def setUp(self):
self.ges = GrammaticalEvolution()
self.ges.set_genotype_length(10)
self.ges.set_population_size(5)
self.ges.set_max_program_length(200)
self.ges.set_bnf(''.join([
'<S> ::=', 'a = <VALUE1>\n', 'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n', '<VALUE2> ::= 1 | 2 | 3 \n'
]))
self.ges.create_genotypes()
self.ges.set_fitness_type(MAX)
# build a fake history
value = .5
for generation in range(3):
# Pretend that the genotypes have run
count = 0
for gene in self.ges.population:
gene._fitness = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
count += 1
self.ges._history.append(deepcopy(self.ges.fitness_list))
def setUp(self):
self.ges = GrammaticalEvolution()
self.ges.set_genotype_length(10)
self.ges.set_population_size(5)
self.ges.set_max_program_length(200)
self.ges.set_bnf(''.join([
'<S> ::=',
'a = <VALUE1>\n',
'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n',
'<VALUE2> ::= 1 | 2 | 3 \n']))
self.ges.create_genotypes()
self.ges.set_fitness_type(MAX)
# build a fake history
value = .5
for generation in range(3):
# Pretend that the genotypes have run
count = 0
for gene in self.ges.population:
gene._fitness = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
count += 1
self.ges._history.append(deepcopy(self.ges.fitness_list))
def test_class_init__(self):
"""
This function tests the initialization of the class.
"""
ges = GrammaticalEvolution()
self.assertEqual(None, ges.stopping_criteria[STOPPING_MAX_GEN])
self.assertEqual(None,
ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE])
self.assertEqual(DEFAULT_CROSSOVER_RATE, ges._crossover_rate)
self.assertEqual(DEFAULT_CHILDEREN_PER_CROSSOVER,
ges._children_per_crossover)
self.assertEqual(DEFAULT_MUTATION_TYPE, ges._mutation_type)
self.assertEqual(DEFAULT_MUTATION_RATE, ges._mutation_rate)
self.assertEqual(DEFAULT_MAX_FITNESS_RATE, ges._max_fitness_rate)
# Parameters for phenotype creation
self.assertEqual(DEFAULT_WRAP, ges._wrap)
self.assertEqual(DEFAULT_EXTEND_GENOTYPE, ges._extend_genotype)
self.assertEqual(DEFAULT_START_GENE_LENGTH, ges._start_gene_length)
self.assertEqual(DEFAULT_MAX_PROGRAM_LENGTH, ges._max_gene_length)
self.assertEqual(DEFAULT_MAX_PROGRAM_LENGTH, ges._max_program_length)
# Parameters for overall process
self.assertEqual(0, ges._generation)
self.assertEqual(FitnessList(CENTER), ges.fitness_list)
self.assertEqual(DEFAULT_FITNESS_FAIL, ges._fitness_fail)
self.assertEqual(DEFAULT_MAINTAIN_HISTORY, ges._maintain_history)
self.assertEqual(DEFAULT_TIMEOUTS, ges._timeouts)
# Parameters used during runtime
self.assertEqual(None, ges.current_g)
self.assertEqual([], ges._fitness_selections)
self.assertEqual([], ges._replacement_selections)
self.assertEqual({}, ges.bnf)
self.assertEqual(0, ges._population_size)
self.assertEqual([], ges.population)
self.assertEqual([], ges._history)
class TestGrammaticalEvolution(unittest.TestCase):
"""
This class tests the GrammaticalEvolution class.
"""
def setUp(self):
self.ges = GrammaticalEvolution()
self.ges.set_genotype_length(10)
self.ges.set_population_size(5)
self.ges.set_max_program_length(200)
self.ges.set_bnf(''.join([
'<S> ::=', 'a = <VALUE1>\n', 'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n', '<VALUE2> ::= 1 | 2 | 3 \n'
]))
self.ges.create_genotypes()
self.ges.set_fitness_type(MAX)
# build a fake history
value = .5
for generation in range(3):
# Pretend that the genotypes have run
count = 0
for gene in self.ges.population:
gene._fitness = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
count += 1
self.ges._history.append(deepcopy(self.ges.fitness_list))
def test_class_init__(self):
"""
This function tests the initialization of the class.
"""
ges = GrammaticalEvolution()
self.assertEqual(None, ges.stopping_criteria[STOPPING_MAX_GEN])
self.assertEqual(None,
ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE])
self.assertEqual(DEFAULT_CROSSOVER_RATE, ges._crossover_rate)
self.assertEqual(DEFAULT_CHILDEREN_PER_CROSSOVER,
ges._children_per_crossover)
self.assertEqual(DEFAULT_MUTATION_TYPE, ges._mutation_type)
self.assertEqual(DEFAULT_MUTATION_RATE, ges._mutation_rate)
self.assertEqual(DEFAULT_MAX_FITNESS_RATE, ges._max_fitness_rate)
# Parameters for phenotype creation
self.assertEqual(DEFAULT_WRAP, ges._wrap)
self.assertEqual(DEFAULT_EXTEND_GENOTYPE, ges._extend_genotype)
self.assertEqual(DEFAULT_START_GENE_LENGTH, ges._start_gene_length)
self.assertEqual(DEFAULT_MAX_PROGRAM_LENGTH, ges._max_gene_length)
self.assertEqual(DEFAULT_MAX_PROGRAM_LENGTH, ges._max_program_length)
# Parameters for overall process
self.assertEqual(0, ges._generation)
self.assertEqual(FitnessList(CENTER), ges.fitness_list)
self.assertEqual(DEFAULT_FITNESS_FAIL, ges._fitness_fail)
self.assertEqual(DEFAULT_MAINTAIN_HISTORY, ges._maintain_history)
self.assertEqual(DEFAULT_TIMEOUTS, ges._timeouts)
# Parameters used during runtime
self.assertEqual(None, ges.current_g)
self.assertEqual([], ges._fitness_selections)
self.assertEqual([], ges._replacement_selections)
self.assertEqual({}, ges.bnf)
self.assertEqual(0, ges._population_size)
self.assertEqual([], ges.population)
self.assertEqual([], ges._history)
def test_set_population_size(self):
"""
This function tests setting the population size.
"""
self.ges.set_population_size(1000)
self.assertEqual(1000, self.ges._population_size)
self.assertEqual(1000, len(self.ges.fitness_list))
self.assertRaises(ValueError, self.ges.set_population_size, 0)
self.assertRaises(ValueError, self.ges.set_population_size, -1)
def test_get_population_size(self):
"""
This function tests getting the population size.
"""
self.ges._population_size = 1000
self.assertEqual(1000, self.ges.get_population_size())
def test_set_genotype_length(self):
"""
This function tests setting the genotype length.
"""
self.ges.set_genotype_length(10, 1000)
self.assertEqual(10, self.ges._start_gene_length)
self.assertEqual(1000, self.ges._max_gene_length)
self.ges.set_genotype_length(10, None)
self.assertEqual(10, self.ges._start_gene_length)
self.assertEqual(10, self.ges._max_gene_length)
self.assertRaises(ValueError, self.ges.set_genotype_length, 1000, 10)
self.assertRaises(ValueError, self.ges.set_genotype_length, -1, 10)
self.assertRaises(ValueError, self.ges.set_genotype_length, 10, -10)
def test_get_genotype_length(self):
"""
This function test getting the genotype length.
