def test_max_score_1():
def objective_function(para):
score = -para["x1"] * para["x1"]
time.sleep(0.01)
return score
search_space = {
"x1": np.arange(0, 100, 0.1),
}
max_score = -9999
c_time = time.time()
opt = HillClimbingOptimizer(
search_space, initialize={"warm_start": [{"x1": 99}]}
)
opt.search(
objective_function, n_iter=100000, max_score=max_score,
)
diff_time = time.time() - c_time
print("\n Results head \n", opt.results.head())
print("\n Results tail \n", opt.results.tail())
print("\nN iter:", len(opt.results))
assert diff_time < 1
def test_max_score_0():
def objective_function(para):
score = -para["x1"] * para["x1"]
return score
search_space = {
"x1": np.arange(0, 100, 0.1),
}
max_score = -9999
opt = HillClimbingOptimizer(
search_space,
initialize={"warm_start": [{"x1": 99}]},
epsilon=0.01,
rand_rest_p=0,
)
opt.search(
objective_function, n_iter=100000, max_score=max_score,
)
print("\n Results head \n", opt.results.head())
print("\n Results tail \n", opt.results.tail())
print("\nN iter:", len(opt.results))
assert -100 > opt.best_score > max_score
from sklearn.tree import DecisionTreeClassifier
from sklearn.datasets import load_wine
from gradient_free_optimizers import HillClimbingOptimizer
data = load_wine()
X, y = data.data, data.target
def model(para):
gbc = DecisionTreeClassifier(
min_samples_split=para["min_samples_split"],
min_samples_leaf=para["min_samples_leaf"],
)
scores = cross_val_score(gbc, X, y, cv=5)
return scores.mean()
search_space = {
"min_samples_split": np.arange(2, 25, 1),
"min_samples_leaf": np.arange(1, 25, 1),
}
opt = HillClimbingOptimizer(search_space)
opt.search(model, n_iter=500, memory=False)
print("\n\nMemory activated:")
opt = HillClimbingOptimizer(search_space)
opt.search(model, n_iter=500, memory=True)
import numpy as np
from gradient_free_optimizers import HillClimbingOptimizer
def convex_function(pos_new):
score = -(pos_new["x1"] * pos_new["x1"] + pos_new["x2"] * pos_new["x2"])
return score
search_space = {
"x1": np.arange(-100, 101, 0.1),
"x2": np.arange(-100, 101, 0.1),
}
opt = HillClimbingOptimizer(search_space)
opt.search(convex_function, n_iter=300000)
}),
({
"crossover_rate": 2
}),
({
"population": 1
}),
({
"population": 2
}),
({
"population": 100
}),
({
"population": [
HillClimbingOptimizer(search_space),
HillClimbingOptimizer(search_space),
HillClimbingOptimizer(search_space),
HillClimbingOptimizer(search_space),
]
}),
({
"rand_rest_p": 0
}),
({
"rand_rest_p": 0.5
}),
({
"rand_rest_p": 1
}),
]
import numpy as np
from gradient_free_optimizers import HillClimbingOptimizer
n_iter = 10
def get_score(pos_new):
x1 = pos_new[0]
return -x1 * x1
space_dim = np.array([100])
init_positions = [np.array([10])]
opt = HillClimbingOptimizer(init_positions, space_dim, opt_para={})
for nth_init in range(len(init_positions)):
pos_new = opt.init_pos(nth_init)
score_new = get_score(pos_new)
opt.evaluate(score_new)
for nth_iter in range(len(init_positions), n_iter):
pos_new = opt.iterate(nth_iter)
score_new = get_score(pos_new)
opt.evaluate(score_new)