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 numpy as np
def init_grid_search(space_dim, n_pos):
positions = []
n_dim = len(space_dim)
p_per_dim = int(np.power(n_pos, 1 / n_dim))
for dim in space_dim:
dim_dist = int(dim / (p_per_dim + 1))
n_points = [n * dim_dist for n in range(1, p_per_dim + 1)]
positions.append(n_points)
pos_mesh = np.array(np.meshgrid(*positions))
positions = list(pos_mesh.T.reshape(-1, n_dim))
diff_pos = n_pos - len(positions)
if diff_pos > 0:
pos_rnd = init_random_search(space_dim, n_pos=diff_pos)
return positions + pos_rnd
return positions
def init_random_search(space_dim, n_pos):
for nth_pos in range(n_pos):
pos = np.random.randint(space_dim, size=space_dim.shape)
positions.append(pos)
SimonBlanke /
Gradient-Free-Optimizers