"""
self.ges._start_gene_length = 1000
self.ges._max_gene_length = 1500
self.assertEqual((1000, 1500), self.ges.get_genotype_length())
def test_set_extend_genotype(self):
"""
This function tests setting the genotype extend flag.
"""
self.ges.set_extend_genotype(True)
self.assertEqual(True, self.ges._extend_genotype)
self.ges.set_extend_genotype(False)
self.assertEqual(False, self.ges._extend_genotype)
self.assertRaises(ValueError, self.ges.set_extend_genotype, 'nottrue')
def test_get_extend_genotype(self):
"""
This function tests getting the genotype extend flag.
"""
self.ges._extend_genotype = True
self.assertEqual(True, self.ges.get_extend_genotype())
self.ges._extend_genotype = False
self.assertEqual(False, self.ges.get_extend_genotype())
def test_set_wrap(self):
"""
This function tests setting the wrap flag.
"""
self.ges.set_wrap(True)
self.assertEqual(True, self.ges._wrap)
self.ges.set_wrap(False)
self.assertEqual(False, self.ges._wrap)
self.assertRaises(ValueError, self.ges.set_wrap, 'nottrue')
def test_get_wrap(self):
"""
This function tests getting the wrap flag.
"""
self.ges._wrap = True
self.assertEqual(True, self.ges.get_wrap())
self.ges._wrap = False
self.assertEqual(False, self.ges._wrap)
def test_set_bnf(self):
sample_bnf = ''.join([
'<S> ::=', 'a = <VALUE1>\n', 'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n', '<VALUE2> ::= 1 | 2 | 3 \n'
])
result_bnf = {
'<S>': [
''.join([
'a = <VALUE1>\n', 'b = <VALUE2>\n', 'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)'
])
],
'<VALUE1>': ['-1', '2', '0'],
'<VALUE2>': ['1', '2', '3']
}
self.ges.set_bnf(sample_bnf)
self.assertEqual(result_bnf['<VALUE1>'], self.ges.bnf['<VALUE1>'])
self.assertEqual(result_bnf['<VALUE1>'], self.ges.bnf['<VALUE1>'])
self.assertEqual(result_bnf['<S>'][0], self.ges.bnf['<S>'][0])
self.assertEqual(result_bnf['<S>'][0], self.ges.bnf['<S>'][0])
self.assertEqual(result_bnf, self.ges.bnf)
def test_get_bnf(self):
"""
This function test getting a BNF.
"""
sample_bnf = ''.join([
'<S> ::=', 'a = <VALUE1>\n', 'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n', '<VALUE2> ::= 1 | 2 | 3 \n'
])
result_bnf = {
'<S>': [
''.join([
'a = <VALUE1>\n', 'b = <VALUE2>\n', 'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)'
])
],
'<VALUE1>': ['-1', '2', '0'],
'<VALUE2>': ['1', '2', '3']
}
self.ges.set_bnf(sample_bnf)
self.assertEqual(result_bnf, self.ges.bnf)
self.assertEqual(result_bnf, self.ges.get_bnf())
def test_set_maintain_history(self):
"""
This function tests setting the maintain fitness list history flag.
"""
self.ges.set_maintain_history(True)
self.assertEqual(True, self.ges._maintain_history)
self.ges.set_maintain_history(False)
self.assertEqual(False, self.ges._maintain_history)
self.assertRaises(ValueError, self.ges.set_maintain_history, 'nottrue')
def test_get_maintain_history(self):
"""
This function tests getting the maintain fitness list history flag.
"""
self.ges._maintain_history = True
self.assertEqual(True, self.ges.get_maintain_history())
self.ges._maintain_history = False
self.assertEqual(False, self.ges.get_maintain_history())
def test_set_max_program_length(self):
"""
This function tests setting the maximum program length.
"""
self.ges.set_max_program_length(1000)
self.assertEqual(1000, self.ges._max_program_length)
self.assertRaises(ValueError, self.ges.set_max_program_length, -1)
def test_get_max_program_length(self):
"""
This function tests getting the maximum program length.
"""
self.ges._maintain_history = True
self.assertEqual(True, self.ges.get_maintain_history())
self.ges._maintain_history = False
self.assertEqual(False, self.ges.get_maintain_history())
def test_set_fitness_fail(self):
"""
This function tests setting the default fitness value for failure.
"""
self.ges.set_fitness_fail(-9999.0)
self.assertEqual(-9999.0, self.ges._fitness_fail)
self.ges.set_fitness_fail(-9999)
self.assertEqual(-9999.0, self.ges._fitness_fail)
self.assertRaises(ValueError, self.ges.set_fitness_fail, 'notfloat')
def test_get_fitness_fail(self):
"""
This function tests getting the default fitness value for failure.
"""
self.ges._fitness_fail = -9999.0
self.assertEqual(-9999.0, self.ges.get_fitness_fail())
def test_set_mutation_type(self):
"""
This function tests setting the default fitness value for failure.
"""
self.ges.set_mutation_type(MUT_TYPE_S)
self.assertEqual(MUT_TYPE_S, self.ges._mutation_type)
self.ges.set_mutation_type(MUT_TYPE_M)
self.assertEqual(MUT_TYPE_M, self.ges._mutation_type)
self.assertRaises(ValueError, self.ges.set_mutation_type, 'wrongtype')
def test_get_mutation_type(self):
"""
This function tests getting the default fitness value for failure.
"""
self.ges._mutation_type = MUT_TYPE_S
self.assertEqual(MUT_TYPE_S, self.ges.get_mutation_type())
def test_set_mut_cross_rate(self):
"""
This function tests the ability to set mutation and cross-over rates.
"""
self.ges.set_mutation_rate(.5)
self.assertEqual(0.5, self.ges._mutation_rate)
# boundaries
self.ges.set_mutation_rate(1.0)
self.assertEqual(1.0, self.ges._mutation_rate)
self.ges.set_mutation_rate(0.0)
self.assertEqual(0.0, self.ges._mutation_rate)
self.assertRaises(ValueError, self.ges.set_mutation_rate, -.01)
self.assertRaises(ValueError, self.ges.set_mutation_rate, 1.01)
# crossover rate
self.ges.set_crossover_rate(.5)
self.assertEqual(0.5, self.ges._crossover_rate)
# boundaries
self.ges.set_crossover_rate(1.0)
self.assertEqual(1.0, self.ges._crossover_rate)
self.ges.set_crossover_rate(0.0)
self.assertEqual(0.0, self.ges._crossover_rate)
self.assertRaises(ValueError, self.ges.set_crossover_rate, -.01)
self.assertRaises(ValueError, self.ges.set_crossover_rate, 1.01)
def test_get_mut_cross_rate(self):
"""
This function tests the ability to get mutation and cross-over rates.
"""
self.ges._mutation_rate = 0.5
self.assertEqual(0.5, self.ges.get_mutation_rate())
self.ges._crossover = 0.2
self.assertEqual(0.2, self.ges.get_crossover_rate())
def test_set_children_per_crossover(self):
"""
This function tests setting the number of children to be generated
upon crossover.
"""
self.ges.set_children_per_crossover(1)
self.assertEqual(1, self.ges._children_per_crossover)
self.ges.set_children_per_crossover(2)
self.assertEqual(2, self.ges._children_per_crossover)
self.assertRaises(ValueError, self.ges.set_children_per_crossover, 0)
self.assertRaises(ValueError, self.ges.set_children_per_crossover, 3)
def test_get_children_per_crossover(self):
"""
This function tests getting the number of children per crossover.
