def test_ball_tree_query_radius(n_samples=100, n_features=10):
rng = check_random_state(0)
X = 2 * rng.random_sample(size=(n_samples, n_features)) - 1
query_pt = np.zeros(n_features, dtype=float)
eps = 1E-15 # roundoff error can cause test to fail
bt = BallTree(X, leaf_size=5)
rad = np.sqrt(((X - query_pt)**2).sum(1))
for r in np.linspace(rad[0], rad[-1], 100):
ind = bt.query_radius([query_pt], r + eps)[0]
i = np.where(rad <= r + eps)[0]
ind.sort()
i.sort()
assert_array_almost_equal(i, ind)
def test_ball_tree_query_radius_distance(n_samples=100, n_features=10):
rng = check_random_state(0)
X = 2 * rng.random_sample(size=(n_samples, n_features)) - 1
query_pt = np.zeros(n_features, dtype=float)
eps = 1E-15 # roundoff error can cause test to fail
bt = BallTree(X, leaf_size=5)
rad = np.sqrt(((X - query_pt)**2).sum(1))
for r in np.linspace(rad[0], rad[-1], 100):
ind, dist = bt.query_radius([query_pt], r + eps, return_distance=True)
ind = ind[0]
dist = dist[0]
d = np.sqrt(((query_pt - X[ind])**2).sum(1))
assert_array_almost_equal(d, dist)