from springbots.springbot import store_xml
from springbots.evolvespringbot import random_springbot
import optparse
if __name__ == "__main__":
# Parses command line
parser = optparse.OptionParser(description="Generate a random Springbots XML genome population")
parser.add_option("-p", "--population", dest="population", default=1,
help="Number of random springbots to create", metavar="NUMBER")
parser.add_option("-n", "--nodes", dest="nodes_num", default=10,
help="Number of nodes", metavar="NUMBER")
parser.add_option("-s", "--springs", dest="springs_num", default=30,
help="Number of springs", metavar="NUMBER")
parser.add_option("-r", "--noderadius", dest="node_radius", default=100,
help="Max distance to create new nodes", metavar="NUMBER")
(options, args) = parser.parse_args()
# Parses integer values
options.population = int(options.population)
options.springs_num = int(options.springs_num)
options.nodes_num = int(options.nodes_num)
options.node_radius = int(options.node_radius)
# Creates a population of random springbots with the chosen parameters and
# writes it to XML output
store_xml(\
[random_springbot(options.nodes_num, options.springs_num, options.node_radius) \
for x in xrange(options.population)])
"--nodes",
dest="nodes_num",
default=10,
help="Number of nodes",
metavar="NUMBER")
parser.add_option("-s",
"--springs",
dest="springs_num",
default=30,
help="Number of springs",
metavar="NUMBER")
parser.add_option("-r",
"--noderadius",
dest="node_radius",
default=100,
help="Max distance to create new nodes",
metavar="NUMBER")
(options, args) = parser.parse_args()
# Parses integer values
options.population = int(options.population)
options.springs_num = int(options.springs_num)
options.nodes_num = int(options.nodes_num)
options.node_radius = int(options.node_radius)
# Creates a population of random springbots with the chosen parameters and
# writes it to XML output
store_xml(\
[random_springbot(options.nodes_num, options.springs_num, options.node_radius) \
for x in range(options.population)])
def serial_evolve(
population,
fitness=fitness.walk,
save_freq=100,
limit=-1,
verbose=False,
graphics=False,
discard_fraction=0.4,
random_insert=0.1,
best=False,
start_iteration=1,
prefix="",
):
"""
Given the initial population 'population', executes
a genetic algorithm to evolve them for best fitness.
Saves the population each 'save_freq' interval(ordered by fitness)
"""
# Test if parameters are correct
if discard_fraction < 0 or random_insert < 0:
raise ValueError("discard_fraction and random_insert must both range from 0 to 1")
elif discard_fraction + random_insert > 1:
raise ValueError("the sum of discard_fraction and random_insert must not be greater than 1")
iter = start_iteration # Initial iteration
# Calculate amount of discarded and random population
discarded = int(len(population) / 2 * discard_fraction)
randoms = int(len(population) / 2 * random_insert)
if verbose:
print "# Initiating simulation with a population of %d specimens." % (len(population))
print "# Evolving for %s:" % (fitness.__name__)
print "# At each iteration %d will be discarded, %d of the remaining will" % (discarded, discarded - randoms),
print " be selected cloned and mutated and %d random springbots will be inserted" % (randoms)
# Transforms population into evolvespringbots
population = [EvolveSpringbot(springbot) for springbot in population]
try:
while population and iter != limit:
if verbose:
print "Iteration %d:" % (iter)
z = 1
fitness_sum = 0
bloodline_len_sum = 0
# Tests fitness for each springbot
for specimen in population:
specimen["fitness"] = fitness(specimen, WIDTH, HEIGHT, graphics)
if verbose:
print '\t%d/%d: "%s"(%d) %.3f' % (
z,
len(population),
specimen["name"],
specimen.generations(),
specimen["fitness"],
)
z += 1
bloodline_len_sum += specimen.generations()
fitness_sum += specimen["fitness"]
if verbose:
print "Bloodline lenght average: %.4f" % (bloodline_len_sum / float(len(population)))
print "Fitness average: %.4f" % (fitness_sum / float(len(population)))
# Now Order population by its fitness
population.sort(cmp=lambda A, B: cmp(A["fitness"], B["fitness"]), reverse=True)
# Discards some of the worse half
for specimen in sample(population[len(population) / 2 :], discarded + randoms):
population.remove(specimen)
# Clones and mutates some of the remaining
for specimen in sample(population, discarded):
child = EvolveSpringbot(specimen).mutate()
child.addBloodline(specimen)
names = child["name"].split()
# Gives a child's name
if len(names) == 1:
child["name"] = names[0] + " " + latimname(2)
elif len(names) == 2:
child["name"] = names[0] + " " + names[1] + " " + latimname(2)
elif len(names) == 3:
child["name"] = names[0] + " " + names[2] + " " + latimname(2)
# Incorporate children into the population
population.append(child)
# Incorporate randoms
population += [EvolveSpringbot(random=True) for x in xrange(randoms)]
# Test if it is time to save population
if iter % save_freq == 0:
# Saves the current population
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness.__name__, len(population), iter)
store_xml(population, filename)
if verbose:
print "# iteration %d saved into %s" % (iter, filename)
# Saves best if asked
if best:
filename = "%s-%s-p%d-best.xml" % (prefix, fitness.__name__, len(population))
store_xml(population[:1], filename)
if verbose:
print "# Best of iteration %d saved into %s" % (iter, filename)
# Increments iteration
iter += 1
except KeyboardInterrupt:
pass
# Order population by its fitness
population.sort(reverse=True)
# Now, saves the current population and quit
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness.__name__, len(population), iter)
store_xml(population, filename)
if verbose:
print
print "# iteration %d saved into %s" % (iter, filename)
print "# terminating..."