"""
self.ges._children_per_crossover = 1
self.assertEqual(1, self.ges.get_children_per_crossover())
def test_set_max_generations(self):
"""
This function tests setting the maximum number of generations.
"""
self.ges.set_max_generations(500)
self.assertEqual(500, self.ges.stopping_criteria[STOPPING_MAX_GEN])
def test_get_max_generations(self):
"""
This function tests getting the maximum number of generations.
"""
self.ges.stopping_criteria[STOPPING_MAX_GEN] = 500
self.assertEqual(500, self.ges.get_max_generations())
def test_set_fitness_type(self):
"""
This function tests setting a fitness type. It also tests the option
of setting a target value at the same time.
"""
self.ges.set_fitness_type(CENTER)
fitl = FitnessList(CENTER)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
fitl = FitnessList(MIN)
self.ges.set_fitness_type(MIN)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
fitl = FitnessList(MAX)
self.ges.set_fitness_type(MAX)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
self.assertRaises(ValueError, self.ges.set_fitness_type, "wrong")
# Include target value
fitl = FitnessList(CENTER, 1.0)
self.ges.set_fitness_type(CENTER, 1.0)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
self.assertEqual(1.0, self.ges.fitness_list._target_value)
self.assertRaises(ValueError, self.ges.set_fitness_type, CENTER,
"test")
def test_get_fitness_type(self):
"""
This function tests getting the maximum number of generations.
"""
self.ges.fitness_list._fitness_type = "min"
self.assertEqual("min", self.ges.get_fitness_type())
def test_set_max_fitness_rate(self):
"""
This function tests the ability to set the maximum fitness rate.
"""
self.ges.set_max_fitness_rate(.5)
self.assertEqual(0.5, self.ges._max_fitness_rate)
# boundaries
self.ges.set_max_fitness_rate(1.0)
self.assertEqual(1.0, self.ges._max_fitness_rate)
self.ges.set_max_fitness_rate(0.0)
self.assertEqual(0.0, self.ges._max_fitness_rate)
self.assertRaises(ValueError, self.ges.set_max_fitness_rate, -.01)
self.assertRaises(ValueError, self.ges.set_max_fitness_rate, 1.01)
def test_get_max_fitness_rate(self):
"""
This function tests the ability to get the maximum fitness rate.
"""
self.ges._max_fitness_rate = 0.5
self.assertEqual(0.5, self.ges.get_max_fitness_rate())
def test_set_fitness_selections(self):
"""
This function tests setting a fitness selection class.
"""
self.ges.set_fitness_selections(Fitness(FitnessList('center')))
self.assertEqual(1, len(self.ges._fitness_selections))
self.ges.set_fitness_selections(Fitness(FitnessList('center')))
self.assertEqual(2, len(self.ges._fitness_selections))
self.ges._fitness_selections = []
self.ges.set_fitness_selections(Fitness(FitnessList('center')),
Fitness(FitnessList('center')))
self.assertEqual(2, len(self.ges._fitness_selections))
self.ges._fitness_selections = []
self.assertRaises(ValueError, self.ges.set_fitness_selections, "wrong")
def test_set_replacement_selections(self):
"""
This function tests setting a replacement selection class.
"""
self.ges.set_replacement_selections(Replacement(FitnessList('center')))
self.assertEqual(1, len(self.ges._replacement_selections))
self.ges.set_replacement_selections(Replacement(FitnessList('center')))
self.assertEqual(2, len(self.ges._replacement_selections))
self.ges._replacement_selections = []
self.ges.set_replacement_selections(Replacement(FitnessList('center')),
Replacement(FitnessList('center')))
self.assertEqual(2, len(self.ges._replacement_selections))
self.ges._replacement_selections = []
self.assertRaises(ValueError, self.ges.set_replacement_selections,
"wrong")
def test_get_fitness_history(self):
"""
This function tests getting the fitness history.
"""
self.ges.set_fitness_type('max')
getfh = self.ges.get_fitness_history
self.assertEqual([4.5, 4.5, 4.5], getfh('best_value'))
self.assertEqual([2.5, 2.5, 2.5], getfh('mean'))
self.assertEqual([0.5, 0.5, 0.5], getfh('min_value'))
self.assertEqual([4.5, 4.5, 4.5], getfh('max_value'))
self.assertEqual([0.5, 0.5, 0.5], getfh('worst_value'))
self.assertEqual([0, 0, 0], getfh('min_member'))
self.assertEqual([4, 4, 4], getfh('max_member'))
self.assertEqual([4, 4, 4], getfh('best_member'))
self.assertEqual([0, 0, 0], getfh('worst_member'))
def test_get_best_member(self):
"""
This function tests getting the best member of the fitness list.
"""
self.assertEqual(self.ges.population[4], self.ges.get_best_member())
def test_get_worst_member(self):
"""
This function tests getting the worst member of the fitness list.
"""
self.assertEqual(self.ges.population[0], self.ges.get_worst_member())
def test_set_timeouts(self):
"""
This function tests setting the timeouts.
"""
self.ges.set_timeouts(5, 10)
self.assertEqual([5, 10], self.ges._timeouts)
self.assertRaises(ValueError, self.ges.set_timeouts, -5, 10)
self.assertRaises(ValueError, self.ges.set_timeouts, 5, -10)
self.assertRaises(ValueError, self.ges.set_timeouts, 5, 'ten')
self.assertRaises(ValueError, self.ges.set_timeouts, 'five', 10)
def test_get_timeouts(self):
"""
This function tests getting the timeouts.
"""
self.ges._timeouts = [5, 10]
self.assertEqual([5, 10], self.ges.get_timeouts())
def test_compute_fitness(self):
"""
This function tests the compute process.
"""
fitness_fail = self.ges._fitness_fail
self.ges._compute_fitness()
# Is each fitness value in the gene
for gene in self.ges.population:
self.assertNotEqual(fitness_fail, gene.get_fitness())
# Is each fitness value in the fitness list
for fitness_value, member_no in self.ges.fitness_list:
self.assertNotEqual(fitness_fail, fitness_value)
def test_run(self):
"""
This function tests running grammatical_evolution start to finish.
"""
self.ges.set_fitness_type(MAX, 20.0)
self.ges.set_max_generations(5)
self.ges._history = []
self.ges.set_maintain_history(True)
self.ges.run()
# number of generations saved.
self.ges.set_fitness_type('max')
getfh = self.ges.get_fitness_history
# Not really sure how to prove a run.
print('best_value', getfh('best_value'))
print('mean', getfh('mean'))
print('min_value', getfh('min_value'))
print('max_value', getfh('max_value'))
print('worst_value', getfh('worst_value'))
print('min_member', getfh('min_member'))
print('max_member', getfh('max_member'))
print('best_member', getfh('best_member'))
print('worst_member', getfh('worst_member'))
self.assertEqual(5, self.ges._generation)
# this is 6, because the counter starts at 0
self.assertEqual(6, len(getfh('best_value')))
def test_create_genotypes(self):
"""
This function test the building of genotypes and passing along the
parameters to each individual.
