def main(str_optimizer_method_gp, str_mlm_method, str_ms_method, int_bo, int_iter, int_init):
int_dim = 2
bounds = utils_benchmarks.get_bounds(INFO_TARGET, int_dim)
list_Y = []
list_time = []
for ind_bo in range(0, int_bo):
print('BO Iteration', ind_bo)
model_bo = bo.BO(bounds, str_optimizer_method_gp=str_optimizer_method_gp, debug=False)
X_final, Y_final, time_final = utils_bo.optimize_many_with_random_init(model_bo, fun_target, int_init, int_iter, str_initial_method_bo='uniform', str_initial_method_ao='uniform', int_samples_ao=100, str_mlm_method=str_mlm_method, str_modelselection_method=str_ms_method, int_seed=77*(ind_bo+1))
list_Y.append(Y_final)
list_time.append(time_final)
arr_Y = np.array(list_Y)
if int_bo == 1:
arr_Y = np.expand_dims(np.squeeze(arr_Y), axis=0)
else:
arr_Y = np.squeeze(arr_Y)
arr_Y = np.expand_dims(arr_Y, axis=0)
arr_time = np.array(list_time)
arr_time = np.expand_dims(arr_time, axis=0)
print(np.array2string(arr_Y, separator=','))
print(np.array2string(arr_time, separator=','))
utils_plotting.plot_minimum(arr_Y, [STR_FUN_TARGET], int_init, True, path_save=None, str_postfix=None)
utils_plotting.plot_minimum_time(arr_time, arr_Y, [STR_FUN_TARGET], int_init, True, path_save=None, str_postfix=None)
return arr_Y, arr_time
def test_optimize_str_modelselection_method():
np.random.seed(42)
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range_1.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range_1, str_modelselection_method='loocv')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps, times = model_bo.optimize(
X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(times, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def main():
# (alpha, )
num_init = 1
model_bo = bo.BO(np.array([[0.1, 2]]), debug=True)
list_Y = []
list_time = []
for _ in range(0, 10):
X_final, Y_final, time_final, _, _ = wrappers_bo_function.run_single_round(
model_bo,
fun_target,
num_init,
10,
str_initial_method_bo='sobol',
str_sampling_method_ao='sobol',
num_samples_ao=100)
list_Y.append(Y_final)
list_time.append(time_final)
arr_Y = np.array(list_Y)
arr_Y = np.expand_dims(np.squeeze(arr_Y), axis=0)
arr_time = np.array(list_time)
arr_time = np.expand_dims(arr_time, axis=0)
utils_plotting.plot_minimum_vs_iter(arr_Y, ['ridge'],
num_init,
True,
path_save=PATH_SAVE,
str_postfix='ridge')
utils_plotting.plot_minimum_vs_time(arr_time,
arr_Y, ['ridge'],
num_init,
True,
path_save=PATH_SAVE,
str_postfix='ridge')
def main():
int_bo = 5
int_iter = 50
int_init = 3
int_dim = 2
bounds = utils_benchmarks.get_bounds(INFO_TARGET, int_dim)
model_bo = bo.BO(bounds, debug=True)
list_Y = []
list_time = []
for ind_bo in range(0, int_bo):
print('BO Iteration', ind_bo)
X_final, Y_final, time_final = utils_bo.optimize_many_with_random_init(model_bo, fun_target, int_init, int_iter, str_initial_method_bo='uniform', str_initial_method_ao='uniform', int_samples_ao=100)
print(X_final)
print(Y_final)
print(time_final)
list_Y.append(Y_final)
list_time.append(time_final)
arr_Y = np.array(list_Y)
arr_Y = np.expand_dims(np.squeeze(arr_Y), axis=0)
arr_time = np.array(list_time)
arr_time = np.expand_dims(arr_time, axis=0)
utils_plotting.plot_minimum(arr_Y, [STR_FUN_TARGET], int_init, True, path_save=PATH_SAVE, str_postfix=STR_FUN_TARGET)
utils_plotting.plot_minimum_time(arr_time, arr_Y, [STR_FUN_TARGET], int_init, True, path_save=PATH_SAVE, str_postfix=STR_FUN_TARGET)
def main():
# (max_depth, n_estimators)
int_init = 3
model_bo = bo.BO(np.array([[1, 10], [100, 500]]), debug=True)
list_Y = []
list_time = []
for _ in range(0, 5):
X_final, Y_final, time_final = utils_bo.optimize_many_with_random_init(
model_bo,
fun_target,
int_init,
20,
str_initial_method_bo='uniform',
str_initial_method_ao='uniform',
int_samples_ao=100)
list_Y.append(Y_final)
list_time.append(time_final)
arr_Y = np.array(list_Y)
arr_Y = np.expand_dims(np.squeeze(arr_Y), axis=0)