def serial_evolve(population, fitness=fitness.walk, save_freq=100,
limit=-1,
verbose=False, graphics=False, discard_fraction=0.4, random_insert=0.1,
best=False, start_iteration = 1, prefix=''):
"""
Given the initial population 'population', executes
a genetic algorithm to evolve them for best fitness.
Saves the population each 'save_freq' interval(ordered by fitness)
"""
# Test if parameters are correct
if discard_fraction < 0 or random_insert < 0:
raise ValueError("discard_fraction and random_insert must both range from 0 to 1")
elif discard_fraction + random_insert > 1:
raise ValueError("the sum of discard_fraction and random_insert must not be greater than 1")
iter = start_iteration # Initial iteration
# Calculate amount of discarded and random population
discarded = int(len(population)/2 * discard_fraction)
randoms = int(len(population)/2 * random_insert)
if verbose:
print("# Initiating simulation with a population of %d specimens." % (len(population)))
print("# Evolving for %s:" % (fitness.__name__))
print("# At each iteration %d will be discarded, %d of the remaining will" % (discarded, discarded-randoms), end=' ')
print(" be selected cloned and mutated and %d random springbots will be inserted" % (randoms))
# Transforms population into evolvespringbots
population = [EvolveSpringbot(springbot) for springbot in population]
try:
while population and iter != limit:
if verbose:
print("Iteration %d:" % (iter))
z = 1
fitness_sum = 0
bloodline_len_sum = 0
# Tests fitness for each springbot
for specimen in population:
specimen['fitness'] = fitness(specimen, WIDTH, HEIGHT, graphics)
if verbose:
print("\t%d/%d: \"%s\"(%d) %.3f" % \
(z, len(population), specimen['name'],
specimen.generations(), specimen['fitness']))
z += 1
bloodline_len_sum += specimen.generations()
fitness_sum += specimen['fitness']
if verbose:
print("Bloodline lenght average: %.4f" % (bloodline_len_sum/float(len(population))))
print("Fitness average: %.4f" % (fitness_sum/float(len(population))))
# Now Order population by its fitness
population.sort(
cmp=lambda A,B: cmp(A['fitness'], B['fitness']), reverse=True)
# Discards some of the worse half
for specimen in sample(population[len(population)//2:], discarded + randoms):
population.remove(specimen)
# Clones and mutates some of the remaining
for specimen in sample(population, discarded):
child = EvolveSpringbot(specimen).mutate()
child.addBloodline(specimen)
names = child['name'].split()
# Gives a child's name
if len(names) == 1:
child['name'] = names[0] + " " + latimname(2)
elif len(names) == 2:
child['name'] = names[0] + " " + names[1] + " " + latimname(2)
elif len(names) == 3:
child['name'] = names[0] + " " + names[2] + " " + latimname(2)
# Incorporate children into the population
population.append(child)
# Incorporate randoms
population += [EvolveSpringbot(random=True) for x in range(randoms)]
# Test if it is time to save population
if iter % save_freq == 0:
# Saves the current population
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness.__name__, len(population), iter)
store_xml(population, filename)
if verbose:
print("# iteration %d saved into %s" % (iter, filename))
# Saves best if asked
if best:
filename = "%s-%s-p%d-best.xml" % (prefix, fitness.__name__, len(population))
store_xml(population[:1], filename)
if verbose:
print("# Best of iteration %d saved into %s" % (iter, filename))
# Increments iteration
iter += 1
except KeyboardInterrupt:
pass
# Order population by its fitness
population.sort(
cmp=lambda A,B: cmp(A['fitness'], B['fitness']), reverse=True)
# Now, saves the current population and quit
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness.__name__, len(population), iter)
store_xml(population, filename)
if verbose:
print()
print("# iteration %d saved into %s" % (iter, filename))
print("# terminating...")
