SimonBlanke / Gradient-Free-Optimizers

    assert diff_time2 < diff_time1 * 0.5

@pytest.mark.parametrize(*optimizers_smbo)

def test_memory_warm_start_1(Optimizer_smbo):

    def objective_function(para):

        time.sleep(0.5)

        score = -para["x1"] * para["x1"]

        return score

    search_space = {

        "x1": np.arange(0, 1, 1),

    }

    n_iter = 2

    c_time1 = time.time()

    opt0 = GridSearchOptimizer(search_space)

    opt0.search(objective_function, n_iter=n_iter)

    diff_time1 = time.time() - c_time1

    c_time2 = time.time()

    opt1 = Optimizer_smbo(search_space)

    opt1.search(objective_function, n_iter=n_iter, memory_warm_start=opt0.search_data)

    diff_time2 = time.time() - c_time2

    print("\n diff_time1 ", diff_time1)

    print("\n diff_time2 ", diff_time2)

    assert diff_time2 < diff_time1 * 0.5

from ._parametrize import optimizers_non_smbo, optimizers_smbo

@pytest.mark.parametrize(*optimizers_non_smbo)

def test_memory_warm_start_0(Optimizer_non_smbo):

    def objective_function(para):

        time.sleep(0.1)

        score = -para["x1"] * para["x1"]

        return score

    search_space = {

        "x1": np.arange(0, 10, 1),

    }

    n_iter = 20

    c_time1 = time.time()

    opt0 = GridSearchOptimizer(search_space)

    opt0.search(objective_function, n_iter=n_iter)

    diff_time1 = time.time() - c_time1

    c_time2 = time.time()

    opt1 = Optimizer_non_smbo(search_space)

    opt1.search(objective_function, n_iter=n_iter, memory_warm_start=opt0.search_data)

    diff_time2 = time.time() - c_time2

    print("\n diff_time1 ", diff_time1)

    print("\n diff_time2 ", diff_time2)

    assert diff_time2 < diff_time1 * 0.5

@pytest.mark.parametrize(*optimizers_smbo)