arr_time = np.array(list_time)
arr_time = np.expand_dims(arr_time, axis=0)
utils_plotting.plot_minimum(arr_Y, ['xgboost'],
int_init,
True,
path_save=PATH_SAVE,
str_postfix='xgboost')
utils_plotting.plot_minimum_time(arr_time,
arr_Y, ['xgboost'],
int_init,
True,
path_save=PATH_SAVE,
str_postfix='xgboost')
def test_optimize_str_optimize_method_bo():
np.random.seed(42)
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range_1.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range_1, str_optimizer_method_bo='L-BFGS-B')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps, times = model_bo.optimize(
X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(times, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
# TODO: add DIRECT test, now it causes an error.
model_bo = bo.BO(arr_range_1, str_optimizer_method_bo='CMA-ES')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps, times = model_bo.optimize(
X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(times, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def main():
num_points = 100
is_fixed_noise = False
bounds = obj_fun.get_bounds()
model_bo = bo.BO(bounds, debug=True)
X_init = model_bo.get_initial('uniform', fun_objective=fun_target, int_samples=num_points)
X_test = bo.get_grid(bounds, 50)
mu, sigma = gp.predict_optimized(X_init, fun_target(X_init), X_test, is_fixed_noise=is_fixed_noise, debug=True)
def test_optimize():
np.random.seed(42)
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range_1.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range_1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(1, Y)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, 1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(np.random.randn(num_X), Y)
with pytest.raises(AssertionError) as error:
model_bo.optimize(np.random.randn(num_X, 1), Y)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, np.random.randn(num_X))
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, np.random.randn(num_X, 2))
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, np.random.randn(3, 1))
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_initial_method=1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_initial_method='abc')
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_mlm_method=1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_mlm_method='abc')
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_modelselection_method=1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_modelselection_method='abc')
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, int_samples='abc')
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, is_normalized='abc')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def main():
str_acq = 'pure_explore'
num_iter = 10
X_train = np.array([
[-5],
[-1],
[1],
[2],
])
num_init = X_train.shape[0]
model_bo = bo.BO(np.array([[-6., 6.]]), str_acq=str_acq, normalize_Y=False)
X_test = np.linspace(-6, 6, 400)
X_test = np.reshape(X_test, (400, 1))
for ind_ in range(1, num_iter + 1):
Y_train = fun_target(X_train)
next_x, dict_info = model_bo.optimize(X_train,
fun_target(X_train),
str_sampling_method='uniform')
cov_X_X = dict_info['cov_X_X']
inv_cov_X_X = dict_info['inv_cov_X_X']
hyps = dict_info['hyps']
mu_test, sigma_test = model_bo.compute_posteriors(
X_train, Y_train, X_test, cov_X_X, inv_cov_X_X, hyps)
acq_test = model_bo.compute_acquisitions(X_test, X_train, Y_train,
cov_X_X, inv_cov_X_X, hyps)
mu_test = np.expand_dims(mu_test, axis=1)
sigma_test = np.expand_dims(sigma_test, axis=1)
acq_test = np.expand_dims(acq_test, axis=1)
X_train = np.vstack((X_train, next_x))
Y_train = fun_target(X_train)
utils_plotting.plot_bo_step(X_train,
Y_train,
X_test,
fun_target(X_test),
mu_test,
sigma_test,
path_save=PATH_SAVE,
str_postfix='bo_{}_'.format(str_acq) +
str(ind_),
num_init=num_init)
utils_plotting.plot_bo_step_with_acq(