def network_evolve(save_freq=100,
limit=-1,
verbose=False,
discard_fraction=0.4,
random_insert=0.1,
best=False,
start_iteration=1,
prefix=''):
"""
Given the initial population 'population', executes
a genetic algorithm to evolve them for best fitness.
Saves the population each 'save_freq' interval(ordered by fitness)
"""
global population, servers, fitness_function
# Test if parameters are correct
if discard_fraction < 0 or random_insert < 0:
raise ValueError(
"discard_fraction and random_insert must both range from 0 to 1")
elif discard_fraction + random_insert > 1:
raise ValueError(
"the sum of discard_fraction and random_insert must not be greater than 1"
)
iteration = start_iteration # Initial iteration
# Calculate amount of discarded and random population
discarded = int(len(population) / 2 * discard_fraction)
randoms = int(len(population) / 2 * random_insert)
if verbose:
print("# Initiating simulation with a population of %d specimens..." %
(len(population)))
print("# Evolving for %s:" % (fitness_function))
print(
"# At each iteration %d will be discarded, %d of the remaining will"
% (discarded, discarded - randoms),
end=' ')
print(
" be selected cloned and mutated and %d random springbots will be inserted"
% (randoms))
# Turn all population into NetworkEvolveSpringbot
population = [
NetworkEvolveSpringbot(springbot) for springbot in population
]
threads = []
try:
while population and iteration != limit:
# (Re)load servers from file
load_servers()
if verbose:
print("Iteration %d:" % (iteration))
fitness_sum = 0
bloodline_len_sum = 0
# Create threads
threads = [FitnessThread(i) for i in range(len(population))]
# Start all threads
for thread in threads:
thread.start()
# Join(waits) all threads
for thread in threads:
thread.join()
if verbose:
specimen = population[thread.index]
print("\t%d/%d: \"%s\"(%d) %.3f" % \
(thread.index+1, len(population), specimen['name'],
specimen.generations(), specimen['fitness']))
bloodline_len_sum += specimen.generations()
fitness_sum += specimen['fitness']
if verbose:
print("Bloodline lenght average: %.4f" %
(bloodline_len_sum / float(len(population))))
print("Fitness average: %.4f" %
(fitness_sum / float(len(population))))
# Now Order population by its fitness
population.sort(cmp=lambda A, B: cmp(A['fitness'], B['fitness']),
reverse=True)
# Discards some of the worse half
for specimen in sample(population[len(population) // 2:],
discarded + randoms):
population.remove(specimen)
# Clones and mutates some of the remaining half
for specimen in sample(population, discarded):
child = NetworkEvolveSpringbot(specimen).mutate()
child.addBloodline(specimen)
names = child['name'].split()
# Gives a child's name
if len(names) == 1:
child['name'] = names[0] + " " + latimname(2)
elif len(names) == 2:
child['name'] = " ".join(
[names[0], names[1], latimname(2)])
elif len(names) == 3:
child['name'] = " ".join(
[names[0], names[2], latimname(2)])
# Incorporate children into the population
population.append(child)
# Incorporate randoms
population += [
NetworkEvolveSpringbot(random=True) for x in range(randoms)
]
# Test if it is time to save population
if iteration % save_freq == 0:
# Saves the current population
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness_function,
len(population), iteration)
store_xml(population, filename)
if verbose:
print("# iteration %d saved into %s" %
(iteration, filename))
# Saves best if asked
if best:
filename = "%s-%s-p%d-best.xml" % (prefix, fitness_function,
len(population))
store_xml(population[:1], filename)
if verbose:
print("# Best of iteration %d saved into %s" %
(iteration, filename))
# Increments iteration
iteration += 1
except KeyboardInterrupt:
pass
if verbose:
print("# waiting for threads...")
