def setUp(self):
# config and mutation
self.config = {
"cartesian": {
"rows": 4,
"columns": 4,
"levels_back": 2,
"num_inputs": 4,
"num_outputs": 4
},
"function_nodes": [{
"type": "FUNCTION",
"name": "ADD",
"arity": 2
}, {
"type": "FUNCTION",
"name": "SUB",
"arity": 2
}, {
"type": "FUNCTION",
"name": "MUL",
"arity": 2
}, {
"type": "FUNCTION",
"name": "DIV",
"arity": 2
}, {
"type": "FUNCTION",
"name": "COS",
"arity": 1
}, {
"type": "FUNCTION",
"name": "SIN",
"arity": 1
}, {
"type": "FUNCTION",
"name": "RAD",
"arity": 1
}],
"mutation": {
"methods": ["POINT_MUTATION"],
"probability": 1.0
}
}
self.mutator = CartesianMutation(self.config)
# make cartesian
self.data = [[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12],
[13, 14, 15, 16]]
self.chromosome = [[0, 0, 2], [0, 0, 3], [3, 4, 5], [0, 1, 2],
[0, 1, 3], [2, 5, 7], [2, 6, 9], [0, 5, 7],
[2, 11, 8], [0, 11, 8]]
self.output_nodes = [4, 9, 12, 13]
self.cartesian = Cartesian(config={},
rows=1,
columns=14,
levels_back=0,
func_nodes=self.chromosome,
input_nodes=self.data,
output_nodes=self.output_nodes)
def setUp(self):
# config and mutation
self.config = {
"cartesian": {
"rows": 4,
"columns": 4,
"levels_back": 2,
"num_inputs": 4,
"num_outputs": 4
},
"function_nodes": [
{"type": "FUNCTION", "name": "ADD", "arity": 2},
{"type": "FUNCTION", "name": "SUB", "arity": 2},
{"type": "FUNCTION", "name": "MUL", "arity": 2},
{"type": "FUNCTION", "name": "DIV", "arity": 2},
{"type": "FUNCTION", "name": "COS", "arity": 1},
{"type": "FUNCTION", "name": "SIN", "arity": 1},
{"type": "FUNCTION", "name": "RAD", "arity": 1}
],
"mutation": {
"methods": ["POINT_MUTATION"],
"probability": 1.0
}
}
self.mutator = CartesianMutation(self.config)
# make cartesian
self.data = [
[1, 2, 3, 4],
[5, 6, 7, 8],
[9, 10, 11, 12],
[13, 14, 15, 16]
]
self.chromosome = [
[0, 0, 2],
[0, 0, 3],
[3, 4, 5],
[0, 1, 2],
[0, 1, 3],
[2, 5, 7],
[2, 6, 9],
[0, 5, 7],
[2, 11, 8],
[0, 11, 8]
]
self.output_nodes = [4, 9, 12, 13]
self.cartesian = Cartesian(
config={},
rows=1,
columns=14,
levels_back=0,
func_nodes=self.chromosome,
input_nodes=self.data,
output_nodes=self.output_nodes
)
class CartesianMutationTests(unittest.TestCase):
def setUp(self):
# config and mutation
self.config = {
"cartesian": {
"rows": 4,
"columns": 4,
"levels_back": 2,
"num_inputs": 4,
"num_outputs": 4
},
"function_nodes": [{
"type": "FUNCTION",
"name": "ADD",
"arity": 2
}, {
"type": "FUNCTION",
"name": "SUB",
"arity": 2
}, {
"type": "FUNCTION",
"name": "MUL",
"arity": 2
}, {
"type": "FUNCTION",
"name": "DIV",
"arity": 2
}, {
"type": "FUNCTION",
"name": "COS",
"arity": 1
}, {
"type": "FUNCTION",
"name": "SIN",
"arity": 1
}, {
"type": "FUNCTION",
"name": "RAD",
"arity": 1
}],
"mutation": {
"methods": ["POINT_MUTATION"],
"probability": 1.0
}
}
self.mutator = CartesianMutation(self.config)
# make cartesian
self.data = [[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12],
[13, 14, 15, 16]]
self.chromosome = [[0, 0, 2], [0, 0, 3], [3, 4, 5], [0, 1, 2],
[0, 1, 3], [2, 5, 7], [2, 6, 9], [0, 5, 7],
[2, 11, 8], [0, 11, 8]]
self.output_nodes = [4, 9, 12, 13]
self.cartesian = Cartesian(config={},
rows=1,
columns=14,
levels_back=0,
func_nodes=self.chromosome,
input_nodes=self.data,
output_nodes=self.output_nodes)
def test_mutate_function_node(self):
num_input_nodes = len(self.cartesian.input_nodes)
num_nodes = len(self.cartesian.graph())