X_train,
Y_train,
X_test,
fun_target(X_test),
mu_test,
sigma_test,
acq_test,
path_save=PATH_SAVE,
str_postfix='bo_{}_'.format(str_acq) + str(ind_),
num_init=num_init)
def test_optimize_many():
np.random.seed(42)
arr_range = np.array([
[-5.0, 5.0],
])
dim_X = arr_range.shape[0]
num_X = 3
num_iter = 10
X = np.random.randn(num_X, dim_X)
fun_target = lambda x: x**2 - 2.0 * x + 1.0
model_bo = bo.BO(arr_range)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(1, fun_target, X, num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo, 1, X, num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo, fun_target, 1, num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo, fun_target, X, 1.2)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo, fun_target, np.random.randn(num_X),
num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo,
fun_target,
X,
num_iter,
str_initial_method_ao=1)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo,
fun_target,
X,
num_iter,
str_initial_method_ao='abc')
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many(model_bo,
fun_target,
X,
num_iter,
int_samples_ao='abc')
X_final, Y_final, time_all_final, time_gp_final, time_acq_final = utils_bo.optimize_many(
model_bo, fun_target, X, num_iter)
assert len(X_final.shape) == 2
assert len(Y_final.shape) == 2
assert len(time_all_final.shape) == 1
assert len(time_gp_final.shape) == 1
assert len(time_acq_final.shape) == 1
assert X_final.shape[1] == dim_X
assert X_final.shape[0] == Y_final.shape[0] == time_all_final.shape[
0] == num_X + num_iter
assert Y_final.shape[1] == 1
assert time_gp_final.shape[0] == time_acq_final.shape[0]
def main():
str_acq = 'ei'
num_iter = 10
X_train = np.array([
[-5],
[-1],
[1],
[2],
])
num_init = X_train.shape[0]
model_bo = bo.BO(np.array([[-6., 6.]]), str_acq=str_acq)
X_test = np.linspace(-6, 6, 400)
X_test = np.reshape(X_test, (400, 1))
for ind_ in range(1, num_iter + 1):
Y_train = fun_target(X_train)
next_x, dict_info = model_bo.optimize(X_train,
fun_target(X_train),
str_initial_method_ao='uniform')
cov_X_X = dict_info['cov_X_X']
inv_cov_X_X = dict_info['inv_cov_X_X']
hyps = dict_info['hyps']
mu_test, sigma_test, Sigma_test = gp.predict_test_(
X_train, Y_train, X_test, cov_X_X, inv_cov_X_X, hyps)
acq_test = acquisition.ei(mu_test.flatten(), sigma_test.flatten(),
Y_train)
acq_test = np.expand_dims(acq_test, axis=1)
X_train = np.vstack((X_train, next_x))
Y_train = fun_target(X_train)
utils_plotting.plot_bo_step(X_train,
Y_train,
X_test,
fun_target(X_test),
mu_test,
sigma_test,
path_save=PATH_SAVE,
str_postfix='bo_{}_'.format(str_acq) +
str(ind_),
int_init=num_init)
utils_plotting.plot_bo_step_acq(X_train,
Y_train,
X_test,
fun_target(X_test),
mu_test,
sigma_test,
acq_test,
path_save=PATH_SAVE,
str_postfix='bo_{}_'.format(str_acq) +
str(ind_),
int_init=num_init)
def main():
num_points = 100
is_fixed_noise = False
model_bo = bo.BO(INFO_TARGET.get('bounds'), debug=True)
X_init = model_bo.get_initial('uniform',
fun_objective=fun_target,
int_samples=num_points)
X_test = bo.get_grid(INFO_TARGET.get('bounds'), 50)
mu, sigma = gp.predict_optimized(X_init,
np.expand_dims(fun_target(X_init),
axis=1),
X_test,
is_fixed_noise=is_fixed_noise,
debug=True)
def test_run_single_round_with_all_initial_information():
np.random.seed(42)
arr_range = np.array([
[-5.0, 5.0],
])
dim_X = arr_range.shape[0]
num_X = 3
num_iter = 10
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
fun_target = lambda x: 2.0 * x + 1.0
model_bo = bo.BO(arr_range)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(1, fun_target, X, Y, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, 1, X, Y, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, 1, Y, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, 1, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, Y, 'abc')
