class TestSpecialEuclidean(geomstats.tests.TestCase): def setUp(self): gs.random.seed(12) self.n = 2 self.group = SpecialEuclidean(n=self.n) self.n_samples = 3 self.point = self.group.random_point(self.n_samples) self.tangent_vec = self.group.to_tangent( gs.random.rand(self.n_samples, self.group.n + 1, self.group.n + 1), self.point) def test_belongs(self): theta = gs.pi / 3 point_1 = gs.array([[gs.cos(theta), -gs.sin(theta), 2.], [gs.sin(theta), gs.cos(theta), 3.], [0., 0., 1.]]) result = self.group.belongs(point_1) self.assertTrue(result) point_2 = gs.array([[gs.cos(theta), -gs.sin(theta), 2.], [gs.sin(theta), gs.cos(theta), 3.], [0., 0., 0.]]) result = self.group.belongs(point_2) self.assertFalse(result) point = gs.array([point_1, point_2]) expected = gs.array([True, False]) result = self.group.belongs(point) self.assertAllClose(result, expected) point = point_1[0] result = self.group.belongs(point) self.assertFalse(result) point = gs.zeros((2, 3)) result = self.group.belongs(point) self.assertFalse(result) point = gs.zeros((2, 2, 3)) result = self.group.belongs(point) self.assertFalse(gs.all(result)) self.assertAllClose(result.shape, (2, )) def test_random_point_and_belongs(self): point = self.group.random_point() result = self.group.belongs(point) self.assertTrue(result) point = self.group.random_point(self.n_samples) result = self.group.belongs(point) expected = gs.array([True] * self.n_samples) self.assertAllClose(result, expected) def test_identity(self): result = self.group.identity expected = gs.eye(self.n + 1) self.assertAllClose(result, expected) def test_is_tangent(self): theta = gs.pi / 3 vec_1 = gs.array([[0., -theta, 2.], [theta, 0., 3.], [0., 0., 0.]]) point = self.group.random_point() tangent_vec = self.group.compose(point, vec_1) result = self.group.is_tangent(tangent_vec, point) self.assertTrue(result) vec_2 = gs.array([[0., -theta, 2.], [theta, 0., 3.], [0., 0., 1.]]) tangent_vec = self.group.compose(point, vec_2) result = self.group.is_tangent(tangent_vec, point) self.assertFalse(result) vec = gs.array([vec_1, vec_2]) expected = gs.array([True, False]) result = self.group.is_tangent(vec) self.assertAllClose(result, expected) def test_to_tangent_vec_vectorization(self): n = self.group.n tangent_vecs = gs.arange(self.n_samples * (n + 1)**2) tangent_vecs = gs.cast(tangent_vecs, gs.float32) tangent_vecs = gs.reshape(tangent_vecs, (self.n_samples, ) + (n + 1, ) * 2) point = self.group.random_point(self.n_samples) tangent_vecs = Matrices.mul(point, tangent_vecs) regularized = self.group.to_tangent(tangent_vecs, point) result = Matrices.mul( Matrices.transpose(point), regularized) + \ Matrices.mul(Matrices.transpose(regularized), point) result = result[:, :n, :n] expected = gs.zeros_like(result) self.assertAllClose(result, expected) def test_compose_and_inverse_matrix_form(self): point = self.group.random_point() inv_point = self.group.inverse(point) result = self.group.compose(point, inv_point) expected = self.group.identity self.assertAllClose(result, expected) if not geomstats.tests.tf_backend(): result = self.group.compose(inv_point, point) expected = self.group.identity self.assertAllClose(result, expected) def test_compose_vectorization(self): n_samples = self.n_samples n_points_a = self.group.random_point(n_samples=n_samples) n_points_b = self.group.random_point(n_samples=n_samples) one_point = self.group.random_point(n_samples=1) result = self.group.compose(one_point, n_points_a) self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) result = self.group.compose(n_points_a, one_point) if not geomstats.tests.tf_backend(): self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) result = self.group.compose(n_points_a, n_points_b) self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) def test_inverse_vectorization(self): n_samples = self.n_samples points = self.group.random_point(n_samples=n_samples) result = self.group.inverse(points) self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) def test_compose_matrix_form(self): point = self.group.random_point() result = self.group.compose(point, self.group.identity) expected = point self.assertAllClose(result, expected) if not geomstats.tests.tf_backend(): # Composition by identity, on the left # Expect the original transformation result = self.group.compose(self.group.identity, point) expected = point self.assertAllClose(result, expected) # Composition of translations (no rotational part) # Expect the sum of the translations point_a = gs.array([[1., 0., 1.], [0., 1., 1.5], [0., 0., 1.]]) point_b = gs.array([[1., 0., 2.], [0., 1., 2.5], [0., 0., 1.]]) result = self.group.compose(point_a, point_b) last_line_0 = gs.array_from_sparse([(0, 2), (1, 2)], [1., 1.], (3, 3)) expected = point_a + point_b * last_line_0 self.assertAllClose(result, expected) def test_left_exp_coincides(self): vector_group = SpecialEuclidean(n=2, point_type='vector') theta = gs.pi / 3 initial_vec = gs.array([theta, 2., 2.]) initial_matrix_vec = self.group.lie_algebra.matrix_representation( initial_vec) vector_exp = vector_group.left_canonical_metric.exp(initial_vec) result = self.group.left_canonical_metric.exp(initial_matrix_vec) expected = vector_group.matrix_from_vector(vector_exp) self.assertAllClose(result, expected) def test_right_exp_coincides(self): vector_group = SpecialEuclidean(n=2, point_type='vector') theta = gs.pi / 2 initial_vec = gs.array([theta, 1., 1.]) initial_matrix_vec = self.group.lie_algebra.matrix_representation( initial_vec) vector_exp = vector_group.right_canonical_metric.exp(initial_vec) result = self.group.right_canonical_metric.exp(initial_matrix_vec, n_steps=25) expected = vector_group.matrix_from_vector(vector_exp) self.assertAllClose(result, expected, atol=1e-6) def test_basis_belongs(self): lie_algebra = self.group.lie_algebra result = lie_algebra.belongs(lie_algebra.basis) self.assertTrue(gs.all(result)) def test_basis_has_the_right_dimension(self): for n in range(2, 5): algebra = SpecialEuclideanMatrixLieAlgebra(n) self.assertEqual(int(n * (n + 1) / 2), algebra.dim) def test_belongs_lie_algebra(self): theta = gs.pi / 3 vec_1 = gs.array([[0., -theta, 2.], [theta, 0., 3.], [0., 0., 0.]]) result = self.group.lie_algebra.belongs(vec_1) expected = True self.assertAllClose(result, expected) vec_2 = gs.array([[0., -theta, 2.], [theta, 0., 3.], [0., 0., 1.]]) result = self.group.lie_algebra.belongs(vec_2) expected = False self.assertAllClose(result, expected) vec = gs.array([vec_1, vec_2]) expected = gs.array([True, False]) result = self.group.lie_algebra.belongs(vec) self.assertAllClose(result, expected) def test_basis_representation_is_correctly_vectorized(self): for n in range(2, 5): algebra = SpecialEuclideanMatrixLieAlgebra(n) shape = gs.shape(algebra.basis_representation(algebra.basis)) dim = int(n * (n + 1) / 2) self.assertAllClose(shape, (dim, dim)) def test_left_metric_wrong_group(self): group = self.group.rotations self.assertRaises( ValueError, lambda: SpecialEuclideanMatrixCannonicalLeftMetric(group)) group = SpecialEuclidean(3, point_type='vector') self.assertRaises( ValueError, lambda: SpecialEuclideanMatrixCannonicalLeftMetric(group)) def test_exp_and_belongs(self): exp = self.group.left_canonical_metric.exp(self.tangent_vec, self.point) result = self.group.belongs(exp) self.assertTrue(gs.all(result)) exp = self.group.left_canonical_metric.exp(self.tangent_vec[0], self.point[0]) result = self.group.belongs(exp) self.assertTrue(result) @geomstats.tests.np_and_tf_only def test_log_and_is_tan(self): exp = self.group.left_canonical_metric.exp(self.tangent_vec, self.point) log = self.group.left_canonical_metric.log(exp, self.point) result = self.group.is_tangent(log, self.point) self.assertTrue(gs.all(result)) exp = self.group.left_canonical_metric.exp(self.tangent_vec[0], self.point[0]) log = self.group.left_canonical_metric.log(exp, self.point) result = self.group.is_tangent(log, self.point) self.assertTrue(gs.all(result)) log = self.group.left_canonical_metric.log(exp, self.point[0]) result = self.group.is_tangent(log, self.point[0]) self.assertTrue(result) @geomstats.tests.np_and_tf_only def test_exp_log(self): exp = self.group.left_canonical_metric.exp(self.tangent_vec, self.point) result = self.group.left_canonical_metric.log(exp, self.point) self.assertAllClose(result, self.tangent_vec) exp = self.group.left_canonical_metric.exp(self.tangent_vec[0], self.point[0]) result = self.group.left_canonical_metric.log(exp, self.point[0]) self.assertAllClose(result, self.tangent_vec[0]) def test_parallel_transport(self): metric = self.group.left_canonical_metric shape = (self.n_samples, self.group.n + 1, self.group.n + 1) results = helper.test_parallel_transport(self.group, metric, shape) for res in results: self.assertTrue(res) def test_lie_algebra_basis_belongs(self): basis = self.group.lie_algebra.basis result = self.group.lie_algebra.belongs(basis) self.assertTrue(gs.all(result)) def test_lie_algebra_projection_and_belongs(self): vec = gs.random.rand(self.n_samples, self.group.n + 1, self.group.n + 1) tangent_vec = self.group.lie_algebra.projection(vec) result = self.group.lie_algebra.belongs(tangent_vec) self.assertTrue(gs.all(result)) def test_basis_representation(self): vec = gs.random.rand(self.n_samples, self.group.dim) tangent_vec = self.group.lie_algebra.matrix_representation(vec) result = self.group.lie_algebra.basis_representation(tangent_vec) self.assertAllClose(result, vec) result = self.group.lie_algebra.basis_representation(tangent_vec[0]) self.assertAllClose(result, vec[0]) def test_metrics_expected_point_type(self): left = self.group.left_canonical_metric right = self.group.right_canonical_metric metric = self.group.metric for m in [left, right, metric]: self.assertTrue(m.default_point_type == 'matrix') def test_metric_left_invariant(self): group = self.group point = group.random_point() expected = group.left_canonical_metric.norm(self.tangent_vec) translated = group.tangent_translation_map(point)(self.tangent_vec) result = group.left_canonical_metric.norm(translated) self.assertAllClose(result, expected) def test_projection_and_belongs(self): shape = (self.n_samples, self.n + 1, self.n + 1) result = helper.test_projection_and_belongs(self.group, shape) for res in result: self.assertTrue(res)
class TestSpecialEuclidean(geomstats.tests.TestCase): def setUp(self): self.n = 2 self.group = SpecialEuclidean(n=self.n) self.n_samples = 4 def test_belongs(self): theta = gs.pi / 3 point_1 = gs.array([[gs.cos(theta), -gs.sin(theta), 2.], [gs.sin(theta), gs.cos(theta), 3.], [0., 0., 1.]]) result = self.group.belongs(point_1) expected = True self.assertAllClose(result, expected) point_2 = gs.array([[gs.cos(theta), -gs.sin(theta), 2.], [gs.sin(theta), gs.cos(theta), 3.], [0., 0., 0.]]) result = self.group.belongs(point_2) expected = False self.assertAllClose(result, expected) point = gs.array([point_1, point_2]) expected = gs.array([True, False]) result = self.group.belongs(point) self.assertAllClose(result, expected) def