def test_multivariate_normal_size_types(self):
# Test for multivariate_normal issue with 'size' argument.
# Check that the multivariate_normal size argument can be a
# numpy integer.
rnd.multivariate_normal([0], [[0]], size=1)
rnd.multivariate_normal([0], [[0]], size=np.int_(1))
rnd.multivariate_normal([0], [[0]], size=np.int64(1))
def test_multivariate_normal(self):
rnd.seed(self.seed, self.brng)
mean = (.123456789, 10)
# Hmm... not even symmetric.
cov = [[1, 0], [1, 0]]
size = (3, 2)
actual = rnd.multivariate_normal(mean, cov, size)
desired = np.array([[[-2.42282709811266, 10.0],
[1.2267795840027274, 10.0]],
[[0.06813924868067336, 10.0],
[1.001190462507746, 10.0]],
[[-1.74157261455869, 10.0],
[1.0400952859037553, 10.0]]])
np.testing.assert_array_almost_equal(actual, desired, decimal=10)
# Check for default size, was raising deprecation warning
actual = rnd.multivariate_normal(mean, cov)
desired = np.array([1.0579899448949994, 10.0])
np.testing.assert_array_almost_equal(actual, desired, decimal=10)
# Check that non positive-semidefinite covariance raises warning
mean = [0, 0]
cov = [[1, 1 + 1e-10], [1 + 1e-10, 1]]
assert_warns(RuntimeWarning, rnd.multivariate_normal, mean, cov)