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, np.random.randn(num_X), Y, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, np.random.randn(num_X), num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, np.random.randn(2, dim_X), Y, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, np.random.randn(num_X, 2), num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, Y, num_iter, str_sampling_method_ao=1)
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, Y, num_iter, str_sampling_method_ao='abc')
with pytest.raises(AssertionError) as error:
package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, Y, num_iter, num_samples_ao='abc')
X_final, Y_final, time_all_final, time_surrogate_final, time_acq_final = package_target.run_single_round_with_all_initial_information(model_bo, fun_target, X, Y, num_iter)
assert len(X_final.shape) == 2
assert len(Y_final.shape) == 2
assert len(time_all_final.shape) == 1
assert len(time_surrogate_final.shape) == 1
assert len(time_acq_final.shape) == 1
assert X_final.shape[1] == dim_X
assert X_final.shape[0] == Y_final.shape[0] == num_X + num_iter
assert time_all_final.shape[0] == num_iter
assert Y_final.shape[1] == 1
assert time_surrogate_final.shape[0] == time_acq_final.shape[0]
def test_get_initial():
np.random.seed(42)
arr_range = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
fun_objective = lambda X: np.sum(X)
model_bo = bo.BO(arr_range)
with pytest.raises(AssertionError) as error:
model_bo.get_initial(1)
with pytest.raises(AssertionError) as error:
model_bo.get_initial('grid', fun_objective=None)
with pytest.raises(AssertionError) as error:
model_bo.get_initial('uniform', fun_objective=1)
with pytest.raises(AssertionError) as error:
model_bo.get_initial('uniform', int_samples='abc')
with pytest.raises(AssertionError) as error:
model_bo.get_initial('uniform', int_seed='abc')
with pytest.raises(NotImplementedError) as error:
model_bo.get_initial('abc')
arr_initials = model_bo.get_initial('grid', fun_objective=fun_objective)
truth_arr_initials = np.array([
[0.0, -2.0, -5.0],
])
assert (np.abs(arr_initials - truth_arr_initials) < TEST_EPSILON).all()
arr_initials = model_bo.get_initial('sobol', int_samples=3)
arr_initials = model_bo.get_initial('sobol', int_samples=3, int_seed=42)
truth_arr_initials = np.array([
[4.84375, 1.3125, 0.46875],
[3.59375, -0.1875, -0.78125],
[8.59375, 1.8125, 4.21875],
])
assert (np.abs(arr_initials - truth_arr_initials) < TEST_EPSILON).all()
arr_initials = model_bo.get_initial('uniform', int_samples=3, int_seed=42)
truth_arr_initials = np.array([
[3.74540119, 1.80285723, 2.31993942],
[5.98658484, -1.37592544, -3.4400548],
[0.58083612, 1.46470458, 1.01115012],
])
assert (np.abs(arr_initials - truth_arr_initials) < TEST_EPSILON).all()
def main():
num_points = 100
fix_noise = False
bounds = obj_fun.get_bounds()
model_bo = bo.BO(bounds, debug=True)
X_init = model_bo.get_initials('uniform', num_points)
X_test = utils_common.get_grids(bounds, 50)
mu, sigma, Sigma = gp.predict_with_optimized_hyps(
X_init,
fun_target(X_init),
X_test,
str_optimizer_method='Nelder-Mead',
fix_noise=fix_noise,
debug=True)
def main():
num_bo = 5
num_iter = 10
num_init = 5
bounds = obj_fun.get_bounds()
model_bo = bo.BO(bounds, debug=True)
list_Y = []
list_time = []
for ind_bo in range(0, num_bo):
print('BO Round', ind_bo + 1)
X_final, Y_final, time_final, _, _ = wrappers.run_single_round(
model_bo,
fun_target,
num_init,
num_iter,
str_initial_method_bo='gaussian',
str_sampling_method_ao='gaussian',
num_samples_ao=100,
seed=42 * (ind_bo + 1))
print(X_final)
print(Y_final)
print(time_final)
list_Y.append(Y_final)
list_time.append(time_final)
bx_best, y_best = utils_bo.get_best_acquisition_by_history(
X_final, Y_final)
print(bx_best, y_best)
arr_Y = np.array(list_Y)
arr_Y = np.expand_dims(np.squeeze(arr_Y), axis=0)
arr_time = np.array(list_time)
arr_time = np.expand_dims(arr_time, axis=0)
utils_plotting.plot_minimum_vs_iter(arr_Y, [STR_FUN_TARGET],
