import numpy as np
from numba import jit # type: ignore [attr-defined]
from numba import njit
[docs]
@njit(
"boolean(float32[:,:], float32[:], float32)", fastmath=True, nogil=True, cache=True
)
def is_point_far_enough(points, tested_point, squared_min_distance):
nbp = points.shape[0]
farenough = True
for i in range(nbp - 1, -1, -1):
if (
(tested_point[0] - points[i][0]) ** 2
+ (tested_point[1] - points[i][1]) ** 2
+ (tested_point[2] - points[i][2]) ** 2
) < squared_min_distance:
farenough = False
break
return farenough
[docs]
@jit(cache=True, nogil=True, nopython=True)
def xyz_to_child_index(xyz, aabb_center):
test = np.greater_equal(xyz - aabb_center, 0).astype(np.int8)
return np.sum(np.left_shift(test, np.array([2, 1, 0])), axis=1)
[docs]
@njit(
"int32[:](float32[:,:], int32[:], float32[:], float32[:], int32)",
cache=True,
nogil=True,
)
def xyz_to_key(xyz, cell_count, aabb_min, inv_aabb_size, shift):
a = ((cell_count * inv_aabb_size) * (xyz - aabb_min)).astype(np.int64)
a = np.minimum(np.maximum(a, 0), cell_count - 1)
a[:, 1] <<= shift
a[:, 2] <<= 2 * shift
return np.sum(a, axis=1).astype(np.int32)