for testing and deploying your application
for finding and fixing issues
for empowering human code reviews
# Author: Simon Blanke
# Email: [email protected]
# License: MIT License
import pytest
import numpy as np
from gradient_free_optimizers import GridSearchOptimizer
def parabola_function(para):
loss = para["x"] * para["x"] + para["y"] * para["y"]
return -loss
search_space = {
"x": np.arange(-100, 100, 1),
"y": np.arange(-100, 100, 1),
}
n_iter = 50
def test_direction_0():
n_initialize = 1
opt = GridSearchOptimizer(
search_space, initialize={"vertices": n_initialize}, direction="orthogonal"
)
opt.search(parabola_function, n_iter=n_iter)
search_data = opt.search_data
print("\n search_data \n", search_data, "\n")
y_data_count = search_data["y"].value_counts().to_dict()
print("\n y_data_count \n", y_data_count, "\n")
assert y_data_count[-100] >= n_iter - n_initialize
SimonBlanke /
Gradient-Free-Optimizers