"""
self.ges.population = []
self.ges.create_genotypes()
self.assertEqual(self.ges._population_size, len(self.ges.population))
for gene in self.ges.population:
self.assertEqual(gene._gene_length, self.ges._start_gene_length)
self.assertEqual(gene._max_gene_length, self.ges._max_gene_length)
self.assertEqual(gene.local_bnf['<member_no>'], [gene.member_no])
self.assertEqual(gene._max_program_length,
self.ges._max_program_length)
self.assertEqual(gene._fitness, self.ges._fitness_fail)
self.assertEqual(gene._extend_genotype, self.ges._extend_genotype)
self.assertEqual(gene._timeouts, self.ges._timeouts)
self.assertEqual(gene._wrap, self.ges._wrap)
def test_perform_endcycle(self):
print("perform_endcycle not yet tested")
def test_evaluate_fitness(self):
# test whether select from fitness selections works
self.assertEqual(.5, self.ges._max_fitness_rate)
self.ges.set_fitness_selections(
FitnessElites(self.ges.fitness_list, .1))
pool = self.ges._evaluate_fitness()
self.assertEqual(1, len(pool))
# test whether max fitness rate works
self.ges.set_fitness_selections(
FitnessElites(self.ges.fitness_list, .75))
pool = self.ges._evaluate_fitness()
self.assertEqual(3, len(pool))
def test_perform_crossovers(self):
# Make a fitness pool
flist = self.ges.population[2:4]
self.assertEqual(2, self.ges._children_per_crossover)
clist = self.ges._perform_crossovers(flist)
self.assertEqual(2, len(clist))
# Does it round fitness pool properly?
flist = self.ges.population[1:4]
clist = self.ges._perform_crossovers(flist)
self.assertEqual(2, len(clist))
# Change children per cross over
self.ges.set_children_per_crossover(1)
clist = self.ges._perform_crossovers(flist)
self.assertEqual(1, len(clist))
def test_crossover(self):
"""
This function tests the crossover process from the level of two genes.
"""
parent1 = self.ges.population[0]
parent2 = self.ges.population[1]
(child1, child2) = self.ges._crossover(parent1, parent2)
pgene1 = parent1.binary_gene
pgene2 = parent2.binary_gene
cgene1 = child1.binary_gene
cgene2 = child2.binary_gene
self.assertEqual(len(pgene1), len(cgene1))
self.assertEqual(len(pgene2), len(cgene2))
# was there any crossover
self.assertNotEqual(pgene1, cgene1)
self.assertNotEqual(pgene2, cgene2)
self.assertNotEqual(pgene1, cgene2)
self.assertNotEqual(pgene2, cgene1)
# compare cross points
for crosspoint in range(len(pgene1)):
if pgene1[crosspoint] != cgene1[crosspoint]:
break
self.assertEqual(len(pgene1[:crosspoint]), len(cgene1[:crosspoint]))
self.assertEqual(len(pgene2[:crosspoint]), len(cgene2[:crosspoint]))
self.assertEqual(len(pgene1[crosspoint:]), len(cgene2[crosspoint:]))
self.assertEqual(len(pgene2[crosspoint:]), len(cgene1[crosspoint:]))
def test_crossover_function(self):
"""
This function tests the crossover process at the level of two strings
of binary numbers.777777777777
"""
crosspoint = 10
child1_binary = '000000000011111'
child2_binary = '111111111100000'
self.assertEqual(
('000000000000000', '111111111111111'),
self.ges._crossover_function(child1_binary, child2_binary,
crosspoint))
def test_perform_mutations(self):
"""
This function tests the process that performs mutations of the genes.
"""
# single
self.ges.population = []
self.ges.set_population_size(100)
self.ges.set_mutation_type('s')
self.ges.set_mutation_rate(.2)
self.ges.create_genotypes()
self.ges._compute_fitness()
pop = deepcopy(self.ges.population)
new_pop = self.ges.population
self.ges._perform_mutations(new_pop)
diffs = 0
for gene in pop:
if new_pop[gene.member_no].binary_gene != gene.binary_gene:
diffs += 1
self.assertNotEqual(0, diffs)
# multiple
self.ges.population = []
self.ges.set_population_size(100)
self.ges.set_mutation_type('m')
self.ges.set_mutation_rate(.01)
self.ges.create_genotypes()
self.ges._compute_fitness()
pop = deepcopy(self.ges.population)
new_pop = self.ges.population
self.ges._perform_mutations(new_pop)
diffs = 0
for gene in pop:
if new_pop[gene.member_no].binary_gene != gene.binary_gene:
diffs += 1
self.assertNotEqual(0, diffs)
def test_perform_replacements(self):
# Existing fitness list
#[[0.5, 0], [1.5, 1], [2.5, 2], [3.5, 3], [4.5, 4]]
# make some replacement selections
# This should replace the member_no 4
self.ges.fitness_list.set_fitness_type(MIN)
self.ges.set_replacement_selections(
ReplacementDeleteWorst(self.ges.fitness_list, 1))
# make a fitness pool
new_member = deepcopy(self.ges.population[2])
new_member.local_bnf['<IAMNEW>'] = ['yes i am']
orig_member_no = new_member.member_no
fitness_pool = [new_member]
# do the replacements
self.ges._perform_replacements(fitness_pool)
# did replacement take place
self.assertEqual('yes i am',
self.ges.population[4].local_bnf['<IAMNEW>'][0])
# was the local bnf for each replaced member update for new member no
self.assertEqual(4, self.ges.population[4].member_no)
# was the generation number bumped up by one
self.assertEqual(1, self.ges.population[4]._generation)
def test_continue_processing(self):
# Max
self.ges.fitness_list.set_fitness_type(MAX)
self.ges.fitness_list.set_target_value(5.0)
self.ges._generation = 0
self.assertEqual(True, self.ges._continue_processing())
self.ges.fitness_list.set_target_value(4.0)
self.assertEqual(False, self.ges._continue_processing())
# Min
self.ges.fitness_list.set_fitness_type(MIN)
self.assertEqual(False, self.ges._continue_processing())
self.ges.fitness_list.set_target_value(-4.0)
self.assertEqual(True, self.ges._continue_processing())
# Center
self.ges.fitness_list.set_fitness_type(CENTER)
self.ges.fitness_list.set_target_value(-.3)
self.assertEqual(True, self.ges._continue_processing())
self.ges.fitness_list.set_target_value(-4.0)
self.assertEqual(True, self.ges._continue_processing())
# Maximum generations
self.ges.stopping_criteria[STOPPING_MAX_GEN] = 1
self.assertEqual(True, self.ges._continue_processing())
self.ges._generation = 1
self.assertEqual(False, self.ges._continue_processing())
self.ges._generation = 0
# Fitness Landscape function, two pretend functions
def fit_landscape(fitness_list):
return True
def fit_landscape1(fitness_list):
return False
self.ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE] = fit_landscape
self.assertEqual(True, self.ges._continue_processing())
self.ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE] = fit_landscape1
self.assertEqual(False, self.ges._continue_processing())
<plus> ::= +
<minus> ::= -
<real> ::= <int-const>.<int-const>
<int-const> ::= <int-const> | 1 | 2 | 3 | 4 | 5 | 6 |
7 | 8 | 9 | 0
<S> ::=
import math
total = 0.0
for i in xrange(100):
value = float(i) / float(100)
total += abs(<expr> - pow(value, 3))
fitness = total
self.set_bnf_variable('<fitness>', fitness)
"""
ges = GrammaticalEvolution()
ges.set_bnf(bnf)
ges.set_genotype_length(start_gene_length=20, max_gene_length=50)
ges.set_population_size(50)
ges.set_wrap(True)
ges.set_max_generations(1000)
ges.set_fitness_type(MIN, .01)
ges.set_max_program_length(500)
ges.set_timeouts(10, 120)
ges.set_fitness_fail(100.0)
ges.set_mutation_rate(.025)
ges.set_fitness_selections(
modelname = self.runtime_resolve('<model_name>', 'str')
net.save(modelname)
self.set_bnf_variable('<saved_name>', modelname)
# This method can be used to look at all the particulars
# of what happened...uses disk space
self.net = net
fitness = mse
self.set_bnf_variable('<fitness>', fitness)
"""
ges = GrammaticalEvolution()
ges.set_bnf(bnf)
ges.set_genotype_length(start_gene_length=100,
max_gene_length=200)
ges.set_population_size(20)
ges.set_max_generations(50)
ges.set_fitness_type('center', 0.01)
ges.set_max_program_length(4000)
ges.set_wrap(True)
ges.set_fitness_fail(2.0)
ges.set_mutation_type('m')
ges.set_max_fitness_rate(.25)
class TestGrammaticalEvolution(unittest.TestCase):
"""
This class tests the GrammaticalEvolution class.
