def test_minimize_callback(self):
"""Callback should run at the end of every iteration."""
variables = np.random.uniform(low=-1.0, high=1.0, size=[1])
saved_objective = [None]
saved_iteration = [None]
iteration_count = [0]
def callback(iteration, objective_value, variables):
del variables
def save(iteration, objective_value):
saved_objective[0] = objective_value
saved_iteration[0] = iteration
iteration_count[0] += 1
return tf.py_function(save, [iteration, objective_value], [])
final_objective, _ = levenberg_marquardt.minimize(
lambda x: x, variables, max_iterations=5, callback=callback)
final_saved_objective = tf.py_function(lambda: saved_objective[0], [],
tf.float64)
final_saved_iteration = tf.py_function(lambda: saved_iteration[0], [],
tf.int32)
final_iteration_count = tf.py_function(lambda: iteration_count[0], [],
tf.int32)
with self.subTest(name="objective_value"):
self.assertAllClose(final_objective, final_saved_objective)
with self.subTest(name="iterations"):
self.assertAllEqual(final_saved_iteration, final_iteration_count)
def test_minimize_preset(self, residuals, variables, max_iterations,
final_objective_value, final_variables):
"""Tests the output of the minimization for some presets."""
objective_value, output_variables = levenberg_marquardt.minimize(
residuals, variables, max_iterations)
with self.subTest(name="objective_value"):
self.assertAllClose(objective_value, final_objective_value)
with self.subTest(name="variables"):
for output_variable, final_variable in zip(output_variables,
final_variables):
self.assertAllClose(output_variable, final_variable)
def test_minimize_linear_residuals_random(self):
"""Optimizing linear residuals should give the minimum in 1 step."""
tensor_size = np.random.randint(1, 3)
tensor_shape = np.random.randint(1, 10, size=(tensor_size)).tolist()
variables = np.random.uniform(low=-1.0, high=1.0, size=tensor_shape)
objective_value, variables = levenberg_marquardt.minimize(
lambda x: x, variables, max_iterations=1)
with self.subTest(name="objective_value"):
self.assertAllClose(objective_value, tf.zeros_like(objective_value))
with self.subTest(name="variables"):
for variable in variables:
self.assertAllClose(variable, tf.zeros_like(variable))
def test_minimize_ill_conditioned_not_raised(self):
"""Optimizing an ill conditioned problem should not raise an exception."""
if not tf.executing_eagerly():
return
def f1(x, y):
return x * y * 10000.0
def f2(x, y):
return x * y * 0.0001
x = (1., )
y = (1., )
try:
self.evaluate(
levenberg_marquardt.minimize(residuals=(f1, f2),
variables=(x, y),
max_iterations=1,
regularizer=1e-20))
except Exception as e: # pylint: disable=broad-except
self.fail("Exception raised: %s" % str(e))
def optimize(*variables): levenberg_marquardt.minimize(lambda x: x, variables, max_iterations)
