def test_log_normal_pdf_compare_with_scipy(self):
"""Check the derivations of the log_normal_pdf."""
from mixture_model.gaussian_mixture import _log_normal_pdf
# Smoke test
random_errors = np.random.rand(1, 2)
unit_vars = np.array([2, 3])
attempt_random_result = _log_normal_pdf(random_errors**2, unit_vars)
assert attempt_random_result is not None
# Comparison with a zero-mean normal distribution.
errors = np.arange(-1, 1, 0.01)
goal_values = norm.logpdf(errors, loc=0, scale=1)
attempt_values = _log_normal_pdf(errors**2, 1)
assert_almost_equal(goal_values, attempt_values)
# Comparison with a zero-mean wide normal distribution.
goal_values = norm.logpdf(errors, loc=0, scale=4)
attempt_values = _log_normal_pdf(errors**2, 4**2)
assert_almost_equal(goal_values, attempt_values)
def test_log_normal_pdf_is_isotropic(self):
"""Test whether the log_normal_pdf ."""
from mixture_model.gaussian_mixture import _log_normal_pdf
# Smoke test
random_errors = np.random.rand(1, 2)
unit_vars = np.array([2, 3])
attempt_random_result = _log_normal_pdf(random_errors**2, unit_vars)
assert attempt_random_result is not None
# Test isotropic property
random_goal_values = \
norm.logpdf(random_errors, loc=[0, 0], scale=np.sqrt(unit_vars))
assert_almost_equal(random_goal_values, attempt_random_result)