num_init,
True,
path_save=PATH_SAVE,
str_postfix=STR_FUN_TARGET)
utils_plotting.plot_minimum_vs_time(arr_time,
arr_Y, [STR_FUN_TARGET],
num_init,
True,
path_save=PATH_SAVE,
str_postfix=STR_FUN_TARGET)
def main():
num_iter = 10
X_train = np.array([
[-5],
[-1],
[1],
[2],
])
num_init = X_train.shape[0]
model_bo = bo.BO(np.array([[-6., 6.]]))
X_test = np.linspace(-6, 6, 400)
X_test = np.reshape(X_test, (400, 1))
for ind_ in range(1, num_iter + 1):
Y_train = fun_target(X_train)
next_x, _, _, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(
X_train, fun_target(X_train), str_initial_method='uniform')
mu_test, sigma_test = gp.predict_test_(X_train, Y_train, X_test,
cov_X_X, inv_cov_X_X, hyps)
acq_test = acquisition.ei(mu_test.flatten(), sigma_test.flatten(),
Y_train)
acq_test = np.expand_dims(acq_test, axis=1)
X_train = np.vstack((X_train, next_x))
Y_train = fun_target(X_train)
utils_plotting.plot_bo_step(X_train,
Y_train,
X_test,
fun_target(X_test),
mu_test,
sigma_test,
path_save=PATH_SAVE,
str_postfix='basics_bo_' + str(ind_),
int_init=num_init)
utils_plotting.plot_bo_step_acq(X_train,
Y_train,
X_test,
fun_target(X_test),
mu_test,
sigma_test,
acq_test,
path_save=PATH_SAVE,
str_postfix='basics_bo_' + str(ind_),
int_init=num_init,
is_acq_axis_small=True)
def test_run_single_round():
np.random.seed(42)
arr_range = np.array([
[-5.0, 5.0],
])
dim_X = arr_range.shape[0]
num_X = 3
num_iter = 10
fun_target = lambda x: x**2 - 2.0 * x + 1.0
model_bo = bo.BO(arr_range, debug=True)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(1, fun_target, num_X, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, 1, num_X, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, 1.2, num_iter)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, 1.2)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, num_iter, str_initial_method_bo=1)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, num_iter, str_initial_method_bo='abc')
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, num_iter, str_initial_method_bo='grid')
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, num_iter, str_sampling_method_ao=1)
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, num_iter, str_sampling_method_ao='abc')
with pytest.raises(AssertionError) as error:
package_target.run_single_round(model_bo, fun_target, num_X, num_iter, seed=1.2)
X_final, Y_final, time_all_final, time_surrogate_final, time_acq_final = package_target.run_single_round(model_bo, fun_target, num_X, num_iter, str_initial_method_bo='uniform')
assert len(X_final.shape) == 2
assert len(Y_final.shape) == 2
assert len(time_all_final.shape) == 1
assert len(time_surrogate_final.shape) == 1
assert len(time_acq_final.shape) == 1
assert X_final.shape[1] == dim_X
assert X_final.shape[0] == Y_final.shape[0] == time_all_final.shape[0] == num_X + num_iter
assert Y_final.shape[1] == 1
assert time_surrogate_final.shape[0] == time_acq_final.shape[0]
def test_optimize_str_modelselection_method():
np.random.seed(42)
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range_1.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range_1, str_modelselection_method='loocv')
next_point, dict_info = model_bo.optimize(X, Y)
next_points = dict_info['next_points']
acquisitions = dict_info['acquisitions']
cov_X_X = dict_info['cov_X_X']
inv_cov_X_X = dict_info['inv_cov_X_X']
hyps = dict_info['hyps']
time_overall = dict_info['time_overall']
time_gp = dict_info['time_gp']
time_acq = dict_info['time_acq']
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(time_overall, float)
assert isinstance(time_gp, float)
assert isinstance(time_acq, float)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def test_load_bo():
# legitimate cases
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
arr_range_2 = np.array([
[0.0, 10.0],
[2.0, 2.0],
[5.0, 5.0],
])