"""
def setUp(self):
self.ges = GrammaticalEvolution()
self.ges.set_genotype_length(10)
self.ges.set_population_size(5)
self.ges.set_max_program_length(200)
self.ges.set_bnf(''.join([
'<S> ::=',
'a = <VALUE1>\n',
'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n',
'<VALUE2> ::= 1 | 2 | 3 \n']))
self.ges.create_genotypes()
self.ges.set_fitness_type(MAX)
# build a fake history
value = .5
for generation in range(3):
# Pretend that the genotypes have run
count = 0
for gene in self.ges.population:
gene._fitness = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
self.ges.fitness_list[count][0] = float(count) + value
count += 1
self.ges._history.append(deepcopy(self.ges.fitness_list))
def test_class_init__(self):
"""
This function tests the initialization of the class.
"""
ges = GrammaticalEvolution()
self.assertEqual(None, ges.stopping_criteria[STOPPING_MAX_GEN])
self.assertEqual(None,
ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE])
self.assertEqual(DEFAULT_CROSSOVER_RATE, ges._crossover_rate)
self.assertEqual(
DEFAULT_CHILDEREN_PER_CROSSOVER, ges._children_per_crossover)
self.assertEqual(DEFAULT_MUTATION_TYPE, ges._mutation_type)
self.assertEqual(DEFAULT_MUTATION_RATE, ges._mutation_rate)
self.assertEqual(DEFAULT_MAX_FITNESS_RATE, ges._max_fitness_rate)
# Parameters for phenotype creation
self.assertEqual(DEFAULT_WRAP, ges._wrap)
self.assertEqual(DEFAULT_EXTEND_GENOTYPE, ges._extend_genotype)
self.assertEqual(DEFAULT_START_GENE_LENGTH, ges._start_gene_length)
self.assertEqual(DEFAULT_MAX_PROGRAM_LENGTH, ges._max_gene_length)
self.assertEqual(DEFAULT_MAX_PROGRAM_LENGTH, ges._max_program_length)
# Parameters for overall process
self.assertEqual(0, ges._generation)
self.assertEqual(FitnessList(CENTER), ges.fitness_list)
self.assertEqual(DEFAULT_FITNESS_FAIL, ges._fitness_fail)
self.assertEqual(DEFAULT_MAINTAIN_HISTORY, ges._maintain_history)
self.assertEqual(DEFAULT_TIMEOUTS, ges._timeouts)
# Parameters used during runtime
self.assertEqual(None, ges.current_g)
self.assertEqual([], ges._fitness_selections)
self.assertEqual([], ges._replacement_selections)
self.assertEqual({}, ges.bnf)
self.assertEqual(0, ges._population_size)
self.assertEqual([], ges.population)
self.assertEqual([], ges._history)
def test_set_population_size(self):
"""
This function tests setting the population size.
"""
self.ges.set_population_size(1000)
self.assertEqual(1000L, self.ges._population_size)
self.assertEqual(1000L, len(self.ges.fitness_list))
self.assertRaises(ValueError, self.ges.set_population_size, 0)
self.assertRaises(ValueError, self.ges.set_population_size, -1)
def test_get_population_size(self):
"""
This function tests getting the population size.
"""
self.ges._population_size = 1000
self.assertEqual(1000L, self.ges.get_population_size())
def test_set_genotype_length(self):
"""
This function tests setting the genotype length.
"""
self.ges.set_genotype_length(10, 1000)
self.assertEqual(10, self.ges._start_gene_length)
self.assertEqual(1000, self.ges._max_gene_length)
self.ges.set_genotype_length(10, None)
self.assertEqual(10, self.ges._start_gene_length)
self.assertEqual(10, self.ges._max_gene_length)
self.assertRaises(ValueError, self.ges.set_genotype_length, 1000, 10)
self.assertRaises(ValueError, self.ges.set_genotype_length, -1, 10)
self.assertRaises(ValueError, self.ges.set_genotype_length, 10, -10)
def test_get_genotype_length(self):
"""
This function test getting the genotype length.
"""
self.ges._start_gene_length = 1000
self.ges._max_gene_length = 1500
self.assertEqual((1000L, 1500L), self.ges.get_genotype_length())
def test_set_extend_genotype(self):
"""
This function tests setting the genotype extend flag.
"""
self.ges.set_extend_genotype(True)
self.assertEqual(True, self.ges._extend_genotype)
self.ges.set_extend_genotype(False)
self.assertEqual(False, self.ges._extend_genotype)
self.assertRaises(ValueError, self.ges.set_extend_genotype, 'nottrue')
def test_get_extend_genotype(self):
"""
This function tests getting the genotype extend flag.
"""
self.ges._extend_genotype = True
self.assertEqual(True, self.ges.get_extend_genotype())
self.ges._extend_genotype = False
self.assertEqual(False, self.ges.get_extend_genotype())
def test_set_wrap(self):
"""
This function tests setting the wrap flag.
"""
self.ges.set_wrap(True)
self.assertEqual(True, self.ges._wrap)
self.ges.set_wrap(False)
self.assertEqual(False, self.ges._wrap)
self.assertRaises(ValueError, self.ges.set_wrap, 'nottrue')
def test_get_wrap(self):
"""
This function tests getting the wrap flag.
"""
self.ges._wrap = True
self.assertEqual(True, self.ges.get_wrap())
self.ges._wrap = False
self.assertEqual(False, self.ges._wrap)
def test_set_bnf(self):
sample_bnf = ''.join([
'<S> ::=',
'a = <VALUE1>\n',
'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n',
'<VALUE2> ::= 1 | 2 | 3 \n'])
result_bnf = {'<S>' : [''.join([
'a = <VALUE1>\n',
'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)'])],
'<VALUE1>' : ['-1', '2', '0'],
'<VALUE2>' : ['1', '2', '3']}
self.ges.set_bnf(sample_bnf)
self.assertEqual(result_bnf['<VALUE1>'], self.ges.bnf['<VALUE1>'])
self.assertEqual(result_bnf['<VALUE1>'], self.ges.bnf['<VALUE1>'])
self.assertEqual(result_bnf['<S>'][0], self.ges.bnf['<S>'][0])
self.assertEqual(result_bnf['<S>'][0], self.ges.bnf['<S>'][0])
self.assertEqual(result_bnf, self.ges.bnf)
def test_get_bnf(self):
"""
This function test getting a BNF.