for i in range(100):
n_addr = random.randint(num_input_nodes, num_nodes - 1)
# before
graph_before = copy.deepcopy(self.cartesian.graph())
# print "BEFORE:", self.cartesian.graph
# mutate
g_index = self.mutator.mutate_function_node(n_addr, self.cartesian)
# after
graph_after = copy.deepcopy(self.cartesian.graph())
# print "AFTER:", self.cartesian.graph
gene_before = graph_before[n_addr][g_index]
gene_after = graph_after[n_addr][g_index]
# asserts
self.assertNotEquals(graph_before, graph_after)
self.assertNotEquals(gene_before, gene_after)
def test_mutate_output_node(self):
num_output_nodes = len(self.cartesian.output_nodes)
for i in range(100):
index = random.randint(0, num_output_nodes - 1)
old_addr = self.cartesian.output_nodes[index]
new_addr = self.mutator.mutate_output_node(index, self.cartesian)
self.assertNotEquals(old_addr, new_addr)
self.assertNotEquals(old_addr, self.cartesian.output_nodes[index])
def test_point_mutation(self):
for i in range(100):
# before
output_before = copy.deepcopy(self.cartesian.output_nodes)
graph_before = copy.deepcopy(self.cartesian.graph)
# print "BEFORE:", graph_before
# mutate
self.mutator.point_mutation(self.cartesian)
# after
output_after = copy.deepcopy(self.cartesian.output_nodes)
graph_after = copy.deepcopy(self.cartesian.graph)
# print "AFTER:", graph_after
# asserts
if self.mutator.index["mutated_node"] == "FUNC_NODE":
self.assertNotEquals(graph_before, graph_after)
elif self.mutator.index["mutated_node"] == "OUTPUT_NODE":
index = self.mutator.index["output_node"]
num_outputs = len(self.cartesian.output_nodes)
self.assertNotEquals(output_before, output_after)
self.assertTrue(index >= 0 and index <= num_outputs - 1)
def test_to_dict(self):
for i in range(100):
# print "PROGRAM:", self.cartesian.program()
self.mutator.mutate(self.cartesian)
# print "PROGRAM:", self.cartesian.program()
# import pprint
# pprint.pprint(self.mutator.to_dict())
mut_dict = self.mutator.to_dict()
before_mut = mut_dict["before_mutation"]
after_mut = mut_dict["after_mutation"]
self.assertNotEquals(before_mut, after_mut)
self.assertIsNotNone(mut_dict["method"])
self.assertIsNotNone(mut_dict["mutation_probability"])
self.assertIsNotNone(mut_dict["random_probability"])
self.assertIsNotNone(mut_dict["mutated"])
self.assertIsNotNone(mut_dict["before_mutation"])
self.assertIsNotNone(mut_dict["after_mutation"])
class CartesianMutationTests(unittest.TestCase):
def setUp(self):
# config and mutation
self.config = {
"cartesian": {
"rows": 4,
"columns": 4,
"levels_back": 2,
"num_inputs": 4,
"num_outputs": 4
},
"function_nodes": [
{"type": "FUNCTION", "name": "ADD", "arity": 2},
{"type": "FUNCTION", "name": "SUB", "arity": 2},
{"type": "FUNCTION", "name": "MUL", "arity": 2},
{"type": "FUNCTION", "name": "DIV", "arity": 2},
{"type": "FUNCTION", "name": "COS", "arity": 1},
{"type": "FUNCTION", "name": "SIN", "arity": 1},
{"type": "FUNCTION", "name": "RAD", "arity": 1}
],
"mutation": {
"methods": ["POINT_MUTATION"],
"probability": 1.0
}
}
self.mutator = CartesianMutation(self.config)
# make cartesian
self.data = [
[1, 2, 3, 4],
[5, 6, 7, 8],
[9, 10, 11, 12],
[13, 14, 15, 16]
]
self.chromosome = [
[0, 0, 2],
[0, 0, 3],
[3, 4, 5],
[0, 1, 2],
[0, 1, 3],
[2, 5, 7],
[2, 6, 9],
[0, 5, 7],
[2, 11, 8],
[0, 11, 8]
]
self.output_nodes = [4, 9, 12, 13]
self.cartesian = Cartesian(
config={},
rows=1,