# wrong cases
arr_range_3 = np.array([
[20.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
arr_range_4 = np.array([
[20.0, 10.0],
[4.0, 2.0],
[10.0, 5.0],
])
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(np.arange(0, 10))
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_3)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_4)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_cov=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_cov='abc')
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_acq=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_acq='abc')
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, is_ard='abc')
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, is_ard=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, prior_mu=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_optimizer_method_gp=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_optimizer_method_gp='abc')
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_optimizer_method_bo=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_optimizer_method_bo='abc')
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_modelselection_method=1)
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, str_modelselection_method='abc')
with pytest.raises(AssertionError) as error:
model_bo = bo.BO(arr_range_1, debug=1)
model_bo = bo.BO(arr_range_1)
model_bo = bo.BO(arr_range_2)
def _acquire_next_point(str_optimizer_method_bo, num_samples):
model_bo = bo.BO(ranges, str_cov='matern52', str_acq='ei', str_optimizer_method_bo=str_optimizer_method_bo)
next_point, _ = model_bo.optimize(X_train, Y_train, num_samples=num_samples)
return next_point
def _get_model_bo(str_optimizer_method_bo):
model_bo = bo.BO(arr_range,
str_cov='matern52',
str_acq='ei',
str_optimizer_method_bo=str_optimizer_method_bo)
return model_bo
def test_optimize_is_normalized():
np.random.seed(42)
arr_range = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range.shape[0]
num_X = 1
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range, str_acq='ei', is_normalized=True)
next_point, dict_info = model_bo.optimize(X, Y)
next_points = dict_info['next_points']
acquisitions = dict_info['acquisitions']
cov_X_X = dict_info['cov_X_X']
inv_cov_X_X = dict_info['inv_cov_X_X']
hyps = dict_info['hyps']
time_overall = dict_info['time_overall']
time_gp = dict_info['time_gp']
time_acq = dict_info['time_acq']
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(time_overall, float)
assert isinstance(time_gp, float)
assert isinstance(time_acq, float)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
X = np.array([
[3.0, 0.0, 1.0],
[2.0, -1.0, 4.0],
[9.0, 1.5, 3.0],
])
Y = np.array([
[100.0],
[100.0],
[100.0],
])
model_bo = bo.BO(arr_range, str_acq='ei', is_normalized=True)
next_point, dict_info = model_bo.optimize(X, Y)
next_points = dict_info['next_points']
acquisitions = dict_info['acquisitions']
cov_X_X = dict_info['cov_X_X']
inv_cov_X_X = dict_info['inv_cov_X_X']
hyps = dict_info['hyps']
time_overall = dict_info['time_overall']
time_gp = dict_info['time_gp']
time_acq = dict_info['time_acq']
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(time_overall, float)
assert isinstance(time_gp, float)
assert isinstance(time_acq, float)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def test_optimize():
np.random.seed(42)
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range_1.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range_1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(1, Y)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, 1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(np.random.randn(num_X), Y)
with pytest.raises(AssertionError) as error:
model_bo.optimize(np.random.randn(num_X, 1), Y)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, np.random.randn(num_X))
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, np.random.randn(num_X, 2))
