def test_minimize(self):
power_signals_d = np.array(
[[0.0, 0.0, 0.0, 0.0],
[1.33389997, 1.40310001, 0.67150003, 0.77249998],
[1.42349994, 1.51800001, 1.43809998, 1.20449996],
[1.52020001, 1.45150006, 1.84809995, 0.99949998]])
rank_k = 4
weights = np.array([0.0, 0.0, 0.97073243, 0.97243198])
tau = 0.9
mu_l = 5e2
solver_type = SolverType.ecos
initial_l_cs_value = np.array(
[[0.12227644, -0.05536519, -0.02796016, 0.11115515],
[0.12183656, -0.06418167, -0.03631565, 0.09248578],
[0.12190038, -0.07035386, -0.03077544, 0.06306368],
[0.12185763, -0.0822263, -0.02468169, 0.03843156]])
initial_r_cs_value = np.array(
[[7.81948013, 11.26965908, 11.43521789, 8.89706298],
[0.18783052, -1.17162576, -1.68541257, -1.14962881],
[0.97275831, 0.99957452, 0.92734892, 0.453427],
[-0.86265428, -3.28835462, -4.00326343, -1.76664483]])
initial_beta_value = 0.0
initial_component_r0 = np.array(
[1.36527916, 2.70624333, 4.04720749, 5.38817165])
expected_l_cs_value = np.array(
[[2.610888e-14, -1.027025e-14, 1.481367e-14, -1.786423e-14],
[6.769088e-02, -5.028329e-14, -4.090143e-14, 1.891483e-13],
[1.353818e-01, -8.877942e-14, 4.614613e-15, -1.047267e-14],
[2.030726e-01, 1.495160e-13, 1.955246e-14, -1.573292e-13]])
expected_r_cs_value = initial_r_cs_value
expected_beta_value = initial_beta_value
left_matrix_minimization = LeftMatrixMinimization(
power_signals_d,
rank_k,
weights,
tau,
mu_l,
solver_type=SolverType.ecos)
actual_l_cs_value, actual_r_cs_value, actual_beta_value =\
left_matrix_minimization.minimize(initial_l_cs_value,
initial_r_cs_value,
initial_beta_value,
initial_component_r0)
# np.testing.assert_array_equal(actual_l_cs_value, expected_l_cs_value)
np.testing.assert_almost_equal(actual_l_cs_value,
expected_l_cs_value,
decimal=6)
np.testing.assert_array_equal(actual_r_cs_value, expected_r_cs_value)
np.testing.assert_array_equal(actual_beta_value, expected_beta_value)
def _get_left_matrix_minimization(self, weights, tau, mu_l):
"""
For dependency injection for testing, i.e. for injecting mock.
"""
if ((not hasattr(self, '_left_matrix_minimization')) or
(self._left_matrix_minimization is None)):
return LeftMatrixMinimization(
self._power_signals_d, self._rank_k, weights, tau, mu_l,
solver_type=self._solver_type)
else: # This must be mock object inject from test
return self._left_matrix_minimization
def test_minimize_with_large_data(self):
input_power_signals_file_path = os.path.abspath(
os.path.join(os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"three_years_power_signals_d_1.csv"))
with open(input_power_signals_file_path) as file:
power_signals_d = np.loadtxt(file, delimiter=',')
rank_k = 6
weights_file_path = os.path.abspath(
os.path.join(os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"three_years_weights.csv"))
with open(weights_file_path) as file:
weights = np.loadtxt(file, delimiter=',')
tau = 0.9
mu_l = 5e2
initial_l_cs_value_file_path = os.path.abspath(
os.path.join(os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"three_years_initial_l_cs_value.csv"))
with open(initial_l_cs_value_file_path) as file:
initial_l_cs_value = np.loadtxt(file, delimiter=',')
initial_r_cs_value_file_path = os.path.abspath(
os.path.join(os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"three_years_initial_r_cs_value.csv"))
with open(initial_r_cs_value_file_path) as file:
initial_r_cs_value = np.loadtxt(file, delimiter=',')
initial_beta_value = 0.0
l_cs_value_after_iteration_1_file_path = os.path.abspath(
os.path.join(
os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"l_cs_value_after_left_matrix_minimization_iteration_1_NEW.csv"
))
with open(l_cs_value_after_iteration_1_file_path) as file:
expected_l_cs_value = np.loadtxt(file, delimiter=',')
r_cs_value_after_iteration_1_file_path = os.path.abspath(
os.path.join(
os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"r_cs_value_after_left_matrix_minimization_iteration_1.csv"))
with open(r_cs_value_after_iteration_1_file_path) as file:
expected_r_cs_value = np.loadtxt(file, delimiter=',')
expected_beta_value = initial_beta_value
initial_r0_value_file_path = os.path.abspath(
os.path.join(os.path.dirname(__file__),
"../../fixtures/left_matrix_minimization",
"three_years_initial_component_r0.csv"))
with open(initial_r0_value_file_path) as file:
initial_component_r0_value = np.loadtxt(file, delimiter=',')
left_matrix_minimization = LeftMatrixMinimization(power_signals_d,
rank_k,
weights,
tau,
mu_l,
solver_type='MOSEK')
try:
actual_l_cs_value, actual_r_cs_value, actual_beta_value =\
left_matrix_minimization.minimize(initial_l_cs_value,
initial_r_cs_value,
initial_beta_value,
initial_component_r0_value)
except cvx.SolverError:
self.skipTest(
"This test uses MOSEK solver" +
"because default ECOS solver fails with large data. " +
"Unless MOSEK is installed, this test fails.")
else:
np.testing.assert_array_almost_equal(actual_l_cs_value,
expected_l_cs_value)
np.testing.assert_array_almost_equal(actual_r_cs_value,
expected_r_cs_value)
np.testing.assert_array_almost_equal(actual_beta_value,
expected_beta_value)