columns=14,
levels_back=0,
func_nodes=self.chromosome,
input_nodes=self.data,
output_nodes=self.output_nodes
)
def test_mutate_function_node(self):
num_input_nodes = len(self.cartesian.input_nodes)
num_nodes = len(self.cartesian.graph())
for i in range(100):
n_addr = random.randint(num_input_nodes, num_nodes - 1)
# before
graph_before = copy.deepcopy(self.cartesian.graph())
# print "BEFORE:", self.cartesian.graph
# mutate
g_index = self.mutator.mutate_function_node(n_addr, self.cartesian)
# after
graph_after = copy.deepcopy(self.cartesian.graph())
# print "AFTER:", self.cartesian.graph
gene_before = graph_before[n_addr][g_index]
gene_after = graph_after[n_addr][g_index]
# asserts
self.assertNotEquals(graph_before, graph_after)
self.assertNotEquals(gene_before, gene_after)
def test_mutate_output_node(self):
num_output_nodes = len(self.cartesian.output_nodes)
for i in range(100):
index = random.randint(0, num_output_nodes - 1)
old_addr = self.cartesian.output_nodes[index]
new_addr = self.mutator.mutate_output_node(index, self.cartesian)
self.assertNotEquals(old_addr, new_addr)
self.assertNotEquals(old_addr, self.cartesian.output_nodes[index])
def test_point_mutation(self):
for i in range(100):
# before
output_before = copy.deepcopy(self.cartesian.output_nodes)
graph_before = copy.deepcopy(self.cartesian.graph)
# print "BEFORE:", graph_before
# mutate
self.mutator.point_mutation(self.cartesian)
# after
output_after = copy.deepcopy(self.cartesian.output_nodes)
graph_after = copy.deepcopy(self.cartesian.graph)
# print "AFTER:", graph_after
# asserts
if self.mutator.index["mutated_node"] == "FUNC_NODE":
self.assertNotEquals(graph_before, graph_after)
elif self.mutator.index["mutated_node"] == "OUTPUT_NODE":
index = self.mutator.index["output_node"]
num_outputs = len(self.cartesian.output_nodes)
self.assertNotEquals(output_before, output_after)
self.assertTrue(index >= 0 and index <= num_outputs - 1)
def test_to_dict(self):
for i in range(100):
# print "PROGRAM:", self.cartesian.program()
self.mutator.mutate(self.cartesian)
# print "PROGRAM:", self.cartesian.program()
# import pprint
# pprint.pprint(self.mutator.to_dict())
mut_dict = self.mutator.to_dict()
before_mut = mut_dict["before_mutation"]
after_mut = mut_dict["after_mutation"]
self.assertNotEquals(before_mut, after_mut)
self.assertIsNotNone(mut_dict["method"])
self.assertIsNotNone(mut_dict["mutation_probability"])
self.assertIsNotNone(mut_dict["random_probability"])
self.assertIsNotNone(mut_dict["mutated"])
self.assertIsNotNone(mut_dict["before_mutation"])
self.assertIsNotNone(mut_dict["after_mutation"])
config["data"]["1.0"] = [1.0 for j in range(rows)]
# import pprint
# pprint.pprint(config)
json_store = JSONStore(config)
functions = [
funcs.add_function, funcs.sub_function, funcs.mul_function,
funcs.div_function, funcs.sin_function, funcs.cos_function,
funcs.rad_function
]
generator = CartesianGenerator(config)
# genetic operators
selection = Selection(config, recorder=json_store)
mutation = CartesianMutation(config)
# run symbolic regression
population = generator.init()
start_time = time.time()
details = play.play_details(population=population,
evaluate=evaluate,
functions=functions,
selection=selection,
mutation=mutation,
print_func=print_func,
plot_func=plot_func,
stop_func=default_stop_func,
config=config)
# play.play_evolution_strategy(details)