test_random_uniform_and_belongs(self): point = self.group.random_uniform() result = self.group.belongs(point) expected = True self.assertAllClose(result, expected) point = self.group.random_uniform(self.n_samples) result = self.group.belongs(point) expected = gs.array([True] * self.n_samples) self.assertAllClose(result, expected) def test_identity(self): result = self.group.identity expected = gs.eye(self.n + 1) self.assertAllClose(result, expected) def test_is_in_lie_algebra(self): theta = gs.pi / 3 vec_1 = gs.array([[0., -theta, 2.], [theta, 0., 3.], [0., 0., 0.]]) result = self.group.is_tangent(vec_1) expected = True self.assertAllClose(result, expected) vec_2 = gs.array([[0., -theta, 2.], [theta, 0., 3.], [0., 0., 1.]]) result = self.group.is_tangent(vec_2) expected = False self.assertAllClose(result, expected) vec = gs.array([vec_1, vec_2]) expected = gs.array([True, False]) result = self.group.is_tangent(vec) self.assertAllClose(result, expected) def test_to_tangent_vec_vectorization(self): n = self.group.n tangent_vecs = gs.arange(self.n_samples * (n + 1)**2) tangent_vecs = gs.cast(tangent_vecs, gs.float32) tangent_vecs = gs.reshape(tangent_vecs, (self.n_samples, ) + (n + 1, ) * 2) point = self.group.random_uniform(self.n_samples) tangent_vecs = self.group.compose(point, tangent_vecs) regularized = self.group.to_tangent(tangent_vecs, point) result = self.group.compose( self.group.transpose(point), regularized) + \ self.group.compose(self.group.transpose(regularized), point) result = result[:, :n, :n] expected = gs.zeros_like(result) self.assertAllClose(result, expected) def test_compose_and_inverse_matrix_form(self): point = self.group.random_uniform() inv_point = self.group.inverse(point) result = self.group.compose(point, inv_point) expected = self.group.identity self.assertAllClose(result, expected) if not geomstats.tests.tf_backend(): result = self.group.compose(inv_point, point) expected = self.group.identity self.assertAllClose(result, expected) def test_compose_vectorization(self): n_samples = self.n_samples n_points_a = self.group.random_uniform(n_samples=n_samples) n_points_b = self.group.random_uniform(n_samples=n_samples) one_point = self.group.random_uniform(n_samples=1) result = self.group.compose(one_point, n_points_a) self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) result = self.group.compose(n_points_a, one_point) if not geomstats.tests.tf_backend(): self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) result = self.group.compose(n_points_a, n_points_b) self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) def test_inverse_vectorization(self): n_samples = self.n_samples points = self.group.random_uniform(n_samples=n_samples) result = self.group.inverse(points) self.assertAllClose(gs.shape(result), (n_samples, ) + (self.group.n + 1, ) * 2) def test_compose_matrix_form(self): point = self.group.random_uniform() result = self.group.compose(point, self.group.identity) expected = point self.assertAllClose(result, expected) if not geomstats.tests.tf_backend(): # Composition by identity, on the left # Expect the original transformation result = self.group.compose(self.group.identity, point) expected = point self.assertAllClose(result, expected) # Composition of translations (no rotational part) # Expect the sum of the translations point_a = gs.array([[1., 0., 1.], [0., 1., 1.5], [0., 0., 1.]]) point_b = gs.array([[1., 0., 2.], [0., 1., 2.5], [0., 0., 1.]]) result = self.group.compose(point_a, point_b) last_line_0 = gs.array_from_sparse([(0, 2), (1, 2)], [1., 1.], (3, 3)) expected = point_a + point_b * last_line_0 self.assertAllClose(result, expected)