"""
sample_bnf = ''.join([
'<S> ::=',
'a = <VALUE1>\n',
'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)\n',
'<VALUE1> ::= -1 | 2 | 0 \n',
'<VALUE2> ::= 1 | 2 | 3 \n'])
result_bnf = {'<S>' : [''.join([
'a = <VALUE1>\n',
'b = <VALUE2>\n',
'fitness = a + b\n',
'self.set_bnf_variable("<fitness>", fitness)'])],
'<VALUE1>' : ['-1', '2', '0'],
'<VALUE2>' : ['1', '2', '3']}
self.ges.set_bnf(sample_bnf)
self.assertEqual(result_bnf, self.ges.bnf)
self.assertEqual(result_bnf, self.ges.get_bnf())
def test_set_maintain_history(self):
"""
This function tests setting the maintain fitness list history flag.
"""
self.ges.set_maintain_history(True)
self.assertEqual(True, self.ges._maintain_history)
self.ges.set_maintain_history(False)
self.assertEqual(False, self.ges._maintain_history)
self.assertRaises(ValueError, self.ges.set_maintain_history, 'nottrue')
def test_get_maintain_history(self):
"""
This function tests getting the maintain fitness list history flag.
"""
self.ges._maintain_history = True
self.assertEqual(True, self.ges.get_maintain_history())
self.ges._maintain_history = False
self.assertEqual(False, self.ges.get_maintain_history())
def test_set_max_program_length(self):
"""
This function tests setting the maximum program length.
"""
self.ges.set_max_program_length(1000)
self.assertEqual(1000, self.ges._max_program_length)
self.assertRaises(ValueError, self.ges.set_max_program_length, -1)
def test_get_max_program_length(self):
"""
This function tests getting the maximum program length.
"""
self.ges._maintain_history = True
self.assertEqual(True, self.ges.get_maintain_history())
self.ges._maintain_history = False
self.assertEqual(False, self.ges.get_maintain_history())
def test_set_fitness_fail(self):
"""
This function tests setting the default fitness value for failure.
"""
self.ges.set_fitness_fail(-9999.0)
self.assertEqual(-9999.0, self.ges._fitness_fail)
self.ges.set_fitness_fail(-9999)
self.assertEqual(-9999.0, self.ges._fitness_fail)
self.assertRaises(ValueError, self.ges.set_fitness_fail, 'notfloat')
def test_get_fitness_fail(self):
"""
This function tests getting the default fitness value for failure.
"""
self.ges._fitness_fail = -9999.0
self.assertEqual(-9999.0, self.ges.get_fitness_fail())
def test_set_mutation_type(self):
"""
This function tests setting the default fitness value for failure.
"""
self.ges.set_mutation_type(MUT_TYPE_S)
self.assertEqual(MUT_TYPE_S, self.ges._mutation_type)
self.ges.set_mutation_type(MUT_TYPE_M)
self.assertEqual(MUT_TYPE_M, self.ges._mutation_type)
self.assertRaises(ValueError, self.ges.set_mutation_type, 'wrongtype')
def test_get_mutation_type(self):
"""
This function tests getting the default fitness value for failure.
"""
self.ges._mutation_type = MUT_TYPE_S
self.assertEqual(MUT_TYPE_S, self.ges.get_mutation_type())
def test_set_mut_cross_rate(self):
"""
This function tests the ability to set mutation and cross-over rates.
"""
self.ges.set_mutation_rate(.5)
self.assertEqual(0.5, self.ges._mutation_rate)
# boundaries
self.ges.set_mutation_rate(1.0)
self.assertEqual(1.0, self.ges._mutation_rate)
self.ges.set_mutation_rate(0.0)
self.assertEqual(0.0, self.ges._mutation_rate)
self.assertRaises(ValueError, self.ges.set_mutation_rate, -.01)
self.assertRaises(ValueError, self.ges.set_mutation_rate, 1.01)
# crossover rate
self.ges.set_crossover_rate(.5)
self.assertEqual(0.5, self.ges._crossover_rate)
# boundaries
self.ges.set_crossover_rate(1.0)
self.assertEqual(1.0, self.ges._crossover_rate)
self.ges.set_crossover_rate(0.0)
self.assertEqual(0.0, self.ges._crossover_rate)
self.assertRaises(ValueError, self.ges.set_crossover_rate, -.01)
self.assertRaises(ValueError, self.ges.set_crossover_rate, 1.01)
def test_get_mut_cross_rate(self):
"""
This function tests the ability to get mutation and cross-over rates.
"""
self.ges._mutation_rate = 0.5
self.assertEqual(0.5, self.ges.get_mutation_rate())
self.ges._crossover = 0.2
self.assertEqual(0.2, self.ges.get_crossover_rate())
def test_set_children_per_crossover(self):
"""
This function tests setting the number of children to be generated
upon crossover.
"""
self.ges.set_children_per_crossover(1)
self.assertEqual(1, self.ges._children_per_crossover)
self.ges.set_children_per_crossover(2)
self.assertEqual(2, self.ges._children_per_crossover)
self.assertRaises(ValueError,
self.ges.set_children_per_crossover, 0)
self.assertRaises(ValueError,
self.ges.set_children_per_crossover, 3)
def test_get_children_per_crossover(self):
"""
This function tests getting the number of children per crossover.
"""
self.ges._children_per_crossover = 1
self.assertEqual(1, self.ges.get_children_per_crossover())
def test_set_max_generations(self):
"""
This function tests setting the maximum number of generations.
"""
self.ges.set_max_generations(500)
self.assertEqual(500, self.ges.stopping_criteria[STOPPING_MAX_GEN])
def test_get_max_generations(self):
"""
This function tests getting the maximum number of generations.
"""
self.ges.stopping_criteria[STOPPING_MAX_GEN] = 500
self.assertEqual(500, self.ges.get_max_generations())
def test_set_fitness_type(self):
"""
This function tests setting a fitness type. It also tests the option
of setting a target value at the same time.
"""
self.ges.set_fitness_type(CENTER)
fitl = FitnessList(CENTER)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
fitl = FitnessList(MIN)
self.ges.set_fitness_type(MIN)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
fitl = FitnessList(MAX)
self.ges.set_fitness_type(MAX)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
self.assertRaises(ValueError, self.ges.set_fitness_type, "wrong")
# Include target value
fitl = FitnessList(CENTER, 1.0)
self.ges.set_fitness_type(CENTER, 1.0)
self.assertEqual(fitl.get_fitness_type(),
self.ges.fitness_list.get_fitness_type())
self.assertEqual(1.0, self.ges.fitness_list._target_value)
self.assertRaises(ValueError,
self.ges.set_fitness_type, CENTER, "test")
def test_get_fitness_type(self):
"""
This function tests getting the maximum number of generations.
"""
self.ges.fitness_list._fitness_type = "min"
self.assertEqual("min", self.ges.get_fitness_type())
def test_set_max_fitness_rate(self):
"""
This function tests the ability to set the maximum fitness rate.
