert 10 in list(opt.results["x1"].values)
def test_attributes_results_4():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function,
n_iter=1,
initialize={},
warm_start=[{
"x1": 0
}])
assert 0 in list(opt.results["x1"].values)
def test_initialize_grid_0():
search_space = {
"x1": np.arange(-1, 2, 1),
}
initialize = {"grid": 1}
opt = RandomSearchOptimizer(search_space, initialize=initialize)
opt.search(objective_function, n_iter=1)
assert abs(opt.best_score) < 0.001
def test_function():
def objective_function(para):
score = -para["x1"] * para["x1"]
return score
search_space = {
"x1": np.arange(-100, 101, 1),
}
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=30)
def test_obj_func_return_dictionary_0():
def objective_function(para):
score = -para["x1"] * para["x1"]
return score, {"_x1_": para["x1"]}
search_space = {
"x1": np.arange(-100, 101, 1),
}
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=30)
assert "_x1_" in list(opt.results.columns)
def test_initialize_warm_start_0():
init = {
"x1": 0,
}
initialize = {"warm_start": [init]}
opt = RandomSearchOptimizer(search_space, initialize=initialize)
opt.search(objective_function, n_iter=1)
# print("\nself.results \n", opt.results)
assert abs(opt.best_score) < 0.001
def test_memory_timeSave_0():
data = load_breast_cancer()
X, y = data.data, data.target
def objective_function(para):
dtc = DecisionTreeClassifier(
min_samples_split=para["min_samples_split"])
scores = cross_val_score(dtc, X, y, cv=5)
return scores.mean()
search_space = {
"min_samples_split": np.arange(2, 20),
}
c_time1 = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
diff_time1 = time.time() - c_time1
c_time2 = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100, memory=False)
diff_time2 = time.time() - c_time2
assert diff_time1 < diff_time2 * 0.8
def test_initialize_warm_start_2():
search_space = {
"x1": np.arange(-10, 10, 1),
}
init = {
"x1": -10,
}
initialize = {"warm_start": [init], "random": 0, "vertices": 0, "grid": 0}
opt = RandomSearchOptimizer(search_space, initialize=initialize)
opt.search(objective_function, n_iter=1)
assert opt.best_para == init
def test_memory_warm_start():
data = load_breast_cancer()
X, y = data.data, data.target
def objective_function(para):
dtc = DecisionTreeClassifier(
max_depth=para["max_depth"],
min_samples_split=para["min_samples_split"],
)
scores = cross_val_score(dtc, X, y, cv=5)
return scores.mean()
search_space = {
"max_depth": np.arange(1, 10),
"min_samples_split": np.arange(2, 20),
}
c_time1 = time.time()
opt0 = RandomSearchOptimizer(search_space)
opt0.search(objective_function, n_iter=300)
diff_time1 = time.time() - c_time1
c_time2 = time.time()
opt1 = RandomSearchOptimizer(search_space)
opt1.search(objective_function, n_iter=300, memory_warm_start=opt0.results)
diff_time2 = time.time() - c_time2
print("\n opt0.results \n", opt0.results)
assert diff_time2 < diff_time1 * 0.5
def test_memory_warm_start_manual():
data = load_breast_cancer()
X, y = data.data, data.target
def objective_function(para):
dtc = GradientBoostingClassifier(n_estimators=para["n_estimators"], )
scores = cross_val_score(dtc, X, y, cv=5)
return scores.mean()
search_space = {
"n_estimators": np.arange(500, 502),
}
c_time_1 = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=1)
diff_time_1 = time.time() - c_time_1
memory_warm_start = pd.DataFrame([[500, 0.9], [501, 0.91]],
columns=["n_estimators", "score"])
c_time = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function,
n_iter=10,
memory_warm_start=memory_warm_start)
diff_time = time.time() - c_time
assert diff_time_1 > diff_time * 0.3
def test_no_random_state_0():
opt0 = RandomSearchOptimizer(search_space, initialize={"random": 1})
opt0.search(objective_function, n_iter=100)
opt1 = RandomSearchOptimizer(search_space, initialize={"random": 1})
opt1.search(objective_function, n_iter=100)
assert abs(opt0.best_score - opt1.best_score) > err
def test_initialize_warm_start_0():
init = {
"x1": 0,
}
initialize = {"warm_start": [init]}
opt = RandomSearchOptimizer(search_space)
opt.search(
