def test_attribute_vertex_list(self):
tree, altitudes = TestAttributes.get_test_tree()
res = hg.attribute_vertex_list(tree)
ref = [[0], [1], [2], [3], [4], [5], [6], [7], [8], [0, 1], [2, 5],
[3, 4], [6, 7], [6, 7, 8], [0, 1, 2, 5], [0, 1, 2, 5, 6, 7, 8],
[0, 1, 2, 5, 6, 7, 8, 3, 4]]
self.assertTrue(len(ref) == len(res))
for i in range(len(ref)):
self.assertTrue(set(ref[i]) == set(res[i]))
def test_attribute_gaussian_region_weights_model_scalar(self):
tree, altitudes = TestAttributes.get_test_tree()
vertex_list = hg.attribute_vertex_list(tree)
np.random.seed(42)
vertex_weights = np.random.rand(tree.num_leaves())
mean, variance = hg.attribute_gaussian_region_weights_model(tree, vertex_weights)
for i in tree.leaves_to_root_iterator():
m = np.mean(vertex_weights[vertex_list[i]])
v = np.var(vertex_weights[vertex_list[i]])
self.assertTrue(np.isclose(m, mean[i]))
self.assertTrue(np.isclose(v, variance[i]))
def test_attribute_gaussian_region_weights_model_vectorial(self):
tree, altitudes = TestAttributes.get_test_tree()
vertex_list = hg.attribute_vertex_list(tree)
np.random.seed(42)
vertex_weights = np.random.rand(tree.num_leaves(), 3)
mean, variance = hg.attribute_gaussian_region_weights_model(tree, vertex_weights)
for i in tree.leaves_to_root_iterator(include_leaves=True):
m = np.mean(vertex_weights[vertex_list[i]], 0)
self.assertTrue(np.allclose(m, mean[i, :]))
# numpy wrongly interprets a single observation with several variables as
# multiple observations of a single variables
if i >= tree.num_leaves():
v = np.cov(vertex_weights[vertex_list[i]], rowvar=False, bias=True)
self.assertTrue(np.allclose(v, variance[i, ...]))
else:
v = np.zeros_like(variance[i, ...])
self.assertTrue(np.allclose(v, variance[i, ...]))