# Join(waits) all threads
for thread in threads:
thread.join()
# Order population by its fitness
population.sort(reverse=True)
# Now, saves the current population and quit
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness_function,
len(population), iteration)
store_xml(population, filename)
if verbose:
print()
print("# iteration %d saved into %s" % (iteration, filename))
print("# terminating...")
def network_evolve(save_freq=100, limit=-1,
verbose=False, discard_fraction=0.4, random_insert=0.1,
best=False, start_iteration = 1, prefix=''):
"""
Given the initial population 'population', executes
a genetic algorithm to evolve them for best fitness.
Saves the population each 'save_freq' interval(ordered by fitness)
"""
global population, servers, fitness_function
# Test if parameters are correct
if discard_fraction < 0 or random_insert < 0:
raise ValueError("discard_fraction and random_insert must both range from 0 to 1")
elif discard_fraction + random_insert > 1:
raise ValueError("the sum of discard_fraction and random_insert must not be greater than 1")
iteration = start_iteration # Initial iteration
# Calculate amount of discarded and random population
discarded = int(len(population)/2 * discard_fraction)
randoms = int(len(population)/2 * random_insert)
if verbose:
print "# Initiating simulation with a population of %d specimens..." % (len(population))
print "# Evolving for %s:" % (fitness_function)
print "# At each iteration %d will be discarded, %d of the remaining will" % (
discarded, discarded-randoms),
print " be selected cloned and mutated and %d random springbots will be inserted" % (
randoms)
# Turn all population into NetworkEvolveSpringbot
population = [NetworkEvolveSpringbot(springbot) for springbot in population]
threads = []
try:
while population and iteration != limit:
# (Re)load servers from file
load_servers()
if verbose:
print "Iteration %d:" % (iteration)
fitness_sum = 0
bloodline_len_sum = 0
# Create threads
threads = [FitnessThread(i) for i in xrange(len(population))]
# Start all threads
for thread in threads:
thread.start()
# Join(waits) all threads
for thread in threads:
thread.join()
if verbose:
specimen = population[thread.index]
print "\t%d/%d: \"%s\"(%d) %.3f" % \
(thread.index+1, len(population), specimen['name'],
specimen.generations(), specimen['fitness'])
bloodline_len_sum +=specimen.generations()
fitness_sum += specimen['fitness']
if verbose:
print "Bloodline lenght average: %.4f" % (bloodline_len_sum/float(len(population)))
print "Fitness average: %.4f" % (fitness_sum/float(len(population)))
# Now Order population by its fitness
population.sort(
cmp=lambda A,B: cmp(A['fitness'], B['fitness']), reverse=True)
# Discards some of the worse half
for specimen in sample(population[len(population)/2:], discarded + randoms):
population.remove(specimen)
# Clones and mutates some of the remaining half
for specimen in sample(population, discarded):
child = NetworkEvolveSpringbot(specimen).mutate()
child.addBloodline(specimen)
names = child['name'].split()
# Gives a child's name
if len(names) == 1:
child['name'] = names[0] + " " + latimname(2)
elif len(names) == 2:
child['name'] = " ".join([names[0], names[1], latimname(2)])
elif len(names) == 3:
child['name'] = " ".join([names[0], names[2], latimname(2)])
# Incorporate children into the population
population.append(child)
# Incorporate randoms
population += [NetworkEvolveSpringbot(random=True) for x in xrange(randoms)]
# Test if it is time to save population
if iteration % save_freq == 0:
# Saves the current population
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness_function, len(population), iteration)
store_xml(population, filename)
if verbose:
print "# iteration %d saved into %s" % (iteration, filename)
# Saves best if asked
if best:
filename = "%s-%s-p%d-best.xml" % (prefix, fitness_function, len(population))
store_xml(population[:1], filename)
if verbose:
print "# Best of iteration %d saved into %s" % (iteration, filename)
# Increments iteration
iteration += 1
except KeyboardInterrupt:
pass
if verbose:
print "# waiting for threads..."
# Join(waits) all threads
for thread in threads:
thread.join()
# Order population by its fitness
population.sort(reverse=True)
# Now, saves the current population and quit
filename = "%s-%s-p%d-i%d.xml" % (prefix, fitness_function, len(population), iteration)
store_xml(population, filename)
if verbose:
print
print "# iteration %d saved into %s" % (iteration, filename)
print "# terminating..."