def test_copy_equals_network():
"""
Test that a copy of an entity has the same network as the original
"""
ent = entity.NeuralEntity()
copy = ent.copy()
assert ent.equal_network(copy)
assert copy.equal_network(ent)
def test_forward_propogation_output_size():
"""
Test that forward propogation through the entity's
neural network returns the correct number of activations
"""
ent = entity.NeuralEntity(0, [5])
inputs = np.array([0.5, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 1, 1])
inputs = np.expand_dims(inputs, 1)
outputs = ent.forward_propagation(inputs)
assert len(outputs) == 5
def test_reproduce():
"""
Test that reproduction creates offspring
"""
ent = entity.NeuralEntity(100, [5])
children = ent.reproduce(5, 0.5)
assert len(children) == 5
for child in children:
assert child.fitness == 0
assert len(child.weights) == len(ent.weights)
assert len(child.biases) == len(ent.biases)
def test_forward_propogation_binary_values():
"""
Test that forward propogation through the entity's
neural network returns a cache with appropriate weights
"""
ent = entity.NeuralEntity(0, [5])
inputs = np.array([0.5, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 1, 1])
inputs = np.expand_dims(inputs, 1)
outputs = ent.forward_propagation(inputs)
for out in outputs:
assert out in (0, 1)
def test_initialise_parameters_size():
"""
Test that parameters in the neural network for
NeuralEntity are set with the right size
"""
ent = entity.NeuralEntity(0, [5])
assert len(ent.weights) == 3
assert len(ent.biases) == 3
assert ent.weights[1].shape == (5, 14)
assert ent.biases[1].shape == (5, 1)
assert ent.weights[2].shape == (5, 5)
assert ent.biases[2].shape == (5, 1)
def test_behaviour_output_correct():
"""
Test that the behaviour call outputs an action and a 3-bit vocal call
"""
ent = entity.NeuralEntity(0, [5])
action, vocal = ent.behaviour(0.4, 0b1111100000, [0, 1, 1])
assert action in [
Action.FORWARDS, Action.LEFT, Action.RIGHT, Action.NOTHING
]
assert len(vocal) == 3
for x in vocal:
assert x in (0, 1)
def test_different_entities_not_equal():
"""
Test that two different entities are not equal
"""
ent = entity.NeuralEntity()
copies = [ent.copy() for _ in range(5)]
copies[0].weights[1] += 1
copies[1].biases[0] = np.array([0])
copies[2].weights.append(None)
copies[3].biases[1] = None
copies[4].biases = copies[4].biases[:-1]
for copy in copies:
assert not copy.equal_network(ent)
assert not ent.equal_network(copy)
def test_copy_new_parameters():
"""
Test that reproduction creates unique parameters
"""
ent = entity.NeuralEntity(100, [5])
copy = ent.copy()
for layer in range(1, len(ent.weights)):
ent.weights[layer] = None
ent.biases[layer] = None
assert copy.weights is not ent.weights
assert copy.biases is not ent.biases
assert copy.fitness == 0
for layer in range(1, len(copy.weights)):
assert not copy.weights[layer] is None
assert not copy.biases[layer] is None
def test_reproduce_new_parameters():
"""
Test that reproduction creates unique parameters
"""
ent = entity.NeuralEntity(100, [5])
children = ent.reproduce(5, 0.5)
assert len(children) == 5
for layer in range(1, len(ent.weights)):
ent.weights[layer] = None
ent.biases[layer] = None
for child in children:
assert not child.weights is ent.weights
assert not child.biases is ent.biases
for layer in range(1, len(child.weights)):
assert not child.weights[layer] is None
assert not child.biases[layer] is None
def test_deeper_parameters_size():
"""
Test that parameters in the neural network for
NeuralEntity are set with the right size when a deeper
network is created
"""
ent = entity.NeuralEntity(0, [5, 10, 5])
assert len(ent.weights) == 5
assert len(ent.biases) == 5
assert ent.weights[1].shape == (5, 14)
assert ent.biases[1].shape == (5, 1)
assert ent.weights[2].shape == (10, 5)
assert ent.biases[2].shape == (10, 1)
assert ent.weights[3].shape == (5, 10)
assert ent.biases[3].shape == (5, 1)
assert ent.weights[4].shape == (5, 5)
assert ent.biases[4].shape == (5, 1)