"""
self.ges.set_max_fitness_rate(.5)
self.assertEqual(0.5, self.ges._max_fitness_rate)
# boundaries
self.ges.set_max_fitness_rate(1.0)
self.assertEqual(1.0, self.ges._max_fitness_rate)
self.ges.set_max_fitness_rate(0.0)
self.assertEqual(0.0, self.ges._max_fitness_rate)
self.assertRaises(ValueError, self.ges.set_max_fitness_rate, -.01)
self.assertRaises(ValueError, self.ges.set_max_fitness_rate, 1.01)
def test_get_max_fitness_rate(self):
"""
This function tests the ability to get the maximum fitness rate.
"""
self.ges._max_fitness_rate = 0.5
self.assertEqual(0.5, self.ges.get_max_fitness_rate())
def test_set_fitness_selections(self):
"""
This function tests setting a fitness selection class.
"""
self.ges.set_fitness_selections(Fitness(FitnessList('center')))
self.assertEqual(1, len(self.ges._fitness_selections))
self.ges.set_fitness_selections(Fitness(FitnessList('center')))
self.assertEqual(2, len(self.ges._fitness_selections))
self.ges._fitness_selections = []
self.ges.set_fitness_selections(Fitness(FitnessList('center')),
Fitness(FitnessList('center')))
self.assertEqual(2, len(self.ges._fitness_selections))
self.ges._fitness_selections = []
self.assertRaises(ValueError, self.ges.set_fitness_selections, "wrong")
def test_set_replacement_selections(self):
"""
This function tests setting a replacement selection class.
"""
self.ges.set_replacement_selections(Replacement(FitnessList('center')))
self.assertEqual(1, len(self.ges._replacement_selections))
self.ges.set_replacement_selections(Replacement(FitnessList('center')))
self.assertEqual(2, len(self.ges._replacement_selections))
self.ges._replacement_selections = []
self.ges.set_replacement_selections(Replacement(FitnessList('center')),
Replacement(FitnessList('center')))
self.assertEqual(2, len(self.ges._replacement_selections))
self.ges._replacement_selections = []
self.assertRaises(ValueError,
self.ges.set_replacement_selections, "wrong")
def test_get_fitness_history(self):
"""
This function tests getting the fitness history.
"""
self.ges.set_fitness_type('max')
getfh = self.ges.get_fitness_history
self.assertEqual([4.5, 4.5, 4.5], getfh('best_value'))
self.assertEqual([2.5, 2.5, 2.5], getfh('mean'))
self.assertEqual([0.5, 0.5, 0.5], getfh('min_value'))
self.assertEqual([4.5, 4.5, 4.5], getfh('max_value'))
self.assertEqual([0.5, 0.5, 0.5], getfh('worst_value'))
self.assertEqual([0, 0, 0], getfh('min_member'))
self.assertEqual([4, 4, 4], getfh('max_member'))
self.assertEqual([4, 4, 4], getfh('best_member'))
self.assertEqual([0, 0, 0], getfh('worst_member'))
def test_get_best_member(self):
"""
This function tests getting the best member of the fitness list.
"""
self.assertEqual(self.ges.population[4], self.ges.get_best_member())
def test_get_worst_member(self):
"""
This function tests getting the worst member of the fitness list.
"""
self.assertEqual(self.ges.population[0], self.ges.get_worst_member())
def test_set_timeouts(self):
"""
This function tests setting the timeouts.
"""
self.ges.set_timeouts(5, 10)
self.assertEqual([5, 10], self.ges._timeouts)
self.assertRaises(ValueError, self.ges.set_timeouts, -5, 10)
self.assertRaises(ValueError, self.ges.set_timeouts, 5, -10)
self.assertRaises(ValueError, self.ges.set_timeouts, 5, 'ten')
self.assertRaises(ValueError, self.ges.set_timeouts, 'five', 10)
def test_get_timeouts(self):
"""
This function tests getting the timeouts.
"""
self.ges._timeouts = [5, 10]
self.assertEqual([5, 10], self.ges.get_timeouts())
def test_compute_fitness(self):
"""
This function tests the compute process.
"""
fitness_fail = self.ges._fitness_fail
self.ges._compute_fitness()
# Is each fitness value in the gene
for gene in self.ges.population:
self.assertNotEqual(fitness_fail, gene.get_fitness())
# Is each fitness value in the fitness list
for fitness_value, member_no in self.ges.fitness_list:
self.assertNotEqual(fitness_fail, fitness_value)
def test_run(self):
"""
This function tests running grammatical_evolution start to finish.
"""
self.ges.set_fitness_type(MAX, 20.0)
self.ges.set_max_generations(5)
self.ges._history = []
self.ges.set_maintain_history(True)
self.ges.run()
# number of generations saved.
self.ges.set_fitness_type('max')
getfh = self.ges.get_fitness_history
# Not really sure how to prove a run.
print 'best_value', getfh('best_value')
print 'mean', getfh('mean')
print 'min_value', getfh('min_value')
print 'max_value', getfh('max_value')
print 'worst_value', getfh('worst_value')
print 'min_member', getfh('min_member')
print 'max_member', getfh('max_member')
print 'best_member', getfh('best_member')
print 'worst_member', getfh('worst_member')
self.assertEqual(5, self.ges._generation)
# this is 6, because the counter starts at 0
self.assertEqual(6, len(getfh('best_value')))
def test_create_genotypes(self):
"""
This function test the building of genotypes and passing along the
parameters to each individual.
"""
self.ges.population = []
self.ges.create_genotypes()
self.assertEqual(self.ges._population_size, len(self.ges.population))
for gene in self.ges.population:
self.assertEqual(gene._gene_length,
self.ges._start_gene_length)
self.assertEqual(gene._max_gene_length,
self.ges._max_gene_length)
self.assertEqual(gene.local_bnf['<member_no>'],
[gene.member_no])
self.assertEqual(gene._max_program_length,
self.ges._max_program_length)
self.assertEqual(gene._fitness,
self.ges._fitness_fail)
self.assertEqual(gene._extend_genotype,
self.ges._extend_genotype)
self.assertEqual(gene._timeouts,
self.ges._timeouts)
self.assertEqual(gene._wrap,
self.ges._wrap)
def test_perform_endcycle(self):
print "perform_endcycle not yet tested"
def test_evaluate_fitness(self):
# test whether select from fitness selections works
self.assertEqual(.5, self.ges._max_fitness_rate)
self.ges.set_fitness_selections(
FitnessElites(self.ges.fitness_list, .1))
pool = self.ges._evaluate_fitness()
self.assertEqual(1, len(pool))
# test whether max fitness rate works
self.ges.set_fitness_selections(
FitnessElites(self.ges.fitness_list, .75))
pool = self.ges._evaluate_fitness()
self.assertEqual(3, len(pool))
def test_perform_crossovers(self):
# Make a fitness pool
flist = self.ges.population[2:4]
self.assertEqual(2, self.ges._children_per_crossover)
clist = self.ges._perform_crossovers(flist)
self.assertEqual(2, len(clist))
# Does it round fitness pool properly?
flist = self.ges.population[1:4]
clist = self.ges._perform_crossovers(flist)
self.assertEqual(2, len(clist))
# Change children per cross over
self.ges.set_children_per_crossover(1)
clist = self.ges._perform_crossovers(flist)
self.assertEqual(1, len(clist))
def test_crossover(self):
"""
This function tests the crossover process from the level of two genes.