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, np.random.randn(3, 1))
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_initial_method_ao=1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_initial_method_ao='abc')
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_mlm_method=1)
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, str_mlm_method='abc')
with pytest.raises(AssertionError) as error:
model_bo.optimize(X, Y, int_samples='abc')
next_point, dict_info = model_bo.optimize(X, Y)
next_points = dict_info['next_points']
acquisitions = dict_info['acquisitions']
cov_X_X = dict_info['cov_X_X']
inv_cov_X_X = dict_info['inv_cov_X_X']
hyps = dict_info['hyps']
time_overall = dict_info['time_overall']
time_gp = dict_info['time_gp']
time_acq = dict_info['time_acq']
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert isinstance(time_overall, float)
assert isinstance(time_gp, float)
assert isinstance(time_acq, float)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def test_optimize_str_acq():
np.random.seed(42)
arr_range_1 = np.array([
[0.0, 10.0],
[-2.0, 2.0],
[-5.0, 5.0],
])
dim_X = arr_range_1.shape[0]
num_X = 5
X = np.random.randn(num_X, dim_X)
Y = np.random.randn(num_X, 1)
model_bo = bo.BO(arr_range_1, str_acq='pi')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
model_bo = bo.BO(arr_range_1, str_acq='ucb')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
model_bo = bo.BO(arr_range_1, str_acq='aei')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
model_bo = bo.BO(arr_range_1, str_acq='pure_exploit')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
model_bo = bo.BO(arr_range_1, str_acq='pure_explore')
next_point, next_points, acquisitions, cov_X_X, inv_cov_X_X, hyps = model_bo.optimize(X, Y)
assert isinstance(next_point, np.ndarray)
assert isinstance(next_points, np.ndarray)
assert isinstance(acquisitions, np.ndarray)
assert isinstance(cov_X_X, np.ndarray)
assert isinstance(inv_cov_X_X, np.ndarray)
assert isinstance(hyps, dict)
assert len(next_point.shape) == 1
assert len(next_points.shape) == 2
assert len(acquisitions.shape) == 1
assert next_point.shape[0] == dim_X
assert next_points.shape[1] == dim_X
assert next_points.shape[0] == acquisitions.shape[0]
def test_optimize_many_with_random_init():
np.random.seed(42)
arr_range = np.array([
[-5.0, 5.0],
])
dim_X = arr_range.shape[0]
num_X = 3
num_iter = 10
fun_target = lambda x: x**2 - 2.0 * x + 1.0
model_bo = bo.BO(arr_range)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(1, fun_target, num_X, num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo, 1, num_X, num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo, fun_target, 1.2,
num_iter)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo, fun_target, num_X,
1.2)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo,
fun_target,
num_X,
num_iter,
str_initial_method_bo=1)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo,
fun_target,
num_X,
num_iter,
str_initial_method_bo='abc')
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo,
fun_target,
num_X,
num_iter,
str_initial_method_bo='grid')
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo,
fun_target,
num_X,
num_iter,
str_initial_method_ao=1)
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo,
fun_target,
num_X,
num_iter,
str_initial_method_ao='abc')
with pytest.raises(AssertionError) as error:
utils_bo.optimize_many_with_random_init(model_bo,
fun_target,
num_X,
num_iter,
int_seed=1.2)
X_final, Y_final, time_final = utils_bo.optimize_many_with_random_init(
model_bo, fun_target, num_X, num_iter, str_initial_method_bo='uniform')
assert len(X_final.shape) == 2
assert len(Y_final.shape) == 2
assert len(time_final.shape) == 1
assert X_final.shape[1] == dim_X
assert X_final.shape[0] == Y_final.shape[0] == time_final.shape[
0] == num_X + num_iter
assert Y_final.shape[1] == 1