objective_function,
n_iter=1,
initialize=initialize,
)
assert abs(opt.best_score) < 0.001
def test_attributes_results_6():
def objective_function(para):
score = -para["x1"] * para["x1"]
return score
search_space = {
"x1": np.arange(0, 10, 1),
}
opt = RandomSearchOptimizer(search_space, initialize={"random": 1})
opt.search(objective_function, n_iter=20, memory=False)
x1_results = list(opt.results["x1"].values)
print("\n x1_results \n", x1_results)
assert len(set(x1_results)) < len(x1_results)
def test_sklearn():
data = load_iris()
X, y = data.data, data.target
def model(para):
knr = KNeighborsClassifier(n_neighbors=para["n_neighbors"])
scores = cross_val_score(knr, X, y, cv=5)
score = scores.mean()
return score
search_space = {
"n_neighbors": np.arange(1, 51, 1),
}
opt = RandomSearchOptimizer(search_space)
opt.search(model, n_iter=30)
def test_initialize_warm_start_1():
search_space = {
"x1": np.arange(-10, 10, 1),
}
init = {
"x1": -10,
}
initialize = {"warm_start": [init]}
opt = RandomSearchOptimizer(search_space)
opt.search(
objective_function,
n_iter=1,
initialize=initialize,
)
assert opt.best_para == init
def test_memory_timeSave_1():
data = load_breast_cancer()
X, y = data.data, data.target
def objective_function(para):
dtc = DecisionTreeClassifier(max_depth=para["max_depth"])
scores = cross_val_score(dtc, X, y, cv=5)
return scores.mean()
search_space = {
"max_depth": np.arange(1, 101),
}
results = pd.DataFrame(np.arange(1, 101), columns=["max_depth"])
results["score"] = 0
c_time1 = time.time()
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=300, memory_warm_start=results)
diff_time1 = time.time() - c_time1
assert diff_time1 < 1
return np.inf
search_space_ = {"x1": np.arange(0, 20, 1)}
def objective_function(para):
score = -para["x1"] * para["x1"]
return score
search_space = {"x1": np.arange(-10, 11, 1)}
search_space2 = {"x1": np.arange(-10, 51, 1)}
search_space3 = {"x1": np.arange(-50, 11, 1)}
opt1 = RandomSearchOptimizer(search_space)
opt2 = RandomSearchOptimizer(search_space2)
opt3 = RandomSearchOptimizer(search_space3)
opt4 = RandomSearchOptimizer(search_space_)
opt5 = RandomSearchOptimizer(search_space_)
opt6 = RandomSearchOptimizer(search_space_)
opt1.search(objective_function, n_iter=30)
opt2.search(objective_function, n_iter=30)
opt3.search(objective_function, n_iter=30)
opt4.search(objective_function_nan, n_iter=30)
opt5.search(objective_function_m_inf, n_iter=30)
opt6.search(objective_function_inf, n_iter=30)
search_data1 = opt1.results
search_data2 = opt2.results
def test_attributes_best_score_0():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert np.inf > opt.best_score > -np.inf
def test_attributes_best_para_0():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert isinstance(opt.best_para, dict)
def test_attributes_best_para_1():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert list(opt.best_para.keys()) == list(search_space.keys())
def test_attributes_iter_times_0():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert isinstance(opt.iter_times, list)
def test_attributes_results_3():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert "score" in list(opt.results.columns)
def test_attributes_results_1():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert set(search_space.keys()) < set(opt.results.columns)
def test_attributes_results_0():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100)
assert isinstance(opt.results, pd.DataFrame)
import numpy as np
from gradient_free_optimizers import RandomSearchOptimizer
def ackley_function(pos_new):
x = pos_new["x1"]
y = pos_new["x2"]
a1 = -20 * np.exp(-0.2 * np.sqrt(0.5 * (x * x + y * y)))
a2 = -np.exp(0.5 * (np.cos(2 * np.pi * x) + np.cos(2 * np.pi * y)))
score = a1 + a2 + 20
return -score
search_space = {
"x1": np.arange(-100, 101, 0.1),
"x2": np.arange(-100, 101, 0.1),
}
opt = RandomSearchOptimizer(search_space)
opt.search(ackley_function, n_iter=30000)
def test_verbosity_5():
opt = RandomSearchOptimizer(search_space)
opt.search(objective_function, n_iter=100, verbosity=[])