"""
parent1 = self.ges.population[0]
parent2 = self.ges.population[1]
(child1, child2) = self.ges._crossover(parent1, parent2)
pgene1 = parent1.binary_gene
pgene2 = parent2.binary_gene
cgene1 = child1.binary_gene
cgene2 = child2.binary_gene
self.assertEqual(len(pgene1), len(cgene1))
self.assertEqual(len(pgene2), len(cgene2))
# was there any crossover
self.assertNotEqual(pgene1, cgene1)
self.assertNotEqual(pgene2, cgene2)
self.assertNotEqual(pgene1, cgene2)
self.assertNotEqual(pgene2, cgene1)
# compare cross points
for crosspoint in range(len(pgene1)):
if pgene1[crosspoint] != cgene1[crosspoint]:
break
self.assertEqual(len(pgene1[:crosspoint]),
len(cgene1[:crosspoint]))
self.assertEqual(len(pgene2[:crosspoint]),
len(cgene2[:crosspoint]))
self.assertEqual(len(pgene1[crosspoint:]),
len(cgene2[crosspoint:]))
self.assertEqual(len(pgene2[crosspoint:]),
len(cgene1[crosspoint:]))
def test_crossover_function(self):
"""
This function tests the crossover process at the level of two strings
of binary numbers.777777777777
"""
crosspoint = 10
child1_binary = '000000000011111'
child2_binary = '111111111100000'
self.assertEqual(('000000000000000', '111111111111111'),
self.ges._crossover_function(child1_binary,
child2_binary, crosspoint))
def test_perform_mutations(self):
"""
This function tests the process that performs mutations of the genes.
"""
# single
self.ges.population = []
self.ges.set_population_size(100)
self.ges.set_mutation_type('s')
self.ges.set_mutation_rate(.2)
self.ges.create_genotypes()
self.ges._compute_fitness()
pop = deepcopy(self.ges.population)
new_pop = self.ges.population
self.ges._perform_mutations(new_pop)
diffs = 0
for gene in pop:
if new_pop[gene.member_no].binary_gene != gene.binary_gene:
diffs += 1
self.assertNotEqual(0, diffs)
# multiple
self.ges.population = []
self.ges.set_population_size(100)
self.ges.set_mutation_type('m')
self.ges.set_mutation_rate(.01)
self.ges.create_genotypes()
self.ges._compute_fitness()
pop = deepcopy(self.ges.population)
new_pop = self.ges.population
self.ges._perform_mutations(new_pop)
diffs = 0
for gene in pop:
if new_pop[gene.member_no].binary_gene != gene.binary_gene:
diffs += 1
self.assertNotEqual(0, diffs)
def test_perform_replacements(self):
# Existing fitness list
#[[0.5, 0], [1.5, 1], [2.5, 2], [3.5, 3], [4.5, 4]]
# make some replacement selections
# This should replace the member_no 4
self.ges.fitness_list.set_fitness_type(MIN)
self.ges.set_replacement_selections(
ReplacementDeleteWorst(self.ges.fitness_list, 1))
# make a fitness pool
new_member = deepcopy(self.ges.population[2])
new_member.local_bnf['<IAMNEW>'] = ['yes i am']
orig_member_no = new_member.member_no
fitness_pool = [new_member]
# do the replacements
self.ges._perform_replacements(fitness_pool)
# did replacement take place
self.assertEqual('yes i am',
self.ges.population[4].local_bnf['<IAMNEW>'][0])
# was the local bnf for each replaced member update for new member no
self.assertEqual(4, self.ges.population[4].member_no)
# was the generation number bumped up by one
self.assertEqual(1, self.ges.population[4]._generation)
def test_continue_processing(self):
# Max
self.ges.fitness_list.set_fitness_type(MAX)
self.ges.fitness_list.set_target_value(5.0)
self.ges._generation = 0
self.assertEqual(True, self.ges._continue_processing())
self.ges.fitness_list.set_target_value(4.0)
self.assertEqual(False, self.ges._continue_processing())
# Min
self.ges.fitness_list.set_fitness_type(MIN)
self.assertEqual(False, self.ges._continue_processing())
self.ges.fitness_list.set_target_value(-4.0)
self.assertEqual(True, self.ges._continue_processing())
# Center
self.ges.fitness_list.set_fitness_type(CENTER)
self.ges.fitness_list.set_target_value(-.3)
self.assertEqual(True, self.ges._continue_processing())
self.ges.fitness_list.set_target_value(-4.0)
self.assertEqual(True, self.ges._continue_processing())
# Maximum generations
self.ges.stopping_criteria[STOPPING_MAX_GEN] = 1
self.assertEqual(True, self.ges._continue_processing())
self.ges._generation = 1
self.assertEqual(False, self.ges._continue_processing())
self.ges._generation = 0
# Fitness Landscape function, two pretend functions
def fit_landscape(fitness_list):
return True
def fit_landscape1(fitness_list):
return False
self.ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE] = fit_landscape
self.assertEqual(True, self.ges._continue_processing())
self.ges.stopping_criteria[STOPPING_FITNESS_LANDSCAPE] = fit_landscape1
self.assertEqual(False, self.ges._continue_processing())
print "mse", mse
modelname = self.runtime_resolve('<model_name>', 'str')
net.save(modelname)
self.set_bnf_variable('<saved_name>', modelname)
# This method can be used to look at all the particulars
# of what happened...uses disk space
self.net = net
fitness = mse
self.set_bnf_variable('<fitness>', fitness)
"""
ges = GrammaticalEvolution()
ges.set_bnf(bnf)
ges.set_genotype_length(start_gene_length=100, max_gene_length=200)
ges.set_population_size(20)
ges.set_max_generations(
1) #controls the maximum generations of the genes i changed here
ges.set_fitness_type('center', 0.01)
ges.set_max_program_length(4000)
ges.set_wrap(True)
ges.set_fitness_fail(2.0)
ges.set_mutation_type('m')
ges.set_max_fitness_rate(.25)
7 | 8 | 9 | 0
<S> ::=
import math
total = 0.0
for i in xrange(100):
value = float(i) / float(100)
total += abs(<expr> - pow(value, 3))
fitness = total
self.set_bnf_variable('<fitness>', fitness)
"""
options = ['+', '-', '*', '/']
codon = 73
choice_number = codon % len(options)
choice = options[choice_number]
ges = GrammaticalEvolution()
ges.set_bnf(bnf)
ges.set_genotype_length(start_gene_length=20,
max_gene_length=50)
ges.set_population_size(50)
ges.set_wrap(True)
ges.set_max_generations(1000)
ges.set_fitness_type(MIN, .01)
ges.set_max_program_length(500)
ges.set_timeouts(10, 120)
ges.set_fitness_fail(100.0)
ges.set_fitness_selections(
modelname = self.runtime_resolve('<model_name>', 'str')
net.save(modelname)
self.set_bnf_variable('<saved_name>', modelname)
# This method can be used to look at all the particulars
# of what happened...uses disk space
self.net = net
fitness = mse
self.set_bnf_variable('<fitness>', fitness)
"""
ges = GrammaticalEvolution()
ges.set_bnf(bnf)
ges.set_genotype_length(start_gene_length=100, max_gene_length=200)
ges.set_population_size(20)
ges.set_max_generations(1) # controls the maximum generations of the genes i changed here
ges.set_fitness_type("center", 0.01)
ges.set_max_program_length(4000)
ges.set_wrap(True)
ges.set_fitness_fail(2.0)
ges.set_mutation_type("m")
ges.set_max_fitness_rate(0.25)
ges.set_mutation_rate(0.025)